feat: v2.9.7 (#4576 )

This pull request primarily updates the GoFrame (`gf`) framework and its related driver dependencies from version `v2.9.6` to `v2.9.7` across the repository. Additionally, it removes the `examples` submodule and updates the contributors image in the `README.MD` to reflect the new version. Dependency updates: * Updated all references to `github.com/gogf/gf/v2` and related driver dependencies from `v2.9.6` to `v2.9.7` in various `go.mod` files throughout the repository, including core modules and contributed drivers/configs. [[1]](diffhunk://#diff-ee0abb9c50b9f91f424349123e31b7b1ba1e1e4f7497250422696c5bda2e74ceL6-R12) [[2]](diffhunk://#diff-cef597d401b6dad225f9e2e431bdde7e53cb60bdf287624cef38a6a7bb9ae7a3L7-R7) [[3]](diffhunk://#diff-970f7eacff9cd97a0d8a00d59ea8041eedaa21c7544c6669aaa58ca692c6b274L6-R6) [[4]](diffhunk://#diff-c23d0ca80cd6588b7df84de8ef84713f0ce0555ba05d2d9e7f5d1e0324b1ed3aL6-R6) [[5]](diffhunk://#diff-aa230a2b1198e6ef8afeb7f48335eb2e2f51d87d918d63c4d891fea612d18ff0L6-R6) [[6]](diffhunk://#diff-86c2390edbede20803cd862908fe95e7207f7dbabd5089ddd4838e1f26e7fecaL6-R6) [[7]](diffhunk://#diff-5e1af33d38ced461fc0e13981d7051e125876d1692efc3aa9cb4b7faa4c18addL7-R7) [[8]](diffhunk://#diff-8c6247829130f219981483ccf25af699a63de99afedeb0dd5c1b7bd8ff0919bdL9-R9) [[9]](diffhunk://#diff-accbd2d37d45e51db3fcb0468043b1e1fd53eeac9e3d3558467ef24444188d2fL7-R7) [[10]](diffhunk://#diff-15fac9b8e76d2782594c91da72f6a6f42fc18e359c3be35bf6564ac3ca09f700L6-R7) [[11]](diffhunk://#diff-8e1a76afd564b6073aac7b02ca59f296ae45a24da3dc4d5c40f18169f48ceba1L6-R6) [[12]](diffhunk://#diff-00a9db26966c21305c72e8f659628dffaff0d6e9dc98a751406d2141d51a5d90L7-R7) [[13]](diffhunk://#diff-2cbf2f66d5cb77d9f4d00e4c0ce45055620fff50c941a588da31729f09a81f1bL6-R7) [[14]](diffhunk://#diff-20a21d07addeea398c4adb76d077875894a73b4b5b181b9df1fafe497d3fc843L6-R6) [[15]](diffhunk://#diff-909670f1c29b0bba24faf1420504b9eacdff124c4cbbec1ddec5de60653ad007L6-R6) [[16]](diffhunk://#diff-8eef5f0c081743f8002e0faba686e838b323cb53b749706ea42e0440aaa793f1L7-R7) [[17]](diffhunk://#diff-82345842a29e8eaffa4f51aab96fa2aa78597e6639fe4b0ece797bc60edacea8L6-R6) [[18]](diffhunk://#diff-23c6a84d45f3b30ae7ab1a95dec0b30329e702923cc74c5344b3606237ddd929L6-R7) Repository maintenance: * Removed the `examples` submodule entry from `.gitmodules`, indicating that the examples are no longer included as a submodule in the repository. Documentation update: * Updated the contributors image in `README.MD` to reference version `v2.9.7` instead of `v2.9.6`.
feat(‎crypto/grsa): Add RSA encryption and decryption function (#4571 )
2026-06-08 02:27:42 +08:00 · 2025-12-27 16:07:23 +08:00 · 2025-12-26 18:18:30 +08:00 · 2025-12-26 16:43:19 +08:00 · 2025-12-26 16:42:06 +08:00 · 2025-12-26 16:37:45 +08:00
242 changed files with 29325 additions and 5913 deletions
--- a/.github/workflows/ci-main.yml
+++ b/.github/workflows/ci-main.yml
@ -54,7 +54,7 @@ jobs:
    # Service containers to run with `code-test`
    services:
      # Etcd service.
-      # docker run -d --name etcd -p 2379:2379 -e ALLOW_NONE_AUTHENTICATION=yes bitnamilegacy/etcd:3.4.24
+      # docker run -p 2379:2379 -e ALLOW_NONE_AUTHENTICATION=yes bitnamilegacy/etcd:3.4.24
      etcd:
        image: bitnamilegacy/etcd:3.4.24
        env:
@ -75,7 +75,7 @@ jobs:
          - 6379:6379

      # MySQL backend server.
-      # docker run -d --name mysql \
+      # docker run \
      # -p 3306:3306 \
      # -e MYSQL_DATABASE=test \
      # -e MYSQL_ROOT_PASSWORD=12345678 \
@ -89,7 +89,7 @@ jobs:
          - 3306:3306

      # MariaDb backend server.
-      # docker run -d --name mariadb \
+      # docker run \
      # -p 3307:3306 \
      # -e MYSQL_DATABASE=test \
      # -e MYSQL_ROOT_PASSWORD=12345678 \
@ -103,7 +103,7 @@ jobs:
          - 3307:3306

      # PostgreSQL backend server.
-      # docker run -d --name postgres \
+      # docker run \
      # -p 5432:5432 \
      # -e POSTGRES_PASSWORD=12345678 \
      # -e POSTGRES_USER=postgres \
@ -150,7 +150,7 @@ jobs:
          --health-retries 10

      # ClickHouse backend server.
-      # docker run -d --name clickhouse \
+      # docker run \
      # -p 9000:9000 -p 8123:8123 -p 9001:9001 \
      # clickhouse/clickhouse-server:24.11.1.2557-alpine
      clickhouse-server:
@ -161,7 +161,7 @@ jobs:
          - 9001:9001

      # Polaris backend server.
-      # docker run -d --name polaris \
+      # docker run \
      # -p 8090:8090 -p 8091:8091 -p 8093:8093 -p 9090:9090 -p 9091:9091 \
      # polarismesh/polaris-standalone:v1.17.2
      polaris:
@ -198,6 +198,17 @@ jobs:
        ports:
          - 5236:5236

+      # openGauss server
+      # docker run --privileged=true -e GS_PASSWORD=UTpass@1234 -p 9950:5432 opengauss/opengauss:7.0.0-RC1.B023
+      gaussdb:
+        image: opengauss/opengauss:7.0.0-RC1.B023
+        env:
+          GS_PASSWORD: UTpass@1234
+          TZ: Asia/Shanghai
+        ports:
+          - 9950:5432
+
+
      zookeeper:
        image: zookeeper:3.8
        ports:
@ -221,6 +232,12 @@ jobs:
        detached: true
        limit-access-to-actor: false

+    - name: Free Disk Space
+      run: |
+        df -h /
+        sudo rm -rf /usr/share/dotnet /usr/local/lib/android /opt/hostedtoolcache/CodeQL /opt/hostedtoolcache/Python || true
+        df -h /
+
    - name: Start Apollo Containers
      run:  docker compose -f ".github/workflows/apollo/docker-compose.yml" up -d --build
      
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@ -17,11 +17,12 @@ jobs:
    steps:
      - name: Checkout Github Code
        uses: actions/checkout@v5
-      
+
      - name: Set Up Golang Environment
        uses: actions/setup-go@v5
        with:
          go-version: 1.25
+          cache: false
      
      - name: Build CLI Binary
        run: |
--- a/.github/workflows/scripts/before_script.sh
+++ b/.github/workflows/scripts/before_script.sh
--- a/.github/workflows/scripts/ci-main-clean.sh
+++ b/.github/workflows/scripts/ci-main-clean.sh
@ -0,0 +1,250 @@
+#!/usr/bin/env bash
+
+dirpath=$1
+
+# Extract the base directory name for pattern matching
+if [ -n "$dirpath" ]; then
+    dirname=$(basename "$dirpath")
+    echo "Cleaning Docker resources for path: $dirpath (pattern: $dirname)"
+    df -h /
+    
+    # Process containers and images based on the directory
+    case "$dirname" in
+        # "mysql")
+        #     echo "Cleaning mysql resources..."
+        #     containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+        #     if [ -n "$containers" ]; then
+        #         echo "Stopping and removing mysql containers..."
+        #         docker stop $containers 2>/dev/null || true
+        #         docker rm -f $containers 2>/dev/null || true
+        #     fi
+        #     docker rmi -f $(docker images -q mysql 2>/dev/null) 2>/dev/null || true
+        #     ;;
+        "mssql")
+            echo "Cleaning mssql resources..."
+            containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+            if [ -n "$containers" ]; then
+                echo "Stopping and removing mssql containers..."
+                docker stop $containers 2>/dev/null || true
+                docker rm -f $containers 2>/dev/null || true
+            fi
+            docker rmi -f $(docker images -q mcr.microsoft.com/mssql/server 2>/dev/null) 2>/dev/null || true
+        ;;
+        "pgsql")
+            echo "Cleaning postgres resources..."
+            containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+            if [ -n "$containers" ]; then
+                echo "Stopping and removing postgres containers..."
+                docker stop $containers 2>/dev/null || true
+                docker rm -f $containers 2>/dev/null || true
+            fi
+            docker rmi -f $(docker images -q postgres 2>/dev/null) 2>/dev/null || true
+        ;;
+        "oracle")
+            echo "Cleaning oracle resources..."
+            containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+            if [ -n "$containers" ]; then
+                echo "Stopping and removing oracle containers..."
+                docker stop $containers 2>/dev/null || true
+                docker rm -f $containers 2>/dev/null || true
+            fi
+            docker rmi -f $(docker images -q loads/oracle-xe-11g-r2 2>/dev/null) 2>/dev/null || true
+        ;;
+        "dm")
+            echo "Cleaning dm resources..."
+            containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+            if [ -n "$containers" ]; then
+                echo "Stopping and removing dm containers..."
+                docker stop $containers 2>/dev/null || true
+                docker rm -f $containers 2>/dev/null || true
+            fi
+            docker rmi -f $(docker images -q loads/dm 2>/dev/null) 2>/dev/null || true
+        ;;
+        "clickhouse")
+            echo "Cleaning clickhouse resources..."
+            containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+            if [ -n "$containers" ]; then
+                echo "Stopping and removing clickhouse containers..."
+                docker stop $containers 2>/dev/null || true
+                docker rm -f $containers 2>/dev/null || true
+            fi
+            docker rmi -f $(docker images -q clickhouse/clickhouse-server 2>/dev/null) 2>/dev/null || true
+        ;;
+        # "redis")
+        #     echo "Cleaning redis resources..."
+        #     containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+        #     if [ -n "$containers" ]; then
+        #         echo "Stopping and removing redis containers..."
+        #         docker stop $containers 2>/dev/null || true
+        #         docker rm -f $containers 2>/dev/null || true
+        #     fi
+        #     docker rmi -f $(docker images -q redis loads/redis loads/redis-sentinel 2>/dev/null) 2>/dev/null || true
+        #     ;;
+        "etcd")
+            echo "Cleaning etcd resources..."
+            containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+            if [ -n "$containers" ]; then
+                echo "Stopping and removing etcd containers..."
+                docker stop $containers 2>/dev/null || true
+                docker rm -f $containers 2>/dev/null || true
+            fi
+            docker rmi -f $(docker images -q bitnamilegacy/etcd 2>/dev/null) 2>/dev/null || true
+        ;;
+        # "consul")
+        #     echo "Cleaning consul resources..."
+        #     containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+        #     if [ -n "$containers" ]; then
+        #         echo "Stopping and removing consul containers..."
+        #         docker stop $containers 2>/dev/null || true
+        #         docker rm -f $containers 2>/dev/null || true
+        #     fi
+        #     docker rmi -f $(docker images -q consul 2>/dev/null) 2>/dev/null || true
+        #     ;;
+        # "nacos")
+        #     echo "Cleaning nacos resources..."
+        #     containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+        #     if [ -n "$containers" ]; then
+        #         echo "Stopping and removing nacos containers..."
+        #         docker stop $containers 2>/dev/null || true
+        #         docker rm -f $containers 2>/dev/null || true
+        #     fi
+        #     docker rmi -f $(docker images -q nacos/nacos-server 2>/dev/null) 2>/dev/null || true
+        #     ;;
+        # "polaris")
+        #     echo "Cleaning polaris resources..."
+        #     containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+        #     if [ -n "$containers" ]; then
+        #         echo "Stopping and removing polaris containers..."
+        #         docker stop $containers 2>/dev/null || true
+        #         docker rm -f $containers 2>/dev/null || true
+        #     fi
+        #     docker rmi -f $(docker images -q polarismesh/polaris-standalone 2>/dev/null) 2>/dev/null || true
+        #     ;;
+        "zookeeper")
+            echo "Cleaning zookeeper resources..."
+            containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+            if [ -n "$containers" ]; then
+                echo "Stopping and removing zookeeper containers..."
+                docker stop $containers 2>/dev/null || true
+                docker rm -f $containers 2>/dev/null || true
+            fi
+            docker rmi -f $(docker images -q zookeeper 2>/dev/null) 2>/dev/null || true
+        ;;
+        # "apollo")
+        #     echo "Cleaning apollo resources..."
+        #     containers=$(docker ps -aq --filter "name=$dirname" 2>/dev/null)
+        #     if [ -n "$containers" ]; then
+        #         echo "Stopping and removing apollo containers..."
+        #         docker stop $containers 2>/dev/null || true
+        #         docker rm -f $containers 2>/dev/null || true
+        #     fi
+        #     docker rmi -f $(docker images -q loads/apollo-quick-start 2>/dev/null) 2>/dev/null || true
+        #     ;;
+        *)
+            # No matching pattern, skip cleanup
+            echo "No specific Docker cleanup rule for '$dirname', skipping cleanup"
+        ;;
+    esac
+    
+    # Remove dangling images and volumes to free up space
+    echo "Removing dangling images and unused volumes..."
+    docker image prune -f 2>/dev/null || true
+    docker volume prune -f 2>/dev/null || true
+    
+    echo "Docker cleanup completed for $dirname"
+    docker system df
+    df -h /
+fi
+
+# df -h /
+# Filesystem      Size  Used Avail Use% Mounted on
+# /dev/root        72G   67G  5.4G  93% /
+# tmpfs           7.9G   84K  7.9G   1% /dev/shm
+# tmpfs           3.2G  2.6M  3.2G   1% /run
+# tmpfs           5.0M     0  5.0M   0% /run/lock
+# /dev/sdb16      881M   62M  758M   8% /boot
+# /dev/sdb15      105M  6.2M   99M   6% /boot/efi
+# /dev/sda1        74G  4.1G   66G   6% /mnt
+# tmpfs           1.6G   12K  1.6G   1% /run/user/1001
+
+# runner@runnervmg1sw1:~/work/gf/gf$ docker system df
+# TYPE            TOTAL     ACTIVE    SIZE      RECLAIMABLE
+# Images          18        11        8.326GB   1.644GB (19%)
+# Containers      11        11        2.692GB   0B (0%)
+# Local Volumes   11        8         665.7MB   211.9MB (31%)
+# Build Cache     0         0         0B        0B
+
+# runner@runnervmg1sw1:~/work/gf/gf$ docker images
+# REPOSITORY                       TAG                               IMAGE ID       CREATED         SIZE
+# alpine/curl                      latest                            99fd43792a61   2 days ago      13.5MB
+# postgres                         17-alpine                         b6bf692a8125   9 days ago      278MB
+# zookeeper                        3.8                               2f26c02b94ca   10 days ago     306MB
+# mariadb                          11.4                              063fb6684f96   10 days ago     332MB
+# mcr.microsoft.com/mssql/server   2022-latest                       a2fbff321505   4 weeks ago     1.61GB
+# clickhouse/clickhouse-server     24.11.1.2557-alpine               2eee9fd3ae74   12 months ago   539MB
+# redis                            7.0                               7705dd2858c1   18 months ago   109MB
+# consul                           1.15                              686495461132   20 months ago   155MB
+# mysql                            5.7                               5107333e08a8   23 months ago   501MB
+# polarismesh/polaris-standalone   v1.17.2                           b7a8cf0a8438   2 years ago     545MB
+# bitnamilegacy/etcd               3.4.24                            74ae5e205ac5   2 years ago     134MB
+# nacos/nacos-server               v2.1.2                            a978644d9246   2 years ago     1.06GB
+# loads/redis                      7.0-sentinel                      6f12d40540ba   3 years ago     114MB
+# loads/dm                         v8.1.2.128_ent_x86_64_ctm_pack4   ccb727ce9dce   3 years ago     432MB
+# loads/redis-sentinel             7.0                               6818c626f5ca   3 years ago     104MB
+# loads/apollo-quick-start         latest                            8490de672148   3 years ago     190MB
+# alpine                           3.8                               c8bccc0af957   5 years ago     4.41MB
+# loads/oracle-xe-11g-r2           11.2.0                            0d19fd2e072e   6 years ago     2.1GB
+
+# runner@runnervmg1sw1:~/work/gf/gf$ docker ps -s
+# CONTAINER ID   IMAGE                                              COMMAND                  CREATED         STATUS                   PORTS                                                                                                                                                                                                                         NAMES                                                                               SIZE
+# 8214f83420c6   zookeeper:3.8                                      "/docker-entrypoint.…"   6 minutes ago   Up 6 minutes             2888/tcp, 3888/tcp, 0.0.0.0:2181->2181/tcp, [::]:2181->2181/tcp, 8080/tcp                                                                                                                                                     d66bac92ae9646f688f70ed4b5176f14_zookeeper38_3a22ef                                 33kB (virtual 306MB)
+# 8938d73842e8   loads/dm:v8.1.2.128_ent_x86_64_ctm_pack4           "/bin/bash /opt/star…"   6 minutes ago   Up 6 minutes             0.0.0.0:5236->5236/tcp, [::]:5236->5236/tcp                                                                                                                                                                                   ca280fbdb86f40c2acf86d7d526c6285_loadsdmv812128_ent_x86_64_ctm_pack4_770a59         844MB (virtual 1.28GB)
+# 0d3a653fe1f2   loads/oracle-xe-11g-r2:11.2.0                      "/bin/sh -c '/usr/sb…"   6 minutes ago   Up 6 minutes             22/tcp, 8080/tcp, 0.0.0.0:1521->1521/tcp, [::]:1521->1521/tcp                                                                                                                                                                 2048856d428c4967b1c35193eb8c9192_loadsoraclexe11gr21120_295d54                      1.3GB (virtual 3.4GB)
+# ca3936189166   polarismesh/polaris-standalone:v1.17.2             "/bin/bash run.sh"       6 minutes ago   Up 6 minutes             0.0.0.0:8090-8091->8090-8091/tcp, [::]:8090-8091->8090-8091/tcp, 8080/tcp, 8100-8101/tcp, 0.0.0.0:8093->8093/tcp, [::]:8093->8093/tcp, 8761/tcp, 15010/tcp, 0.0.0.0:9090-9091->9090-9091/tcp, [::]:9090-9091->9090-9091/tcp   cbd43dceef754e2d8aab507e33167be7_polarismeshpolarisstandalonev1172_ca40b6           299MB (virtual 844MB)
+# 26169dad485e   clickhouse/clickhouse-server:24.11.1.2557-alpine   "/entrypoint.sh"         6 minutes ago   Up 6 minutes             0.0.0.0:8123->8123/tcp, [::]:8123->8123/tcp, 0.0.0.0:9000-9001->9000-9001/tcp, [::]:9000-9001->9000-9001/tcp, 9009/tcp                                                                                                        f1c7766fbe36401792a6f735d7acf123_clickhouseclickhouseserver241112557alpine_cfc034   338kB (virtual 539MB)
+# 04689a1d581f   mcr.microsoft.com/mssql/server:2022-latest         "/opt/mssql/bin/laun…"   6 minutes ago   Up 6 minutes (healthy)   0.0.0.0:1433->1433/tcp, [::]:1433->1433/tcp                                                                                                                                                                                   41d685349a7640b28230db8d0f60efe7_mcrmicrosoftcommssqlserver2022latest_fe29fb        108MB (virtual 1.72GB)
+# d5fbc5f811af   postgres:17-alpine                                 "docker-entrypoint.s…"   6 minutes ago   Up 6 minutes (healthy)   0.0.0.0:5432->5432/tcp, [::]:5432->5432/tcp                                                                                                                                                                                   2783be71b5ce417ab9a31428e7b4d8f2_postgres17alpine_c60840                            63B (virtual 278MB)
+# da96a7ad7a01   mariadb:11.4                                       "docker-entrypoint.s…"   7 minutes ago   Up 7 minutes             0.0.0.0:3307->3306/tcp, [::]:3307->3306/tcp                                                                                                                                                                                   45eed646fa6c4a698893ee11cda95a4c_mariadb114_3a9cd6                                  2B (virtual 332MB)
+# 27ba1904ba3a   mysql:5.7                                          "docker-entrypoint.s…"   7 minutes ago   Up 7 minutes             0.0.0.0:3306->3306/tcp, [::]:3306->3306/tcp, 33060/tcp                                                                                                                                                                        ea6d7a4c207d427a95b5ae0db91fdf56_mysql57_c21053                                     4B (virtual 501MB)
+# 518e785d1bb6   redis:7.0                                          "docker-entrypoint.s…"   7 minutes ago   Up 7 minutes (healthy)   0.0.0.0:6379->6379/tcp, [::]:6379->6379/tcp                                                                                                                                                                                   af6044fc849e441bbc6c48f7a5ec5fec_redis70_b11994                                     0B (virtual 109MB)
+# 7495ec2cd8e3   bitnamilegacy/etcd:3.4.24                          "/opt/bitnami/script…"   7 minutes ago   Up 7 minutes             0.0.0.0:2379->2379/tcp, [::]:2379->2379/tcp, 2380/tcp                                                                                                                                                                         49f2a2a6bf3a4fae842cc950bbc3658a_bitnamilegacyetcd3424_1265e1                       145MB (virtual 279MB)
+
+# runner@runnervmg1sw1:~/work/gf/gf$ du -ah --max-depth=1 /usr | sort -n
+# 4.0K    /usr/games
+# 4.0K    /usr/lib64
+# 6.6G    /usr/lib
+# 9.3G    /usr/share
+# 15M     /usr/lib32
+# 24G     /usr/local
+# 41G     /usr
+# 95M     /usr/sbin
+# 156M    /usr/include
+# 158M    /usr/src
+# 402M    /usr/libexec
+# 841M    /usr/bin
+
+# runner@runnervmg1sw1:~/work/gf/gf$ du -ah --max-depth=1 /opt | sort -n                                                                        
+# 4.0K    /opt/pipx_bin
+# 5.8G    /opt/hostedtoolcache
+# 8.5G    /opt
+# 12K     /opt/containerd
+# 14M     /opt/hca
+# 16K     /opt/post-generation
+# 217M    /opt/runner-cache
+# 243M    /opt/actionarchivecache
+# 374M    /opt/google
+# 515M    /opt/pipx
+# 655M    /opt/az
+# 783M    /opt/microsoft
+
+# runner@runnervmg1sw1:~/work/gf/gf$ du -ah --max-depth=1 /opt/hostedtoolcache/ | sort -n
+# 1.1G    /opt/hostedtoolcache/go
+# 1.6G    /opt/hostedtoolcache/CodeQL
+# 1.9G    /opt/hostedtoolcache/Python
+# 5.8G    /opt/hostedtoolcache/
+# 9.9M    /opt/hostedtoolcache/protoc
+# 24K     /opt/hostedtoolcache/Java_Temurin-Hotspot_jdk
+# 217M    /opt/hostedtoolcache/Ruby
+# 520M    /opt/hostedtoolcache/PyPy
+# 574M    /opt/hostedtoolcache/node
+
--- a/.github/workflows/scripts/ci-main.sh
+++ b/.github/workflows/scripts/ci-main.sh
@ -6,56 +6,54 @@ coverage=$1
 for file in `find . -name go.mod`; do
    dirpath=$(dirname $file)
    echo $dirpath
-
-    # ignore mssql tests as its docker service failed
-    # TODO remove this ignoring codes after the mssql docker service OK
-    if [ "mssql" = $(basename $dirpath) ]; then
-        continue 1
-    fi
-
+    
    # package kubecm was moved to sub ci procedure.
    if [ "kubecm" = $(basename $dirpath) ]; then
        continue 1
    fi
-
+    
    # examples directory was moved to sub ci procedure.
    if [[ $dirpath =~ "/examples/" ]]; then
        continue 1
    fi
-
+    
+    if [[ $file =~ "/testdata/" ]]; then
+        echo "ignore testdata path $file"
+        continue 1
+    fi
+    
    # Check if it's a contrib directory
    if [[ $dirpath =~ "/contrib/" ]]; then
        # Check if go version meets the requirement
        if ! go version | grep -qE "go${LATEST_GO_VERSION}"; then
            echo "ignore path $dirpath as go version is not ${LATEST_GO_VERSION}: $(go version)"
+            # clean docker containers and images to free disk space
+            # bash .github/workflows/scripts/ci-main-clean.sh "$dirpath"
            continue 1
        fi
    fi
-
-    if [[ $file =~ "/testdata/" ]]; then
-        echo "ignore testdata path $file"
-        continue 1
-    fi
-
-    if [[ $dirpath = "." ]]; then
-        # No space left on device error sometimes occurs in CI pipelines, so clean the cache before tests.
-        go clean -cache
-    fi
-
+    
+    # if [[ $dirpath = "." ]]; then
+    #     # No space left on device error sometimes occurs in CI pipelines, so clean the cache before tests.
+    #     go clean -cache
+    # fi
+    
    cd $dirpath
    go mod tidy
    go build ./...
-
+    
    # test with coverage
    if [ "${coverage}" = "coverage" ]; then
-      go test ./... -race -coverprofile=coverage.out -covermode=atomic -coverpkg=./...,github.com/gogf/gf/... || exit 1
-
-      if grep -q "/gogf/gf/.*/v2" go.mod; then
-        sed -i "s/gogf\/gf\(\/.*\)\/v2/gogf\/gf\/v2\1/g" coverage.out
-      fi
+        go test ./... -count=1 -race -coverprofile=coverage.out -covermode=atomic -coverpkg=./...,github.com/gogf/gf/... || exit 1
+        
+        if grep -q "/gogf/gf/.*/v2" go.mod; then
+            sed -i "s/gogf\/gf\(\/.*\)\/v2/gogf\/gf\/v2\1/g" coverage.out
+        fi
    else
-      go test ./... -race || exit 1
+        go test ./... -count=1 -race || exit 1
    fi
-
+    
    cd -
+    # clean docker containers and images to free disk space
+    # bash .github/workflows/scripts/ci-main-clean.sh "$dirpath"
 done
--- a/.github/workflows/scripts/ci-sub.sh
+++ b/.github/workflows/scripts/ci-sub.sh
--- a/.gitmodules
+++ b/.gitmodules
@ -1,3 +0,0 @@
-[submodule "examples"]
-	path = examples
-	url = git@github.com:gogf/examples.git
--- a/1
+++ b/1
@ -6,6 +6,7 @@ tidy:
 	./.make_tidy.sh

 # execute "golangci-lint" to check code style
+# go install github.com/golangci/golangci-lint/v2/cmd/golangci-lint@latest
 .PHONY: lint
 lint:
 	golangci-lint run -c .golangci.yml
--- a/README.MD
+++ b/README.MD
@ -1,3 +1,4 @@
+English | [简体中文](README.zh_CN.MD)

 <div align=center>
 <img src="https://goframe.org/img/logo_full.png" width="300" alt="goframe gf logo"/>
@ -23,6 +24,12 @@

 A powerful framework for faster, easier, and more efficient project development.

+## Installation
+
+```bash
+go get -u github.com/gogf/gf/v2
+```
+
 ## Documentation

 - Official Site: [https://goframe.org](https://goframe.org)
@ -31,13 +38,14 @@ A powerful framework for faster, easier, and more efficient project development.
 - Mirror Site: [Github Pages](https://pages.goframe.org)
 - Mirror Site: [Offline Docs](https://github.com/gogf/goframe.org-pdf?tab=readme-ov-file#%E6%9C%80%E6%96%B0%E7%89%88%E6%9C%AC)
 - GoDoc API: [https://pkg.go.dev/github.com/gogf/gf/v2](https://pkg.go.dev/github.com/gogf/gf/v2)
+- Doc Source: [https://github.com/gogf/gf-site](https://github.com/gogf/gf-site)

 ## Contributors

 💖 [Thanks to all the contributors who made GoFrame possible](https://github.com/gogf/gf/graphs/contributors) 💖

 <a href="https://github.com/gogf/gf/graphs/contributors">
-<img src="https://goframe.org/img/contributors.svg?version=v2.9.5" alt="goframe contributors"/>
+<img src="https://goframe.org/img/contributors.svg?version=v2.9.7" alt="goframe contributors"/>
 </a>

 ## License
--- a/README.zh_CN.MD
+++ b/README.zh_CN.MD
@ -0,0 +1,53 @@
+[English](README.MD) | 简体中文
+
+<div align=center>
+<img src="https://goframe.org/img/logo_full.png" width="300" alt="goframe gf logo"/>
+
+[![Go Reference](https://pkg.go.dev/badge/github.com/gogf/gf/v2.svg)](https://pkg.go.dev/github.com/gogf/gf/v2)
+[![GoFrame CI](https://github.com/gogf/gf/actions/workflows/ci-main.yml/badge.svg)](https://github.com/gogf/gf/actions/workflows/ci-main.yml)
+[![OpenSSF Scorecard](https://api.securityscorecards.dev/projects/github.com/gogf/gf/badge)](https://scorecard.dev/viewer/?uri=github.com/gogf/gf)
+[![CII Best Practices](https://bestpractices.coreinfrastructure.org/projects/9233/badge)](https://bestpractices.coreinfrastructure.org/projects/9233)
+[![Go Report Card](https://goreportcard.com/badge/github.com/gogf/gf/v2)](https://goreportcard.com/report/github.com/gogf/gf/v2)
+[![Code Coverage](https://codecov.io/gh/gogf/gf/branch/master/graph/badge.svg)](https://codecov.io/gh/gogf/gf)
+[![Production Ready](https://img.shields.io/badge/production-ready-blue.svg?style=flat)](https://github.com/gogf/gf)
+[![License](https://img.shields.io/github/license/gogf/gf.svg?style=flat)](https://github.com/gogf/gf)
+
+[![Release](https://img.shields.io/github/v/release/gogf/gf?style=flat)](https://github.com/gogf/gf/releases)
+[![GitHub pull requests](https://img.shields.io/github/issues-pr/gogf/gf?style=flat)](https://github.com/gogf/gf/pulls)
+[![GitHub closed pull requests](https://img.shields.io/github/issues-pr-closed/gogf/gf?style=flat)](https://github.com/gogf/gf/pulls?q=is%3Apr+is%3Aclosed)
+[![GitHub issues](https://img.shields.io/github/issues/gogf/gf?style=flat)](https://github.com/gogf/gf/issues)
+[![GitHub closed issues](https://img.shields.io/github/issues-closed/gogf/gf?style=flat)](https://github.com/gogf/gf/issues?q=is%3Aissue+is%3Aclosed)
+![Stars](https://img.shields.io/github/stars/gogf/gf?style=flat)
+![Forks](https://img.shields.io/github/forks/gogf/gf?style=flat)
+
+</div>
+
+一个强大的框架，为了更快、更轻松、更高效的项目开发。
+
+## 安装
+
+```bash
+go get -u github.com/gogf/gf/v2
+```
+
+## 文档
+
+- 官方网站: [https://goframe.org](https://goframe.org)
+- 官方网站(en): [https://goframe.org/en](https://goframe.org/en)
+- 国内镜像: [https://goframe.org.cn](https://goframe.org.cn)
+- 镜像网站: [Github Pages](https://pages.goframe.org)
+- 镜像网站: [离线文档](https://github.com/gogf/goframe.org-pdf?tab=readme-ov-file#%E6%9C%80%E6%96%B0%E7%89%88%E6%9C%AC)
+- Go包文档: [https://pkg.go.dev/github.com/gogf/gf/v2](https://pkg.go.dev/github.com/gogf/gf/v2)
+- 文档源码: [https://github.com/gogf/gf-site](https://github.com/gogf/gf-site)
+
+## 贡献者
+
+💖 [感谢所有使 GoFrame 成为可能的贡献者](https://github.com/gogf/gf/graphs/contributors) 💖
+
+<a href="https://github.com/gogf/gf/graphs/contributors">
+<img src="https://goframe.org/img/contributors.svg?version=v2.9.5" alt="goframe contributors"/>
+</a>
+
+## 许可证
+
+`GoFrame` 采用 [MIT License](LICENSE) 许可，100% 免费和开源，永久保持。
--- a/cmd/gf/README.MD
+++ b/cmd/gf/README.MD
@ -1,3 +1,5 @@
+English | [简体中文](README.zh_CN.MD)
+
 # gf

 `gf` is a powerful CLI tool for building [GoFrame](https://goframe.org) application with convenience.
@ -21,18 +23,18 @@ You can also install `gf` tool using pre-built binaries: <https://github.com/gog

 3. Database support

-   |     DB     | builtin support |                                                                             remarks                                                                              |
-   |:----------:|:---------------:|:----------------------------------------------------------------------------------------------------------------------------------------------------------------:|
-   |   mysql    |       yes       |                                                                                -                                                                                 |
-   |  mariadb   |       yes       |                                                                                -                                                                                 |
-   |    tidb    |       yes       |                                                                                -                                                                                 |
-   |   mssql    |       yes       |                                                                                -                                                                                 |
-   |   oracle   |       yes       |                                                                                -                                                                                 |
-   |   pgsql    |       yes       |                                                                                -                                                                                 |
-   |   sqlite   |       yes       |                                                                                -                                                                                 |
-   | sqlitecgo  |       no        | to support sqlite database on 32bit architecture systems, manually add package import to the [source codes](./internal/cmd/cmd_gen_dao.go) and do the building.  |
-   | clickhouse |       no        |                              manually add package import to the [source codes](./internal/cmd/cmd_gen_dao.go) and do the building.                               |
-   |     dm     |       no        |                              manually add package import to the [source codes](./internal/cmd/cmd_gen_dao.go) and do the building.                               |
+   |     DB     | builtin support |                                                                             remarks                                                                             |
+   | :--------: | :-------------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------: |
+   |   mysql    |       yes       |                                                                                -                                                                                |
+   |  mariadb   |       yes       |                                                                                -                                                                                |
+   |    tidb    |       yes       |                                                                                -                                                                                |
+   |   mssql    |       yes       |                                                                                -                                                                                |
+   |   oracle   |       yes       |                                                                                -                                                                                |
+   |   pgsql    |       yes       |                                                                                -                                                                                |
+   |   sqlite   |       yes       |                                                                                -                                                                                |
+   | sqlitecgo  |       no        | to support sqlite database on 32bit architecture systems, manually add package import to the [source codes](./internal/cmd/cmd_gen_dao.go) and do the building. |
+   | clickhouse |       yes       |                                                                                -                                                                                |
+   |     dm     |       no        |                              manually add package import to the [source codes](./internal/cmd/cmd_gen_dao.go) and do the building.                              |

 ## 2) Manually Install

@ -43,30 +45,31 @@ go install github.com/gogf/gf/cmd/gf/v2@v2.5.5 # certain version(should be >= v2

 ## 2. Commands

-```html
-$ gf
+```shell
+$ gf -h
 USAGE
    gf COMMAND [OPTION]

 COMMAND
-   up         upgrade GoFrame version/tool to latest one in current project
-   env        show current Golang environment variables
-   fix        auto fixing codes after upgrading to new GoFrame version
-   run        running go codes with hot-compiled-like feature
-   gen        automatically generate go files for dao/do/entity/pb/pbentity
-   tpl        template parsing and building commands
-   init       create and initialize an empty GoFrame project
-   pack       packing any file/directory to a resource file, or a go file
-   build      cross-building go project for lots of platforms
-   docker     build docker image for current GoFrame project
-   install    install gf binary to system (might need root/admin permission)
-   version    show version information of current binary
+    up         upgrade GoFrame version/tool to latest one in current project
+    env        show current Golang environment variables
+    fix        auto fixing codes after upgrading to new GoFrame version
+    run        running go codes with hot-compiled-like feature
+    gen        automatically generate go files for dao/do/entity/pb/pbentity
+    tpl        template parsing and building commands
+    init       create and initialize an empty GoFrame project
+    pack       packing any file/directory to a resource file, or a go file
+    build      cross-building go project for lots of platforms
+    docker     build docker image for current GoFrame project
+    install    install gf binary to system (might need root/admin permission)
+    version    show version information of current binary
+    doc        download https://pages.goframe.org/ to run locally

 OPTION
-   -y, --yes       all yes for all command without prompt ask
-   -v, --version   show version information of current binary
-   -d, --debug     show internal detailed debugging information
-   -h, --help      more information about this command
+    -y, --yes       all yes for all command without prompt ask
+    -v, --version   show version information of current binary
+    -d, --debug     show internal detailed debugging information
+    -h, --help      more information about this command

 ADDITIONAL
    Use "gf COMMAND -h" for details about a command.
--- a/cmd/gf/README.zh_CN.MD
+++ b/cmd/gf/README.zh_CN.MD
@ -0,0 +1,82 @@
+[English](README.MD) | 简体中文
+
+# gf
+
+`gf` 是一个强大的 CLI 工具，用于便捷地构建 [GoFrame](https://goframe.org) 应用程序。
+
+## 1. 安装
+
+## 1) 预编译二进制文件
+
+您也可以使用预构建的二进制文件安装 `gf` 工具：<https://github.com/gogf/gf/releases>
+
+1. `Mac` & `Linux`
+
+   ```shell
+    wget -O gf https://github.com/gogf/gf/releases/latest/download/gf_$(go env GOOS)_$(go env GOARCH) && chmod +x gf && ./gf install -y && rm ./gf
+   ```
+
+   > 如果您使用 `zsh`，您可能需要通过命令 `alias gf=gf` 重命名别名以解决 `gf` 和 `git fetch` 之间的冲突。
+
+2. `Windows`
+   手动下载，在命令行中执行，然后按照说明操作。
+
+3. 数据库支持
+
+   |     数据库     | 内置支持 |                                                                             说明                                                                             |
+   | :--------: | :-------------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------: |
+   |   mysql    |       是       |                                                                                -                                                                                |
+   |  mariadb   |       是       |                                                                                -                                                                                |
+   |    tidb    |       是       |                                                                                -                                                                                |
+   |   mssql    |       是       |                                                                                -                                                                                |
+   |   oracle   |       是       |                                                                                -                                                                                |
+   |   pgsql    |       是       |                                                                                -                                                                                |
+   |   sqlite   |       是       |                                                                                -                                                                                |
+   | sqlitecgo  |       否        | 要在 32 位架构系统上支持 sqlite 数据库，请手动向[源代码](./internal/cmd/cmd_gen_dao.go)添加包导入并进行构建。 |
+   | clickhouse |       是        |                                                                                -                                                                                |
+   |     dm     |       否        |                              手动向[源代码](./internal/cmd/cmd_gen_dao.go)添加包导入并进行构建。                              |
+
+## 2) 手动安装
+
+```shell
+go install github.com/gogf/gf/cmd/gf/v2@latest # 最新版本
+go install github.com/gogf/gf/cmd/gf/v2@v2.5.5 # 特定版本(应该 >= v2.5.5)
+```
+
+## 2. 命令
+
+```shell
+$ gf -h
+用法
+    gf 命令 [选项]
+
+命令
+    up         升级项目中的 GoFrame 版本/工具到最新版本
+    env        显示当前 Golang 环境变量
+    fix        升级到新 GoFrame 版本后自动修复代码
+    run        运行 go 代码，具有热编译功能
+    gen        自动生成 dao/do/entity/pb/pbentity 的 go 文件
+    tpl        模板解析和构建命令
+    init       创建并初始化一个空的 GoFrame 项目
+    pack       将任何文件/目录打包到资源文件或 go 文件
+    build      为多个平台交叉编译 go 项目
+    docker     为当前 GoFrame 项目构建 docker 镜像
+    install    将 gf 二进制文件安装到系统(可能需要 root/admin 权限)
+    version    显示当前二进制文件的版本信息
+    doc        下载 https://pages.goframe.org/ 本地运行
+
+选项
+    -y, --yes       对所有命令都使用 yes，不再提示
+    -v, --version   显示当前二进制文件的版本信息
+    -d, --debug     显示内部详细的调试信息
+    -h, --help      显示此命令的更多信息
+
+附加信息
+    使用 "gf 命令 -h" 获取有关命令的详细信息。
+```
+
+## 3. 常见问题
+
+### 1). 命令 `gf run` 返回 `pipe: too many open files`
+
+请使用 `ulimit -n 65535` 扩大系统配置以增加当前终端 shell 会话的最大打开文件数，然后再运行 `gf run`。
--- a/cmd/gf/go.mod
+++ b/cmd/gf/go.mod
@ -3,13 +3,13 @@ module github.com/gogf/gf/cmd/gf/v2
 go 1.23.0

 require (
-	github.com/gogf/gf/contrib/drivers/clickhouse/v2 v2.9.5
-	github.com/gogf/gf/contrib/drivers/mssql/v2 v2.9.5
-	github.com/gogf/gf/contrib/drivers/mysql/v2 v2.9.5
-	github.com/gogf/gf/contrib/drivers/oracle/v2 v2.9.5
-	github.com/gogf/gf/contrib/drivers/pgsql/v2 v2.9.5
-	github.com/gogf/gf/contrib/drivers/sqlite/v2 v2.9.5
-	github.com/gogf/gf/v2 v2.9.5
+	github.com/gogf/gf/contrib/drivers/clickhouse/v2 v2.9.7
+	github.com/gogf/gf/contrib/drivers/mssql/v2 v2.9.7
+	github.com/gogf/gf/contrib/drivers/mysql/v2 v2.9.7
+	github.com/gogf/gf/contrib/drivers/oracle/v2 v2.9.7
+	github.com/gogf/gf/contrib/drivers/pgsql/v2 v2.9.7
+	github.com/gogf/gf/contrib/drivers/sqlite/v2 v2.9.7
+	github.com/gogf/gf/v2 v2.9.7
 	github.com/gogf/selfupdate v0.0.0-20231215043001-5c48c528462f
 	github.com/olekukonko/tablewriter v1.1.0
 	github.com/schollz/progressbar/v3 v3.15.0
@ -23,7 +23,7 @@ require (
 	github.com/ClickHouse/clickhouse-go/v2 v2.0.15 // indirect
 	github.com/clbanning/mxj/v2 v2.7.0 // indirect
 	github.com/dustin/go-humanize v1.0.1 // indirect
-	github.com/emirpasic/gods v1.18.1 // indirect
+	github.com/emirpasic/gods/v2 v2.0.0-alpha // indirect
 	github.com/fatih/color v1.18.0 // indirect
 	github.com/fsnotify/fsnotify v1.9.0 // indirect
 	github.com/glebarez/go-sqlite v1.21.2 // indirect
--- a/cmd/gf/go.sum
+++ b/cmd/gf/go.sum
@ -27,8 +27,8 @@ github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/dustin/go-humanize v1.0.1 h1:GzkhY7T5VNhEkwH0PVJgjz+fX1rhBrR7pRT3mDkpeCY=
 github.com/dustin/go-humanize v1.0.1/go.mod h1:Mu1zIs6XwVuF/gI1OepvI0qD18qycQx+mFykh5fBlto=
-github.com/emirpasic/gods v1.18.1 h1:FXtiHYKDGKCW2KzwZKx0iC0PQmdlorYgdFG9jPXJ1Bc=
-github.com/emirpasic/gods v1.18.1/go.mod h1:8tpGGwCnJ5H4r6BWwaV6OrWmMoPhUl5jm/FMNAnJvWQ=
+github.com/emirpasic/gods/v2 v2.0.0-alpha h1:dwFlh8pBg1VMOXWGipNMRt8v96dKAIvBehtCt6OtunU=
+github.com/emirpasic/gods/v2 v2.0.0-alpha/go.mod h1:W0y4M2dtBB9U5z3YlghmpuUhiaZT2h6yoeE+C1sCp6A=
 github.com/fatih/color v1.18.0 h1:S8gINlzdQ840/4pfAwic/ZE0djQEH3wM94VfqLTZcOM=
 github.com/fatih/color v1.18.0/go.mod h1:4FelSpRwEGDpQ12mAdzqdOukCy4u8WUtOY6lkT/6HfU=
 github.com/fsnotify/fsnotify v1.9.0 h1:2Ml+OJNzbYCTzsxtv8vKSFD9PbJjmhYF14k/jKC7S9k=
@ -46,20 +46,6 @@ github.com/go-ole/go-ole v1.2.6/go.mod h1:pprOEPIfldk/42T2oK7lQ4v4JSDwmV0As9GaiU
 github.com/go-sql-driver/mysql v1.4.0/go.mod h1:zAC/RDZ24gD3HViQzih4MyKcchzm+sOG5ZlKdlhCg5w=
 github.com/go-sql-driver/mysql v1.7.1 h1:lUIinVbN1DY0xBg0eMOzmmtGoHwWBbvnWubQUrtU8EI=
 github.com/go-sql-driver/mysql v1.7.1/go.mod h1:OXbVy3sEdcQ2Doequ6Z5BW6fXNQTmx+9S1MCJN5yJMI=
-github.com/gogf/gf/contrib/drivers/clickhouse/v2 v2.9.5 h1:WTbuQvbtOSddi2GtiobM/FCRUr5god7yzlEQP7WGOM8=
-github.com/gogf/gf/contrib/drivers/clickhouse/v2 v2.9.5/go.mod h1:/lnvHd9+8VmjDLdgIczzRTJA1tZYY5AA8bdcaexi/ao=
-github.com/gogf/gf/contrib/drivers/mssql/v2 v2.9.5 h1:hgkgzbi6j8tDf+UpjG01fLqnAchD3913RZ37ruY9TqE=
-github.com/gogf/gf/contrib/drivers/mssql/v2 v2.9.5/go.mod h1:Sy0DQNJ/xEL4snJyWqcflmYGr2jE8Tn9mynxqbe2Dds=
-github.com/gogf/gf/contrib/drivers/mysql/v2 v2.9.5 h1:0+ZBYhi4sqwxXwL+hIBpp06a7G4m5nmjskQ3NNb8qYc=
-github.com/gogf/gf/contrib/drivers/mysql/v2 v2.9.5/go.mod h1:vyB7J/uJcLCrHD5lfFBzxhEEMkePIRzfhd33EcsuLa0=
-github.com/gogf/gf/contrib/drivers/oracle/v2 v2.9.5 h1:eCPD7oteF/gurCU5ZfAhFBU7E1vI85cnoHKRdckn9Vc=
-github.com/gogf/gf/contrib/drivers/oracle/v2 v2.9.5/go.mod h1:e5sxdxw3OrAFB+4VdYt3Wbm0G7LP5wofNRKj3JHIots=
-github.com/gogf/gf/contrib/drivers/pgsql/v2 v2.9.5 h1:FCgvTLBuhPX7HE6YyR/dbNASoiKv72jpVS5SqoYYJTw=
-github.com/gogf/gf/contrib/drivers/pgsql/v2 v2.9.5/go.mod h1:/Lz1qzhYN3ogt5aMFoIfjp82SbeuzjpXCqQhFu4Z3MI=
-github.com/gogf/gf/contrib/drivers/sqlite/v2 v2.9.5 h1:MooWqn5qLMfB105PlBvd2Z7J3KdVBsjYxULtb42u308=
-github.com/gogf/gf/contrib/drivers/sqlite/v2 v2.9.5/go.mod h1:pr68Q85xhRnJ5PhyGte/LiFJ/m+998RPjW+Jn+w+xYo=
-github.com/gogf/gf/v2 v2.9.5 h1:1scfOdHbMP854oQaiLejl+eL+c4xfuvtWmmZiDJxbKs=
-github.com/gogf/gf/v2 v2.9.5/go.mod h1:VUb5eyJKpvW77O/dXsbbLNO/Kjrg0UycIiq0lRiBjjo=
 github.com/gogf/selfupdate v0.0.0-20231215043001-5c48c528462f h1:7xfXR/BhG3JDqO1s45n65Oyx9t4E/UqDOXep6jXdLCM=
 github.com/gogf/selfupdate v0.0.0-20231215043001-5c48c528462f/go.mod h1:HnYoio6S7VaFJdryKcD/r9HgX+4QzYfr00XiXUo/xz0=
 github.com/gogo/protobuf v1.3.2/go.mod h1:P1XiOD3dCwIKUDQYPy72D8LYyHL2YPYrpS2s69NZV8Q=
--- a/cmd/gf/go.work
+++ b/cmd/gf/go.work
@ -14,5 +14,9 @@ replace (
 	github.com/gogf/gf/contrib/drivers/oracle/v2 => ../../contrib/drivers/oracle
 	github.com/gogf/gf/contrib/drivers/pgsql/v2 => ../../contrib/drivers/pgsql
 	github.com/gogf/gf/contrib/drivers/sqlite/v2 => ../../contrib/drivers/sqlite
+	github.com/gogf/gf/contrib/drivers/mariadb/v2 => ../../contrib/drivers/mariadb
+	github.com/gogf/gf/contrib/drivers/tidb/v2 => ../../contrib/drivers/tidb
+	github.com/gogf/gf/contrib/drivers/oceanbase/v2 => ../../contrib/drivers/oceanbase
+	github.com/gogf/gf/contrib/drivers/gaussdb/v2 => ../../contrib/drivers/gaussdb
 	github.com/gogf/gf/v2 => ../../
 )
--- a/cmd/gf/internal/cmd/cmd_init.go
+++ b/cmd/gf/internal/cmd/cmd_init.go
@ -7,9 +7,11 @@
 package cmd

 import (
+	"bufio"
 	"context"
 	"fmt"
 	"os"
+	"strconv"
 	"strings"

 	"github.com/gogf/gf/v2/frame/g"
@ -20,6 +22,7 @@ import (
 	"github.com/gogf/gf/v2/text/gstr"
 	"github.com/gogf/gf/v2/util/gtag"

+	"github.com/gogf/gf/cmd/gf/v2/internal/cmd/geninit"
 	"github.com/gogf/gf/cmd/gf/v2/internal/utility/allyes"
 	"github.com/gogf/gf/cmd/gf/v2/internal/utility/mlog"
 	"github.com/gogf/gf/cmd/gf/v2/internal/utility/utils"
@ -44,6 +47,12 @@ const (
 gf init my-project
 gf init my-mono-repo -m
 gf init my-mono-repo -a
+gf init my-project -u
+gf init my-project -g "github.com/myorg/myproject"
+gf init -r github.com/gogf/template-single my-project
+gf init -r github.com/gogf/template-single my-project -s
+gf init -r github.com/gogf/examples/httpserver/jwt my-jwt
+gf init -i
 `
 	cInitNameBrief = `
 name for the project. It will create a folder with NAME in current directory.
@ -55,6 +64,16 @@ The NAME will also be the module name for the project.
 	cInitGitignore = ".gitignore"
 )

+// defaultTemplates is the list of predefined templates for interactive selection
+var defaultTemplates = []struct {
+	Name string
+	Repo string
+	Desc string
+}{
+	{"template-single", "github.com/gogf/template-single", "Single project template"},
+	{"template-mono", "github.com/gogf/template-mono", "Mono-repo project template"},
+}
+
 func init() {
 	gtag.Sets(g.MapStrStr{
 		`cInitBrief`:     cInitBrief,
@ -64,17 +83,86 @@ func init() {
 }

 type cInitInput struct {
-	g.Meta  `name:"init"`
-	Name    string `name:"NAME" arg:"true" v:"required" brief:"{cInitNameBrief}"`
-	Mono    bool   `name:"mono" short:"m" brief:"initialize a mono-repo instead a single-repo" orphan:"true"`
-	MonoApp bool   `name:"monoApp" short:"a" brief:"initialize a mono-repo-app instead a single-repo" orphan:"true"`
-	Update  bool   `name:"update" short:"u" brief:"update to the latest goframe version" orphan:"true"`
-	Module  string `name:"module" short:"g" brief:"custom go module"`
+	g.Meta      `name:"init"`
+	Name        string `name:"NAME" arg:"true" brief:"{cInitNameBrief}"`
+	Mono        bool   `name:"mono" short:"m" brief:"initialize a mono-repo instead a single-repo" orphan:"true"`
+	MonoApp     bool   `name:"monoApp" short:"a" brief:"initialize a mono-repo-app instead a single-repo" orphan:"true"`
+	Update      bool   `name:"update" short:"u" brief:"update to the latest goframe version" orphan:"true"`
+	Module      string `name:"module" short:"g" brief:"custom go module"`
+	Repo        string `name:"repo" short:"r" brief:"remote repository URL for template download"`
+	SelectVer   bool   `name:"select" short:"s" brief:"enable interactive version selection for remote template" orphan:"true"`
+	Interactive bool   `name:"interactive" short:"i" brief:"enable interactive mode to select template" orphan:"true"`
 }

 type cInitOutput struct{}

 func (c cInit) Index(ctx context.Context, in cInitInput) (out *cInitOutput, err error) {
+	// Check if using remote template mode
+	if in.Repo != "" || in.Interactive {
+		return c.initFromRemote(ctx, in)
+	}
+
+	// If no name provided and no remote mode, enter interactive mode
+	if in.Name == "" {
+		return c.initInteractive(ctx, in)
+	}
+
+	// Default: use built-in template
+	return c.initFromBuiltin(ctx, in)
+}
+
+// initFromRemote initializes project from remote repository
+func (c cInit) initFromRemote(ctx context.Context, in cInitInput) (out *cInitOutput, err error) {
+	repo := in.Repo
+	name := in.Name
+
+	// If interactive mode and no repo specified, let user select
+	if in.Interactive && repo == "" {
+		var modPath string
+		var upgradeDeps bool
+		repo, name, modPath, upgradeDeps, err = interactiveSelectTemplate()
+		if err != nil {
+			return nil, err
+		}
+		if modPath != "" {
+			in.Module = modPath
+		}
+		if upgradeDeps {
+			in.Update = true
+		}
+	}
+
+	if repo == "" {
+		return nil, fmt.Errorf("repository URL is required for remote template mode")
+	}
+
+	// Default name to repo basename if empty
+	if name == "" {
+		name = gfile.Basename(repo)
+		mlog.Printf("Using repository basename as project name: %s", name)
+	}
+
+	mlog.Print("initializing from remote template...")
+
+	opts := &geninit.ProcessOptions{
+		SelectVersion: in.SelectVer,
+		ModulePath:    in.Module,
+		UpgradeDeps:   in.Update,
+	}
+
+	if err = geninit.Process(ctx, repo, name, opts); err != nil {
+		return nil, err
+	}
+
+	mlog.Print("initialization done!")
+	if name != "" && name != "." {
+		mlog.Printf(`you can now run "cd %s && gf run main.go" to start your journey, enjoy!`, name)
+	}
+	return
+}
+
+// initFromBuiltin initializes project from built-in template
+func (c cInit) initFromBuiltin(ctx context.Context, in cInitInput) (out *cInitOutput, err error) {
 	var overwrote = false
 	if !gfile.IsEmpty(in.Name) && !allyes.Check() {
 		s := gcmd.Scanf(`the folder "%s" is not empty, files might be overwrote, continue? [y/n]: `, in.Name)
@ -180,3 +268,170 @@ func (c cInit) Index(ctx context.Context, in cInitInput) (out *cInitOutput, err
 	}
 	return
 }
+
+// initInteractive enters interactive mode when no arguments provided
+func (c cInit) initInteractive(ctx context.Context, in cInitInput) (out *cInitOutput, err error) {
+	reader := bufio.NewReader(os.Stdin)
+
+	// Ask user which mode to use
+	fmt.Println("\nPlease select initialization mode:")
+	fmt.Println(strings.Repeat("-", 50))
+	fmt.Println("  [1] Built-in template (default)")
+	fmt.Println("  [2] Remote template")
+	fmt.Println(strings.Repeat("-", 50))
+
+	fmt.Print("Select mode [1-2] (default: 1): ")
+	input, err := reader.ReadString('\n')
+	if err != nil {
+		mlog.Fatalf("failed to read input: %v", err)
+		return
+	}
+	input = strings.TrimSpace(input)
+
+	if input == "2" {
+		in.Interactive = true
+		return c.initFromRemote(ctx, in)
+	}
+
+	// Built-in template mode
+	fmt.Println("\nPlease select project type:")
+	fmt.Println(strings.Repeat("-", 50))
+	fmt.Println("  [1] Single project (default)")
+	fmt.Println("  [2] Mono-repo project")
+	fmt.Println("  [3] Mono-repo app")
+	fmt.Println(strings.Repeat("-", 50))
+
+	fmt.Print("Select type [1-3] (default: 1): ")
+	input, err = reader.ReadString('\n')
+	if err != nil {
+		mlog.Fatalf("failed to read input: %v", err)
+		return
+	}
+	input = strings.TrimSpace(input)
+
+	switch input {
+	case "2":
+		in.Mono = true
+	case "3":
+		in.MonoApp = true
+	}
+
+	// Get project name
+	for {
+		fmt.Print("Enter project name: ")
+		input, err = reader.ReadString('\n')
+		if err != nil {
+			mlog.Fatalf("failed to read input: %v", err)
+			return
+		}
+		in.Name = strings.TrimSpace(input)
+		if in.Name != "" {
+			break
+		}
+		fmt.Println("Project name cannot be empty")
+	}
+
+	// Get module path (optional)
+	fmt.Printf("Enter Go module path (leave empty to use \"%s\"): ", in.Name)
+	input, err = reader.ReadString('\n')
+	if err != nil {
+		mlog.Fatalf("failed to read input: %v", err)
+		return
+	}
+	in.Module = strings.TrimSpace(input)
+
+	// Ask about update
+	fmt.Print("Update to latest GoFrame version? [y/N]: ")
+	input, err = reader.ReadString('\n')
+	if err != nil {
+		mlog.Fatalf("failed to read input: %v", err)
+		return
+	}
+	input = strings.TrimSpace(strings.ToLower(input))
+	in.Update = input == "y" || input == "yes"
+
+	fmt.Println()
+	return c.initFromBuiltin(ctx, in)
+}
+
+// interactiveSelectTemplate prompts user to select a template interactively
+func interactiveSelectTemplate() (repo, name, modPath string, upgradeDeps bool, err error) {
+	reader := bufio.NewReader(os.Stdin)
+
+	// 1. Select template
+	fmt.Println("\nPlease select a project template:")
+	fmt.Println(strings.Repeat("-", 50))
+	for i, t := range defaultTemplates {
+		fmt.Printf("  [%d] %s - %s\n", i+1, t.Name, t.Desc)
+	}
+	fmt.Printf("  [%d] Custom repository URL\n", len(defaultTemplates)+1)
+	fmt.Println(strings.Repeat("-", 50))
+
+	for {
+		fmt.Printf("Select template [1-%d]: ", len(defaultTemplates)+1)
+		input, err := reader.ReadString('\n')
+		if err != nil {
+			return "", "", "", false, fmt.Errorf("failed to read template selection: %w", err)
+		}
+		input = strings.TrimSpace(input)
+
+		idx, e := strconv.Atoi(input)
+		if e != nil || idx < 1 || idx > len(defaultTemplates)+1 {
+			fmt.Printf("Invalid selection, please enter a number between 1-%d\n", len(defaultTemplates)+1)
+			continue
+		}
+
+		if idx <= len(defaultTemplates) {
+			repo = defaultTemplates[idx-1].Repo
+			fmt.Printf("Selected: %s\n\n", repo)
+		} else {
+			// Custom URL
+			fmt.Print("Enter repository URL: ")
+			input, err = reader.ReadString('\n')
+			if err != nil {
+				return "", "", "", false, fmt.Errorf("failed to read repository URL: %w", err)
+			}
+			repo = strings.TrimSpace(input)
+			if repo == "" {
+				fmt.Println("Repository URL cannot be empty")
+				continue
+			}
+		}
+		break
+	}
+
+	// 2. Enter project name
+	for {
+		fmt.Print("Enter project name: ")
+		input, err := reader.ReadString('\n')
+		if err != nil {
+			return "", "", "", false, fmt.Errorf("failed to read project name: %w", err)
+		}
+		name = strings.TrimSpace(input)
+		if name == "" {
+			fmt.Println("Project name cannot be empty")
+			continue
+		}
+		break
+	}
+
+	// 3. Enter module path (optional)
+	fmt.Printf("Enter Go module path (leave empty to use \"%s\"): ", name)
+	input, err := reader.ReadString('\n')
+	if err != nil {
+		return "", "", "", false, fmt.Errorf("failed to read module path: %w", err)
+	}
+	modPath = strings.TrimSpace(input)
+
+	// 4. Ask about upgrade
+	fmt.Print("Upgrade dependencies to latest (go get -u)? [y/N]: ")
+	input, err = reader.ReadString('\n')
+	if err != nil {
+		return "", "", "", false, fmt.Errorf("failed to read upgrade confirmation: %w", err)
+	}
+	input = strings.TrimSpace(strings.ToLower(input))
+	upgradeDeps = input == "y" || input == "yes"
+
+	fmt.Println()
+	return repo, name, modPath, upgradeDeps, nil
+}
--- a/cmd/gf/internal/cmd/cmd_run.go
+++ b/cmd/gf/internal/cmd/cmd_run.go
@ -33,12 +33,18 @@ type cRun struct {
 	g.Meta `name:"run" usage:"{cRunUsage}" brief:"{cRunBrief}" eg:"{cRunEg}" dc:"{cRunDc}"`
 }

+type watchPath struct {
+	Path      string
+	Recursive bool
+}
+
 type cRunApp struct {
-	File       string   // Go run file name.
-	Path       string   // Directory storing built binary.
-	Options    string   // Extra "go run" options.
-	Args       string   // Custom arguments.
-	WatchPaths []string // Watch paths for live reload.
+	File           string   // Go run file name.
+	Path           string   // Directory storing built binary.
+	Options        string   // Extra "go run" options.
+	Args           string   // Custom arguments.
+	WatchPaths     []string // Watch paths for live reload.
+	IgnorePatterns []string // Custom ignore patterns.
 }

 const (
@ -48,43 +54,47 @@ const (
 gf run main.go
 gf run main.go --args "server -p 8080"
 gf run main.go -mod=vendor
-gf run main.go -w "manifest/config/*.yaml"
+gf run main.go -w internal,api
+gf run main.go -i ".git,node_modules"
 `
 	cRunDc = `
 The "run" command is used for running go codes with hot-compiled-like feature,
 which compiles and runs the go codes asynchronously when codes change.
 `
-	cRunFileBrief       = `building file path.`
-	cRunPathBrief       = `output directory path for built binary file. it's "./" in default`
-	cRunExtraBrief      = `the same options as "go run"/"go build" except some options as follows defined`
-	cRunArgsBrief       = `custom arguments for your process`
-	cRunWatchPathsBrief = `watch additional paths for live reload, separated by ",". i.e. "manifest/config/*.yaml"`
+	cRunFileBrief          = `building file path.`
+	cRunPathBrief          = `output directory path for built binary file. it's "./" in default`
+	cRunExtraBrief         = `the same options as "go run"/"go build" except some options as follows defined`
+	cRunArgsBrief          = `custom arguments for your process`
+	cRunWatchPathsBrief    = `watch additional paths for live reload, separated by ",". i.e. "internal,api"`
+	cRunIgnorePatternBrief = `custom ignore patterns for watch, separated by ",". i.e. ".git,node_modules". default patterns: node_modules, vendor, .*, _*. Glob syntax: "*" matches any chars, "?" matches single char, "[abc]" matches char class. Note: patterns match directory names only, not paths`
 )

 var process *gproc.Process

 func init() {
 	gtag.Sets(g.MapStrStr{
-		`cRunUsage`:           cRunUsage,
-		`cRunBrief`:           cRunBrief,
-		`cRunEg`:              cRunEg,
-		`cRunDc`:              cRunDc,
-		`cRunFileBrief`:       cRunFileBrief,
-		`cRunPathBrief`:       cRunPathBrief,
-		`cRunExtraBrief`:      cRunExtraBrief,
-		`cRunArgsBrief`:       cRunArgsBrief,
-		`cRunWatchPathsBrief`: cRunWatchPathsBrief,
+		`cRunUsage`:              cRunUsage,
+		`cRunBrief`:              cRunBrief,
+		`cRunEg`:                 cRunEg,
+		`cRunDc`:                 cRunDc,
+		`cRunFileBrief`:          cRunFileBrief,
+		`cRunPathBrief`:          cRunPathBrief,
+		`cRunExtraBrief`:         cRunExtraBrief,
+		`cRunArgsBrief`:          cRunArgsBrief,
+		`cRunWatchPathsBrief`:    cRunWatchPathsBrief,
+		`cRunIgnorePatternBrief`: cRunIgnorePatternBrief,
 	})
 }

 type (
 	cRunInput struct {
-		g.Meta     `name:"run" config:"gfcli.run"`
-		File       string   `name:"FILE"       arg:"true" brief:"{cRunFileBrief}" v:"required"`
-		Path       string   `name:"path"       short:"p"  brief:"{cRunPathBrief}" d:"./"`
-		Extra      string   `name:"extra"      short:"e"  brief:"{cRunExtraBrief}"`
-		Args       string   `name:"args"       short:"a"  brief:"{cRunArgsBrief}"`
-		WatchPaths []string `name:"watchPaths" short:"w"  brief:"{cRunWatchPathsBrief}"`
+		g.Meta         `name:"run" config:"gfcli.run"`
+		File           string   `name:"FILE"           arg:"true" brief:"{cRunFileBrief}" v:"required"`
+		Path           string   `name:"path"           short:"p"  brief:"{cRunPathBrief}" d:"./"`
+		Extra          string   `name:"extra"          short:"e"  brief:"{cRunExtraBrief}"`
+		Args           string   `name:"args"           short:"a"  brief:"{cRunArgsBrief}"`
+		WatchPaths     []string `name:"watchPaths"     short:"w"  brief:"{cRunWatchPathsBrief}"`
+		IgnorePatterns []string `name:"ignorePatterns" short:"i"  brief:"{cRunIgnorePatternBrief}"`
 	}
 	cRunOutput struct{}
 )
@ -101,17 +111,25 @@ func (c cRun) Index(ctx context.Context, in cRunInput) (out *cRunOutput, err err
 		mlog.Fatalf(`command "go" not found in your environment, please install golang first to proceed this command`)
 	}

-	if len(in.WatchPaths) == 1 {
-		in.WatchPaths = strings.Split(in.WatchPaths[0], ",")
+	// Parse comma-separated values in WatchPaths
+	if len(in.WatchPaths) > 0 {
+		in.WatchPaths = parseCommaSeparatedArgs(in.WatchPaths)
 		mlog.Printf("watchPaths: %v", in.WatchPaths)
 	}

+	// Parse comma-separated values in IgnorePatterns
+	if len(in.IgnorePatterns) > 0 {
+		in.IgnorePatterns = parseCommaSeparatedArgs(in.IgnorePatterns)
+		mlog.Printf("ignorePatterns: %v", in.IgnorePatterns)
+	}
+
 	app := &cRunApp{
-		File:       in.File,
-		Path:       filepath.FromSlash(in.Path),
-		Options:    in.Extra,
-		Args:       in.Args,
-		WatchPaths: in.WatchPaths,
+		File:           in.File,
+		Path:           filepath.FromSlash(in.Path),
+		Options:        in.Extra,
+		Args:           in.Args,
+		WatchPaths:     in.WatchPaths,
+		IgnorePatterns: in.IgnorePatterns,
 	}
 	dirty := gtype.NewBool()

@ -121,6 +139,7 @@ func (c cRun) Index(ctx context.Context, in cRunInput) (out *cRunOutput, err err
 			return
 		}

+		// Check if the file extension is 'go'.
 		if gfile.ExtName(event.Path) != "go" {
 			return
 		}
@ -138,15 +157,11 @@ func (c cRun) Index(ctx context.Context, in cRunInput) (out *cRunOutput, err err
 		})
 	}

-	if len(app.WatchPaths) > 0 {
-		for _, path := range app.WatchPaths {
-			_, err = gfsnotify.Add(gfile.RealPath(path), callbackFunc)
-			if err != nil {
-				mlog.Fatal(err)
-			}
-		}
-	} else {
-		_, err = gfsnotify.Add(gfile.RealPath("."), callbackFunc)
+	// Get directories to watch (recursive or non-recursive monitoring).
+	watchPaths := app.getWatchPaths()
+	for _, wp := range watchPaths {
+		option := gfsnotify.WatchOption{NoRecursive: !wp.Recursive}
+		_, err = gfsnotify.Add(wp.Path, callbackFunc, option)
 		if err != nil {
 			mlog.Fatal(err)
 		}
@ -249,35 +264,181 @@ func (app *cRunApp) End(ctx context.Context, sig os.Signal, outputPath string) {
 }

 func (app *cRunApp) genOutputPath() (outputPath string) {
-	var renamePath string
 	outputPath = gfile.Join(app.Path, gfile.Name(app.File))
 	if runtime.GOOS == "windows" {
 		outputPath += ".exe"
 		if gfile.Exists(outputPath) {
-			renamePath = outputPath + "~"
+			renamePath := outputPath + "~"
 			if err := gfile.Rename(outputPath, renamePath); err != nil {
 				mlog.Print(err)
 			}
+			// Clean up the renamed old binary file
+			defer func() {
+				if gfile.Exists(renamePath) {
+					_ = gfile.Remove(renamePath)
+				}
+			}()
 		}
 	}
 	return filepath.FromSlash(outputPath)
 }

-func matchWatchPaths(watchPaths []string, eventPath string) bool {
-	for _, path := range watchPaths {
-		absPath, err := filepath.Abs(path)
-		if err != nil {
-			mlog.Printf("match watchPath '%s' error: %s", path, err.Error())
+// getWatchPaths uses DFS to find the minimal set of directories to watch.
+// Rule: if a directory and all its descendants have no ignored subdirectories, watch it;
+// otherwise, recurse into valid children and watch the current directory non-recursively.
+func (app *cRunApp) getWatchPaths() []watchPath {
+	roots := []string{"."}
+	if len(app.WatchPaths) > 0 {
+		roots = app.WatchPaths
+	}
+
+	// Use custom ignore patterns if provided, otherwise use default.
+	ignorePatterns := defaultIgnorePatterns
+	if len(app.IgnorePatterns) > 0 {
+		ignorePatterns = app.IgnorePatterns
+	}
+
+	var watchPaths []watchPath
+
+	for _, root := range roots {
+		absRoot := gfile.RealPath(root)
+		if absRoot == "" {
+			mlog.Printf("watch path '%s' not found, skipping", root)
 			continue
 		}
-		matched, err := filepath.Match(absPath, eventPath)
-		if err != nil {
-			mlog.Printf("match watchPath '%s' error: %s", path, err.Error())
+		if isIgnoredDirName(absRoot, ignorePatterns) {
 			continue
 		}
-		if matched {
+		app.collectWatchPaths(absRoot, ignorePatterns, &watchPaths)
+	}
+
+	if len(watchPaths) == 0 {
+		mlog.Printf("no directories to watch, using current directory")
+		if absCur := gfile.RealPath("."); absCur != "" {
+			return []watchPath{{Path: absCur, Recursive: true}}
+		}
+		return []watchPath{{Path: ".", Recursive: true}}
+	}
+
+	mlog.Printf("watching %d paths", len(watchPaths))
+	for _, wp := range watchPaths {
+		recursiveStr := "recursive"
+		if !wp.Recursive {
+			recursiveStr = "non-recursive"
+		}
+		mlog.Debugf("  - %s (%s)", wp.Path, recursiveStr)
+	}
+	return watchPaths
+}
+
+// collectWatchPaths performs a DFS traversal to collect the minimal set of directories to watch.
+// Returns true if the directory or any of its descendants contains ignored directories.
+// Rule: if a directory has no ignored descendants at any depth, watch it recursively;
+// otherwise, watch it non-recursively and recurse into valid children.
+func (app *cRunApp) collectWatchPaths(dir string, ignorePatterns []string, watchPaths *[]watchPath) bool {
+	entries, err := gfile.ScanDir(dir, "*", false)
+	if err != nil {
+		mlog.Printf("scan directory '%s' error: %s", dir, err.Error())
+		// If we can't scan the directory, add it to watch list as fallback
+		*watchPaths = append(*watchPaths, watchPath{Path: dir, Recursive: true})
+		return false
+	}
+
+	// First pass: identify valid subdirectories and check for directly ignored children
+	var validSubDirs []string
+	hasIgnoredChild := false
+	for _, entry := range entries {
+		if !gfile.IsDir(entry) {
+			continue
+		}
+		if isIgnoredDirName(entry, ignorePatterns) {
+			hasIgnoredChild = true
+		} else {
+			validSubDirs = append(validSubDirs, entry)
+		}
+	}
+
+	// If already has ignored child, we know this dir needs non-recursive watch
+	if hasIgnoredChild {
+		*watchPaths = append(*watchPaths, watchPath{Path: dir, Recursive: false})
+		for _, subDir := range validSubDirs {
+			app.collectWatchPaths(subDir, ignorePatterns, watchPaths)
+		}
+		return true
+	}
+
+	// No ignored children, but need to check descendants recursively
+	// Collect results from all subdirectories first
+	subResults := make([]bool, len(validSubDirs))
+	subWatchPaths := make([][]watchPath, len(validSubDirs))
+	hasIgnoredDescendant := false
+
+	for i, subDir := range validSubDirs {
+		var subPaths []watchPath
+		subResults[i] = app.collectWatchPaths(subDir, ignorePatterns, &subPaths)
+		subWatchPaths[i] = subPaths
+		if subResults[i] {
+			hasIgnoredDescendant = true
+		}
+	}
+
+	if !hasIgnoredDescendant {
+		// No ignored descendants at any depth, watch this directory recursively
+		*watchPaths = append(*watchPaths, watchPath{Path: dir, Recursive: true})
+		return false
+	}
+
+	// Has ignored descendants, watch current directory non-recursively
+	// and add all collected subdirectory watch paths
+	*watchPaths = append(*watchPaths, watchPath{Path: dir, Recursive: false})
+	for _, subPaths := range subWatchPaths {
+		*watchPaths = append(*watchPaths, subPaths...)
+	}
+	return true
+}
+
+// defaultIgnorePatterns contains glob patterns for directory names that should be ignored when watching.
+// These directories typically contain third-party code or non-source files.
+// Supported glob syntax (filepath.Match):
+//   - "*" matches any sequence of non-separator characters
+//   - "?" matches any single non-separator character
+//   - "[abc]" matches any character in the bracket
+//   - "[a-z]" matches any character in the range
+//   - "[^abc]" or "[!abc]" matches any character not in the bracket
+//
+// Note: patterns match directory base names only, not full paths (no "/" or path separators allowed).
+var defaultIgnorePatterns = []string{
+	"node_modules",
+	"vendor",
+	".*", // All hidden directories (covers .git, .svn, .hg, .idea, .vscode, etc.)
+	"_*", // Directories starting with underscore
+}
+
+// isIgnoredDirName checks if a directory name matches any ignored pattern.
+// It accepts either a full path or just the directory name, but only matches against the base name.
+// Note: patterns should not contain "/" as they only match directory names, not paths.
+func isIgnoredDirName(name string, ignorePatterns []string) bool {
+	baseName := gfile.Basename(name)
+	for _, pattern := range ignorePatterns {
+		if matched, _ := filepath.Match(pattern, baseName); matched {
 			return true
 		}
 	}
 	return false
 }
+
+// parseCommaSeparatedArgs parses command line arguments that may contain comma-separated values.
+// It handles both single argument with commas (e.g., "a,b,c") and multiple arguments.
+func parseCommaSeparatedArgs(args []string) []string {
+	var result []string
+	for _, arg := range args {
+		parts := strings.Split(arg, ",")
+		for _, part := range parts {
+			trimmed := strings.TrimSpace(part)
+			if trimmed != "" {
+				result = append(result, trimmed)
+			}
+		}
+	}
+	return result
+}
--- a/cmd/gf/internal/cmd/cmd_z_unit_run_test.go
+++ b/cmd/gf/internal/cmd/cmd_z_unit_run_test.go
@ -0,0 +1,336 @@
+// Copyright GoFrame gf Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package cmd
+
+import (
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/gogf/gf/v2/os/gfile"
+	"github.com/gogf/gf/v2/test/gtest"
+)
+
+func Test_cRunApp_getWatchPaths_Basic(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		app := &cRunApp{
+			WatchPaths: []string{"."},
+		}
+		watchPaths := app.getWatchPaths()
+
+		t.AssertGT(len(watchPaths), 0)
+		for _, v := range watchPaths {
+			t.Log(v)
+		}
+	})
+}
+
+func Test_cRunApp_getWatchPaths_EmptyWatchPaths(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		app := &cRunApp{
+			WatchPaths: []string{},
+		}
+		watchPaths := app.getWatchPaths()
+
+		// Should default to current directory "."
+		t.AssertGT(len(watchPaths), 0)
+	})
+}
+
+func Test_cRunApp_getWatchPaths_CustomIgnorePattern(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		app := &cRunApp{
+			WatchPaths:     []string{"testdata"},
+			IgnorePatterns: []string{"2572"},
+		}
+		watchPaths := app.getWatchPaths()
+
+		// Ensure the "2572" directory is not watched directly.
+		for _, wp := range watchPaths {
+			t.Log("watch path:", wp)
+			t.Assert(strings.HasSuffix(wp.Path, "2572"), false)
+		}
+		t.AssertGT(len(watchPaths), 0)
+	})
+}
+
+func Test_cRunApp_getWatchPaths_WithIgnoredDirectories(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		// Create a temporary directory structure for testing
+		tempDir := gfile.Temp("gf_run_test")
+		defer gfile.Remove(tempDir)
+
+		// Create directory structure:
+		// tempDir/
+		//   ├── src/
+		//   │   ├── api/
+		//   │   └── internal/
+		//   ├── vendor/  <-- ignored
+		//   └── node_modules/  <-- ignored
+		gfile.Mkdir(filepath.Join(tempDir, "src", "api"))
+		gfile.Mkdir(filepath.Join(tempDir, "src", "internal"))
+		gfile.Mkdir(filepath.Join(tempDir, "vendor"))
+		gfile.Mkdir(filepath.Join(tempDir, "node_modules"))
+
+		app := &cRunApp{
+			WatchPaths: []string{tempDir},
+		}
+		watchPaths := app.getWatchPaths()
+
+		// Should watch tempDir non-recursively (to catch top-level files) and src recursively
+		t.Assert(len(watchPaths), 2)
+		// First path is tempDir (non-recursive)
+		t.Assert(watchPaths[0].Path, tempDir)
+		t.Assert(watchPaths[0].Recursive, false)
+		// Second path is src (recursive, since it has no ignored descendants)
+		t.Assert(watchPaths[1].Path, filepath.Join(tempDir, "src"))
+		t.Assert(watchPaths[1].Recursive, true)
+	})
+}
+
+func Test_cRunApp_getWatchPaths_NoIgnoredDirectories(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		// Create a temporary directory structure without ignored directories
+		tempDir := gfile.Temp("gf_run_test_no_ignore")
+		defer gfile.Remove(tempDir)
+
+		// Create directory structure without ignored patterns:
+		// tempDir/
+		//   ├── src/
+		//   │   ├── api/
+		//   │   └── internal/
+		gfile.Mkdir(filepath.Join(tempDir, "src", "api"))
+		gfile.Mkdir(filepath.Join(tempDir, "src", "internal"))
+
+		app := &cRunApp{
+			WatchPaths: []string{tempDir},
+		}
+		watchPaths := app.getWatchPaths()
+
+		// Should watch the root directory recursively since no ignored directories exist
+		t.Assert(len(watchPaths), 1)
+		t.Assert(watchPaths[0].Path, tempDir)
+		t.Assert(watchPaths[0].Recursive, true)
+	})
+}
+
+func Test_cRunApp_getWatchPaths_CustomIgnorePatterns(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		// Create a temporary directory structure
+		tempDir := gfile.Temp("gf_run_test_custom_ignore")
+		defer gfile.Remove(tempDir)
+
+		// Create directory structure:
+		// tempDir/
+		//   ├── src/
+		//   │   ├── api/
+		//   │   └── internal/
+		//   ├── build/  <-- ignored
+		//   └── dist/  <-- ignored
+		gfile.Mkdir(filepath.Join(tempDir, "src", "api"))
+		gfile.Mkdir(filepath.Join(tempDir, "src", "internal"))
+		gfile.Mkdir(filepath.Join(tempDir, "build"))
+		gfile.Mkdir(filepath.Join(tempDir, "dist"))
+
+		app := &cRunApp{
+			WatchPaths:     []string{tempDir},
+			IgnorePatterns: []string{"build", "dist"},
+		}
+		watchPaths := app.getWatchPaths()
+
+		// Should watch tempDir non-recursively and src recursively
+		t.Assert(len(watchPaths), 2)
+		t.Assert(watchPaths[0].Path, tempDir)
+		t.Assert(watchPaths[0].Recursive, false)
+		t.Assert(watchPaths[1].Path, filepath.Join(tempDir, "src"))
+		t.Assert(watchPaths[1].Recursive, true)
+	})
+}
+
+func Test_cRunApp_getWatchPaths_DeepNestedStructure(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		// Create a deep nested directory structure
+		tempDir := gfile.Temp("gf_run_test_deep")
+		defer gfile.Remove(tempDir)
+
+		// Create deep directory structure:
+		// tempDir/
+		//   ├── a/
+		//   │   ├── b/
+		//   │   │   └── c/
+		//   │   └── vendor/  <-- ignored
+		//   └── d/
+		gfile.Mkdir(filepath.Join(tempDir, "a", "b", "c"))
+		gfile.Mkdir(filepath.Join(tempDir, "a", "vendor"))
+		gfile.Mkdir(filepath.Join(tempDir, "d"))
+
+		app := &cRunApp{
+			WatchPaths: []string{tempDir},
+		}
+		watchPaths := app.getWatchPaths()
+
+		// Should watch individual valid directories due to ignored vendor directory
+		t.AssertGT(len(watchPaths), 0)
+
+		// Verify that vendor directory is not in watch list
+		for _, wp := range watchPaths {
+			t.Assert(strings.Contains(wp.Path, "vendor"), false)
+		}
+	})
+}
+
+func Test_cRunApp_getWatchPaths_MultipleRoots(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		// Create multiple temporary directories
+		tempDir1 := gfile.Temp("gf_run_test_multi1")
+		tempDir2 := gfile.Temp("gf_run_test_multi2")
+		defer gfile.Remove(tempDir1)
+		defer gfile.Remove(tempDir2)
+
+		gfile.Mkdir(filepath.Join(tempDir1, "src"))
+		gfile.Mkdir(filepath.Join(tempDir2, "api"))
+
+		app := &cRunApp{
+			WatchPaths: []string{tempDir1, tempDir2},
+		}
+		watchPaths := app.getWatchPaths()
+
+		// Should watch both root directories recursively
+		t.Assert(len(watchPaths), 2)
+
+		// Both directories should be in the watch list
+		foundDir1, foundDir2 := false, false
+		for _, wp := range watchPaths {
+			if wp.Path == tempDir1 {
+				foundDir1 = true
+				t.Assert(wp.Recursive, true)
+			}
+			if wp.Path == tempDir2 {
+				foundDir2 = true
+				t.Assert(wp.Recursive, true)
+			}
+		}
+		t.Assert(foundDir1, true)
+		t.Assert(foundDir2, true)
+	})
+}
+
+func Test_cRunApp_getWatchPaths_NonExistentDirectory(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		app := &cRunApp{
+			WatchPaths: []string{"/non/existent/path"},
+		}
+		watchPaths := app.getWatchPaths()
+
+		// Should fall back to current directory when no valid paths found
+		t.AssertGT(len(watchPaths), 0)
+
+		// Should contain current directory
+		currentDir, _ := os.Getwd()
+		foundCurrentDir := false
+		for _, wp := range watchPaths {
+			if wp.Path == currentDir {
+				foundCurrentDir = true
+				break
+			}
+		}
+		t.Assert(foundCurrentDir, true)
+	})
+}
+
+func Test_isIgnoredDirName(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		// Test default ignore patterns
+		t.Assert(isIgnoredDirName("node_modules", defaultIgnorePatterns), true)
+		t.Assert(isIgnoredDirName("vendor", defaultIgnorePatterns), true)
+		t.Assert(isIgnoredDirName(".git", defaultIgnorePatterns), true)
+		t.Assert(isIgnoredDirName("_private", defaultIgnorePatterns), true)
+		t.Assert(isIgnoredDirName("src", defaultIgnorePatterns), false)
+		t.Assert(isIgnoredDirName("api", defaultIgnorePatterns), false)
+
+		// Test custom ignore patterns
+		customPatterns := []string{"build", "dist", "*.tmp"}
+		t.Assert(isIgnoredDirName("build", customPatterns), true)
+		t.Assert(isIgnoredDirName("dist", customPatterns), true)
+		t.Assert(isIgnoredDirName("test.tmp", customPatterns), true)
+		t.Assert(isIgnoredDirName("src", customPatterns), false)
+	})
+}
+
+func Test_cRunApp_getWatchPaths_DeeplyNestedIgnore(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		// Create a temporary directory structure with deeply nested ignored directory
+		tempDir := gfile.Temp("gf_run_test_deeply_nested")
+		defer gfile.Remove(tempDir)
+
+		// Create directory structure:
+		// tempDir/
+		//   ├── a/
+		//   │   ├── b/
+		//   │   │   ├── c/
+		//   │   │   │   └── vendor/  <-- deeply nested ignored (4 levels)
+		//   │   │   └── d/
+		//   │   └── e/
+		//   └── f/
+		gfile.Mkdir(filepath.Join(tempDir, "a", "b", "c", "vendor"))
+		gfile.Mkdir(filepath.Join(tempDir, "a", "b", "d"))
+		gfile.Mkdir(filepath.Join(tempDir, "a", "e"))
+		gfile.Mkdir(filepath.Join(tempDir, "f"))
+
+		app := &cRunApp{
+			WatchPaths: []string{tempDir},
+		}
+		watchPaths := app.getWatchPaths()
+
+		// Expected watch paths:
+		// 1. tempDir (non-recursive) - has ignored descendant
+		// 2. a (non-recursive) - has ignored descendant in b/c/vendor
+		// 3. b (non-recursive) - has ignored descendant in c/vendor
+		// 4. c (non-recursive) - has ignored child vendor
+		// 5. d (recursive) - no ignored descendants
+		// 6. e (recursive) - no ignored descendants
+		// 7. f (recursive) - no ignored descendants
+
+		t.AssertGT(len(watchPaths), 0)
+
+		// Verify vendor is not in watch paths
+		for _, wp := range watchPaths {
+			t.Assert(strings.Contains(wp.Path, "vendor"), false)
+		}
+
+		// Find specific paths and verify their recursive flags
+		foundF := false
+		for _, wp := range watchPaths {
+			if wp.Path == filepath.Join(tempDir, "f") {
+				foundF = true
+				t.Assert(wp.Recursive, true) // f should be recursive (no ignored descendants)
+			}
+		}
+		t.Assert(foundF, true)
+	})
+}
+
+func Test_cRunApp_getWatchPaths_EmptyDirectory(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		// Create an empty temporary directory
+		tempDir := gfile.Temp("gf_run_test_empty")
+		defer gfile.Remove(tempDir)
+
+		gfile.Mkdir(tempDir)
+
+		app := &cRunApp{
+			WatchPaths: []string{tempDir},
+		}
+		watchPaths := app.getWatchPaths()
+
+		// Empty directory should be watched recursively (no ignored descendants)
+		t.Assert(len(watchPaths), 1)
+		t.Assert(watchPaths[0].Path, tempDir)
+		t.Assert(watchPaths[0].Recursive, true)
+	})
+}
--- a/cmd/gf/internal/cmd/gendao/gendao.go
+++ b/cmd/gf/internal/cmd/gendao/gendao.go
@ -104,6 +104,10 @@ var (
 		"smallmoney": {
 			Type: "float64",
 		},
+		"uuid": {
+			Type:   "uuid.UUID",
+			Import: "github.com/google/uuid",
+		},
 	}

 	// tablewriter Options
--- a/cmd/gf/internal/cmd/gendao/gendao_structure.go
+++ b/cmd/gf/internal/cmd/gendao/gendao_structure.go
@ -98,7 +98,6 @@ func generateStructFieldDefinition(
 		err              error
 		localTypeName    gdb.LocalType
 		localTypeNameStr string
-		jsonTag          = gstr.CaseConvert(field.Name, gstr.CaseTypeMatch(in.JsonCase))
 	)

 	if in.TypeMapping != nil && len(in.TypeMapping) > 0 {
@ -156,6 +155,8 @@ func generateStructFieldDefinition(
 		"    #" + formatFieldName(newFiledName, FieldNameCaseCamel),
 		" #" + localTypeNameStr,
 	}
+
+	jsonTag := gstr.CaseConvert(newFiledName, gstr.CaseTypeMatch(in.JsonCase))
 	attrLines = append(attrLines, fmt.Sprintf(` #%sjson:"%s"`, tagKey, jsonTag))
 	// orm tag
 	if !in.IsDo {
--- a/cmd/gf/internal/cmd/geninit/geninit.go
+++ b/cmd/gf/internal/cmd/geninit/geninit.go
@ -0,0 +1,236 @@
+// Copyright GoFrame gf Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package geninit
+
+import (
+	"context"
+	"path/filepath"
+
+	"github.com/gogf/gf/v2/os/gfile"
+	"github.com/gogf/gf/v2/text/gstr"
+
+	"github.com/gogf/gf/cmd/gf/v2/internal/utility/mlog"
+)
+
+// ProcessOptions contains options for the Process function
+type ProcessOptions struct {
+	SelectVersion bool   // Enable interactive version selection
+	ModulePath    string // Custom go.mod module path (e.g., github.com/xxx/xxx)
+	UpgradeDeps   bool   // Upgrade dependencies to latest (go get -u ./...)
+}
+
+// Process handles the template generation flow from remote repository
+func Process(ctx context.Context, repo, name string, opts *ProcessOptions) error {
+	if opts == nil {
+		opts = &ProcessOptions{}
+	}
+
+	// 0. Check Go environment first
+	mlog.Print("Checking Go environment...")
+	goEnv, err := CheckGoEnv(ctx)
+	if err != nil {
+		mlog.Printf("Go environment check failed: %v", err)
+		return err
+	}
+	mlog.Printf("Go environment OK (version: %s)", goEnv.GOVERSION)
+
+	// Check if this is a git subdirectory URL
+	if IsSubdirRepo(repo) {
+		return processGitSubdir(ctx, repo, name, opts)
+	}
+
+	// Try Go module download first, fallback to git subdirectory if it fails
+	// This handles edge cases where the heuristic may be incorrect
+	err = processGoModule(ctx, repo, name, opts)
+	if err != nil {
+		mlog.Printf("Go module download failed, trying git subdirectory mode: %v", err)
+		mlog.Print("Note: If this is a git subdirectory, you can force git mode by using a full git URL")
+
+		// If Go module download fails, try git subdirectory as fallback
+		// This handles cases where the heuristic incorrectly classified a git subdir as Go module
+		if IsSubdirRepo(repo) {
+			mlog.Print("Falling back to git subdirectory download...")
+			return processGitSubdir(ctx, repo, name, opts)
+		}
+	}
+
+	return err
+}
+
+// processGoModule handles standard Go module download via go get
+func processGoModule(ctx context.Context, repo, name string, opts *ProcessOptions) error {
+	// Extract module path (without version)
+	modulePath := repo
+	specifiedVersion := ""
+	if gstr.Contains(repo, "@") {
+		parts := gstr.Split(repo, "@")
+		modulePath = parts[0]
+		specifiedVersion = parts[1]
+	}
+
+	// Default name to repo basename if empty
+	if name == "" {
+		name = filepath.Base(modulePath)
+	}
+
+	// Determine the target module path for go.mod
+	targetModulePath := name
+	if opts.ModulePath != "" {
+		targetModulePath = opts.ModulePath
+	}
+
+	// 1. Determine version to use
+	var targetVersion string
+	if specifiedVersion != "" {
+		// User specified version
+		targetVersion = specifiedVersion
+		mlog.Printf("Using specified version: %s", targetVersion)
+	} else if opts.SelectVersion {
+		// Interactive version selection
+		mlog.Print("Fetching available versions...")
+		versionInfo, err := GetModuleVersions(ctx, modulePath)
+		if err != nil {
+			mlog.Printf("Failed to get versions: %v", err)
+			return err
+		}
+
+		targetVersion, err = SelectVersion(ctx, versionInfo.Versions, modulePath)
+		if err != nil {
+			mlog.Printf("Version selection failed: %v", err)
+			return err
+		}
+	} else {
+		// Default: use latest version
+		mlog.Print("Fetching latest version...")
+		latest, err := GetLatestVersion(ctx, modulePath)
+		if err != nil {
+			mlog.Printf("Failed to get latest version, will try @latest tag: %v", err)
+			targetVersion = "latest"
+		} else {
+			targetVersion = latest
+			mlog.Printf("Latest version: %s", targetVersion)
+		}
+	}
+
+	// 2. Download Template with determined version
+	repoWithVersion := modulePath + "@" + targetVersion
+	srcDir, err := downloadTemplate(ctx, repoWithVersion)
+	if err != nil {
+		mlog.Printf("Download failed: %v", err)
+		return err
+	}
+
+	mlog.Debugf("Template located at: %s", srcDir)
+
+	// 3. Generate Project
+	if err := generateProject(ctx, srcDir, name, modulePath, targetModulePath); err != nil {
+		mlog.Printf("Generation failed: %v", err)
+		return err
+	}
+
+	// 4. Handle dependencies
+	var projectDir string
+	if name == "." {
+		projectDir = gfile.Pwd()
+	} else {
+		projectDir = filepath.Join(gfile.Pwd(), name)
+	}
+	if opts.UpgradeDeps {
+		// Upgrade all dependencies to latest
+		if err := upgradeDependencies(ctx, projectDir); err != nil {
+			mlog.Printf("Failed to upgrade dependencies: %v", err)
+		}
+	} else {
+		// Default: just tidy dependencies
+		if err := tidyDependencies(ctx, projectDir); err != nil {
+			mlog.Printf("Failed to tidy dependencies: %v", err)
+		}
+	}
+
+	return nil
+}
+
+// processGitSubdir handles git subdirectory download via sparse checkout
+func processGitSubdir(ctx context.Context, repo, name string, opts *ProcessOptions) error {
+	mlog.Print("Detected subdirectory URL, using git sparse checkout...")
+
+	// Check if git is available
+	gitVersion, err := CheckGitEnv(ctx)
+	if err != nil {
+		mlog.Printf("Git is required for subdirectory templates: %v", err)
+		return err
+	}
+	mlog.Printf("Git available (%s)", gitVersion)
+
+	// Download via git sparse checkout
+	srcDir, gitInfo, err := downloadGitSubdir(ctx, repo)
+	if err != nil {
+		mlog.Printf("Git download failed: %v", err)
+		return err
+	}
+
+	// Clean up temp directory after generation
+	// The temp dir is parent of parent of srcDir (tempDir/repo/subpath)
+	tempDir := filepath.Dir(filepath.Dir(srcDir))
+	if tempDir != "" && gfile.Exists(tempDir) && gstr.Contains(tempDir, "gf-init-git") {
+		defer func() {
+			if err := gfile.Remove(tempDir); err != nil {
+				mlog.Debugf("Failed to remove temp directory %s: %v", tempDir, err)
+			} else {
+				mlog.Debugf("Cleaned up temp directory: %s", tempDir)
+			}
+		}()
+	}
+
+	// Default name to subpath basename if empty
+	if name == "" {
+		name = filepath.Base(gitInfo.SubPath)
+	}
+
+	// Get original module name from go.mod (might be "main" or something else)
+	oldModule := GetModuleNameFromGoMod(srcDir)
+	if oldModule == "" {
+		// Fallback: construct from git info
+		oldModule = gitInfo.Host + "/" + gitInfo.Owner + "/" + gitInfo.Repo + "/" + gitInfo.SubPath
+	}
+
+	// Determine the target module path for go.mod
+	targetModulePath := name
+	if opts.ModulePath != "" {
+		targetModulePath = opts.ModulePath
+	}
+
+	mlog.Debugf("Template located at: %s", srcDir)
+	mlog.Debugf("Original module: %s", oldModule)
+
+	// Generate Project
+	if err := generateProject(ctx, srcDir, name, oldModule, targetModulePath); err != nil {
+		mlog.Printf("Generation failed: %v", err)
+		return err
+	}
+
+	// Handle dependencies
+	var projectDir string
+	if name == "." {
+		projectDir = gfile.Pwd()
+	} else {
+		projectDir = filepath.Join(gfile.Pwd(), name)
+	}
+	if opts.UpgradeDeps {
+		// Upgrade all dependencies to latest
+		if err := upgradeDependencies(ctx, projectDir); err != nil {
+			mlog.Printf("Failed to upgrade dependencies: %v", err)
+		}
+	} else {
+		// Default: just tidy dependencies
+		if err := tidyDependencies(ctx, projectDir); err != nil {
+			mlog.Printf("Failed to tidy dependencies: %v", err)
+		}
+	}
+
+	return nil
+}
--- a/cmd/gf/internal/cmd/geninit/geninit_ast.go
+++ b/cmd/gf/internal/cmd/geninit/geninit_ast.go
@ -0,0 +1,126 @@
+// Copyright GoFrame gf Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package geninit
+
+import (
+	"bytes"
+	"context"
+	"go/ast"
+	"go/parser"
+	"go/printer"
+	"go/token"
+	"os"
+	"path/filepath"
+	"strings"
+
+	"github.com/gogf/gf/v2/os/gfile"
+
+	"github.com/gogf/gf/cmd/gf/v2/internal/utility/mlog"
+)
+
+// ASTReplacer handles import path replacement using Go AST
+type ASTReplacer struct {
+	oldModule string
+	newModule string
+	fset      *token.FileSet
+}
+
+// NewASTReplacer creates a new AST-based import replacer
+func NewASTReplacer(oldModule, newModule string) *ASTReplacer {
+	return &ASTReplacer{
+		oldModule: oldModule,
+		newModule: newModule,
+		fset:      token.NewFileSet(),
+	}
+}
+
+// ReplaceInFile replaces import paths in a single Go file
+func (r *ASTReplacer) ReplaceInFile(ctx context.Context, filePath string) error {
+	// Read file content
+	content := gfile.GetContents(filePath)
+	if content == "" {
+		return nil
+	}
+
+	// Parse the file
+	file, err := parser.ParseFile(r.fset, filePath, content, parser.ParseComments)
+	if err != nil {
+		mlog.Debugf("Failed to parse %s: %v", filePath, err)
+		return nil // Skip files that can't be parsed
+	}
+
+	// Track if any changes were made
+	changed := false
+
+	// Traverse and modify imports
+	ast.Inspect(file, func(n ast.Node) bool {
+		switch x := n.(type) {
+		case *ast.ImportSpec:
+			if x.Path != nil {
+				importPath := strings.Trim(x.Path.Value, `"`)
+				if strings.HasPrefix(importPath, r.oldModule) {
+					// Replace only the leading module prefix for clarity and correctness.
+					newPath := r.newModule + strings.TrimPrefix(importPath, r.oldModule)
+					x.Path.Value = `"` + newPath + `"`
+					changed = true
+					mlog.Debugf("Replaced import: %s -> %s in %s", importPath, newPath, filePath)
+				}
+			}
+		}
+		return true
+	})
+
+	if !changed {
+		return nil
+	}
+
+	// Write back to file
+	var buf bytes.Buffer
+	// Use default printer configuration to match gofmt output
+	cfg := &printer.Config{}
+	if err := cfg.Fprint(&buf, r.fset, file); err != nil {
+		return err
+	}
+
+	return gfile.PutContents(filePath, buf.String())
+}
+
+// ReplaceInDir replaces import paths in all Go files in a directory (recursively)
+func (r *ASTReplacer) ReplaceInDir(ctx context.Context, dir string) error {
+	mlog.Printf("Replacing imports: %s -> %s", r.oldModule, r.newModule)
+
+	// Find all .go files
+	files, err := findGoFiles(dir)
+	if err != nil {
+		return err
+	}
+
+	for _, file := range files {
+		if err := r.ReplaceInFile(ctx, file); err != nil {
+			mlog.Printf("Failed to process %s: %v", file, err)
+		}
+	}
+
+	return nil
+}
+
+// findGoFiles recursively finds all .go files in a directory
+func findGoFiles(dir string) ([]string, error) {
+	var files []string
+
+	err := filepath.Walk(dir, func(path string, info os.FileInfo, err error) error {
+		if err != nil {
+			return err
+		}
+		if !info.IsDir() && strings.HasSuffix(path, ".go") {
+			files = append(files, path)
+		}
+		return nil
+	})
+
+	return files, err
+}
--- a/cmd/gf/internal/cmd/geninit/geninit_downloader.go
+++ b/cmd/gf/internal/cmd/geninit/geninit_downloader.go
@ -0,0 +1,111 @@
+// Copyright GoFrame gf Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package geninit
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"os"
+	"os/exec"
+	"strings"
+
+	"github.com/gogf/gf/v2/os/gfile"
+	"github.com/gogf/gf/v2/text/gstr"
+
+	"github.com/gogf/gf/cmd/gf/v2/internal/utility/mlog"
+)
+
+// downloadTemplate fetches the remote repository using go get
+func downloadTemplate(ctx context.Context, repo string) (string, error) {
+	// 1. Create a temporary directory workspace
+	tempDir := gfile.Temp("gf-init-cli")
+	if tempDir == "" {
+		return "", fmt.Errorf("failed to create temporary directory")
+	}
+	if err := gfile.Mkdir(tempDir); err != nil {
+		return "", err
+	}
+	defer func() {
+		if err := gfile.Remove(tempDir); err != nil {
+			mlog.Debugf("Failed to remove temp directory %s: %v", tempDir, err)
+		}
+	}() // Clean up the temp workspace
+
+	mlog.Debugf("Using temp workspace: %s", tempDir)
+
+	// 2. Initialize a temp go module to perform go get
+	// We run commands inside the temp directory
+	if err := runCmd(ctx, tempDir, "go", "mod", "init", "temp"); err != nil {
+		return "", err
+	}
+
+	// 3. Run go get <repo>
+	// Try different version strategies: original -> @latest -> @master
+	moduleName := repo
+	if gstr.Contains(repo, "@") {
+		moduleName = gstr.Split(repo, "@")[0]
+	}
+
+	var downloadErrs []string
+	versionsToTry := []string{repo}
+	if !gstr.Contains(repo, "@") {
+		versionsToTry = append(versionsToTry, repo+"@latest", repo+"@master")
+	}
+
+	var successRepo string
+	for _, tryRepo := range versionsToTry {
+		mlog.Printf("Downloading template %s...", tryRepo)
+		if err := runCmd(ctx, tempDir, "go", "get", tryRepo); err == nil {
+			successRepo = tryRepo
+			break
+		} else {
+			downloadErrs = append(downloadErrs, fmt.Sprintf("%s: %v", tryRepo, err))
+			mlog.Debugf("Failed to download %s, trying next...", tryRepo)
+		}
+	}
+
+	if successRepo == "" {
+		errMsg := "all download attempts failed"
+		if len(downloadErrs) > 0 {
+			errMsg = strings.Join(downloadErrs, "; ")
+		}
+		return "", fmt.Errorf("failed to download repo %s: %s", repo, errMsg)
+	}
+
+	// 4. Find the local path using go list -m -json <repo>
+	listCmd := exec.CommandContext(ctx, "go", "list", "-m", "-json", moduleName)
+	listCmd.Dir = tempDir
+	output, err := listCmd.Output()
+	if err != nil {
+		if exitErr, ok := err.(*exec.ExitError); ok {
+			return "", fmt.Errorf("go list failed: %s", string(exitErr.Stderr))
+		}
+		return "", fmt.Errorf("failed to locate module path: %w", err)
+	}
+
+	var modInfo struct {
+		Dir string `json:"Dir"`
+	}
+	if err := json.Unmarshal(output, &modInfo); err != nil {
+		return "", fmt.Errorf("failed to parse go list output: %w", err)
+	}
+
+	if modInfo.Dir == "" {
+		return "", fmt.Errorf("module directory not found for %s", repo)
+	}
+
+	return modInfo.Dir, nil
+}
+
+func runCmd(ctx context.Context, dir string, name string, args ...string) error {
+	cmd := exec.CommandContext(ctx, name, args...)
+	cmd.Dir = dir
+	cmd.Stdout = os.Stdout
+	cmd.Stderr = os.Stderr
+	return cmd.Run()
+}
--- a/cmd/gf/internal/cmd/geninit/geninit_env.go
+++ b/cmd/gf/internal/cmd/geninit/geninit_env.go
@ -0,0 +1,90 @@
+// Copyright GoFrame gf Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package geninit
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"os/exec"
+	"strings"
+
+	"github.com/gogf/gf/cmd/gf/v2/internal/utility/mlog"
+)
+
+// GoEnv represents Go environment variables
+type GoEnv struct {
+	GOVERSION   string `json:"GOVERSION"`
+	GOROOT      string `json:"GOROOT"`
+	GOPATH      string `json:"GOPATH"`
+	GOMODCACHE  string `json:"GOMODCACHE"`
+	GOPROXY     string `json:"GOPROXY"`
+	GO111MODULE string `json:"GO111MODULE"`
+}
+
+// CheckGoEnv verifies Go is installed and properly configured
+func CheckGoEnv(ctx context.Context) (*GoEnv, error) {
+	// 1. Check if go binary exists
+	goPath, err := exec.LookPath("go")
+	if err != nil {
+		return nil, fmt.Errorf("go is not installed or not in PATH: %w", err)
+	}
+	mlog.Debugf("Found go binary at: %s", goPath)
+
+	// 2. Get go env as JSON
+	cmd := exec.CommandContext(ctx, "go", "env", "-json")
+	output, err := cmd.Output()
+	if err != nil {
+		if exitErr, ok := err.(*exec.ExitError); ok {
+			return nil, fmt.Errorf("go env failed: %s", string(exitErr.Stderr))
+		}
+		return nil, fmt.Errorf("failed to run go env: %w", err)
+	}
+
+	// 3. Parse JSON output
+	var env GoEnv
+	if err := json.Unmarshal(output, &env); err != nil {
+		return nil, fmt.Errorf("failed to parse go env output: %w", err)
+	}
+
+	// 4. Validate critical environment variables
+	if env.GOROOT == "" {
+		return nil, fmt.Errorf("GOROOT is not set")
+	}
+	if env.GOMODCACHE == "" && env.GOPATH == "" {
+		return nil, fmt.Errorf("neither GOMODCACHE nor GOPATH is set")
+	}
+
+	mlog.Debugf("Go Version: %s", env.GOVERSION)
+	mlog.Debugf("GOROOT: %s", env.GOROOT)
+	mlog.Debugf("GOMODCACHE: %s", env.GOMODCACHE)
+	mlog.Debugf("GOPROXY: %s", env.GOPROXY)
+
+	return &env, nil
+}
+
+// CheckGitEnv verifies Git is installed and returns its version
+func CheckGitEnv(ctx context.Context) (string, error) {
+	// 1. Check if git binary exists
+	gitPath, err := exec.LookPath("git")
+	if err != nil {
+		return "", fmt.Errorf("git is not installed or not in PATH: %w", err)
+	}
+	mlog.Debugf("Found git binary at: %s", gitPath)
+
+	// 2. Get git version
+	cmd := exec.CommandContext(ctx, "git", "--version")
+	output, err := cmd.Output()
+	if err != nil {
+		return "", fmt.Errorf("failed to get git version: %w", err)
+	}
+
+	version := strings.TrimSpace(string(output))
+	mlog.Debugf("Git version: %s", version)
+
+	return version, nil
+}
--- a/cmd/gf/internal/cmd/geninit/geninit_generator.go
+++ b/cmd/gf/internal/cmd/geninit/geninit_generator.go
@ -0,0 +1,110 @@
+// Copyright GoFrame gf Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package geninit
+
+import (
+	"context"
+	"fmt"
+	"path/filepath"
+
+	"github.com/gogf/gf/v2/os/gfile"
+	"github.com/gogf/gf/v2/text/gstr"
+
+	"github.com/gogf/gf/cmd/gf/v2/internal/utility/mlog"
+)
+
+// generateProject copies the template to the destination and performs cleanup
+// oldModule: original module path from template
+// newModule: target module path for go.mod (can be different from project name)
+func generateProject(ctx context.Context, srcPath, name, oldModule, newModule string) error {
+	pwd := gfile.Pwd()
+
+	dstPath := filepath.Join(pwd, name)
+	if name == "." {
+		dstPath = pwd
+	}
+
+	if gfile.Exists(dstPath) && !gfile.IsEmpty(dstPath) {
+		return fmt.Errorf("target directory %s is not empty", dstPath)
+	}
+
+	mlog.Printf("Generating project in %s...", dstPath)
+
+	// 1. Copy files
+	if err := gfile.Copy(srcPath, dstPath); err != nil {
+		return err
+	}
+
+	// 2. Clean up .git directory
+	gitDir := filepath.Join(dstPath, ".git")
+	if gfile.Exists(gitDir) {
+		if err := gfile.Remove(gitDir); err != nil {
+			mlog.Debugf("Failed to remove .git directory: %v", err)
+		}
+	}
+
+	// 3. Clean up go.work and go.work.sum (workspace files should not be in generated project)
+	for _, workFile := range []string{"go.work", "go.work.sum"} {
+		workPath := filepath.Join(dstPath, workFile)
+		if gfile.Exists(workPath) {
+			if err := gfile.Remove(workPath); err != nil {
+				mlog.Printf("Failed to remove %s: %v", workFile, err)
+			} else {
+				mlog.Debugf("Removed %s", workFile)
+			}
+		}
+	}
+
+	// 4. Update go.mod module name
+	goModPath := filepath.Join(dstPath, "go.mod")
+	if gfile.Exists(goModPath) {
+		content := gfile.GetContents(goModPath)
+		lines := gstr.Split(content, "\n")
+		if len(lines) > 0 && gstr.HasPrefix(lines[0], "module ") {
+			lines[0] = "module " + newModule
+			newContent := gstr.Join(lines, "\n")
+			if err := gfile.PutContents(goModPath, newContent); err != nil {
+				mlog.Printf("Failed to update go.mod: %v", err)
+			}
+		}
+	}
+
+	// 5. Use AST to replace import paths in all Go files
+	if oldModule != "" && oldModule != newModule {
+		replacer := NewASTReplacer(oldModule, newModule)
+		if err := replacer.ReplaceInDir(ctx, dstPath); err != nil {
+			return fmt.Errorf("failed to replace imports: %w", err)
+		}
+	}
+
+	mlog.Print("Project generated successfully!")
+	return nil
+}
+
+// tidyDependencies runs go mod tidy in the project directory
+func tidyDependencies(ctx context.Context, projectDir string) error {
+	mlog.Print("Tidying dependencies (go mod tidy)...")
+	if err := runCmd(ctx, projectDir, "go", "mod", "tidy"); err != nil {
+		return fmt.Errorf("go mod tidy failed: %w", err)
+	}
+	mlog.Print("Dependencies tidied successfully!")
+	return nil
+}
+
+// upgradeDependencies runs go get -u ./... to upgrade all dependencies to latest
+func upgradeDependencies(ctx context.Context, projectDir string) error {
+	mlog.Print("Upgrading dependencies to latest (go get -u ./...)...")
+	if err := runCmd(ctx, projectDir, "go", "get", "-u", "./..."); err != nil {
+		return fmt.Errorf("go get -u failed: %w", err)
+	}
+	// Run tidy again after upgrade
+	if err := runCmd(ctx, projectDir, "go", "mod", "tidy"); err != nil {
+		return fmt.Errorf("go mod tidy after upgrade failed: %w", err)
+	}
+	mlog.Print("Dependencies upgraded successfully!")
+	return nil
+}
--- a/cmd/gf/internal/cmd/geninit/geninit_git_downloader.go
+++ b/cmd/gf/internal/cmd/geninit/geninit_git_downloader.go
@ -0,0 +1,241 @@
+// Copyright GoFrame gf Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package geninit
+
+import (
+	"context"
+	"fmt"
+	"path/filepath"
+	"strings"
+
+	"github.com/gogf/gf/v2/os/gfile"
+	"github.com/gogf/gf/v2/text/gstr"
+
+	"github.com/gogf/gf/cmd/gf/v2/internal/utility/mlog"
+)
+
+// GitRepoInfo holds parsed git repository information
+type GitRepoInfo struct {
+	Host     string // e.g., github.com
+	Owner    string // e.g., gogf
+	Repo     string // e.g., examples
+	Branch   string // e.g., main (default: main)
+	SubPath  string // e.g., httpserver/jwt
+	CloneURL string // e.g., https://github.com/gogf/examples.git
+}
+
+// ParseGitURL parses a git URL and extracts repository info
+// Supports formats:
+//   - github.com/owner/repo
+//   - github.com/owner/repo/subdir/path
+//   - github.com/owner/repo/tree/branch/subdir/path (from GitHub web URL)
+func ParseGitURL(url string) (*GitRepoInfo, error) {
+	// Remove protocol prefix if present
+	url = strings.TrimPrefix(url, "https://")
+	url = strings.TrimPrefix(url, "http://")
+	url = strings.TrimSuffix(url, ".git")
+
+	// Remove version suffix like @v1.0.0
+	if idx := strings.Index(url, "@"); idx != -1 {
+		url = url[:idx]
+	}
+
+	parts := strings.Split(url, "/")
+	if len(parts) < 3 {
+		return nil, fmt.Errorf("invalid git URL: %s", url)
+	}
+
+	info := &GitRepoInfo{
+		Host:   parts[0],
+		Owner:  parts[1],
+		Repo:   parts[2],
+		Branch: "main", // default branch
+	}
+
+	// Check for /tree/branch/ pattern (GitHub web URL)
+	if len(parts) > 4 && parts[3] == "tree" {
+		info.Branch = parts[4]
+		if len(parts) > 5 {
+			info.SubPath = strings.Join(parts[5:], "/")
+		}
+	} else if len(parts) > 3 {
+		// Direct subpath: github.com/owner/repo/subdir/path
+		info.SubPath = strings.Join(parts[3:], "/")
+	}
+
+	info.CloneURL = fmt.Sprintf("https://%s/%s/%s.git", info.Host, info.Owner, info.Repo)
+
+	return info, nil
+}
+
+// IsSubdirRepo checks if the URL points to a subdirectory of a repository
+// Returns false for Go module paths (which may have /vN suffix or nested module paths)
+// Note: This uses heuristics that may have false positives/negatives in edge cases
+func IsSubdirRepo(url string) bool {
+	info, err := ParseGitURL(url)
+	if err != nil {
+		return false
+	}
+	if info.SubPath == "" {
+		return false
+	}
+
+	// Check if this looks like a Go module path rather than a git subdirectory
+	// Go modules can have nested paths like github.com/owner/repo/cmd/tool/v2
+	// We should try to resolve it as a Go module first
+
+	// If the URL can be resolved as a Go module, it's not a subdir repo
+	// We use a heuristic: check if the full path looks like a valid Go module
+	// by checking if it ends with /vN (major version) or contains common module patterns
+
+	// Remove version suffix for checking
+	cleanURL := url
+	if before, _, ok := strings.Cut(url, "@"); ok {
+		cleanURL = before
+	}
+
+	// Check if the path ends with /vN (Go module major version)
+	parts := strings.Split(cleanURL, "/")
+	if len(parts) > 0 {
+		lastPart := parts[len(parts)-1]
+		if len(lastPart) >= 2 && lastPart[0] == 'v' {
+			// Check if it's v2, v3, etc.
+			if _, err := fmt.Sscanf(lastPart, "v%d", new(int)); err == nil {
+				// This looks like a Go module with major version suffix
+				// It could be either a versioned module or a subdir ending in vN
+				// We'll treat it as a Go module and let go get handle it
+				mlog.Debugf("URL %s detected as Go module (ends with /vN)", url)
+				return false
+			}
+		}
+	}
+
+	// For GitHub URLs, check if the subpath could be a nested Go module
+	// Common patterns: cmd/*, internal/*, pkg/*, contrib/*
+	subPathParts := strings.Split(info.SubPath, "/")
+	if len(subPathParts) > 0 {
+		firstPart := subPathParts[0]
+		// These are common Go module nesting patterns
+		if firstPart == "cmd" || firstPart == "contrib" || firstPart == "tools" {
+			// This might be a nested Go module, not a simple subdirectory
+			// Let go get try first
+			mlog.Debugf("URL %s detected as Go module (starts with common pattern)", url)
+			return false
+		}
+	}
+
+	mlog.Debugf("URL %s detected as git subdirectory", url)
+	return true
+}
+
+// downloadGitSubdir downloads a subdirectory from a git repository using sparse checkout
+func downloadGitSubdir(ctx context.Context, repoURL string) (string, *GitRepoInfo, error) {
+	info, err := ParseGitURL(repoURL)
+	if err != nil {
+		return "", nil, err
+	}
+
+	if info.SubPath == "" {
+		return "", nil, fmt.Errorf("not a subdirectory URL: %s", repoURL)
+	}
+
+	// Create temp directory for clone
+	tempDir := gfile.Temp("gf-init-git")
+	if tempDir == "" {
+		return "", nil, fmt.Errorf("failed to create temporary directory")
+	}
+	if err := gfile.Mkdir(tempDir); err != nil {
+		return "", nil, err
+	}
+
+	cloneDir := filepath.Join(tempDir, info.Repo)
+	mlog.Debugf("Using git temp workspace: %s", tempDir)
+	mlog.Printf("Cloning %s (sparse checkout: %s)...", info.CloneURL, info.SubPath)
+
+	// 1. Clone with no checkout, filter, and sparse
+	if err := runCmd(ctx, tempDir, "git", "clone", "--filter=blob:none", "--no-checkout", "--sparse", info.CloneURL); err != nil {
+		// Fallback: try without filter for older git versions
+		mlog.Debugf("Sparse clone failed, trying full clone...")
+		if err := gfile.Remove(cloneDir); err != nil {
+			mlog.Debugf("Failed to remove clone directory: %v", err)
+		}
+		if err := runCmd(ctx, tempDir, "git", "clone", "--no-checkout", info.CloneURL); err != nil {
+			if err := gfile.Remove(tempDir); err != nil {
+				mlog.Debugf("Failed to remove temp directory: %v", err)
+			}
+			return "", nil, fmt.Errorf("git clone failed: %w", err)
+		}
+	}
+
+	// 2. Set sparse-checkout to the subpath
+	if err := runCmd(ctx, cloneDir, "git", "sparse-checkout", "set", info.SubPath); err != nil {
+		// Fallback for older git: use sparse-checkout init + set
+		mlog.Debugf("sparse-checkout set failed, trying legacy method...")
+		if err := runCmd(ctx, cloneDir, "git", "sparse-checkout", "init", "--cone"); err != nil {
+			if err := gfile.Remove(tempDir); err != nil {
+				mlog.Debugf("Failed to remove temp directory: %v", err)
+			}
+			return "", nil, fmt.Errorf("git sparse-checkout init (legacy) failed: %w", err)
+		}
+		if err := runCmd(ctx, cloneDir, "git", "sparse-checkout", "set", info.SubPath); err != nil {
+			if err := gfile.Remove(tempDir); err != nil {
+				mlog.Debugf("Failed to remove temp directory: %v", err)
+			}
+			return "", nil, fmt.Errorf("git sparse-checkout set (legacy) failed: %w", err)
+		}
+	}
+
+	// 3. Checkout the branch
+	if err := runCmd(ctx, cloneDir, "git", "checkout", info.Branch); err != nil {
+		// Try master if main fails
+		if info.Branch == "main" {
+			mlog.Debugf("Branch 'main' not found, trying 'master'...")
+			info.Branch = "master"
+			if err := runCmd(ctx, cloneDir, "git", "checkout", "master"); err != nil {
+				if err := gfile.Remove(tempDir); err != nil {
+					mlog.Debugf("Failed to remove temp directory: %v", err)
+				}
+				return "", nil, fmt.Errorf("git checkout failed: %w", err)
+			}
+		} else {
+			if err := gfile.Remove(tempDir); err != nil {
+				mlog.Debugf("Failed to remove temp directory: %v", err)
+			}
+			return "", nil, fmt.Errorf("git checkout failed: %w", err)
+		}
+	}
+
+	// Return the path to the subdirectory
+	subDirPath := filepath.Join(cloneDir, info.SubPath)
+	if !gfile.Exists(subDirPath) {
+		if err := gfile.Remove(tempDir); err != nil {
+			mlog.Debugf("Failed to remove temp directory: %v", err)
+		}
+		return "", nil, fmt.Errorf("subdirectory not found: %s", info.SubPath)
+	}
+
+	mlog.Debugf("Subdirectory located at: %s", subDirPath)
+	return subDirPath, info, nil
+}
+
+// GetModuleNameFromGoMod reads module name from go.mod file
+func GetModuleNameFromGoMod(dir string) string {
+	goModPath := filepath.Join(dir, "go.mod")
+	if !gfile.Exists(goModPath) {
+		return ""
+	}
+
+	content := gfile.GetContents(goModPath)
+	lines := gstr.Split(content, "\n")
+	for _, line := range lines {
+		line = strings.TrimSpace(line)
+		if after, ok := strings.CutPrefix(line, "module "); ok {
+			return strings.TrimSpace(after)
+		}
+	}
+	return ""
+}
--- a/cmd/gf/internal/cmd/geninit/geninit_selector.go
+++ b/cmd/gf/internal/cmd/geninit/geninit_selector.go
@ -0,0 +1,99 @@
+// Copyright GoFrame gf Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package geninit
+
+import (
+	"bufio"
+	"context"
+	"fmt"
+	"os"
+	"strconv"
+	"strings"
+
+	"github.com/gogf/gf/cmd/gf/v2/internal/utility/mlog"
+)
+
+// SelectVersion prompts user to select a version interactively
+func SelectVersion(ctx context.Context, versions []string, modulePath string) (string, error) {
+	if len(versions) == 0 {
+		return "", fmt.Errorf("no versions available for selection")
+	}
+
+	if len(versions) == 1 {
+		mlog.Printf("Only one version available: %s", versions[0])
+		return versions[0], nil
+	}
+
+	// Display available versions
+	fmt.Printf("\nAvailable versions for %s:\n", modulePath)
+	fmt.Println(strings.Repeat("-", 40))
+
+	// Show versions with index (newest first)
+	maxDisplay := 20 // Limit display to avoid overwhelming output
+	displayCount := len(versions)
+	if displayCount > maxDisplay {
+		displayCount = maxDisplay
+	}
+
+	for i := 0; i < displayCount; i++ {
+		marker := ""
+		if i == 0 {
+			marker = " (latest)"
+		}
+		fmt.Printf("  [%2d] %s%s\n", i+1, versions[i], marker)
+	}
+
+	if len(versions) > maxDisplay {
+		fmt.Printf("  ... and %d more versions\n", len(versions)-maxDisplay)
+	}
+
+	fmt.Println(strings.Repeat("-", 40))
+
+	// Prompt for selection
+	reader := bufio.NewReader(os.Stdin)
+	for {
+		fmt.Printf("Select version [1-%d] or enter version string (default: 1 for latest): ", displayCount)
+
+		input, err := reader.ReadString('\n')
+		if err != nil {
+			return "", fmt.Errorf("failed to read input: %w", err)
+		}
+
+		input = strings.TrimSpace(input)
+
+		// Default to latest
+		if input == "" {
+			fmt.Printf("Selected: %s (latest)\n", versions[0])
+			return versions[0], nil
+		}
+
+		// Try parsing as number first
+		idx, err := strconv.Atoi(input)
+		if err == nil {
+			// Valid number - check if in range
+			if idx >= 1 && idx <= len(versions) {
+				// Allow selection from all versions, not just displayed ones
+				selected := versions[idx-1]
+				fmt.Printf("Selected: %s\n", selected)
+				return selected, nil
+			} else if idx < 1 || idx > displayCount {
+				fmt.Printf("Invalid selection. Please enter a number between 1 and %d, or type a version string.\n", displayCount)
+				continue
+			}
+		} else {
+			// Try matching the input as a version string (e.g., "v1.2.3")
+			for _, v := range versions {
+				if v == input || strings.Contains(v, input) {
+					fmt.Printf("Selected: %s\n", v)
+					return v, nil
+				}
+			}
+			fmt.Printf("Version '%s' not found. Please select by number or type a valid version string.\n", input)
+			continue
+		}
+	}
+}
--- a/cmd/gf/internal/cmd/geninit/geninit_version.go
+++ b/cmd/gf/internal/cmd/geninit/geninit_version.go
@ -0,0 +1,138 @@
+// Copyright GoFrame gf Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package geninit
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"os/exec"
+	"sort"
+	"strings"
+
+	"golang.org/x/mod/semver"
+
+	"github.com/gogf/gf/v2/os/gfile"
+
+	"github.com/gogf/gf/cmd/gf/v2/internal/utility/mlog"
+)
+
+// VersionInfo contains module version information
+type VersionInfo struct {
+	Module   string   `json:"module"`
+	Versions []string `json:"versions"`
+	Latest   string   `json:"latest"`
+}
+
+// GetModuleVersions fetches available versions for a Go module
+func GetModuleVersions(ctx context.Context, modulePath string) (*VersionInfo, error) {
+	// Create a temporary directory for go list
+	tempDir := gfile.Temp("gf-init-version")
+	if tempDir == "" {
+		return nil, fmt.Errorf("failed to create temporary directory for go list")
+	}
+	if err := gfile.Mkdir(tempDir); err != nil {
+		return nil, err
+	}
+	defer func() {
+		if err := gfile.Remove(tempDir); err != nil {
+			mlog.Debugf("Failed to remove temp directory: %v", err)
+		}
+	}()
+
+	// Initialize a temp go module
+	if err := runCmd(ctx, tempDir, "go", "mod", "init", "temp"); err != nil {
+		return nil, fmt.Errorf("failed to init temp module: %w", err)
+	}
+
+	// Get versions using go list -m -versions
+	cmd := exec.CommandContext(ctx, "go", "list", "-m", "-versions", modulePath)
+	cmd.Dir = tempDir
+	output, err := cmd.Output()
+	if err != nil {
+		// Try with @latest to see if module exists
+		mlog.Debugf("go list -versions failed, trying @latest: %v", err)
+		return getLatestOnly(ctx, tempDir, modulePath)
+	}
+
+	// Parse output: "module/path v1.0.0 v1.1.0 v2.0.0"
+	parts := strings.Fields(strings.TrimSpace(string(output)))
+	if len(parts) < 1 {
+		return nil, fmt.Errorf("no version information found for %s", modulePath)
+	}
+
+	info := &VersionInfo{
+		Module:   parts[0],
+		Versions: []string{},
+	}
+
+	if len(parts) > 1 {
+		info.Versions = parts[1:]
+		// Sort versions in descending order (newest first)
+		sort.Slice(info.Versions, func(i, j int) bool {
+			return semver.Compare(info.Versions[i], info.Versions[j]) > 0
+		})
+		info.Latest = info.Versions[0]
+	}
+
+	// If no tagged versions, try to get latest
+	if len(info.Versions) == 0 {
+		latestInfo, err := getLatestOnly(ctx, tempDir, modulePath)
+		if err != nil {
+			return nil, err
+		}
+		info.Latest = latestInfo.Latest
+		if latestInfo.Latest != "" {
+			info.Versions = []string{latestInfo.Latest}
+		}
+	}
+
+	return info, nil
+}
+
+// getLatestOnly gets only the latest version when go list -versions fails
+func getLatestOnly(ctx context.Context, tempDir, modulePath string) (*VersionInfo, error) {
+	// Try go list -m modulePath@latest
+	cmd := exec.CommandContext(ctx, "go", "list", "-m", "-json", modulePath+"@latest")
+	cmd.Dir = tempDir
+	output, err := cmd.Output()
+	if err != nil {
+		// Try without @latest
+		cmd = exec.CommandContext(ctx, "go", "list", "-m", "-json", modulePath)
+		cmd.Dir = tempDir
+		output, err = cmd.Output()
+		if err != nil {
+			return nil, fmt.Errorf("failed to get module info for %s: %w", modulePath, err)
+		}
+	}
+
+	var modInfo struct {
+		Path    string `json:"Path"`
+		Version string `json:"Version"`
+	}
+	if err := json.Unmarshal(output, &modInfo); err != nil {
+		return nil, fmt.Errorf("failed to parse module info: %w", err)
+	}
+
+	return &VersionInfo{
+		Module:   modInfo.Path,
+		Versions: []string{modInfo.Version},
+		Latest:   modInfo.Version,
+	}, nil
+}
+
+// GetLatestVersion returns the latest version of a module
+func GetLatestVersion(ctx context.Context, modulePath string) (string, error) {
+	info, err := GetModuleVersions(ctx, modulePath)
+	if err != nil {
+		return "", err
+	}
+	if info.Latest == "" {
+		return "", fmt.Errorf("no version found for %s", modulePath)
+	}
+	return info.Latest, nil
+}
--- a/cmd/gf/internal/cmd/testdata/build/varmap/go.mod
+++ b/cmd/gf/internal/cmd/testdata/build/varmap/go.mod
@ -4,7 +4,7 @@ go 1.23.0

 toolchain go1.24.6

-require github.com/gogf/gf/v2 v2.9.5
+require github.com/gogf/gf/v2 v2.9.6

 require (
 	go.opentelemetry.io/otel v1.38.0 // indirect
--- a/cmd/gf/internal/cmd/testdata/build/varmap/go.sum
+++ b/cmd/gf/internal/cmd/testdata/build/varmap/go.sum
@ -4,8 +4,8 @@ github.com/clbanning/mxj/v2 v2.7.0 h1:WA/La7UGCanFe5NpHF0Q3DNtnCsVoxbPKuyBNHWRyM
 github.com/clbanning/mxj/v2 v2.7.0/go.mod h1:hNiWqW14h+kc+MdF9C6/YoRfjEJoR3ou6tn/Qo+ve2s=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
-github.com/emirpasic/gods v1.18.1 h1:FXtiHYKDGKCW2KzwZKx0iC0PQmdlorYgdFG9jPXJ1Bc=
-github.com/emirpasic/gods v1.18.1/go.mod h1:8tpGGwCnJ5H4r6BWwaV6OrWmMoPhUl5jm/FMNAnJvWQ=
+github.com/emirpasic/gods/v2 v2.0.0-alpha h1:dwFlh8pBg1VMOXWGipNMRt8v96dKAIvBehtCt6OtunU=
+github.com/emirpasic/gods/v2 v2.0.0-alpha/go.mod h1:W0y4M2dtBB9U5z3YlghmpuUhiaZT2h6yoeE+C1sCp6A=
 github.com/fatih/color v1.18.0 h1:S8gINlzdQ840/4pfAwic/ZE0djQEH3wM94VfqLTZcOM=
 github.com/fatih/color v1.18.0/go.mod h1:4FelSpRwEGDpQ12mAdzqdOukCy4u8WUtOY6lkT/6HfU=
 github.com/fsnotify/fsnotify v1.9.0 h1:2Ml+OJNzbYCTzsxtv8vKSFD9PbJjmhYF14k/jKC7S9k=
--- a/cmd/gf/internal/cmd/testdata/fix/fix25_content.go
+++ b/cmd/gf/internal/cmd/testdata/fix/fix25_content.go
@ -1,13 +1,10 @@
 package testdata

 import (
-	"fmt"
 	"testing"
-	"time"

 	"github.com/gogf/gf/v2/frame/g"
 	"github.com/gogf/gf/v2/net/ghttp"
-	"github.com/gogf/gf/v2/test/gtest"
 	"github.com/gogf/gf/v2/util/guid"
 )

--- a/container/gmap/gmap_hash_any_any_map.go
+++ b/container/gmap/gmap_hash_any_any_map.go
@ -7,245 +7,143 @@
 package gmap

 import (
-	"reflect"
+	"sync"

 	"github.com/gogf/gf/v2/container/gvar"
-	"github.com/gogf/gf/v2/internal/deepcopy"
-	"github.com/gogf/gf/v2/internal/empty"
-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
-	"github.com/gogf/gf/v2/util/gconv"
 )

 // AnyAnyMap wraps map type `map[any]any` and provides more map features.
 type AnyAnyMap struct {
-	mu   rwmutex.RWMutex
-	data map[any]any
+	*KVMap[any, any]
+	once sync.Once
 }

 // NewAnyAnyMap creates and returns an empty hash map.
 // The parameter `safe` is used to specify whether using map in concurrent-safety,
 // which is false in default.
 func NewAnyAnyMap(safe ...bool) *AnyAnyMap {
-	return &AnyAnyMap{
-		mu:   rwmutex.Create(safe...),
-		data: make(map[any]any),
+	m := &AnyAnyMap{
+		KVMap: NewKVMap[any, any](safe...),
 	}
+	return m
 }

 // NewAnyAnyMapFrom creates and returns a hash map from given map `data`.
 // Note that, the param `data` map will be set as the underlying data map(no deep copy),
 // there might be some concurrent-safe issues when changing the map outside.
 func NewAnyAnyMapFrom(data map[any]any, safe ...bool) *AnyAnyMap {
-	return &AnyAnyMap{
-		mu:   rwmutex.Create(safe...),
-		data: data,
+	m := &AnyAnyMap{
+		KVMap: NewKVMapFrom(data, safe...),
 	}
+	return m
+}
+
+// lazyInit lazily initializes the map.
+func (m *AnyAnyMap) lazyInit() {
+	m.once.Do(func() {
+		if m.KVMap == nil {
+			m.KVMap = NewKVMap[any, any](false)
+		}
+	})
 }

 // Iterator iterates the hash map readonly with custom callback function `f`.
 // If `f` returns true, then it continues iterating; or false to stop.
 func (m *AnyAnyMap) Iterator(f func(k any, v any) bool) {
-	for k, v := range m.Map() {
-		if !f(k, v) {
-			break
-		}
-	}
+	m.lazyInit()
+	m.KVMap.Iterator(f)
 }

 // Clone returns a new hash map with copy of current map data.
 func (m *AnyAnyMap) Clone(safe ...bool) *AnyAnyMap {
-	return NewFrom(m.MapCopy(), safe...)
+	m.lazyInit()
+	return NewAnyAnyMapFrom(m.MapCopy(), safe...)
 }

 // Map returns the underlying data map.
 // Note that, if it's in concurrent-safe usage, it returns a copy of underlying data,
 // or else a pointer to the underlying data.
 func (m *AnyAnyMap) Map() map[any]any {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	if !m.mu.IsSafe() {
-		return m.data
-	}
-	data := make(map[any]any, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.Map()
 }

 // MapCopy returns a shallow copy of the underlying data of the hash map.
 func (m *AnyAnyMap) MapCopy() map[any]any {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[any]any, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.MapCopy()
 }

 // MapStrAny returns a copy of the underlying data of the map as map[string]any.
 func (m *AnyAnyMap) MapStrAny() map[string]any {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[string]any, len(m.data))
-	for k, v := range m.data {
-		data[gconv.String(k)] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.MapStrAny()
 }

 // FilterEmpty deletes all key-value pair of which the value is empty.
 // Values like: 0, nil, false, "", len(slice/map/chan) == 0 are considered empty.
 func (m *AnyAnyMap) FilterEmpty() {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for k, v := range m.data {
-		if empty.IsEmpty(v) {
-			delete(m.data, k)
-		}
-	}
+	m.lazyInit()
+	m.KVMap.FilterEmpty()
 }

 // FilterNil deletes all key-value pair of which the value is nil.
 func (m *AnyAnyMap) FilterNil() {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for k, v := range m.data {
-		if empty.IsNil(v) {
-			delete(m.data, k)
-		}
-	}
+	m.lazyInit()
+	m.KVMap.FilterNil()
 }

 // Set sets key-value to the hash map.
 func (m *AnyAnyMap) Set(key any, value any) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = make(map[any]any)
-	}
-	m.data[key] = value
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Set(key, value)
 }

 // Sets batch sets key-values to the hash map.
 func (m *AnyAnyMap) Sets(data map[any]any) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = data
-	} else {
-		for k, v := range data {
-			m.data[k] = v
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Sets(data)
 }

 // Search searches the map with given `key`.
 // Second return parameter `found` is true if key was found, otherwise false.
 func (m *AnyAnyMap) Search(key any) (value any, found bool) {
-	m.mu.RLock()
-	if m.data != nil {
-		value, found = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Search(key)
 }

 // Get returns the value by given `key`.
 func (m *AnyAnyMap) Get(key any) (value any) {
-	m.mu.RLock()
-	if m.data != nil {
-		value = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Get(key)
 }

 // Pop retrieves and deletes an item from the map.
 func (m *AnyAnyMap) Pop() (key, value any) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for key, value = range m.data {
-		delete(m.data, key)
-		return
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Pop()
 }

 // Pops retrieves and deletes `size` items from the map.
 // It returns all items if size == -1.
 func (m *AnyAnyMap) Pops(size int) map[any]any {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if size > len(m.data) || size == -1 {
-		size = len(m.data)
-	}
-	if size == 0 {
-		return nil
-	}
-	var (
-		index  = 0
-		newMap = make(map[any]any, size)
-	)
-	for k, v := range m.data {
-		delete(m.data, k)
-		newMap[k] = v
-		index++
-		if index == size {
-			break
-		}
-	}
-	return newMap
-}
-
-// doSetWithLockCheck checks whether value of the key exists with mutex.Lock,
-// if not exists, set value to the map with given `key`,
-// or else just return the existing value.
-//
-// When setting value, if `value` is type of `func() interface {}`,
-// it will be executed with mutex.Lock of the hash map,
-// and its return value will be set to the map with `key`.
-//
-// It returns value with given `key`.
-func (m *AnyAnyMap) doSetWithLockCheck(key any, value any) any {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[any]any)
-	}
-	if v, ok := m.data[key]; ok {
-		return v
-	}
-	if f, ok := value.(func() any); ok {
-		value = f()
-	}
-	if value != nil {
-		m.data[key] = value
-	}
-	return value
+	m.lazyInit()
+	return m.KVMap.Pops(size)
 }

 // GetOrSet returns the value by key,
 // or sets value with given `value` if it does not exist and then returns this value.
 func (m *AnyAnyMap) GetOrSet(key any, value any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, value)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSet(key, value)
 }

 // GetOrSetFunc returns the value by key,
 // or sets value with returned value of callback function `f` if it does not exist
 // and then returns this value.
 func (m *AnyAnyMap) GetOrSetFunc(key any, f func() any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f())
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFunc(key, f)
 }

 // GetOrSetFuncLock returns the value by key,
@ -255,55 +153,50 @@ func (m *AnyAnyMap) GetOrSetFunc(key any, f func() any) any {
 // GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`
 // with mutex.Lock of the hash map.
 func (m *AnyAnyMap) GetOrSetFuncLock(key any, f func() any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFuncLock(key, f)
 }

 // GetVar returns a Var with the value by given `key`.
 // The returned Var is un-concurrent safe.
 func (m *AnyAnyMap) GetVar(key any) *gvar.Var {
-	return gvar.New(m.Get(key))
+	m.lazyInit()
+	return m.KVMap.GetVar(key)
 }

 // GetVarOrSet returns a Var with result from GetOrSet.
 // The returned Var is un-concurrent safe.
 func (m *AnyAnyMap) GetVarOrSet(key any, value any) *gvar.Var {
-	return gvar.New(m.GetOrSet(key, value))
+	m.lazyInit()
+	return m.KVMap.GetVarOrSet(key, value)
 }

 // GetVarOrSetFunc returns a Var with result from GetOrSetFunc.
 // The returned Var is un-concurrent safe.
 func (m *AnyAnyMap) GetVarOrSetFunc(key any, f func() any) *gvar.Var {
-	return gvar.New(m.GetOrSetFunc(key, f))
+	m.lazyInit()
+	return m.KVMap.GetVarOrSetFunc(key, f)
 }

 // GetVarOrSetFuncLock returns a Var with result from GetOrSetFuncLock.
 // The returned Var is un-concurrent safe.
 func (m *AnyAnyMap) GetVarOrSetFuncLock(key any, f func() any) *gvar.Var {
-	return gvar.New(m.GetOrSetFuncLock(key, f))
+	m.lazyInit()
+	return m.KVMap.GetVarOrSetFuncLock(key, f)
 }

 // SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *AnyAnyMap) SetIfNotExist(key any, value any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, value)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExist(key, value)
 }

 // SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *AnyAnyMap) SetIfNotExistFunc(key any, f func() any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f())
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFunc(key, f)
 }

 // SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
@ -312,119 +205,76 @@ func (m *AnyAnyMap) SetIfNotExistFunc(key any, f func() any) bool {
 // SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
 // it executes function `f` with mutex.Lock of the hash map.
 func (m *AnyAnyMap) SetIfNotExistFuncLock(key any, f func() any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFuncLock(key, f)
 }

 // Remove deletes value from map by given `key`, and return this deleted value.
 func (m *AnyAnyMap) Remove(key any) (value any) {
-	m.mu.Lock()
-	if m.data != nil {
-		var ok bool
-		if value, ok = m.data[key]; ok {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Remove(key)
 }

 // Removes batch deletes values of the map by keys.
 func (m *AnyAnyMap) Removes(keys []any) {
-	m.mu.Lock()
-	if m.data != nil {
-		for _, key := range keys {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Removes(keys)
 }

 // Keys returns all keys of the map as a slice.
 func (m *AnyAnyMap) Keys() []any {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	var (
-		keys  = make([]any, len(m.data))
-		index = 0
-	)
-	for key := range m.data {
-		keys[index] = key
-		index++
-	}
-	return keys
+	m.lazyInit()
+	return m.KVMap.Keys()
 }

 // Values returns all values of the map as a slice.
 func (m *AnyAnyMap) Values() []any {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	var (
-		values = make([]any, len(m.data))
-		index  = 0
-	)
-	for _, value := range m.data {
-		values[index] = value
-		index++
-	}
-	return values
+	m.lazyInit()
+	return m.KVMap.Values()
 }

 // Contains checks whether a key exists.
 // It returns true if the `key` exists, or else false.
 func (m *AnyAnyMap) Contains(key any) bool {
-	var ok bool
-	m.mu.RLock()
-	if m.data != nil {
-		_, ok = m.data[key]
-	}
-	m.mu.RUnlock()
-	return ok
+	m.lazyInit()
+	return m.KVMap.Contains(key)
 }

 // Size returns the size of the map.
 func (m *AnyAnyMap) Size() int {
-	m.mu.RLock()
-	length := len(m.data)
-	m.mu.RUnlock()
-	return length
+	m.lazyInit()
+	return m.KVMap.Size()
 }

 // IsEmpty checks whether the map is empty.
 // It returns true if map is empty, or else false.
 func (m *AnyAnyMap) IsEmpty() bool {
-	return m.Size() == 0
+	m.lazyInit()
+	return m.KVMap.IsEmpty()
 }

 // Clear deletes all data of the map, it will remake a new underlying data map.
 func (m *AnyAnyMap) Clear() {
-	m.mu.Lock()
-	m.data = make(map[any]any)
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Clear()
 }

 // Replace the data of the map with given `data`.
 func (m *AnyAnyMap) Replace(data map[any]any) {
-	m.mu.Lock()
-	m.data = data
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Replace(data)
 }

 // LockFunc locks writing with given callback function `f` within RWMutex.Lock.
 func (m *AnyAnyMap) LockFunc(f func(m map[any]any)) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.LockFunc(f)
 }

 // RLockFunc locks reading with given callback function `f` within RWMutex.RLock.
 func (m *AnyAnyMap) RLockFunc(f func(m map[any]any)) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.RLockFunc(f)
 }

 // Flip exchanges key-value of the map to value-key.
@ -441,19 +291,8 @@ func (m *AnyAnyMap) Flip() {
 // Merge merges two hash maps.
 // The `other` map will be merged into the map `m`.
 func (m *AnyAnyMap) Merge(other *AnyAnyMap) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = other.MapCopy()
-		return
-	}
-	if other != m {
-		other.mu.RLock()
-		defer other.mu.RUnlock()
-	}
-	for k, v := range other.data {
-		m.data[k] = v
-	}
+	m.lazyInit()
+	m.KVMap.Merge(other.KVMap)
 }

 // String returns the map as a string.
@ -461,79 +300,40 @@ func (m *AnyAnyMap) String() string {
 	if m == nil {
 		return ""
 	}
-	b, _ := m.MarshalJSON()
-	return string(b)
+	m.lazyInit()
+	return m.KVMap.String()
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (m AnyAnyMap) MarshalJSON() ([]byte, error) {
-	return json.Marshal(gconv.Map(m.Map()))
+	m.lazyInit()
+	return m.KVMap.MarshalJSON()
 }

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (m *AnyAnyMap) UnmarshalJSON(b []byte) error {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[any]any)
-	}
-	var data map[string]any
-	if err := json.UnmarshalUseNumber(b, &data); err != nil {
-		return err
-	}
-	for k, v := range data {
-		m.data[k] = v
-	}
-	return nil
+	m.lazyInit()
+	return m.KVMap.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for map.
 func (m *AnyAnyMap) UnmarshalValue(value any) (err error) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[any]any)
-	}
-	for k, v := range gconv.Map(value) {
-		m.data[k] = v
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.UnmarshalValue(value)
 }

 // DeepCopy implements interface for deep copy of current type.
 func (m *AnyAnyMap) DeepCopy() any {
-	if m == nil {
-		return nil
+	m.lazyInit()
+	return &AnyAnyMap{
+		KVMap: m.KVMap.DeepCopy().(*KVMap[any, any]),
 	}
-
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[any]any, len(m.data))
-	for k, v := range m.data {
-		data[k] = deepcopy.Copy(v)
-	}
-	return NewFrom(data, m.mu.IsSafe())
 }

 // IsSubOf checks whether the current map is a sub-map of `other`.
 func (m *AnyAnyMap) IsSubOf(other *AnyAnyMap) bool {
-	if m == other {
-		return true
-	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	for key, value := range m.data {
-		otherValue, ok := other.data[key]
-		if !ok {
-			return false
-		}
-		if otherValue != value {
-			return false
-		}
-	}
-	return true
+	m.lazyInit()
+	return m.KVMap.IsSubOf(other.KVMap)
 }

 // Diff compares current map `m` with map `other` and returns their different keys.
@ -541,22 +341,6 @@ func (m *AnyAnyMap) IsSubOf(other *AnyAnyMap) bool {
 // The returned `removedKeys` are the keys that are in map `other` but not in map `m`.
 // The returned `updatedKeys` are the keys that are both in map `m` and `other` but their values and not equal (`!=`).
 func (m *AnyAnyMap) Diff(other *AnyAnyMap) (addedKeys, removedKeys, updatedKeys []any) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-
-	for key := range m.data {
-		if _, ok := other.data[key]; !ok {
-			removedKeys = append(removedKeys, key)
-		} else if !reflect.DeepEqual(m.data[key], other.data[key]) {
-			updatedKeys = append(updatedKeys, key)
-		}
-	}
-	for key := range other.data {
-		if _, ok := m.data[key]; !ok {
-			addedKeys = append(addedKeys, key)
-		}
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Diff(other.KVMap)
 }
--- a/container/gmap/gmap_hash_int_any_map.go
+++ b/container/gmap/gmap_hash_int_any_map.go
@ -8,244 +8,143 @@
 package gmap

 import (
-	"reflect"
+	"sync"

 	"github.com/gogf/gf/v2/container/gvar"
-	"github.com/gogf/gf/v2/internal/deepcopy"
-	"github.com/gogf/gf/v2/internal/empty"
-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
 	"github.com/gogf/gf/v2/util/gconv"
 )

 // IntAnyMap implements map[int]any with RWMutex that has switch.
 type IntAnyMap struct {
-	mu   rwmutex.RWMutex
-	data map[int]any
+	*KVMap[int, any]
+	once sync.Once
 }

 // NewIntAnyMap returns an empty IntAnyMap object.
 // The parameter `safe` is used to specify whether using map in concurrent-safety,
 // which is false in default.
 func NewIntAnyMap(safe ...bool) *IntAnyMap {
-	return &IntAnyMap{
-		mu:   rwmutex.Create(safe...),
-		data: make(map[int]any),
+	m := &IntAnyMap{
+		KVMap: NewKVMap[int, any](safe...),
 	}
+	return m
 }

 // NewIntAnyMapFrom creates and returns a hash map from given map `data`.
 // Note that, the param `data` map will be set as the underlying data map(no deep copy),
 // there might be some concurrent-safe issues when changing the map outside.
 func NewIntAnyMapFrom(data map[int]any, safe ...bool) *IntAnyMap {
-	return &IntAnyMap{
-		mu:   rwmutex.Create(safe...),
-		data: data,
+	m := &IntAnyMap{
+		KVMap: NewKVMapFrom(data, safe...),
 	}
+	return m
+}
+
+// lazyInit lazily initializes the map.
+func (m *IntAnyMap) lazyInit() {
+	m.once.Do(func() {
+		if m.KVMap == nil {
+			m.KVMap = NewKVMap[int, any](false)
+		}
+	})
 }

 // Iterator iterates the hash map readonly with custom callback function `f`.
 // If `f` returns true, then it continues iterating; or false to stop.
 func (m *IntAnyMap) Iterator(f func(k int, v any) bool) {
-	for k, v := range m.Map() {
-		if !f(k, v) {
-			break
-		}
-	}
+	m.lazyInit()
+	m.KVMap.Iterator(f)
 }

 // Clone returns a new hash map with copy of current map data.
-func (m *IntAnyMap) Clone() *IntAnyMap {
-	return NewIntAnyMapFrom(m.MapCopy(), m.mu.IsSafe())
+func (m *IntAnyMap) Clone(safe ...bool) *IntAnyMap {
+	m.lazyInit()
+	return NewIntAnyMapFrom(m.MapCopy(), safe...)
 }

 // Map returns the underlying data map.
 // Note that, if it's in concurrent-safe usage, it returns a copy of underlying data,
 // or else a pointer to the underlying data.
 func (m *IntAnyMap) Map() map[int]any {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	if !m.mu.IsSafe() {
-		return m.data
-	}
-	data := make(map[int]any, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.Map()
 }

 // MapStrAny returns a copy of the underlying data of the map as map[string]any.
 func (m *IntAnyMap) MapStrAny() map[string]any {
-	m.mu.RLock()
-	data := make(map[string]any, len(m.data))
-	for k, v := range m.data {
-		data[gconv.String(k)] = v
-	}
-	m.mu.RUnlock()
-	return data
+	m.lazyInit()
+	return m.KVMap.MapStrAny()
 }

 // MapCopy returns a copy of the underlying data of the hash map.
 func (m *IntAnyMap) MapCopy() map[int]any {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[int]any, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.MapCopy()
 }

 // FilterEmpty deletes all key-value pair of which the value is empty.
 // Values like: 0, nil, false, "", len(slice/map/chan) == 0 are considered empty.
 func (m *IntAnyMap) FilterEmpty() {
-	m.mu.Lock()
-	for k, v := range m.data {
-		if empty.IsEmpty(v) {
-			delete(m.data, k)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.FilterEmpty()
 }

 // FilterNil deletes all key-value pair of which the value is nil.
 func (m *IntAnyMap) FilterNil() {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for k, v := range m.data {
-		if empty.IsNil(v) {
-			delete(m.data, k)
-		}
-	}
+	m.lazyInit()
+	m.KVMap.FilterNil()
 }

 // Set sets key-value to the hash map.
 func (m *IntAnyMap) Set(key int, val any) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = make(map[int]any)
-	}
-	m.data[key] = val
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Set(key, val)
 }

 // Sets batch sets key-values to the hash map.
 func (m *IntAnyMap) Sets(data map[int]any) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = data
-	} else {
-		for k, v := range data {
-			m.data[k] = v
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Sets(data)
 }

 // Search searches the map with given `key`.
 // Second return parameter `found` is true if key was found, otherwise false.
 func (m *IntAnyMap) Search(key int) (value any, found bool) {
-	m.mu.RLock()
-	if m.data != nil {
-		value, found = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Search(key)
 }

 // Get returns the value by given `key`.
 func (m *IntAnyMap) Get(key int) (value any) {
-	m.mu.RLock()
-	if m.data != nil {
-		value = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Get(key)
 }

 // Pop retrieves and deletes an item from the map.
 func (m *IntAnyMap) Pop() (key int, value any) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for key, value = range m.data {
-		delete(m.data, key)
-		return
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Pop()
 }

 // Pops retrieves and deletes `size` items from the map.
 // It returns all items if size == -1.
 func (m *IntAnyMap) Pops(size int) map[int]any {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if size > len(m.data) || size == -1 {
-		size = len(m.data)
-	}
-	if size == 0 {
-		return nil
-	}
-	var (
-		index  = 0
-		newMap = make(map[int]any, size)
-	)
-	for k, v := range m.data {
-		delete(m.data, k)
-		newMap[k] = v
-		index++
-		if index == size {
-			break
-		}
-	}
-	return newMap
-}
-
-// doSetWithLockCheck checks whether value of the key exists with mutex.Lock,
-// if not exists, set value to the map with given `key`,
-// or else just return the existing value.
-//
-// When setting value, if `value` is type of `func() interface {}`,
-// it will be executed with mutex.Lock of the hash map,
-// and its return value will be set to the map with `key`.
-//
-// It returns value with given `key`.
-func (m *IntAnyMap) doSetWithLockCheck(key int, value any) any {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[int]any)
-	}
-	if v, ok := m.data[key]; ok {
-		return v
-	}
-	if f, ok := value.(func() any); ok {
-		value = f()
-	}
-	if value != nil {
-		m.data[key] = value
-	}
-	return value
+	m.lazyInit()
+	return m.KVMap.Pops(size)
 }

 // GetOrSet returns the value by key,
 // or sets value with given `value` if it does not exist and then returns this value.
 func (m *IntAnyMap) GetOrSet(key int, value any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, value)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSet(key, value)
 }

 // GetOrSetFunc returns the value by key,
 // or sets value with returned value of callback function `f` if it does not exist and returns this value.
 func (m *IntAnyMap) GetOrSetFunc(key int, f func() any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f())
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFunc(key, f)
 }

 // GetOrSetFuncLock returns the value by key,
@ -254,55 +153,50 @@ func (m *IntAnyMap) GetOrSetFunc(key int, f func() any) any {
 // GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`
 // with mutex.Lock of the hash map.
 func (m *IntAnyMap) GetOrSetFuncLock(key int, f func() any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFuncLock(key, f)
 }

 // GetVar returns a Var with the value by given `key`.
 // The returned Var is un-concurrent safe.
 func (m *IntAnyMap) GetVar(key int) *gvar.Var {
-	return gvar.New(m.Get(key))
+	m.lazyInit()
+	return m.KVMap.GetVar(key)
 }

 // GetVarOrSet returns a Var with result from GetVarOrSet.
 // The returned Var is un-concurrent safe.
 func (m *IntAnyMap) GetVarOrSet(key int, value any) *gvar.Var {
-	return gvar.New(m.GetOrSet(key, value))
+	m.lazyInit()
+	return m.KVMap.GetVarOrSet(key, value)
 }

 // GetVarOrSetFunc returns a Var with result from GetOrSetFunc.
 // The returned Var is un-concurrent safe.
 func (m *IntAnyMap) GetVarOrSetFunc(key int, f func() any) *gvar.Var {
-	return gvar.New(m.GetOrSetFunc(key, f))
+	m.lazyInit()
+	return m.KVMap.GetVarOrSetFunc(key, f)
 }

 // GetVarOrSetFuncLock returns a Var with result from GetOrSetFuncLock.
 // The returned Var is un-concurrent safe.
 func (m *IntAnyMap) GetVarOrSetFuncLock(key int, f func() any) *gvar.Var {
-	return gvar.New(m.GetOrSetFuncLock(key, f))
+	m.lazyInit()
+	return m.KVMap.GetVarOrSetFuncLock(key, f)
 }

 // SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *IntAnyMap) SetIfNotExist(key int, value any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, value)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExist(key, value)
 }

 // SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *IntAnyMap) SetIfNotExistFunc(key int, f func() any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f())
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFunc(key, f)
 }

 // SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
@ -311,119 +205,76 @@ func (m *IntAnyMap) SetIfNotExistFunc(key int, f func() any) bool {
 // SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
 // it executes function `f` with mutex.Lock of the hash map.
 func (m *IntAnyMap) SetIfNotExistFuncLock(key int, f func() any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFuncLock(key, f)
 }

 // Removes batch deletes values of the map by keys.
 func (m *IntAnyMap) Removes(keys []int) {
-	m.mu.Lock()
-	if m.data != nil {
-		for _, key := range keys {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Removes(keys)
 }

 // Remove deletes value from map by given `key`, and return this deleted value.
 func (m *IntAnyMap) Remove(key int) (value any) {
-	m.mu.Lock()
-	if m.data != nil {
-		var ok bool
-		if value, ok = m.data[key]; ok {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Remove(key)
 }

 // Keys returns all keys of the map as a slice.
 func (m *IntAnyMap) Keys() []int {
-	m.mu.RLock()
-	var (
-		keys  = make([]int, len(m.data))
-		index = 0
-	)
-	for key := range m.data {
-		keys[index] = key
-		index++
-	}
-	m.mu.RUnlock()
-	return keys
+	m.lazyInit()
+	return m.KVMap.Keys()
 }

 // Values returns all values of the map as a slice.
 func (m *IntAnyMap) Values() []any {
-	m.mu.RLock()
-	var (
-		values = make([]any, len(m.data))
-		index  = 0
-	)
-	for _, value := range m.data {
-		values[index] = value
-		index++
-	}
-	m.mu.RUnlock()
-	return values
+	m.lazyInit()
+	return m.KVMap.Values()
 }

 // Contains checks whether a key exists.
 // It returns true if the `key` exists, or else false.
 func (m *IntAnyMap) Contains(key int) bool {
-	var ok bool
-	m.mu.RLock()
-	if m.data != nil {
-		_, ok = m.data[key]
-	}
-	m.mu.RUnlock()
-	return ok
+	m.lazyInit()
+	return m.KVMap.Contains(key)
 }

 // Size returns the size of the map.
 func (m *IntAnyMap) Size() int {
-	m.mu.RLock()
-	length := len(m.data)
-	m.mu.RUnlock()
-	return length
+	m.lazyInit()
+	return m.KVMap.Size()
 }

 // IsEmpty checks whether the map is empty.
 // It returns true if map is empty, or else false.
 func (m *IntAnyMap) IsEmpty() bool {
-	return m.Size() == 0
+	m.lazyInit()
+	return m.KVMap.IsEmpty()
 }

 // Clear deletes all data of the map, it will remake a new underlying data map.
 func (m *IntAnyMap) Clear() {
-	m.mu.Lock()
-	m.data = make(map[int]any)
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Clear()
 }

 // Replace the data of the map with given `data`.
 func (m *IntAnyMap) Replace(data map[int]any) {
-	m.mu.Lock()
-	m.data = data
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Replace(data)
 }

 // LockFunc locks writing with given callback function `f` within RWMutex.Lock.
 func (m *IntAnyMap) LockFunc(f func(m map[int]any)) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.LockFunc(f)
 }

 // RLockFunc locks reading with given callback function `f` within RWMutex.RLock.
 func (m *IntAnyMap) RLockFunc(f func(m map[int]any)) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.RLockFunc(f)
 }

 // Flip exchanges key-value of the map to value-key.
@ -440,19 +291,8 @@ func (m *IntAnyMap) Flip() {
 // Merge merges two hash maps.
 // The `other` map will be merged into the map `m`.
 func (m *IntAnyMap) Merge(other *IntAnyMap) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = other.MapCopy()
-		return
-	}
-	if other != m {
-		other.mu.RLock()
-		defer other.mu.RUnlock()
-	}
-	for k, v := range other.data {
-		m.data[k] = v
-	}
+	m.lazyInit()
+	m.KVMap.Merge(other.KVMap)
 }

 // String returns the map as a string.
@ -460,81 +300,40 @@ func (m *IntAnyMap) String() string {
 	if m == nil {
 		return ""
 	}
-	b, _ := m.MarshalJSON()
-	return string(b)
+	m.lazyInit()
+	return m.KVMap.String()
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (m IntAnyMap) MarshalJSON() ([]byte, error) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	return json.Marshal(m.data)
+	m.lazyInit()
+	return m.KVMap.MarshalJSON()
 }

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (m *IntAnyMap) UnmarshalJSON(b []byte) error {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[int]any)
-	}
-	if err := json.UnmarshalUseNumber(b, &m.data); err != nil {
-		return err
-	}
-	return nil
+	m.lazyInit()
+	return m.KVMap.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for map.
 func (m *IntAnyMap) UnmarshalValue(value any) (err error) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[int]any)
-	}
-	switch value.(type) {
-	case string, []byte:
-		return json.UnmarshalUseNumber(gconv.Bytes(value), &m.data)
-	default:
-		for k, v := range gconv.Map(value) {
-			m.data[gconv.Int(k)] = v
-		}
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.UnmarshalValue(value)
 }

 // DeepCopy implements interface for deep copy of current type.
 func (m *IntAnyMap) DeepCopy() any {
-	if m == nil {
-		return nil
+	m.lazyInit()
+	return &IntAnyMap{
+		KVMap: m.KVMap.DeepCopy().(*KVMap[int, any]),
 	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[int]any, len(m.data))
-	for k, v := range m.data {
-		data[k] = deepcopy.Copy(v)
-	}
-	return NewIntAnyMapFrom(data, m.mu.IsSafe())
 }

 // IsSubOf checks whether the current map is a sub-map of `other`.
 func (m *IntAnyMap) IsSubOf(other *IntAnyMap) bool {
-	if m == other {
-		return true
-	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	for key, value := range m.data {
-		otherValue, ok := other.data[key]
-		if !ok {
-			return false
-		}
-		if otherValue != value {
-			return false
-		}
-	}
-	return true
+	m.lazyInit()
+	return m.KVMap.IsSubOf(other.KVMap)
 }

 // Diff compares current map `m` with map `other` and returns their different keys.
@ -542,22 +341,6 @@ func (m *IntAnyMap) IsSubOf(other *IntAnyMap) bool {
 // The returned `removedKeys` are the keys that are in map `other` but not in map `m`.
 // The returned `updatedKeys` are the keys that are both in map `m` and `other` but their values and not equal (`!=`).
 func (m *IntAnyMap) Diff(other *IntAnyMap) (addedKeys, removedKeys, updatedKeys []int) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-
-	for key := range m.data {
-		if _, ok := other.data[key]; !ok {
-			removedKeys = append(removedKeys, key)
-		} else if !reflect.DeepEqual(m.data[key], other.data[key]) {
-			updatedKeys = append(updatedKeys, key)
-		}
-	}
-	for key := range other.data {
-		if _, ok := m.data[key]; !ok {
-			addedKeys = append(addedKeys, key)
-		}
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Diff(other.KVMap)
 }
--- a/container/gmap/gmap_hash_int_int_map.go
+++ b/container/gmap/gmap_hash_int_int_map.go
@ -6,17 +6,12 @@

 package gmap

-import (
-	"github.com/gogf/gf/v2/internal/empty"
-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
-	"github.com/gogf/gf/v2/util/gconv"
-)
+import "sync"

 // IntIntMap implements map[int]int with RWMutex that has switch.
 type IntIntMap struct {
-	mu   rwmutex.RWMutex
-	data map[int]int
+	*KVMap[int, int]
+	once sync.Once
 }

 // NewIntIntMap returns an empty IntIntMap object.
@ -24,8 +19,7 @@ type IntIntMap struct {
 // which is false in default.
 func NewIntIntMap(safe ...bool) *IntIntMap {
 	return &IntIntMap{
-		mu:   rwmutex.Create(safe...),
-		data: make(map[int]int),
+		KVMap: NewKVMap[int, int](safe...),
 	}
 }

@ -34,193 +28,109 @@ func NewIntIntMap(safe ...bool) *IntIntMap {
 // there might be some concurrent-safe issues when changing the map outside.
 func NewIntIntMapFrom(data map[int]int, safe ...bool) *IntIntMap {
 	return &IntIntMap{
-		mu:   rwmutex.Create(safe...),
-		data: data,
+		KVMap: NewKVMapFrom(data, safe...),
 	}
 }

+// lazyInit lazily initializes the map.
+func (m *IntIntMap) lazyInit() {
+	m.once.Do(func() {
+		if m.KVMap == nil {
+			m.KVMap = NewKVMap[int, int](false)
+		}
+	})
+}
+
 // Iterator iterates the hash map readonly with custom callback function `f`.
 // If `f` returns true, then it continues iterating; or false to stop.
 func (m *IntIntMap) Iterator(f func(k int, v int) bool) {
-	for k, v := range m.Map() {
-		if !f(k, v) {
-			break
-		}
-	}
+	m.lazyInit()
+	m.KVMap.Iterator(f)
 }

 // Clone returns a new hash map with copy of current map data.
-func (m *IntIntMap) Clone() *IntIntMap {
-	return NewIntIntMapFrom(m.MapCopy(), m.mu.IsSafe())
+func (m *IntIntMap) Clone(safe ...bool) *IntIntMap {
+	m.lazyInit()
+	return &IntIntMap{KVMap: m.KVMap.Clone(safe...)}
 }

 // Map returns the underlying data map.
 // Note that, if it's in concurrent-safe usage, it returns a copy of underlying data,
 // or else a pointer to the underlying data.
 func (m *IntIntMap) Map() map[int]int {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	if !m.mu.IsSafe() {
-		return m.data
-	}
-	data := make(map[int]int, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.Map()
 }

 // MapStrAny returns a copy of the underlying data of the map as map[string]any.
 func (m *IntIntMap) MapStrAny() map[string]any {
-	m.mu.RLock()
-	data := make(map[string]any, len(m.data))
-	for k, v := range m.data {
-		data[gconv.String(k)] = v
-	}
-	m.mu.RUnlock()
-	return data
+	m.lazyInit()
+	return m.KVMap.MapStrAny()
 }

 // MapCopy returns a copy of the underlying data of the hash map.
 func (m *IntIntMap) MapCopy() map[int]int {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[int]int, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.MapCopy()
 }

 // FilterEmpty deletes all key-value pair of which the value is empty.
 // Values like: 0, nil, false, "", len(slice/map/chan) == 0 are considered empty.
 func (m *IntIntMap) FilterEmpty() {
-	m.mu.Lock()
-	for k, v := range m.data {
-		if empty.IsEmpty(v) {
-			delete(m.data, k)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.FilterEmpty()
 }

 // Set sets key-value to the hash map.
 func (m *IntIntMap) Set(key int, val int) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = make(map[int]int)
-	}
-	m.data[key] = val
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Set(key, val)
 }

 // Sets batch sets key-values to the hash map.
 func (m *IntIntMap) Sets(data map[int]int) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = data
-	} else {
-		for k, v := range data {
-			m.data[k] = v
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Sets(data)
 }

 // Search searches the map with given `key`.
 // Second return parameter `found` is true if key was found, otherwise false.
 func (m *IntIntMap) Search(key int) (value int, found bool) {
-	m.mu.RLock()
-	if m.data != nil {
-		value, found = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Search(key)
 }

 // Get returns the value by given `key`.
 func (m *IntIntMap) Get(key int) (value int) {
-	m.mu.RLock()
-	if m.data != nil {
-		value = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Get(key)
 }

 // Pop retrieves and deletes an item from the map.
 func (m *IntIntMap) Pop() (key, value int) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for key, value = range m.data {
-		delete(m.data, key)
-		return
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Pop()
 }

 // Pops retrieves and deletes `size` items from the map.
 // It returns all items if size == -1.
 func (m *IntIntMap) Pops(size int) map[int]int {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if size > len(m.data) || size == -1 {
-		size = len(m.data)
-	}
-	if size == 0 {
-		return nil
-	}
-	var (
-		index  = 0
-		newMap = make(map[int]int, size)
-	)
-	for k, v := range m.data {
-		delete(m.data, k)
-		newMap[k] = v
-		index++
-		if index == size {
-			break
-		}
-	}
-	return newMap
-}
-
-// doSetWithLockCheck checks whether value of the key exists with mutex.Lock,
-// if not exists, set value to the map with given `key`,
-// or else just return the existing value.
-//
-// It returns value with given `key`.
-func (m *IntIntMap) doSetWithLockCheck(key int, value int) int {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[int]int)
-	}
-	if v, ok := m.data[key]; ok {
-		return v
-	}
-	m.data[key] = value
-	return value
+	m.lazyInit()
+	return m.KVMap.Pops(size)
 }

 // GetOrSet returns the value by key,
 // or sets value with given `value` if it does not exist and then returns this value.
 func (m *IntIntMap) GetOrSet(key int, value int) int {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, value)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSet(key, value)
 }

 // GetOrSetFunc returns the value by key,
 // or sets value with returned value of callback function `f` if it does not exist and returns this value.
 func (m *IntIntMap) GetOrSetFunc(key int, f func() int) int {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f())
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFunc(key, f)
 }

 // GetOrSetFuncLock returns the value by key,
@ -229,41 +139,22 @@ func (m *IntIntMap) GetOrSetFunc(key int, f func() int) int {
 // GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`
 // with mutex.Lock of the hash map.
 func (m *IntIntMap) GetOrSetFuncLock(key int, f func() int) int {
-	if v, ok := m.Search(key); !ok {
-		m.mu.Lock()
-		defer m.mu.Unlock()
-		if m.data == nil {
-			m.data = make(map[int]int)
-		}
-		if v, ok = m.data[key]; ok {
-			return v
-		}
-		v = f()
-		m.data[key] = v
-		return v
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFuncLock(key, f)
 }

 // SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *IntIntMap) SetIfNotExist(key int, value int) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, value)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExist(key, value)
 }

 // SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *IntIntMap) SetIfNotExistFunc(key int, f func() int) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f())
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFunc(key, f)
 }

 // SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
@ -272,126 +163,76 @@ func (m *IntIntMap) SetIfNotExistFunc(key int, f func() int) bool {
 // SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
 // it executes function `f` with mutex.Lock of the hash map.
 func (m *IntIntMap) SetIfNotExistFuncLock(key int, f func() int) bool {
-	if !m.Contains(key) {
-		m.mu.Lock()
-		defer m.mu.Unlock()
-		if m.data == nil {
-			m.data = make(map[int]int)
-		}
-		if _, ok := m.data[key]; !ok {
-			m.data[key] = f()
-		}
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFuncLock(key, f)
 }

 // Removes batch deletes values of the map by keys.
 func (m *IntIntMap) Removes(keys []int) {
-	m.mu.Lock()
-	if m.data != nil {
-		for _, key := range keys {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Removes(keys)
 }

 // Remove deletes value from map by given `key`, and return this deleted value.
 func (m *IntIntMap) Remove(key int) (value int) {
-	m.mu.Lock()
-	if m.data != nil {
-		var ok bool
-		if value, ok = m.data[key]; ok {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Remove(key)
 }

 // Keys returns all keys of the map as a slice.
 func (m *IntIntMap) Keys() []int {
-	m.mu.RLock()
-	var (
-		keys  = make([]int, len(m.data))
-		index = 0
-	)
-	for key := range m.data {
-		keys[index] = key
-		index++
-	}
-	m.mu.RUnlock()
-	return keys
+	m.lazyInit()
+	return m.KVMap.Keys()
 }

 // Values returns all values of the map as a slice.
 func (m *IntIntMap) Values() []int {
-	m.mu.RLock()
-	var (
-		values = make([]int, len(m.data))
-		index  = 0
-	)
-	for _, value := range m.data {
-		values[index] = value
-		index++
-	}
-	m.mu.RUnlock()
-	return values
+	m.lazyInit()
+	return m.KVMap.Values()
 }

 // Contains checks whether a key exists.
 // It returns true if the `key` exists, or else false.
 func (m *IntIntMap) Contains(key int) bool {
-	var ok bool
-	m.mu.RLock()
-	if m.data != nil {
-		_, ok = m.data[key]
-	}
-	m.mu.RUnlock()
-	return ok
+	m.lazyInit()
+	return m.KVMap.Contains(key)
 }

 // Size returns the size of the map.
 func (m *IntIntMap) Size() int {
-	m.mu.RLock()
-	length := len(m.data)
-	m.mu.RUnlock()
-	return length
+	m.lazyInit()
+	return m.KVMap.Size()
 }

 // IsEmpty checks whether the map is empty.
 // It returns true if map is empty, or else false.
 func (m *IntIntMap) IsEmpty() bool {
-	return m.Size() == 0
+	m.lazyInit()
+	return m.KVMap.IsEmpty()
 }

 // Clear deletes all data of the map, it will remake a new underlying data map.
 func (m *IntIntMap) Clear() {
-	m.mu.Lock()
-	m.data = make(map[int]int)
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Clear()
 }

 // Replace the data of the map with given `data`.
 func (m *IntIntMap) Replace(data map[int]int) {
-	m.mu.Lock()
-	m.data = data
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Replace(data)
 }

 // LockFunc locks writing with given callback function `f` within RWMutex.Lock.
 func (m *IntIntMap) LockFunc(f func(m map[int]int)) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.LockFunc(f)
 }

 // RLockFunc locks reading with given callback function `f` within RWMutex.RLock.
 func (m *IntIntMap) RLockFunc(f func(m map[int]int)) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.RLockFunc(f)
 }

 // Flip exchanges key-value of the map to value-key.
@ -408,19 +249,8 @@ func (m *IntIntMap) Flip() {
 // Merge merges two hash maps.
 // The `other` map will be merged into the map `m`.
 func (m *IntIntMap) Merge(other *IntIntMap) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = other.MapCopy()
-		return
-	}
-	if other != m {
-		other.mu.RLock()
-		defer other.mu.RUnlock()
-	}
-	for k, v := range other.data {
-		m.data[k] = v
-	}
+	m.lazyInit()
+	m.KVMap.Merge(other.KVMap)
 }

 // String returns the map as a string.
@ -428,81 +258,40 @@ func (m *IntIntMap) String() string {
 	if m == nil {
 		return ""
 	}
-	b, _ := m.MarshalJSON()
-	return string(b)
+	m.lazyInit()
+	return m.KVMap.String()
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (m IntIntMap) MarshalJSON() ([]byte, error) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	return json.Marshal(m.data)
+	m.lazyInit()
+	return m.KVMap.MarshalJSON()
 }

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (m *IntIntMap) UnmarshalJSON(b []byte) error {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[int]int)
-	}
-	if err := json.UnmarshalUseNumber(b, &m.data); err != nil {
-		return err
-	}
-	return nil
+	m.lazyInit()
+	return m.KVMap.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for map.
 func (m *IntIntMap) UnmarshalValue(value any) (err error) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[int]int)
-	}
-	switch value.(type) {
-	case string, []byte:
-		return json.UnmarshalUseNumber(gconv.Bytes(value), &m.data)
-	default:
-		for k, v := range gconv.Map(value) {
-			m.data[gconv.Int(k)] = gconv.Int(v)
-		}
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.UnmarshalValue(value)
 }

 // DeepCopy implements interface for deep copy of current type.
 func (m *IntIntMap) DeepCopy() any {
-	if m == nil {
-		return nil
+	m.lazyInit()
+	return &IntIntMap{
+		KVMap: m.KVMap.DeepCopy().(*KVMap[int, int]),
 	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[int]int, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return NewIntIntMapFrom(data, m.mu.IsSafe())
 }

 // IsSubOf checks whether the current map is a sub-map of `other`.
 func (m *IntIntMap) IsSubOf(other *IntIntMap) bool {
-	if m == other {
-		return true
-	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	for key, value := range m.data {
-		otherValue, ok := other.data[key]
-		if !ok {
-			return false
-		}
-		if otherValue != value {
-			return false
-		}
-	}
-	return true
+	m.lazyInit()
+	return m.KVMap.IsSubOf(other.KVMap)
 }

 // Diff compares current map `m` with map `other` and returns their different keys.
@ -510,22 +299,6 @@ func (m *IntIntMap) IsSubOf(other *IntIntMap) bool {
 // The returned `removedKeys` are the keys that are in map `other` but not in map `m`.
 // The returned `updatedKeys` are the keys that are both in map `m` and `other` but their values and not equal (`!=`).
 func (m *IntIntMap) Diff(other *IntIntMap) (addedKeys, removedKeys, updatedKeys []int) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-
-	for key := range m.data {
-		if _, ok := other.data[key]; !ok {
-			removedKeys = append(removedKeys, key)
-		} else if m.data[key] != other.data[key] {
-			updatedKeys = append(updatedKeys, key)
-		}
-	}
-	for key := range other.data {
-		if _, ok := m.data[key]; !ok {
-			addedKeys = append(addedKeys, key)
-		}
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Diff(other.KVMap)
 }
--- a/container/gmap/gmap_hash_int_str_map.go
+++ b/container/gmap/gmap_hash_int_str_map.go
@ -7,16 +7,15 @@
 package gmap

 import (
-	"github.com/gogf/gf/v2/internal/empty"
-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
+	"sync"
+
 	"github.com/gogf/gf/v2/util/gconv"
 )

 // IntStrMap implements map[int]string with RWMutex that has switch.
 type IntStrMap struct {
-	mu   rwmutex.RWMutex
-	data map[int]string
+	*KVMap[int, string]
+	once sync.Once
 }

 // NewIntStrMap returns an empty IntStrMap object.
@ -24,8 +23,7 @@ type IntStrMap struct {
 // which is false in default.
 func NewIntStrMap(safe ...bool) *IntStrMap {
 	return &IntStrMap{
-		mu:   rwmutex.Create(safe...),
-		data: make(map[int]string),
+		KVMap: NewKVMap[int, string](safe...),
 	}
 }

@ -34,193 +32,109 @@ func NewIntStrMap(safe ...bool) *IntStrMap {
 // there might be some concurrent-safe issues when changing the map outside.
 func NewIntStrMapFrom(data map[int]string, safe ...bool) *IntStrMap {
 	return &IntStrMap{
-		mu:   rwmutex.Create(safe...),
-		data: data,
+		KVMap: NewKVMapFrom(data, safe...),
 	}
 }

+// lazyInit lazily initializes the map.
+func (m *IntStrMap) lazyInit() {
+	m.once.Do(func() {
+		if m.KVMap == nil {
+			m.KVMap = NewKVMap[int, string](false)
+		}
+	})
+}
+
 // Iterator iterates the hash map readonly with custom callback function `f`.
 // If `f` returns true, then it continues iterating; or false to stop.
 func (m *IntStrMap) Iterator(f func(k int, v string) bool) {
-	for k, v := range m.Map() {
-		if !f(k, v) {
-			break
-		}
-	}
+	m.lazyInit()
+	m.KVMap.Iterator(f)
 }

 // Clone returns a new hash map with copy of current map data.
-func (m *IntStrMap) Clone() *IntStrMap {
-	return NewIntStrMapFrom(m.MapCopy(), m.mu.IsSafe())
+func (m *IntStrMap) Clone(safe ...bool) *IntStrMap {
+	m.lazyInit()
+	return &IntStrMap{KVMap: m.KVMap.Clone(safe...)}
 }

 // Map returns the underlying data map.
 // Note that, if it's in concurrent-safe usage, it returns a copy of underlying data,
 // or else a pointer to the underlying data.
 func (m *IntStrMap) Map() map[int]string {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	if !m.mu.IsSafe() {
-		return m.data
-	}
-	data := make(map[int]string, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.Map()
 }

 // MapStrAny returns a copy of the underlying data of the map as map[string]any.
 func (m *IntStrMap) MapStrAny() map[string]any {
-	m.mu.RLock()
-	data := make(map[string]any, len(m.data))
-	for k, v := range m.data {
-		data[gconv.String(k)] = v
-	}
-	m.mu.RUnlock()
-	return data
+	m.lazyInit()
+	return m.KVMap.MapStrAny()
 }

 // MapCopy returns a copy of the underlying data of the hash map.
 func (m *IntStrMap) MapCopy() map[int]string {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[int]string, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.MapCopy()
 }

 // FilterEmpty deletes all key-value pair of which the value is empty.
 // Values like: 0, nil, false, "", len(slice/map/chan) == 0 are considered empty.
 func (m *IntStrMap) FilterEmpty() {
-	m.mu.Lock()
-	for k, v := range m.data {
-		if empty.IsEmpty(v) {
-			delete(m.data, k)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.FilterEmpty()
 }

 // Set sets key-value to the hash map.
 func (m *IntStrMap) Set(key int, val string) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = make(map[int]string)
-	}
-	m.data[key] = val
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Set(key, val)
 }

 // Sets batch sets key-values to the hash map.
 func (m *IntStrMap) Sets(data map[int]string) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = data
-	} else {
-		for k, v := range data {
-			m.data[k] = v
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Sets(data)
 }

 // Search searches the map with given `key`.
 // Second return parameter `found` is true if key was found, otherwise false.
 func (m *IntStrMap) Search(key int) (value string, found bool) {
-	m.mu.RLock()
-	if m.data != nil {
-		value, found = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Search(key)
 }

 // Get returns the value by given `key`.
 func (m *IntStrMap) Get(key int) (value string) {
-	m.mu.RLock()
-	if m.data != nil {
-		value = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Get(key)
 }

 // Pop retrieves and deletes an item from the map.
 func (m *IntStrMap) Pop() (key int, value string) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for key, value = range m.data {
-		delete(m.data, key)
-		return
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Pop()
 }

 // Pops retrieves and deletes `size` items from the map.
 // It returns all items if size == -1.
 func (m *IntStrMap) Pops(size int) map[int]string {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if size > len(m.data) || size == -1 {
-		size = len(m.data)
-	}
-	if size == 0 {
-		return nil
-	}
-	var (
-		index  = 0
-		newMap = make(map[int]string, size)
-	)
-	for k, v := range m.data {
-		delete(m.data, k)
-		newMap[k] = v
-		index++
-		if index == size {
-			break
-		}
-	}
-	return newMap
-}
-
-// doSetWithLockCheck checks whether value of the key exists with mutex.Lock,
-// if not exists, set value to the map with given `key`,
-// or else just return the existing value.
-//
-// It returns value with given `key`.
-func (m *IntStrMap) doSetWithLockCheck(key int, value string) string {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[int]string)
-	}
-	if v, ok := m.data[key]; ok {
-		return v
-	}
-	m.data[key] = value
-	return value
+	m.lazyInit()
+	return m.KVMap.Pops(size)
 }

 // GetOrSet returns the value by key,
 // or sets value with given `value` if it does not exist and then returns this value.
 func (m *IntStrMap) GetOrSet(key int, value string) string {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, value)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSet(key, value)
 }

 // GetOrSetFunc returns the value by key,
 // or sets value with returned value of callback function `f` if it does not exist and returns this value.
 func (m *IntStrMap) GetOrSetFunc(key int, f func() string) string {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f())
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFunc(key, f)
 }

 // GetOrSetFuncLock returns the value by key,
@ -229,41 +143,22 @@ func (m *IntStrMap) GetOrSetFunc(key int, f func() string) string {
 // GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`
 // with mutex.Lock of the hash map.
 func (m *IntStrMap) GetOrSetFuncLock(key int, f func() string) string {
-	if v, ok := m.Search(key); !ok {
-		m.mu.Lock()
-		defer m.mu.Unlock()
-		if m.data == nil {
-			m.data = make(map[int]string)
-		}
-		if v, ok = m.data[key]; ok {
-			return v
-		}
-		v = f()
-		m.data[key] = v
-		return v
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFuncLock(key, f)
 }

 // SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *IntStrMap) SetIfNotExist(key int, value string) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, value)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExist(key, value)
 }

 // SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *IntStrMap) SetIfNotExistFunc(key int, f func() string) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f())
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFunc(key, f)
 }

 // SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
@ -272,126 +167,76 @@ func (m *IntStrMap) SetIfNotExistFunc(key int, f func() string) bool {
 // SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
 // it executes function `f` with mutex.Lock of the hash map.
 func (m *IntStrMap) SetIfNotExistFuncLock(key int, f func() string) bool {
-	if !m.Contains(key) {
-		m.mu.Lock()
-		defer m.mu.Unlock()
-		if m.data == nil {
-			m.data = make(map[int]string)
-		}
-		if _, ok := m.data[key]; !ok {
-			m.data[key] = f()
-		}
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFuncLock(key, f)
 }

 // Removes batch deletes values of the map by keys.
 func (m *IntStrMap) Removes(keys []int) {
-	m.mu.Lock()
-	if m.data != nil {
-		for _, key := range keys {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Removes(keys)
 }

 // Remove deletes value from map by given `key`, and return this deleted value.
 func (m *IntStrMap) Remove(key int) (value string) {
-	m.mu.Lock()
-	if m.data != nil {
-		var ok bool
-		if value, ok = m.data[key]; ok {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Remove(key)
 }

 // Keys returns all keys of the map as a slice.
 func (m *IntStrMap) Keys() []int {
-	m.mu.RLock()
-	var (
-		keys  = make([]int, len(m.data))
-		index = 0
-	)
-	for key := range m.data {
-		keys[index] = key
-		index++
-	}
-	m.mu.RUnlock()
-	return keys
+	m.lazyInit()
+	return m.KVMap.Keys()
 }

 // Values returns all values of the map as a slice.
 func (m *IntStrMap) Values() []string {
-	m.mu.RLock()
-	var (
-		values = make([]string, len(m.data))
-		index  = 0
-	)
-	for _, value := range m.data {
-		values[index] = value
-		index++
-	}
-	m.mu.RUnlock()
-	return values
+	m.lazyInit()
+	return m.KVMap.Values()
 }

 // Contains checks whether a key exists.
 // It returns true if the `key` exists, or else false.
 func (m *IntStrMap) Contains(key int) bool {
-	var ok bool
-	m.mu.RLock()
-	if m.data != nil {
-		_, ok = m.data[key]
-	}
-	m.mu.RUnlock()
-	return ok
+	m.lazyInit()
+	return m.KVMap.Contains(key)
 }

 // Size returns the size of the map.
 func (m *IntStrMap) Size() int {
-	m.mu.RLock()
-	length := len(m.data)
-	m.mu.RUnlock()
-	return length
+	m.lazyInit()
+	return m.KVMap.Size()
 }

 // IsEmpty checks whether the map is empty.
 // It returns true if map is empty, or else false.
 func (m *IntStrMap) IsEmpty() bool {
-	return m.Size() == 0
+	m.lazyInit()
+	return m.KVMap.IsEmpty()
 }

 // Clear deletes all data of the map, it will remake a new underlying data map.
 func (m *IntStrMap) Clear() {
-	m.mu.Lock()
-	m.data = make(map[int]string)
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Clear()
 }

 // Replace the data of the map with given `data`.
 func (m *IntStrMap) Replace(data map[int]string) {
-	m.mu.Lock()
-	m.data = data
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Replace(data)
 }

 // LockFunc locks writing with given callback function `f` within RWMutex.Lock.
 func (m *IntStrMap) LockFunc(f func(m map[int]string)) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.LockFunc(f)
 }

 // RLockFunc locks reading with given callback function `f` within RWMutex.RLock.
 func (m *IntStrMap) RLockFunc(f func(m map[int]string)) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.RLockFunc(f)
 }

 // Flip exchanges key-value of the map to value-key.
@ -408,19 +253,8 @@ func (m *IntStrMap) Flip() {
 // Merge merges two hash maps.
 // The `other` map will be merged into the map `m`.
 func (m *IntStrMap) Merge(other *IntStrMap) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = other.MapCopy()
-		return
-	}
-	if other != m {
-		other.mu.RLock()
-		defer other.mu.RUnlock()
-	}
-	for k, v := range other.data {
-		m.data[k] = v
-	}
+	m.lazyInit()
+	m.KVMap.Merge(other.KVMap)
 }

 // String returns the map as a string.
@ -428,81 +262,40 @@ func (m *IntStrMap) String() string {
 	if m == nil {
 		return ""
 	}
-	b, _ := m.MarshalJSON()
-	return string(b)
+	m.lazyInit()
+	return m.KVMap.String()
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (m IntStrMap) MarshalJSON() ([]byte, error) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	return json.Marshal(m.data)
+	m.lazyInit()
+	return m.KVMap.MarshalJSON()
 }

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (m *IntStrMap) UnmarshalJSON(b []byte) error {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[int]string)
-	}
-	if err := json.UnmarshalUseNumber(b, &m.data); err != nil {
-		return err
-	}
-	return nil
+	m.lazyInit()
+	return m.KVMap.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for map.
 func (m *IntStrMap) UnmarshalValue(value any) (err error) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[int]string)
-	}
-	switch value.(type) {
-	case string, []byte:
-		return json.UnmarshalUseNumber(gconv.Bytes(value), &m.data)
-	default:
-		for k, v := range gconv.Map(value) {
-			m.data[gconv.Int(k)] = gconv.String(v)
-		}
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.UnmarshalValue(value)
 }

 // DeepCopy implements interface for deep copy of current type.
 func (m *IntStrMap) DeepCopy() any {
-	if m == nil {
-		return nil
+	m.lazyInit()
+	return &IntStrMap{
+		KVMap: m.KVMap.DeepCopy().(*KVMap[int, string]),
 	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[int]string, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return NewIntStrMapFrom(data, m.mu.IsSafe())
 }

 // IsSubOf checks whether the current map is a sub-map of `other`.
 func (m *IntStrMap) IsSubOf(other *IntStrMap) bool {
-	if m == other {
-		return true
-	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	for key, value := range m.data {
-		otherValue, ok := other.data[key]
-		if !ok {
-			return false
-		}
-		if otherValue != value {
-			return false
-		}
-	}
-	return true
+	m.lazyInit()
+	return m.KVMap.IsSubOf(other.KVMap)
 }

 // Diff compares current map `m` with map `other` and returns their different keys.
@ -510,22 +303,6 @@ func (m *IntStrMap) IsSubOf(other *IntStrMap) bool {
 // The returned `removedKeys` are the keys that are in map `other` but not in map `m`.
 // The returned `updatedKeys` are the keys that are both in map `m` and `other` but their values and not equal (`!=`).
 func (m *IntStrMap) Diff(other *IntStrMap) (addedKeys, removedKeys, updatedKeys []int) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-
-	for key := range m.data {
-		if _, ok := other.data[key]; !ok {
-			removedKeys = append(removedKeys, key)
-		} else if m.data[key] != other.data[key] {
-			updatedKeys = append(updatedKeys, key)
-		}
-	}
-	for key := range other.data {
-		if _, ok := m.data[key]; !ok {
-			addedKeys = append(addedKeys, key)
-		}
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Diff(other.KVMap)
 }
--- a/container/gmap/gmap_hash_k_v_map.go
+++ b/container/gmap/gmap_hash_k_v_map.go
@ -0,0 +1,582 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package gmap
+
+import (
+	"reflect"
+
+	"github.com/gogf/gf/v2/container/gvar"
+	"github.com/gogf/gf/v2/internal/deepcopy"
+	"github.com/gogf/gf/v2/internal/empty"
+	"github.com/gogf/gf/v2/internal/json"
+	"github.com/gogf/gf/v2/internal/rwmutex"
+	"github.com/gogf/gf/v2/util/gconv"
+)
+
+// KVMap wraps map type `map[K]V` and provides more map features.
+type KVMap[K comparable, V any] struct {
+	mu   rwmutex.RWMutex
+	data map[K]V
+}
+
+// NewKVMap creates and returns an empty hash map.
+// The parameter `safe` is used to specify whether to use the map in concurrent-safety mode,
+// which is false by default.
+func NewKVMap[K comparable, V any](safe ...bool) *KVMap[K, V] {
+	return NewKVMapFrom(make(map[K]V), safe...)
+}
+
+// NewKVMapFrom creates and returns a hash map from given map `data`.
+// Note that, the param `data` map will be set as the underlying data map (no deep copy),
+// there might be some concurrent-safe issues when changing the map outside.
+func NewKVMapFrom[K comparable, V any](data map[K]V, safe ...bool) *KVMap[K, V] {
+	m := &KVMap[K, V]{
+		mu:   rwmutex.Create(safe...),
+		data: data,
+	}
+	return m
+}
+
+// Iterator iterates the hash map readonly with custom callback function `f`.
+// If `f` returns true, then it continues iterating; or false to stop.
+func (m *KVMap[K, V]) Iterator(f func(k K, v V) bool) {
+	for k, v := range m.Map() {
+		if !f(k, v) {
+			break
+		}
+	}
+}
+
+// Clone returns a new hash map with copy of current map data.
+func (m *KVMap[K, V]) Clone(safe ...bool) *KVMap[K, V] {
+	if len(safe) == 0 {
+		return NewKVMapFrom(m.MapCopy(), m.mu.IsSafe())
+	}
+	return NewKVMapFrom(m.MapCopy(), safe...)
+}
+
+// Map returns the underlying data map.
+// Note that, if it's in concurrent-safe usage, it returns a copy of underlying data,
+// or else a pointer to the underlying data.
+func (m *KVMap[K, V]) Map() map[K]V {
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	if !m.mu.IsSafe() {
+		return m.data
+	}
+	data := make(map[K]V, len(m.data))
+	for k, v := range m.data {
+		data[k] = v
+	}
+	return data
+}
+
+// MapCopy returns a shallow copy of the underlying data of the hash map.
+func (m *KVMap[K, V]) MapCopy() map[K]V {
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	data := make(map[K]V, len(m.data))
+	for k, v := range m.data {
+		data[k] = v
+	}
+	return data
+}
+
+// MapStrAny returns a copy of the underlying data of the map as map[string]any.
+func (m *KVMap[K, V]) MapStrAny() map[string]any {
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	data := make(map[string]any, len(m.data))
+	for k, v := range m.data {
+		data[gconv.String(k)] = v
+	}
+	return data
+}
+
+// FilterEmpty deletes all key-value pair of which the value is empty.
+// Values like: 0, nil, false, "", len(slice/map/chan) == 0 are considered empty.
+func (m *KVMap[K, V]) FilterEmpty() {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	for k, v := range m.data {
+		if empty.IsEmpty(v) {
+			delete(m.data, k)
+		}
+	}
+}
+
+// FilterNil deletes all key-value pair of which the value is nil.
+func (m *KVMap[K, V]) FilterNil() {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	for k, v := range m.data {
+		if empty.IsNil(v) {
+			delete(m.data, k)
+		}
+	}
+}
+
+// Set sets key-value to the hash map.
+func (m *KVMap[K, V]) Set(key K, value V) {
+	m.mu.Lock()
+	if m.data == nil {
+		m.data = make(map[K]V)
+	}
+	m.data[key] = value
+	m.mu.Unlock()
+}
+
+// Sets batch sets key-values to the hash map.
+func (m *KVMap[K, V]) Sets(data map[K]V) {
+	m.mu.Lock()
+	if m.data == nil {
+		m.data = data
+	} else {
+		for k, v := range data {
+			m.data[k] = v
+		}
+	}
+	m.mu.Unlock()
+}
+
+// Search searches the map with given `key`.
+// Second return parameter `found` is true if key was found, otherwise false.
+func (m *KVMap[K, V]) Search(key K) (value V, found bool) {
+	m.mu.RLock()
+	if m.data != nil {
+		value, found = m.data[key]
+	}
+	m.mu.RUnlock()
+	return
+}
+
+// Get returns the value by given `key`.
+func (m *KVMap[K, V]) Get(key K) (value V) {
+	m.mu.RLock()
+	if m.data != nil {
+		value = m.data[key]
+	}
+	m.mu.RUnlock()
+	return
+}
+
+// Pop retrieves and deletes an item from the map.
+func (m *KVMap[K, V]) Pop() (key K, value V) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	for key, value = range m.data {
+		delete(m.data, key)
+		return
+	}
+	return
+}
+
+// Pops retrieves and deletes `size` items from the map.
+// It returns all items if size == -1.
+func (m *KVMap[K, V]) Pops(size int) map[K]V {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if size > len(m.data) || size == -1 {
+		size = len(m.data)
+	}
+	if size == 0 {
+		return nil
+	}
+	var (
+		index  = 0
+		newMap = make(map[K]V, size)
+	)
+	for k, v := range m.data {
+		delete(m.data, k)
+		newMap[k] = v
+		index++
+		if index == size {
+			break
+		}
+	}
+	return newMap
+}
+
+// doSetWithLockCheck checks whether value of the key exists with mutex.Lock,
+// if not exists, set value to the map with given `key`,
+// or else just return the existing value.
+//
+// It returns value with given `key`.
+func (m *KVMap[K, V]) doSetWithLockCheck(key K, value V) (val V, ok bool) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if m.data == nil {
+		m.data = make(map[K]V)
+	}
+
+	if v, ok := m.data[key]; ok {
+		return v, true
+	}
+
+	if any(value) != nil {
+		m.data[key] = value
+	}
+	return value, false
+}
+
+// GetOrSet returns the value by key,
+// or sets value with given `value` if it does not exist and then returns this value.
+func (m *KVMap[K, V]) GetOrSet(key K, value V) V {
+	v, _ := m.doSetWithLockCheck(key, value)
+	return v
+}
+
+// GetOrSetFunc returns the value by key,
+// or sets value with returned value of callback function `f` if it does not exist
+// and then returns this value.
+func (m *KVMap[K, V]) GetOrSetFunc(key K, f func() V) V {
+	v, _ := m.doSetWithLockCheck(key, f())
+	return v
+}
+
+// GetOrSetFuncLock returns the value by key,
+// or sets value with returned value of callback function `f` if it does not exist
+// and then returns this value.
+//
+// GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`
+// with mutex.Lock of the hash map.
+func (m *KVMap[K, V]) GetOrSetFuncLock(key K, f func() V) V {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.data == nil {
+		m.data = make(map[K]V)
+	}
+	if v, ok := m.data[key]; ok {
+		return v
+	}
+	value := f()
+	if any(value) != nil {
+		m.data[key] = value
+	}
+	return value
+}
+
+// GetVar returns a Var with the value by given `key`.
+// The returned Var is un-concurrent safe.
+func (m *KVMap[K, V]) GetVar(key K) *gvar.Var {
+	return gvar.New(m.Get(key))
+}
+
+// GetVarOrSet returns a Var with result from GetOrSet.
+// The returned Var is un-concurrent safe.
+func (m *KVMap[K, V]) GetVarOrSet(key K, value V) *gvar.Var {
+	return gvar.New(m.GetOrSet(key, value))
+}
+
+// GetVarOrSetFunc returns a Var with result from GetOrSetFunc.
+// The returned Var is un-concurrent safe.
+func (m *KVMap[K, V]) GetVarOrSetFunc(key K, f func() V) *gvar.Var {
+	return gvar.New(m.GetOrSetFunc(key, f))
+}
+
+// GetVarOrSetFuncLock returns a Var with result from GetOrSetFuncLock.
+// The returned Var is un-concurrent safe.
+func (m *KVMap[K, V]) GetVarOrSetFuncLock(key K, f func() V) *gvar.Var {
+	return gvar.New(m.GetOrSetFuncLock(key, f))
+}
+
+// SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
+// It returns false if `key` exists, and `value` would be ignored.
+func (m *KVMap[K, V]) SetIfNotExist(key K, value V) bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.data == nil {
+		m.data = make(map[K]V)
+	}
+	if _, ok := m.data[key]; !ok {
+		m.data[key] = value
+		return true
+	}
+	return false
+}
+
+// SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
+// It returns false if `key` exists, and `value` would be ignored.
+func (m *KVMap[K, V]) SetIfNotExistFunc(key K, f func() V) bool {
+	if !m.Contains(key) {
+		return m.SetIfNotExist(key, f())
+	}
+	return false
+}
+
+// SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
+// It returns false if `key` exists, and `value` would be ignored.
+//
+// SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
+// it executes function `f` with mutex.Lock of the hash map.
+func (m *KVMap[K, V]) SetIfNotExistFuncLock(key K, f func() V) bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.data == nil {
+		m.data = make(map[K]V)
+	}
+	if _, ok := m.data[key]; !ok {
+		m.data[key] = f()
+		return true
+	}
+	return false
+}
+
+// Remove deletes value from map by given `key`, and return this deleted value.
+func (m *KVMap[K, V]) Remove(key K) (value V) {
+	m.mu.Lock()
+	if m.data != nil {
+		var ok bool
+		if value, ok = m.data[key]; ok {
+			delete(m.data, key)
+		}
+	}
+	m.mu.Unlock()
+	return
+}
+
+// Removes batch deletes values of the map by keys.
+func (m *KVMap[K, V]) Removes(keys []K) {
+	m.mu.Lock()
+	if m.data != nil {
+		for _, key := range keys {
+			delete(m.data, key)
+		}
+	}
+	m.mu.Unlock()
+}
+
+// Keys returns all keys of the map as a slice.
+func (m *KVMap[K, V]) Keys() []K {
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	var (
+		keys  = make([]K, len(m.data))
+		index = 0
+	)
+	for key := range m.data {
+		keys[index] = key
+		index++
+	}
+	return keys
+}
+
+// Values returns all values of the map as a slice.
+func (m *KVMap[K, V]) Values() []V {
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	var (
+		values = make([]V, len(m.data))
+		index  = 0
+	)
+	for _, value := range m.data {
+		values[index] = value
+		index++
+	}
+	return values
+}
+
+// Contains checks whether a key exists.
+// It returns true if the `key` exists, or else false.
+func (m *KVMap[K, V]) Contains(key K) bool {
+	var ok bool
+	m.mu.RLock()
+	if m.data != nil {
+		_, ok = m.data[key]
+	}
+	m.mu.RUnlock()
+	return ok
+}
+
+// Size returns the size of the map.
+func (m *KVMap[K, V]) Size() int {
+	m.mu.RLock()
+	length := len(m.data)
+	m.mu.RUnlock()
+	return length
+}
+
+// IsEmpty checks whether the map is empty.
+// It returns true if map is empty, or else false.
+func (m *KVMap[K, V]) IsEmpty() bool {
+	return m.Size() == 0
+}
+
+// Clear deletes all data of the map, it will remake a new underlying data map.
+func (m *KVMap[K, V]) Clear() {
+	m.mu.Lock()
+	m.data = make(map[K]V)
+	m.mu.Unlock()
+}
+
+// Replace the data of the map with given `data`.
+func (m *KVMap[K, V]) Replace(data map[K]V) {
+	m.mu.Lock()
+	m.data = data
+	m.mu.Unlock()
+}
+
+// LockFunc locks writing with given callback function `f` within RWMutex.Lock.
+func (m *KVMap[K, V]) LockFunc(f func(m map[K]V)) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	f(m.data)
+}
+
+// RLockFunc locks reading with given callback function `f` within RWMutex.RLock.
+func (m *KVMap[K, V]) RLockFunc(f func(m map[K]V)) {
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	f(m.data)
+}
+
+// Flip exchanges key-value of the map to value-key.
+func (m *KVMap[K, V]) Flip() {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	n := make(map[K]V, len(m.data))
+	for k, v := range m.data {
+		var (
+			k0 K
+			v0 V
+		)
+		if err := gconv.Scan(v, &k0); err != nil {
+			continue
+		}
+		if err := gconv.Scan(k, &v0); err != nil {
+			continue
+		}
+		n[k0] = v0
+	}
+	m.data = n
+}
+
+// Merge merges two hash maps.
+// The `other` map will be merged into the map `m`.
+func (m *KVMap[K, V]) Merge(other *KVMap[K, V]) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.data == nil {
+		m.data = other.MapCopy()
+		return
+	}
+	if other != m {
+		other.mu.RLock()
+		defer other.mu.RUnlock()
+	}
+	for k, v := range other.data {
+		m.data[k] = v
+	}
+}
+
+// String returns the map as a string.
+func (m *KVMap[K, V]) String() string {
+	if m == nil {
+		return ""
+	}
+	b, _ := m.MarshalJSON()
+	return string(b)
+}
+
+// MarshalJSON implements the interface MarshalJSON for json.Marshal.
+func (m KVMap[K, V]) MarshalJSON() ([]byte, error) {
+	return json.Marshal(gconv.Map(m.Map()))
+}
+
+// UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
+func (m *KVMap[K, V]) UnmarshalJSON(b []byte) error {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.data == nil {
+		m.data = make(map[K]V)
+	}
+	var data map[string]V
+	if err := json.UnmarshalUseNumber(b, &data); err != nil {
+		return err
+	}
+	if err := gconv.Scan(data, &m.data); err != nil {
+		return err
+	}
+	return nil
+}
+
+// UnmarshalValue is an interface implement which sets any type of value for map.
+func (m *KVMap[K, V]) UnmarshalValue(value any) (err error) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.data == nil {
+		m.data = make(map[K]V)
+	}
+	data := gconv.Map(value)
+	if err := gconv.Scan(data, &m.data); err != nil {
+		return err
+	}
+	return
+}
+
+// DeepCopy implements interface for deep copy of current type.
+func (m *KVMap[K, V]) DeepCopy() any {
+	if m == nil {
+		return nil
+	}
+
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	data := make(map[K]V, len(m.data))
+	for k, v := range m.data {
+		data[k] = deepcopy.Copy(v).(V)
+	}
+	return NewKVMapFrom(data, m.mu.IsSafe())
+}
+
+// IsSubOf checks whether the current map is a sub-map of `other`.
+func (m *KVMap[K, V]) IsSubOf(other *KVMap[K, V]) bool {
+	if m == other {
+		return true
+	}
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	other.mu.RLock()
+	defer other.mu.RUnlock()
+	for key, value := range m.data {
+		otherValue, ok := other.data[key]
+		if !ok {
+			return false
+		}
+
+		if !reflect.DeepEqual(otherValue, value) {
+			return false
+		}
+	}
+	return true
+}
+
+// Diff compares current map `m` with map `other` and returns their different keys.
+// The returned `addedKeys` are the keys that are in map `m` but not in map `other`.
+// The returned `removedKeys` are the keys that are in map `other` but not in map `m`.
+// The returned `updatedKeys` are the keys that are both in map `m` and `other` but their values and not equal (`!=`).
+func (m *KVMap[K, V]) Diff(other *KVMap[K, V]) (addedKeys, removedKeys, updatedKeys []K) {
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	other.mu.RLock()
+	defer other.mu.RUnlock()
+
+	for key := range m.data {
+		if _, ok := other.data[key]; !ok {
+			removedKeys = append(removedKeys, key)
+		} else if !reflect.DeepEqual(m.data[key], other.data[key]) {
+			updatedKeys = append(updatedKeys, key)
+		}
+	}
+	for key := range other.data {
+		if _, ok := m.data[key]; !ok {
+			addedKeys = append(addedKeys, key)
+		}
+	}
+	return
+}
--- a/container/gmap/gmap_hash_str_any_map.go
+++ b/container/gmap/gmap_hash_str_any_map.go
@ -8,71 +8,66 @@
 package gmap

 import (
-	"reflect"
+	"sync"

 	"github.com/gogf/gf/v2/container/gvar"
-	"github.com/gogf/gf/v2/internal/deepcopy"
-	"github.com/gogf/gf/v2/internal/empty"
-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
 	"github.com/gogf/gf/v2/util/gconv"
 )

 // StrAnyMap implements map[string]any with RWMutex that has switch.
 type StrAnyMap struct {
-	mu   rwmutex.RWMutex
-	data map[string]any
+	*KVMap[string, any]
+	once sync.Once
 }

 // NewStrAnyMap returns an empty StrAnyMap object.
 // The parameter `safe` is used to specify whether using map in concurrent-safety,
 // which is false in default.
 func NewStrAnyMap(safe ...bool) *StrAnyMap {
-	return &StrAnyMap{
-		mu:   rwmutex.Create(safe...),
-		data: make(map[string]any),
+	m := &StrAnyMap{
+		KVMap: NewKVMap[string, any](safe...),
 	}
+	return m
 }

 // NewStrAnyMapFrom creates and returns a hash map from given map `data`.
 // Note that, the param `data` map will be set as the underlying data map(no deep copy),
 // there might be some concurrent-safe issues when changing the map outside.
 func NewStrAnyMapFrom(data map[string]any, safe ...bool) *StrAnyMap {
-	return &StrAnyMap{
-		mu:   rwmutex.Create(safe...),
-		data: data,
+	m := &StrAnyMap{
+		KVMap: NewKVMapFrom(data, safe...),
 	}
+	return m
+}
+
+// lazyInit lazily initializes the map.
+func (m *StrAnyMap) lazyInit() {
+	m.once.Do(func() {
+		if m.KVMap == nil {
+			m.KVMap = NewKVMap[string, any](false)
+		}
+	})
 }

 // Iterator iterates the hash map readonly with custom callback function `f`.
 // If `f` returns true, then it continues iterating; or false to stop.
 func (m *StrAnyMap) Iterator(f func(k string, v any) bool) {
-	for k, v := range m.Map() {
-		if !f(k, v) {
-			break
-		}
-	}
+	m.lazyInit()
+	m.KVMap.Iterator(f)
 }

 // Clone returns a new hash map with copy of current map data.
-func (m *StrAnyMap) Clone() *StrAnyMap {
-	return NewStrAnyMapFrom(m.MapCopy(), m.mu.IsSafe())
+func (m *StrAnyMap) Clone(safe ...bool) *StrAnyMap {
+	m.lazyInit()
+	return NewStrAnyMapFrom(m.MapCopy(), safe...)
 }

 // Map returns the underlying data map.
 // Note that, if it's in concurrent-safe usage, it returns a copy of underlying data,
 // or else a pointer to the underlying data.
 func (m *StrAnyMap) Map() map[string]any {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	if !m.mu.IsSafe() {
-		return m.data
-	}
-	data := make(map[string]any, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.Map()
 }

 // MapStrAny returns a copy of the underlying data of the map as map[string]any.
@ -82,165 +77,74 @@ func (m *StrAnyMap) MapStrAny() map[string]any {

 // MapCopy returns a copy of the underlying data of the hash map.
 func (m *StrAnyMap) MapCopy() map[string]any {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[string]any, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.MapCopy()
 }

 // FilterEmpty deletes all key-value pair of which the value is empty.
 // Values like: 0, nil, false, "", len(slice/map/chan) == 0 are considered empty.
 func (m *StrAnyMap) FilterEmpty() {
-	m.mu.Lock()
-	for k, v := range m.data {
-		if empty.IsEmpty(v) {
-			delete(m.data, k)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.FilterEmpty()
 }

 // FilterNil deletes all key-value pair of which the value is nil.
 func (m *StrAnyMap) FilterNil() {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for k, v := range m.data {
-		if empty.IsNil(v) {
-			delete(m.data, k)
-		}
-	}
+	m.lazyInit()
+	m.KVMap.FilterNil()
 }

 // Set sets key-value to the hash map.
 func (m *StrAnyMap) Set(key string, val any) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = make(map[string]any)
-	}
-	m.data[key] = val
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Set(key, val)
 }

 // Sets batch sets key-values to the hash map.
 func (m *StrAnyMap) Sets(data map[string]any) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = data
-	} else {
-		for k, v := range data {
-			m.data[k] = v
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Sets(data)
 }

 // Search searches the map with given `key`.
 // Second return parameter `found` is true if key was found, otherwise false.
 func (m *StrAnyMap) Search(key string) (value any, found bool) {
-	m.mu.RLock()
-	if m.data != nil {
-		value, found = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Search(key)
 }

 // Get returns the value by given `key`.
 func (m *StrAnyMap) Get(key string) (value any) {
-	m.mu.RLock()
-	if m.data != nil {
-		value = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Get(key)
 }

 // Pop retrieves and deletes an item from the map.
 func (m *StrAnyMap) Pop() (key string, value any) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for key, value = range m.data {
-		delete(m.data, key)
-		return
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Pop()
 }

 // Pops retrieves and deletes `size` items from the map.
 // It returns all items if size == -1.
 func (m *StrAnyMap) Pops(size int) map[string]any {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if size > len(m.data) || size == -1 {
-		size = len(m.data)
-	}
-	if size == 0 {
-		return nil
-	}
-	var (
-		index  = 0
-		newMap = make(map[string]any, size)
-	)
-	for k, v := range m.data {
-		delete(m.data, k)
-		newMap[k] = v
-		index++
-		if index == size {
-			break
-		}
-	}
-	return newMap
-}
-
-// doSetWithLockCheck checks whether value of the key exists with mutex.Lock,
-// if not exists, set value to the map with given `key`,
-// or else just return the existing value.
-//
-// When setting value, if `value` is type of `func() interface {}`,
-// it will be executed with mutex.Lock of the hash map,
-// and its return value will be set to the map with `key`.
-//
-// It returns value with given `key`.
-func (m *StrAnyMap) doSetWithLockCheck(key string, value any) any {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[string]any)
-	}
-	if v, ok := m.data[key]; ok {
-		return v
-	}
-	if f, ok := value.(func() any); ok {
-		value = f()
-	}
-	if value != nil {
-		m.data[key] = value
-	}
-	return value
+	m.lazyInit()
+	return m.KVMap.Pops(size)
 }

 // GetOrSet returns the value by key,
 // or sets value with given `value` if it does not exist and then returns this value.
 func (m *StrAnyMap) GetOrSet(key string, value any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, value)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSet(key, value)
 }

 // GetOrSetFunc returns the value by key,
 // or sets value with returned value of callback function `f` if it does not exist
 // and then returns this value.
 func (m *StrAnyMap) GetOrSetFunc(key string, f func() any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f())
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFunc(key, f)
 }

 // GetOrSetFuncLock returns the value by key,
@ -250,55 +154,50 @@ func (m *StrAnyMap) GetOrSetFunc(key string, f func() any) any {
 // GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`
 // with mutex.Lock of the hash map.
 func (m *StrAnyMap) GetOrSetFuncLock(key string, f func() any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFuncLock(key, f)
 }

 // GetVar returns a Var with the value by given `key`.
 // The returned Var is un-concurrent safe.
 func (m *StrAnyMap) GetVar(key string) *gvar.Var {
-	return gvar.New(m.Get(key))
+	m.lazyInit()
+	return m.KVMap.GetVar(key)
 }

 // GetVarOrSet returns a Var with result from GetVarOrSet.
 // The returned Var is un-concurrent safe.
 func (m *StrAnyMap) GetVarOrSet(key string, value any) *gvar.Var {
-	return gvar.New(m.GetOrSet(key, value))
+	m.lazyInit()
+	return m.KVMap.GetVarOrSet(key, value)
 }

 // GetVarOrSetFunc returns a Var with result from GetOrSetFunc.
 // The returned Var is un-concurrent safe.
 func (m *StrAnyMap) GetVarOrSetFunc(key string, f func() any) *gvar.Var {
-	return gvar.New(m.GetOrSetFunc(key, f))
+	m.lazyInit()
+	return m.KVMap.GetVarOrSetFunc(key, f)
 }

 // GetVarOrSetFuncLock returns a Var with result from GetOrSetFuncLock.
 // The returned Var is un-concurrent safe.
 func (m *StrAnyMap) GetVarOrSetFuncLock(key string, f func() any) *gvar.Var {
-	return gvar.New(m.GetOrSetFuncLock(key, f))
+	m.lazyInit()
+	return m.KVMap.GetVarOrSetFuncLock(key, f)
 }

 // SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *StrAnyMap) SetIfNotExist(key string, value any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, value)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExist(key, value)
 }

 // SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *StrAnyMap) SetIfNotExistFunc(key string, f func() any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f())
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFunc(key, f)
 }

 // SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
@ -307,119 +206,76 @@ func (m *StrAnyMap) SetIfNotExistFunc(key string, f func() any) bool {
 // SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
 // it executes function `f` with mutex.Lock of the hash map.
 func (m *StrAnyMap) SetIfNotExistFuncLock(key string, f func() any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFuncLock(key, f)
 }

 // Removes batch deletes values of the map by keys.
 func (m *StrAnyMap) Removes(keys []string) {
-	m.mu.Lock()
-	if m.data != nil {
-		for _, key := range keys {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Removes(keys)
 }

 // Remove deletes value from map by given `key`, and return this deleted value.
 func (m *StrAnyMap) Remove(key string) (value any) {
-	m.mu.Lock()
-	if m.data != nil {
-		var ok bool
-		if value, ok = m.data[key]; ok {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Remove(key)
 }

 // Keys returns all keys of the map as a slice.
 func (m *StrAnyMap) Keys() []string {
-	m.mu.RLock()
-	var (
-		keys  = make([]string, len(m.data))
-		index = 0
-	)
-	for key := range m.data {
-		keys[index] = key
-		index++
-	}
-	m.mu.RUnlock()
-	return keys
+	m.lazyInit()
+	return m.KVMap.Keys()
 }

 // Values returns all values of the map as a slice.
 func (m *StrAnyMap) Values() []any {
-	m.mu.RLock()
-	var (
-		values = make([]any, len(m.data))
-		index  = 0
-	)
-	for _, value := range m.data {
-		values[index] = value
-		index++
-	}
-	m.mu.RUnlock()
-	return values
+	m.lazyInit()
+	return m.KVMap.Values()
 }

 // Contains checks whether a key exists.
 // It returns true if the `key` exists, or else false.
 func (m *StrAnyMap) Contains(key string) bool {
-	var ok bool
-	m.mu.RLock()
-	if m.data != nil {
-		_, ok = m.data[key]
-	}
-	m.mu.RUnlock()
-	return ok
+	m.lazyInit()
+	return m.KVMap.Contains(key)
 }

 // Size returns the size of the map.
 func (m *StrAnyMap) Size() int {
-	m.mu.RLock()
-	length := len(m.data)
-	m.mu.RUnlock()
-	return length
+	m.lazyInit()
+	return m.KVMap.Size()
 }

 // IsEmpty checks whether the map is empty.
 // It returns true if map is empty, or else false.
 func (m *StrAnyMap) IsEmpty() bool {
-	return m.Size() == 0
+	m.lazyInit()
+	return m.KVMap.IsEmpty()
 }

 // Clear deletes all data of the map, it will remake a new underlying data map.
 func (m *StrAnyMap) Clear() {
-	m.mu.Lock()
-	m.data = make(map[string]any)
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Clear()
 }

 // Replace the data of the map with given `data`.
 func (m *StrAnyMap) Replace(data map[string]any) {
-	m.mu.Lock()
-	m.data = data
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Replace(data)
 }

 // LockFunc locks writing with given callback function `f` within RWMutex.Lock.
 func (m *StrAnyMap) LockFunc(f func(m map[string]any)) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.LockFunc(f)
 }

 // RLockFunc locks reading with given callback function `f` within RWMutex.RLock.
 func (m *StrAnyMap) RLockFunc(f func(m map[string]any)) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.RLockFunc(f)
 }

 // Flip exchanges key-value of the map to value-key.
@ -436,19 +292,8 @@ func (m *StrAnyMap) Flip() {
 // Merge merges two hash maps.
 // The `other` map will be merged into the map `m`.
 func (m *StrAnyMap) Merge(other *StrAnyMap) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = other.MapCopy()
-		return
-	}
-	if other != m {
-		other.mu.RLock()
-		defer other.mu.RUnlock()
-	}
-	for k, v := range other.data {
-		m.data[k] = v
-	}
+	m.lazyInit()
+	m.KVMap.Merge(other.KVMap)
 }

 // String returns the map as a string.
@ -456,71 +301,40 @@ func (m *StrAnyMap) String() string {
 	if m == nil {
 		return ""
 	}
-	b, _ := m.MarshalJSON()
-	return string(b)
+	m.lazyInit()
+	return m.KVMap.String()
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (m StrAnyMap) MarshalJSON() ([]byte, error) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	return json.Marshal(m.data)
+	m.lazyInit()
+	return m.KVMap.MarshalJSON()
 }

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (m *StrAnyMap) UnmarshalJSON(b []byte) error {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[string]any)
-	}
-	if err := json.UnmarshalUseNumber(b, &m.data); err != nil {
-		return err
-	}
-	return nil
+	m.lazyInit()
+	return m.KVMap.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for map.
 func (m *StrAnyMap) UnmarshalValue(value any) (err error) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	m.data = gconv.Map(value)
-	return
+	m.lazyInit()
+	return m.KVMap.UnmarshalValue(value)
 }

 // DeepCopy implements interface for deep copy of current type.
 func (m *StrAnyMap) DeepCopy() any {
-	if m == nil {
-		return nil
+	m.lazyInit()
+	return &StrAnyMap{
+		KVMap: m.KVMap.DeepCopy().(*KVMap[string, any]),
 	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[string]any, len(m.data))
-	for k, v := range m.data {
-		data[k] = deepcopy.Copy(v)
-	}
-	return NewStrAnyMapFrom(data, m.mu.IsSafe())
 }

 // IsSubOf checks whether the current map is a sub-map of `other`.
 func (m *StrAnyMap) IsSubOf(other *StrAnyMap) bool {
-	if m == other {
-		return true
-	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	for key, value := range m.data {
-		otherValue, ok := other.data[key]
-		if !ok {
-			return false
-		}
-		if otherValue != value {
-			return false
-		}
-	}
-	return true
+	m.lazyInit()
+	return m.KVMap.IsSubOf(other.KVMap)
 }

 // Diff compares current map `m` with map `other` and returns their different keys.
@ -528,22 +342,6 @@ func (m *StrAnyMap) IsSubOf(other *StrAnyMap) bool {
 // The returned `removedKeys` are the keys that are in map `other` but not in map `m`.
 // The returned `updatedKeys` are the keys that are both in map `m` and `other` but their values and not equal (`!=`).
 func (m *StrAnyMap) Diff(other *StrAnyMap) (addedKeys, removedKeys, updatedKeys []string) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-
-	for key := range m.data {
-		if _, ok := other.data[key]; !ok {
-			removedKeys = append(removedKeys, key)
-		} else if !reflect.DeepEqual(m.data[key], other.data[key]) {
-			updatedKeys = append(updatedKeys, key)
-		}
-	}
-	for key := range other.data {
-		if _, ok := m.data[key]; !ok {
-			addedKeys = append(addedKeys, key)
-		}
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Diff(other.KVMap)
 }
--- a/container/gmap/gmap_hash_str_int_map.go
+++ b/container/gmap/gmap_hash_str_int_map.go
@ -8,16 +8,15 @@
 package gmap

 import (
-	"github.com/gogf/gf/v2/internal/empty"
-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
+	"sync"
+
 	"github.com/gogf/gf/v2/util/gconv"
 )

 // StrIntMap implements map[string]int with RWMutex that has switch.
 type StrIntMap struct {
-	mu   rwmutex.RWMutex
-	data map[string]int
+	*KVMap[string, int]
+	once sync.Once
 }

 // NewStrIntMap returns an empty StrIntMap object.
@ -25,8 +24,7 @@ type StrIntMap struct {
 // which is false in default.
 func NewStrIntMap(safe ...bool) *StrIntMap {
 	return &StrIntMap{
-		mu:   rwmutex.Create(safe...),
-		data: make(map[string]int),
+		KVMap: NewKVMap[string, int](safe...),
 	}
 }

@ -35,195 +33,110 @@ func NewStrIntMap(safe ...bool) *StrIntMap {
 // there might be some concurrent-safe issues when changing the map outside.
 func NewStrIntMapFrom(data map[string]int, safe ...bool) *StrIntMap {
 	return &StrIntMap{
-		mu:   rwmutex.Create(safe...),
-		data: data,
+		KVMap: NewKVMapFrom(data, safe...),
 	}
 }

+// lazyInit lazily initializes the map.
+func (m *StrIntMap) lazyInit() {
+	m.once.Do(func() {
+		if m.KVMap == nil {
+			m.KVMap = NewKVMap[string, int](false)
+		}
+	})
+}
+
 // Iterator iterates the hash map readonly with custom callback function `f`.
 // If `f` returns true, then it continues iterating; or false to stop.
 func (m *StrIntMap) Iterator(f func(k string, v int) bool) {
-	for k, v := range m.Map() {
-		if !f(k, v) {
-			break
-		}
-	}
+	m.lazyInit()
+	m.KVMap.Iterator(f)
 }

 // Clone returns a new hash map with copy of current map data.
-func (m *StrIntMap) Clone() *StrIntMap {
-	return NewStrIntMapFrom(m.MapCopy(), m.mu.IsSafe())
+func (m *StrIntMap) Clone(safe ...bool) *StrIntMap {
+	m.lazyInit()
+	return &StrIntMap{KVMap: m.KVMap.Clone(safe...)}
 }

 // Map returns the underlying data map.
 // Note that, if it's in concurrent-safe usage, it returns a copy of underlying data,
 // or else a pointer to the underlying data.
 func (m *StrIntMap) Map() map[string]int {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	if !m.mu.IsSafe() {
-		return m.data
-	}
-	data := make(map[string]int, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.Map()
 }

 // MapStrAny returns a copy of the underlying data of the map as map[string]any.
 func (m *StrIntMap) MapStrAny() map[string]any {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[string]any, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.MapStrAny()
 }

 // MapCopy returns a copy of the underlying data of the hash map.
 func (m *StrIntMap) MapCopy() map[string]int {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[string]int, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.MapCopy()
 }

 // FilterEmpty deletes all key-value pair of which the value is empty.
 // Values like: 0, nil, false, "", len(slice/map/chan) == 0 are considered empty.
 func (m *StrIntMap) FilterEmpty() {
-	m.mu.Lock()
-	for k, v := range m.data {
-		if empty.IsEmpty(v) {
-			delete(m.data, k)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.FilterEmpty()
 }

 // Set sets key-value to the hash map.
 func (m *StrIntMap) Set(key string, val int) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = make(map[string]int)
-	}
-	m.data[key] = val
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Set(key, val)
 }

 // Sets batch sets key-values to the hash map.
 func (m *StrIntMap) Sets(data map[string]int) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = data
-	} else {
-		for k, v := range data {
-			m.data[k] = v
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Sets(data)
 }

 // Search searches the map with given `key`.
 // Second return parameter `found` is true if key was found, otherwise false.
 func (m *StrIntMap) Search(key string) (value int, found bool) {
-	m.mu.RLock()
-	if m.data != nil {
-		value, found = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Search(key)
 }

 // Get returns the value by given `key`.
 func (m *StrIntMap) Get(key string) (value int) {
-	m.mu.RLock()
-	if m.data != nil {
-		value = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Get(key)
 }

 // Pop retrieves and deletes an item from the map.
 func (m *StrIntMap) Pop() (key string, value int) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for key, value = range m.data {
-		delete(m.data, key)
-		return
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Pop()
 }

 // Pops retrieves and deletes `size` items from the map.
 // It returns all items if size == -1.
 func (m *StrIntMap) Pops(size int) map[string]int {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if size > len(m.data) || size == -1 {
-		size = len(m.data)
-	}
-	if size == 0 {
-		return nil
-	}
-	var (
-		index  = 0
-		newMap = make(map[string]int, size)
-	)
-	for k, v := range m.data {
-		delete(m.data, k)
-		newMap[k] = v
-		index++
-		if index == size {
-			break
-		}
-	}
-	return newMap
-}
-
-// doSetWithLockCheck checks whether value of the key exists with mutex.Lock,
-// if not exists, set value to the map with given `key`,
-// or else just return the existing value.
-//
-// It returns value with given `key`.
-func (m *StrIntMap) doSetWithLockCheck(key string, value int) int {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = make(map[string]int)
-	}
-	if v, ok := m.data[key]; ok {
-		m.mu.Unlock()
-		return v
-	}
-	m.data[key] = value
-	m.mu.Unlock()
-	return value
+	m.lazyInit()
+	return m.KVMap.Pops(size)
 }

 // GetOrSet returns the value by key,
 // or sets value with given `value` if it does not exist and then returns this value.
 func (m *StrIntMap) GetOrSet(key string, value int) int {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, value)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSet(key, value)
 }

 // GetOrSetFunc returns the value by key,
 // or sets value with returned value of callback function `f` if it does not exist
 // and then returns this value.
 func (m *StrIntMap) GetOrSetFunc(key string, f func() int) int {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f())
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFunc(key, f)
 }

 // GetOrSetFuncLock returns the value by key,
@ -233,41 +146,22 @@ func (m *StrIntMap) GetOrSetFunc(key string, f func() int) int {
 // GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`
 // with mutex.Lock of the hash map.
 func (m *StrIntMap) GetOrSetFuncLock(key string, f func() int) int {
-	if v, ok := m.Search(key); !ok {
-		m.mu.Lock()
-		defer m.mu.Unlock()
-		if m.data == nil {
-			m.data = make(map[string]int)
-		}
-		if v, ok = m.data[key]; ok {
-			return v
-		}
-		v = f()
-		m.data[key] = v
-		return v
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFuncLock(key, f)
 }

 // SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *StrIntMap) SetIfNotExist(key string, value int) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, value)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExist(key, value)
 }

 // SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *StrIntMap) SetIfNotExistFunc(key string, f func() int) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f())
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFunc(key, f)
 }

 // SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
@ -276,126 +170,76 @@ func (m *StrIntMap) SetIfNotExistFunc(key string, f func() int) bool {
 // SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
 // it executes function `f` with mutex.Lock of the hash map.
 func (m *StrIntMap) SetIfNotExistFuncLock(key string, f func() int) bool {
-	if !m.Contains(key) {
-		m.mu.Lock()
-		defer m.mu.Unlock()
-		if m.data == nil {
-			m.data = make(map[string]int)
-		}
-		if _, ok := m.data[key]; !ok {
-			m.data[key] = f()
-		}
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFuncLock(key, f)
 }

 // Removes batch deletes values of the map by keys.
 func (m *StrIntMap) Removes(keys []string) {
-	m.mu.Lock()
-	if m.data != nil {
-		for _, key := range keys {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Removes(keys)
 }

 // Remove deletes value from map by given `key`, and return this deleted value.
 func (m *StrIntMap) Remove(key string) (value int) {
-	m.mu.Lock()
-	if m.data != nil {
-		var ok bool
-		if value, ok = m.data[key]; ok {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Remove(key)
 }

 // Keys returns all keys of the map as a slice.
 func (m *StrIntMap) Keys() []string {
-	m.mu.RLock()
-	var (
-		keys  = make([]string, len(m.data))
-		index = 0
-	)
-	for key := range m.data {
-		keys[index] = key
-		index++
-	}
-	m.mu.RUnlock()
-	return keys
+	m.lazyInit()
+	return m.KVMap.Keys()
 }

 // Values returns all values of the map as a slice.
 func (m *StrIntMap) Values() []int {
-	m.mu.RLock()
-	var (
-		values = make([]int, len(m.data))
-		index  = 0
-	)
-	for _, value := range m.data {
-		values[index] = value
-		index++
-	}
-	m.mu.RUnlock()
-	return values
+	m.lazyInit()
+	return m.KVMap.Values()
 }

 // Contains checks whether a key exists.
 // It returns true if the `key` exists, or else false.
 func (m *StrIntMap) Contains(key string) bool {
-	var ok bool
-	m.mu.RLock()
-	if m.data != nil {
-		_, ok = m.data[key]
-	}
-	m.mu.RUnlock()
-	return ok
+	m.lazyInit()
+	return m.KVMap.Contains(key)
 }

 // Size returns the size of the map.
 func (m *StrIntMap) Size() int {
-	m.mu.RLock()
-	length := len(m.data)
-	m.mu.RUnlock()
-	return length
+	m.lazyInit()
+	return m.KVMap.Size()
 }

 // IsEmpty checks whether the map is empty.
 // It returns true if map is empty, or else false.
 func (m *StrIntMap) IsEmpty() bool {
-	return m.Size() == 0
+	m.lazyInit()
+	return m.KVMap.IsEmpty()
 }

 // Clear deletes all data of the map, it will remake a new underlying data map.
 func (m *StrIntMap) Clear() {
-	m.mu.Lock()
-	m.data = make(map[string]int)
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Clear()
 }

 // Replace the data of the map with given `data`.
 func (m *StrIntMap) Replace(data map[string]int) {
-	m.mu.Lock()
-	m.data = data
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Replace(data)
 }

 // LockFunc locks writing with given callback function `f` within RWMutex.Lock.
 func (m *StrIntMap) LockFunc(f func(m map[string]int)) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.LockFunc(f)
 }

 // RLockFunc locks reading with given callback function `f` within RWMutex.RLock.
 func (m *StrIntMap) RLockFunc(f func(m map[string]int)) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.RLockFunc(f)
 }

 // Flip exchanges key-value of the map to value-key.
@ -412,19 +256,8 @@ func (m *StrIntMap) Flip() {
 // Merge merges two hash maps.
 // The `other` map will be merged into the map `m`.
 func (m *StrIntMap) Merge(other *StrIntMap) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = other.MapCopy()
-		return
-	}
-	if other != m {
-		other.mu.RLock()
-		defer other.mu.RUnlock()
-	}
-	for k, v := range other.data {
-		m.data[k] = v
-	}
+	m.lazyInit()
+	m.KVMap.Merge(other.KVMap)
 }

 // String returns the map as a string.
@ -432,81 +265,40 @@ func (m *StrIntMap) String() string {
 	if m == nil {
 		return ""
 	}
-	b, _ := m.MarshalJSON()
-	return string(b)
+	m.lazyInit()
+	return m.KVMap.String()
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (m StrIntMap) MarshalJSON() ([]byte, error) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	return json.Marshal(m.data)
+	m.lazyInit()
+	return m.KVMap.MarshalJSON()
 }

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (m *StrIntMap) UnmarshalJSON(b []byte) error {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[string]int)
-	}
-	if err := json.UnmarshalUseNumber(b, &m.data); err != nil {
-		return err
-	}
-	return nil
+	m.lazyInit()
+	return m.KVMap.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for map.
 func (m *StrIntMap) UnmarshalValue(value any) (err error) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[string]int)
-	}
-	switch value.(type) {
-	case string, []byte:
-		return json.UnmarshalUseNumber(gconv.Bytes(value), &m.data)
-	default:
-		for k, v := range gconv.Map(value) {
-			m.data[k] = gconv.Int(v)
-		}
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.UnmarshalValue(value)
 }

 // DeepCopy implements interface for deep copy of current type.
 func (m *StrIntMap) DeepCopy() any {
-	if m == nil {
-		return nil
+	m.lazyInit()
+	return &StrIntMap{
+		KVMap: m.KVMap.DeepCopy().(*KVMap[string, int]),
 	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[string]int, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return NewStrIntMapFrom(data, m.mu.IsSafe())
 }

 // IsSubOf checks whether the current map is a sub-map of `other`.
 func (m *StrIntMap) IsSubOf(other *StrIntMap) bool {
-	if m == other {
-		return true
-	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	for key, value := range m.data {
-		otherValue, ok := other.data[key]
-		if !ok {
-			return false
-		}
-		if otherValue != value {
-			return false
-		}
-	}
-	return true
+	m.lazyInit()
+	return m.KVMap.IsSubOf(other.KVMap)
 }

 // Diff compares current map `m` with map `other` and returns their different keys.
@ -514,22 +306,6 @@ func (m *StrIntMap) IsSubOf(other *StrIntMap) bool {
 // The returned `removedKeys` are the keys that are in map `other` but not in map `m`.
 // The returned `updatedKeys` are the keys that are both in map `m` and `other` but their values and not equal (`!=`).
 func (m *StrIntMap) Diff(other *StrIntMap) (addedKeys, removedKeys, updatedKeys []string) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-
-	for key := range m.data {
-		if _, ok := other.data[key]; !ok {
-			removedKeys = append(removedKeys, key)
-		} else if m.data[key] != other.data[key] {
-			updatedKeys = append(updatedKeys, key)
-		}
-	}
-	for key := range other.data {
-		if _, ok := m.data[key]; !ok {
-			addedKeys = append(addedKeys, key)
-		}
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Diff(other.KVMap)
 }
--- a/container/gmap/gmap_hash_str_str_map.go
+++ b/container/gmap/gmap_hash_str_str_map.go
@ -7,17 +7,12 @@

 package gmap

-import (
-	"github.com/gogf/gf/v2/internal/empty"
-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
-	"github.com/gogf/gf/v2/util/gconv"
-)
+import "sync"

 // StrStrMap implements map[string]string with RWMutex that has switch.
 type StrStrMap struct {
-	mu   rwmutex.RWMutex
-	data map[string]string
+	*KVMap[string, string]
+	once sync.Once
 }

 // NewStrStrMap returns an empty StrStrMap object.
@ -25,8 +20,7 @@ type StrStrMap struct {
 // which is false in default.
 func NewStrStrMap(safe ...bool) *StrStrMap {
 	return &StrStrMap{
-		data: make(map[string]string),
-		mu:   rwmutex.Create(safe...),
+		KVMap: NewKVMap[string, string](safe...),
 	}
 }

@ -35,194 +29,110 @@ func NewStrStrMap(safe ...bool) *StrStrMap {
 // there might be some concurrent-safe issues when changing the map outside.
 func NewStrStrMapFrom(data map[string]string, safe ...bool) *StrStrMap {
 	return &StrStrMap{
-		mu:   rwmutex.Create(safe...),
-		data: data,
+		KVMap: NewKVMapFrom(data, safe...),
 	}
 }

+// lazyInit lazily initializes the map.
+func (m *StrStrMap) lazyInit() {
+	m.once.Do(func() {
+		if m.KVMap == nil {
+			m.KVMap = NewKVMap[string, string](false)
+		}
+	})
+}
+
 // Iterator iterates the hash map readonly with custom callback function `f`.
 // If `f` returns true, then it continues iterating; or false to stop.
 func (m *StrStrMap) Iterator(f func(k string, v string) bool) {
-	for k, v := range m.Map() {
-		if !f(k, v) {
-			break
-		}
-	}
+	m.lazyInit()
+	m.KVMap.Iterator(f)
 }

 // Clone returns a new hash map with copy of current map data.
-func (m *StrStrMap) Clone() *StrStrMap {
-	return NewStrStrMapFrom(m.MapCopy(), m.mu.IsSafe())
+func (m *StrStrMap) Clone(safe ...bool) *StrStrMap {
+	m.lazyInit()
+	return &StrStrMap{KVMap: m.KVMap.Clone(safe...)}
 }

 // Map returns the underlying data map.
 // Note that, if it's in concurrent-safe usage, it returns a copy of underlying data,
 // or else a pointer to the underlying data.
 func (m *StrStrMap) Map() map[string]string {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	if !m.mu.IsSafe() {
-		return m.data
-	}
-	data := make(map[string]string, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.Map()
 }

 // MapStrAny returns a copy of the underlying data of the map as map[string]any.
 func (m *StrStrMap) MapStrAny() map[string]any {
-	m.mu.RLock()
-	data := make(map[string]any, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	m.mu.RUnlock()
-	return data
+	m.lazyInit()
+	return m.KVMap.MapStrAny()
 }

 // MapCopy returns a copy of the underlying data of the hash map.
 func (m *StrStrMap) MapCopy() map[string]string {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[string]string, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return data
+	m.lazyInit()
+	return m.KVMap.MapCopy()
 }

 // FilterEmpty deletes all key-value pair of which the value is empty.
 // Values like: 0, nil, false, "", len(slice/map/chan) == 0 are considered empty.
 func (m *StrStrMap) FilterEmpty() {
-	m.mu.Lock()
-	for k, v := range m.data {
-		if empty.IsEmpty(v) {
-			delete(m.data, k)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.FilterEmpty()
 }

 // Set sets key-value to the hash map.
 func (m *StrStrMap) Set(key string, val string) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = make(map[string]string)
-	}
-	m.data[key] = val
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Set(key, val)
 }

 // Sets batch sets key-values to the hash map.
 func (m *StrStrMap) Sets(data map[string]string) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = data
-	} else {
-		for k, v := range data {
-			m.data[k] = v
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Sets(data)
 }

 // Search searches the map with given `key`.
 // Second return parameter `found` is true if key was found, otherwise false.
 func (m *StrStrMap) Search(key string) (value string, found bool) {
-	m.mu.RLock()
-	if m.data != nil {
-		value, found = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Search(key)
 }

 // Get returns the value by given `key`.
 func (m *StrStrMap) Get(key string) (value string) {
-	m.mu.RLock()
-	if m.data != nil {
-		value = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Get(key)
 }

 // Pop retrieves and deletes an item from the map.
 func (m *StrStrMap) Pop() (key, value string) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for key, value = range m.data {
-		delete(m.data, key)
-		return
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Pop()
 }

 // Pops retrieves and deletes `size` items from the map.
 // It returns all items if size == -1.
 func (m *StrStrMap) Pops(size int) map[string]string {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if size > len(m.data) || size == -1 {
-		size = len(m.data)
-	}
-	if size == 0 {
-		return nil
-	}
-	var (
-		index  = 0
-		newMap = make(map[string]string, size)
-	)
-	for k, v := range m.data {
-		delete(m.data, k)
-		newMap[k] = v
-		index++
-		if index == size {
-			break
-		}
-	}
-	return newMap
-}
-
-// doSetWithLockCheck checks whether value of the key exists with mutex.Lock,
-// if not exists, set value to the map with given `key`,
-// or else just return the existing value.
-//
-// It returns value with given `key`.
-func (m *StrStrMap) doSetWithLockCheck(key string, value string) string {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[string]string)
-	}
-	if v, ok := m.data[key]; ok {
-		return v
-	}
-	m.data[key] = value
-	return value
+	m.lazyInit()
+	return m.KVMap.Pops(size)
 }

 // GetOrSet returns the value by key,
 // or sets value with given `value` if it does not exist and then returns this value.
 func (m *StrStrMap) GetOrSet(key string, value string) string {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, value)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSet(key, value)
 }

 // GetOrSetFunc returns the value by key,
 // or sets value with returned value of callback function `f` if it does not exist
 // and then returns this value.
 func (m *StrStrMap) GetOrSetFunc(key string, f func() string) string {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f())
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFunc(key, f)
 }

 // GetOrSetFuncLock returns the value by key,
@ -232,41 +142,22 @@ func (m *StrStrMap) GetOrSetFunc(key string, f func() string) string {
 // GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`
 // with mutex.Lock of the hash map.
 func (m *StrStrMap) GetOrSetFuncLock(key string, f func() string) string {
-	if v, ok := m.Search(key); !ok {
-		m.mu.Lock()
-		defer m.mu.Unlock()
-		if m.data == nil {
-			m.data = make(map[string]string)
-		}
-		if v, ok = m.data[key]; ok {
-			return v
-		}
-		v = f()
-		m.data[key] = v
-		return v
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.KVMap.GetOrSetFuncLock(key, f)
 }

 // SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *StrStrMap) SetIfNotExist(key string, value string) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, value)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExist(key, value)
 }

 // SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *StrStrMap) SetIfNotExistFunc(key string, f func() string) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f())
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFunc(key, f)
 }

 // SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
@ -275,126 +166,76 @@ func (m *StrStrMap) SetIfNotExistFunc(key string, f func() string) bool {
 // SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
 // it executes function `f` with mutex.Lock of the hash map.
 func (m *StrStrMap) SetIfNotExistFuncLock(key string, f func() string) bool {
-	if !m.Contains(key) {
-		m.mu.Lock()
-		defer m.mu.Unlock()
-		if m.data == nil {
-			m.data = make(map[string]string)
-		}
-		if _, ok := m.data[key]; !ok {
-			m.data[key] = f()
-		}
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.KVMap.SetIfNotExistFuncLock(key, f)
 }

 // Removes batch deletes values of the map by keys.
 func (m *StrStrMap) Removes(keys []string) {
-	m.mu.Lock()
-	if m.data != nil {
-		for _, key := range keys {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Removes(keys)
 }

 // Remove deletes value from map by given `key`, and return this deleted value.
 func (m *StrStrMap) Remove(key string) (value string) {
-	m.mu.Lock()
-	if m.data != nil {
-		var ok bool
-		if value, ok = m.data[key]; ok {
-			delete(m.data, key)
-		}
-	}
-	m.mu.Unlock()
-	return
+	m.lazyInit()
+	return m.KVMap.Remove(key)
 }

 // Keys returns all keys of the map as a slice.
 func (m *StrStrMap) Keys() []string {
-	m.mu.RLock()
-	var (
-		keys  = make([]string, len(m.data))
-		index = 0
-	)
-	for key := range m.data {
-		keys[index] = key
-		index++
-	}
-	m.mu.RUnlock()
-	return keys
+	m.lazyInit()
+	return m.KVMap.Keys()
 }

 // Values returns all values of the map as a slice.
 func (m *StrStrMap) Values() []string {
-	m.mu.RLock()
-	var (
-		values = make([]string, len(m.data))
-		index  = 0
-	)
-	for _, value := range m.data {
-		values[index] = value
-		index++
-	}
-	m.mu.RUnlock()
-	return values
+	m.lazyInit()
+	return m.KVMap.Values()
 }

 // Contains checks whether a key exists.
 // It returns true if the `key` exists, or else false.
 func (m *StrStrMap) Contains(key string) bool {
-	var ok bool
-	m.mu.RLock()
-	if m.data != nil {
-		_, ok = m.data[key]
-	}
-	m.mu.RUnlock()
-	return ok
+	m.lazyInit()
+	return m.KVMap.Contains(key)
 }

 // Size returns the size of the map.
 func (m *StrStrMap) Size() int {
-	m.mu.RLock()
-	length := len(m.data)
-	m.mu.RUnlock()
-	return length
+	m.lazyInit()
+	return m.KVMap.Size()
 }

 // IsEmpty checks whether the map is empty.
 // It returns true if map is empty, or else false.
 func (m *StrStrMap) IsEmpty() bool {
-	return m.Size() == 0
+	m.lazyInit()
+	return m.KVMap.IsEmpty()
 }

 // Clear deletes all data of the map, it will remake a new underlying data map.
 func (m *StrStrMap) Clear() {
-	m.mu.Lock()
-	m.data = make(map[string]string)
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Clear()
 }

 // Replace the data of the map with given `data`.
 func (m *StrStrMap) Replace(data map[string]string) {
-	m.mu.Lock()
-	m.data = data
-	m.mu.Unlock()
+	m.lazyInit()
+	m.KVMap.Replace(data)
 }

 // LockFunc locks writing with given callback function `f` within RWMutex.Lock.
 func (m *StrStrMap) LockFunc(f func(m map[string]string)) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.LockFunc(f)
 }

 // RLockFunc locks reading with given callback function `f` within RWMutex.RLock.
 func (m *StrStrMap) RLockFunc(f func(m map[string]string)) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	f(m.data)
+	m.lazyInit()
+	m.KVMap.RLockFunc(f)
 }

 // Flip exchanges key-value of the map to value-key.
@ -411,19 +252,8 @@ func (m *StrStrMap) Flip() {
 // Merge merges two hash maps.
 // The `other` map will be merged into the map `m`.
 func (m *StrStrMap) Merge(other *StrStrMap) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = other.MapCopy()
-		return
-	}
-	if other != m {
-		other.mu.RLock()
-		defer other.mu.RUnlock()
-	}
-	for k, v := range other.data {
-		m.data[k] = v
-	}
+	m.lazyInit()
+	m.KVMap.Merge(other.KVMap)
 }

 // String returns the map as a string.
@ -431,71 +261,40 @@ func (m *StrStrMap) String() string {
 	if m == nil {
 		return ""
 	}
-	b, _ := m.MarshalJSON()
-	return string(b)
+	m.lazyInit()
+	return m.KVMap.String()
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (m StrStrMap) MarshalJSON() ([]byte, error) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	return json.Marshal(m.data)
+	m.lazyInit()
+	return m.KVMap.MarshalJSON()
 }

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (m *StrStrMap) UnmarshalJSON(b []byte) error {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[string]string)
-	}
-	if err := json.UnmarshalUseNumber(b, &m.data); err != nil {
-		return err
-	}
-	return nil
+	m.lazyInit()
+	return m.KVMap.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for map.
 func (m *StrStrMap) UnmarshalValue(value any) (err error) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	m.data = gconv.MapStrStr(value)
-	return
+	m.lazyInit()
+	return m.KVMap.UnmarshalValue(value)
 }

 // DeepCopy implements interface for deep copy of current type.
 func (m *StrStrMap) DeepCopy() any {
-	if m == nil {
-		return nil
+	m.lazyInit()
+	return &StrStrMap{
+		KVMap: m.KVMap.DeepCopy().(*KVMap[string, string]),
 	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[string]string, len(m.data))
-	for k, v := range m.data {
-		data[k] = v
-	}
-	return NewStrStrMapFrom(data, m.mu.IsSafe())
 }

 // IsSubOf checks whether the current map is a sub-map of `other`.
 func (m *StrStrMap) IsSubOf(other *StrStrMap) bool {
-	if m == other {
-		return true
-	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	for key, value := range m.data {
-		otherValue, ok := other.data[key]
-		if !ok {
-			return false
-		}
-		if otherValue != value {
-			return false
-		}
-	}
-	return true
+	m.lazyInit()
+	return m.KVMap.IsSubOf(other.KVMap)
 }

 // Diff compares current map `m` with map `other` and returns their different keys.
@ -503,22 +302,6 @@ func (m *StrStrMap) IsSubOf(other *StrStrMap) bool {
 // The returned `removedKeys` are the keys that are in map `other` but not in map `m`.
 // The returned `updatedKeys` are the keys that are both in map `m` and `other` but their values and not equal (`!=`).
 func (m *StrStrMap) Diff(other *StrStrMap) (addedKeys, removedKeys, updatedKeys []string) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-
-	for key := range m.data {
-		if _, ok := other.data[key]; !ok {
-			removedKeys = append(removedKeys, key)
-		} else if m.data[key] != other.data[key] {
-			updatedKeys = append(updatedKeys, key)
-		}
-	}
-	for key := range other.data {
-		if _, ok := m.data[key]; !ok {
-			addedKeys = append(addedKeys, key)
-		}
-	}
-	return
+	m.lazyInit()
+	return m.KVMap.Diff(other.KVMap)
 }
--- a/container/gmap/gmap_list_k_v_map.go
+++ b/container/gmap/gmap_list_k_v_map.go
@ -0,0 +1,654 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package gmap
+
+import (
+	"bytes"
+	"fmt"
+
+	"github.com/gogf/gf/v2/container/glist"
+	"github.com/gogf/gf/v2/container/gvar"
+	"github.com/gogf/gf/v2/internal/deepcopy"
+	"github.com/gogf/gf/v2/internal/empty"
+	"github.com/gogf/gf/v2/internal/json"
+	"github.com/gogf/gf/v2/internal/rwmutex"
+	"github.com/gogf/gf/v2/util/gconv"
+)
+
+// ListKVMap is a map that preserves insertion-order.
+//
+// It is backed by a hash table to store values and doubly-linked list to store ordering.
+//
+// Thread-safety is optional and controlled by the `safe` parameter during initialization.
+//
+// Reference: http://en.wikipedia.org/wiki/Associative_array
+type ListKVMap[K comparable, V any] struct {
+	mu   rwmutex.RWMutex
+	data map[K]*glist.TElement[*gListKVMapNode[K, V]]
+	list *glist.TList[*gListKVMapNode[K, V]]
+}
+
+type gListKVMapNode[K comparable, V any] struct {
+	key   K
+	value V
+}
+
+// NewListKVMap returns an empty link map.
+// ListKVMap is backed by a hash table to store values and doubly-linked list to store ordering.
+// The parameter `safe` is used to specify whether using map in concurrent-safety,
+// which is false in default.
+func NewListKVMap[K comparable, V any](safe ...bool) *ListKVMap[K, V] {
+	return &ListKVMap[K, V]{
+		mu:   rwmutex.Create(safe...),
+		data: make(map[K]*glist.TElement[*gListKVMapNode[K, V]]),
+		list: glist.NewT[*gListKVMapNode[K, V]](),
+	}
+}
+
+// NewListKVMapFrom returns a link map from given map `data`.
+// Note that, the param `data` map will be copied to the underlying data structure,
+// so changes to the original map will not affect the link map.
+func NewListKVMapFrom[K comparable, V any](data map[K]V, safe ...bool) *ListKVMap[K, V] {
+	m := NewListKVMap[K, V](safe...)
+	m.Sets(data)
+	return m
+}
+
+// Iterator is alias of IteratorAsc.
+func (m *ListKVMap[K, V]) Iterator(f func(key K, value V) bool) {
+	m.IteratorAsc(f)
+}
+
+// IteratorAsc iterates the map readonly in ascending order with given callback function `f`.
+// If `f` returns true, then it continues iterating; or false to stop.
+func (m *ListKVMap[K, V]) IteratorAsc(f func(key K, value V) bool) {
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	if m.list != nil {
+		m.list.IteratorAsc(func(e *glist.TElement[*gListKVMapNode[K, V]]) bool {
+			return f(e.Value.key, e.Value.value)
+		})
+	}
+}
+
+// IteratorDesc iterates the map readonly in descending order with given callback function `f`.
+// If `f` returns true, then it continues iterating; or false to stop.
+func (m *ListKVMap[K, V]) IteratorDesc(f func(key K, value V) bool) {
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	if m.list != nil {
+		m.list.IteratorDesc(func(e *glist.TElement[*gListKVMapNode[K, V]]) bool {
+			return f(e.Value.key, e.Value.value)
+		})
+	}
+}
+
+// Clone returns a new link map with copy of current map data.
+func (m *ListKVMap[K, V]) Clone(safe ...bool) *ListKVMap[K, V] {
+	return NewListKVMapFrom(m.Map(), safe...)
+}
+
+// Clear deletes all data of the map, it will remake a new underlying data map.
+func (m *ListKVMap[K, V]) Clear() {
+	m.mu.Lock()
+	m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+	m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	m.mu.Unlock()
+}
+
+// Replace the data of the map with given `data`.
+func (m *ListKVMap[K, V]) Replace(data map[K]V) {
+	m.mu.Lock()
+	m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+	m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	for key, value := range data {
+		m.data[key] = m.list.PushBack(&gListKVMapNode[K, V]{key, value})
+	}
+	m.mu.Unlock()
+}
+
+// Map returns a copy of the underlying data of the map.
+func (m *ListKVMap[K, V]) Map() map[K]V {
+	m.mu.RLock()
+	var data map[K]V
+	if m.list != nil {
+		data = make(map[K]V, len(m.data))
+		m.list.IteratorAsc(func(e *glist.TElement[*gListKVMapNode[K, V]]) bool {
+			data[e.Value.key] = e.Value.value
+			return true
+		})
+	}
+	m.mu.RUnlock()
+	return data
+}
+
+// MapStrAny returns a copy of the underlying data of the map as map[string]any.
+func (m *ListKVMap[K, V]) MapStrAny() map[string]any {
+	m.mu.RLock()
+	var data map[string]any
+	if m.list != nil {
+		data = make(map[string]any, len(m.data))
+		m.list.IteratorAsc(func(e *glist.TElement[*gListKVMapNode[K, V]]) bool {
+			data[gconv.String(e.Value.key)] = e.Value.value
+			return true
+		})
+	}
+	m.mu.RUnlock()
+	return data
+}
+
+// FilterEmpty deletes all key-value pair of which the value is empty.
+func (m *ListKVMap[K, V]) FilterEmpty() {
+	m.mu.Lock()
+	if m.list != nil {
+		var keys = make([]K, 0, m.list.Size())
+		m.list.IteratorAsc(func(e *glist.TElement[*gListKVMapNode[K, V]]) bool {
+			if empty.IsEmpty(e.Value.value) {
+				keys = append(keys, e.Value.key)
+			}
+			return true
+		})
+
+		if len(keys) > 0 {
+			for _, key := range keys {
+				if e, ok := m.data[key]; ok {
+					delete(m.data, key)
+					m.list.Remove(e)
+				}
+			}
+		}
+	}
+	m.mu.Unlock()
+}
+
+// Set sets key-value to the map.
+func (m *ListKVMap[K, V]) Set(key K, value V) {
+	m.mu.Lock()
+	if m.data == nil {
+		m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+		m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	}
+	if e, ok := m.data[key]; !ok {
+		m.data[key] = m.list.PushBack(&gListKVMapNode[K, V]{key, value})
+	} else {
+		e.Value = &gListKVMapNode[K, V]{key, value}
+	}
+	m.mu.Unlock()
+}
+
+// Sets batch sets key-values to the map.
+func (m *ListKVMap[K, V]) Sets(data map[K]V) {
+	m.mu.Lock()
+	if m.data == nil {
+		m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+		m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	}
+	for key, value := range data {
+		if e, ok := m.data[key]; !ok {
+			m.data[key] = m.list.PushBack(&gListKVMapNode[K, V]{key, value})
+		} else {
+			e.Value = &gListKVMapNode[K, V]{key, value}
+		}
+	}
+	m.mu.Unlock()
+}
+
+// Search searches the map with given `key`.
+// Second return parameter `found` is true if key was found, otherwise false.
+func (m *ListKVMap[K, V]) Search(key K) (value V, found bool) {
+	m.mu.RLock()
+	if m.data != nil {
+		if e, ok := m.data[key]; ok {
+			value = e.Value.value
+			found = ok
+		}
+	}
+	m.mu.RUnlock()
+	return
+}
+
+// Get returns the value by given `key`.
+func (m *ListKVMap[K, V]) Get(key K) (value V) {
+	m.mu.RLock()
+	if m.data != nil {
+		if e, ok := m.data[key]; ok {
+			value = e.Value.value
+		}
+	}
+	m.mu.RUnlock()
+	return
+}
+
+// Pop retrieves and deletes an item from the map.
+func (m *ListKVMap[K, V]) Pop() (key K, value V) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	for k, e := range m.data {
+		value = e.Value.value
+		delete(m.data, k)
+		m.list.Remove(e)
+		return k, value
+	}
+	return
+}
+
+// Pops retrieves and deletes `size` items from the map.
+// It returns all items if size == -1.
+func (m *ListKVMap[K, V]) Pops(size int) map[K]V {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if size > len(m.data) || size == -1 {
+		size = len(m.data)
+	}
+	if size == 0 {
+		return nil
+	}
+	index := 0
+	newMap := make(map[K]V, size)
+	for k, e := range m.data {
+		value := e.Value.value
+		delete(m.data, k)
+		m.list.Remove(e)
+		newMap[k] = value
+		index++
+		if index == size {
+			break
+		}
+	}
+	return newMap
+}
+
+// doSetWithLockCheck checks whether value of the key exists with mutex.Lock,
+// if not exists, set value to the map with given `key`,
+// or else just return the existing value.
+//
+// It returns value with given `key`.
+func (m *ListKVMap[K, V]) doSetWithLockCheck(key K, value V) V {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	return m.doSetWithLockCheckWithoutLock(key, value)
+}
+
+func (m *ListKVMap[K, V]) doSetWithLockCheckWithoutLock(key K, value V) V {
+	if m.data == nil {
+		m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+		m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	}
+	if e, ok := m.data[key]; ok {
+		return e.Value.value
+	}
+	if any(value) != nil {
+		m.data[key] = m.list.PushBack(&gListKVMapNode[K, V]{key, value})
+	}
+	return value
+}
+
+// GetOrSet returns the value by key,
+// or sets value with given `value` if it does not exist and then returns this value.
+func (m *ListKVMap[K, V]) GetOrSet(key K, value V) V {
+	if v, ok := m.Search(key); !ok {
+		return m.doSetWithLockCheck(key, value)
+	} else {
+		return v
+	}
+}
+
+// GetOrSetFunc returns the value by key,
+// or sets value with returned value of callback function `f` if it does not exist
+// and then returns this value.
+func (m *ListKVMap[K, V]) GetOrSetFunc(key K, f func() V) V {
+	if v, ok := m.Search(key); !ok {
+		return m.doSetWithLockCheck(key, f())
+	} else {
+		return v
+	}
+}
+
+// GetOrSetFuncLock returns the value by key,
+// or sets value with returned value of callback function `f` if it does not exist
+// and then returns this value.
+//
+// GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`
+// with mutex.Lock of the map.
+func (m *ListKVMap[K, V]) GetOrSetFuncLock(key K, f func() V) V {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if m.data == nil {
+		m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+		m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	}
+	if e, ok := m.data[key]; ok {
+		return e.Value.value
+	}
+	value := f()
+	if any(value) != nil {
+		m.data[key] = m.list.PushBack(&gListKVMapNode[K, V]{key, value})
+	}
+	return value
+}
+
+// GetVar returns a Var with the value by given `key`.
+// The returned Var is un-concurrent safe.
+func (m *ListKVMap[K, V]) GetVar(key K) *gvar.Var {
+	return gvar.New(m.Get(key))
+}
+
+// GetVarOrSet returns a Var with result from GetVarOrSet.
+// The returned Var is un-concurrent safe.
+func (m *ListKVMap[K, V]) GetVarOrSet(key K, value V) *gvar.Var {
+	return gvar.New(m.GetOrSet(key, value))
+}
+
+// GetVarOrSetFunc returns a Var with result from GetOrSetFunc.
+// The returned Var is un-concurrent safe.
+func (m *ListKVMap[K, V]) GetVarOrSetFunc(key K, f func() V) *gvar.Var {
+	return gvar.New(m.GetOrSetFunc(key, f))
+}
+
+// GetVarOrSetFuncLock returns a Var with result from GetOrSetFuncLock.
+// The returned Var is un-concurrent safe.
+func (m *ListKVMap[K, V]) GetVarOrSetFuncLock(key K, f func() V) *gvar.Var {
+	return gvar.New(m.GetOrSetFuncLock(key, f))
+}
+
+// SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
+// It returns false if `key` exists, and `value` would be ignored.
+func (m *ListKVMap[K, V]) SetIfNotExist(key K, value V) bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if m.data == nil {
+		m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+		m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	}
+	if _, ok := m.data[key]; ok {
+		return false
+	}
+	if any(value) != nil {
+		m.data[key] = m.list.PushBack(&gListKVMapNode[K, V]{key, value})
+	}
+	return true
+}
+
+// SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
+// It returns false if `key` exists, and `value` would be ignored.
+func (m *ListKVMap[K, V]) SetIfNotExistFunc(key K, f func() V) bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if m.data == nil {
+		m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+		m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	}
+	if _, ok := m.data[key]; ok {
+		return false
+	}
+	value := f()
+	if any(value) != nil {
+		m.data[key] = m.list.PushBack(&gListKVMapNode[K, V]{key, value})
+	}
+	return true
+}
+
+// SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
+// It returns false if `key` exists, and `value` would be ignored.
+//
+// SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
+// it executes function `f` with mutex.Lock of the map.
+func (m *ListKVMap[K, V]) SetIfNotExistFuncLock(key K, f func() V) bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if m.data == nil {
+		m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+		m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	}
+	if _, ok := m.data[key]; ok {
+		return false
+	}
+	value := f()
+	if any(value) != nil {
+		m.data[key] = m.list.PushBack(&gListKVMapNode[K, V]{key, value})
+	}
+	return true
+}
+
+// Remove deletes value from map by given `key`, and return this deleted value.
+func (m *ListKVMap[K, V]) Remove(key K) (value V) {
+	m.mu.Lock()
+	if m.data != nil {
+		if e, ok := m.data[key]; ok {
+			value = e.Value.value
+			delete(m.data, key)
+			m.list.Remove(e)
+		}
+	}
+	m.mu.Unlock()
+	return
+}
+
+// Removes batch deletes values of the map by keys.
+func (m *ListKVMap[K, V]) Removes(keys []K) {
+	m.mu.Lock()
+	if m.data != nil {
+		for _, key := range keys {
+			if e, ok := m.data[key]; ok {
+				delete(m.data, key)
+				m.list.Remove(e)
+			}
+		}
+	}
+	m.mu.Unlock()
+}
+
+// Keys returns all keys of the map as a slice in ascending order.
+func (m *ListKVMap[K, V]) Keys() []K {
+	m.mu.RLock()
+	var (
+		keys  = make([]K, m.list.Len())
+		index = 0
+	)
+	if m.list != nil {
+		m.list.IteratorAsc(func(e *glist.TElement[*gListKVMapNode[K, V]]) bool {
+			keys[index] = e.Value.key
+			index++
+			return true
+		})
+	}
+	m.mu.RUnlock()
+	return keys
+}
+
+// Values returns all values of the map as a slice.
+func (m *ListKVMap[K, V]) Values() []V {
+	m.mu.RLock()
+	var (
+		values = make([]V, m.list.Len())
+		index  = 0
+	)
+	if m.list != nil {
+		m.list.IteratorAsc(func(e *glist.TElement[*gListKVMapNode[K, V]]) bool {
+			values[index] = e.Value.value
+			index++
+			return true
+		})
+	}
+	m.mu.RUnlock()
+	return values
+}
+
+// Contains checks whether a key exists.
+// It returns true if the `key` exists, or else false.
+func (m *ListKVMap[K, V]) Contains(key K) (ok bool) {
+	m.mu.RLock()
+	if m.data != nil {
+		_, ok = m.data[key]
+	}
+	m.mu.RUnlock()
+	return
+}
+
+// Size returns the size of the map.
+func (m *ListKVMap[K, V]) Size() (size int) {
+	m.mu.RLock()
+	size = len(m.data)
+	m.mu.RUnlock()
+	return
+}
+
+// IsEmpty checks whether the map is empty.
+// It returns true if map is empty, or else false.
+func (m *ListKVMap[K, V]) IsEmpty() bool {
+	return m.Size() == 0
+}
+
+// Flip exchanges key-value of the map to value-key.
+func (m *ListKVMap[K, V]) Flip() error {
+	data := m.Map()
+	m.Clear()
+	for key, value := range data {
+		var (
+			newKey   K
+			newValue V
+		)
+
+		if err := gconv.Scan(value, &newKey); err != nil {
+			return err
+		}
+
+		if err := gconv.Scan(key, &newValue); err != nil {
+			return err
+		}
+		m.Set(newKey, newValue)
+	}
+
+	return nil
+}
+
+// Merge merges two link maps.
+// The `other` map will be merged into the map `m`.
+func (m *ListKVMap[K, V]) Merge(other *ListKVMap[K, V]) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.data == nil {
+		m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+		m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	}
+	if other != m {
+		other.mu.RLock()
+		defer other.mu.RUnlock()
+	}
+	var node *gListKVMapNode[K, V]
+	other.list.IteratorAsc(func(e *glist.TElement[*gListKVMapNode[K, V]]) bool {
+		node = e.Value
+		if e, ok := m.data[node.key]; !ok {
+			m.data[node.key] = m.list.PushBack(&gListKVMapNode[K, V]{node.key, node.value})
+		} else {
+			e.Value = &gListKVMapNode[K, V]{node.key, node.value}
+		}
+		return true
+	})
+}
+
+// String returns the map as a string.
+func (m *ListKVMap[K, V]) String() string {
+	if m == nil {
+		return ""
+	}
+	b, _ := m.MarshalJSON()
+	return string(b)
+}
+
+// MarshalJSON implements the interface MarshalJSON for json.Marshal.
+func (m ListKVMap[K, V]) MarshalJSON() (jsonBytes []byte, err error) {
+	if m.data == nil {
+		return []byte("{}"), nil
+	}
+	buffer := bytes.NewBuffer(nil)
+	buffer.WriteByte('{')
+	m.Iterator(func(key K, value V) bool {
+		valueBytes, valueJSONErr := json.Marshal(value)
+		if valueJSONErr != nil {
+			err = valueJSONErr
+			return false
+		}
+		if buffer.Len() > 1 {
+			buffer.WriteByte(',')
+		}
+		fmt.Fprintf(buffer, `"%v":%s`, key, valueBytes)
+		return true
+	})
+	buffer.WriteByte('}')
+	return buffer.Bytes(), nil
+}
+
+// UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
+func (m *ListKVMap[K, V]) UnmarshalJSON(b []byte) error {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.data == nil {
+		m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+		m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	}
+	var data map[string]V
+	if err := json.UnmarshalUseNumber(b, &data); err != nil {
+		return err
+	}
+	var kvData map[K]V
+	if err := gconv.Scan(data, &kvData); err != nil {
+		return err
+	}
+	for key, value := range kvData {
+		if e, ok := m.data[key]; !ok {
+			m.data[key] = m.list.PushBack(&gListKVMapNode[K, V]{key, value})
+		} else {
+			e.Value = &gListKVMapNode[K, V]{key, value}
+		}
+	}
+	return nil
+}
+
+// UnmarshalValue is an interface implement which sets any type of value for map.
+func (m *ListKVMap[K, V]) UnmarshalValue(value any) (err error) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.data == nil {
+		m.data = make(map[K]*glist.TElement[*gListKVMapNode[K, V]])
+		m.list = glist.NewT[*gListKVMapNode[K, V]]()
+	}
+	var dataMap map[K]V
+	if err = gconv.Scan(value, &dataMap); err != nil {
+		return
+	}
+	for k, v := range dataMap {
+		if e, ok := m.data[k]; !ok {
+			m.data[k] = m.list.PushBack(&gListKVMapNode[K, V]{k, v})
+		} else {
+			e.Value = &gListKVMapNode[K, V]{k, v}
+		}
+	}
+	return
+}
+
+// DeepCopy implements interface for deep copy of current type.
+func (m *ListKVMap[K, V]) DeepCopy() any {
+	if m == nil {
+		return nil
+	}
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	data := make(map[K]V, len(m.data))
+	if m.list != nil {
+		m.list.IteratorAsc(func(e *glist.TElement[*gListKVMapNode[K, V]]) bool {
+			data[e.Value.key] = deepcopy.Copy(e.Value.value).(V)
+			return true
+		})
+	}
+	return NewListKVMapFrom(data, m.mu.IsSafe())
+}
--- a/container/gmap/gmap_list_map.go
+++ b/container/gmap/gmap_list_map.go
@ -7,15 +7,9 @@
 package gmap

 import (
-	"bytes"
-	"fmt"
+	"sync"

-	"github.com/gogf/gf/v2/container/glist"
 	"github.com/gogf/gf/v2/container/gvar"
-	"github.com/gogf/gf/v2/internal/deepcopy"
-	"github.com/gogf/gf/v2/internal/empty"
-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
 	"github.com/gogf/gf/v2/util/gconv"
 )

@ -27,15 +21,11 @@ import (
 //
 // Reference: http://en.wikipedia.org/wiki/Associative_array
 type ListMap struct {
-	mu   rwmutex.RWMutex
-	data map[any]*glist.Element
-	list *glist.List
+	*ListKVMap[any, any]
+	once sync.Once
 }

-type gListMapNode struct {
-	key   any
-	value any
-}
+type gListMapNode = gListKVMapNode[any, any]

 // NewListMap returns an empty link map.
 // ListMap is backed by a hash table to store values and doubly-linked list to store ordering.
@ -43,9 +33,7 @@ type gListMapNode struct {
 // which is false in default.
 func NewListMap(safe ...bool) *ListMap {
 	return &ListMap{
-		mu:   rwmutex.Create(safe...),
-		data: make(map[any]*glist.Element),
-		list: glist.New(),
+		ListKVMap: NewListKVMap[any, any](safe...),
 	}
 }

@ -58,6 +46,15 @@ func NewListMapFrom(data map[any]any, safe ...bool) *ListMap {
 	return m
 }

+// lazyInit lazily initializes the list map.
+func (m *ListMap) lazyInit() {
+	m.once.Do(func() {
+		if m.ListKVMap == nil {
+			m.ListKVMap = NewListKVMap[any, any](false)
+		}
+	})
+}
+
 // Iterator is alias of IteratorAsc.
 func (m *ListMap) Iterator(f func(key, value any) bool) {
 	m.IteratorAsc(f)
@ -66,29 +63,15 @@ func (m *ListMap) Iterator(f func(key, value any) bool) {
 // IteratorAsc iterates the map readonly in ascending order with given callback function `f`.
 // If `f` returns true, then it continues iterating; or false to stop.
 func (m *ListMap) IteratorAsc(f func(key any, value any) bool) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	if m.list != nil {
-		var node *gListMapNode
-		m.list.IteratorAsc(func(e *glist.Element) bool {
-			node = e.Value.(*gListMapNode)
-			return f(node.key, node.value)
-		})
-	}
+	m.lazyInit()
+	m.ListKVMap.IteratorAsc(f)
 }

 // IteratorDesc iterates the map readonly in descending order with given callback function `f`.
 // If `f` returns true, then it continues iterating; or false to stop.
 func (m *ListMap) IteratorDesc(f func(key any, value any) bool) {
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	if m.list != nil {
-		var node *gListMapNode
-		m.list.IteratorDesc(func(e *glist.Element) bool {
-			node = e.Value.(*gListMapNode)
-			return f(node.key, node.value)
-		})
-	}
+	m.lazyInit()
+	m.ListKVMap.IteratorDesc(f)
 }

 // Clone returns a new link map with copy of current map data.
@ -98,232 +81,85 @@ func (m *ListMap) Clone(safe ...bool) *ListMap {

 // Clear deletes all data of the map, it will remake a new underlying data map.
 func (m *ListMap) Clear() {
-	m.mu.Lock()
-	m.data = make(map[any]*glist.Element)
-	m.list = glist.New()
-	m.mu.Unlock()
+	m.lazyInit()
+	m.ListKVMap.Clear()
 }

 // Replace the data of the map with given `data`.
 func (m *ListMap) Replace(data map[any]any) {
-	m.mu.Lock()
-	m.data = make(map[any]*glist.Element)
-	m.list = glist.New()
-	for key, value := range data {
-		if e, ok := m.data[key]; !ok {
-			m.data[key] = m.list.PushBack(&gListMapNode{key, value})
-		} else {
-			e.Value = &gListMapNode{key, value}
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.ListKVMap.Replace(data)
 }

 // Map returns a copy of the underlying data of the map.
 func (m *ListMap) Map() map[any]any {
-	m.mu.RLock()
-	var node *gListMapNode
-	var data map[any]any
-	if m.list != nil {
-		data = make(map[any]any, len(m.data))
-		m.list.IteratorAsc(func(e *glist.Element) bool {
-			node = e.Value.(*gListMapNode)
-			data[node.key] = node.value
-			return true
-		})
-	}
-	m.mu.RUnlock()
-	return data
+	m.lazyInit()
+	return m.ListKVMap.Map()
 }

 // MapStrAny returns a copy of the underlying data of the map as map[string]any.
 func (m *ListMap) MapStrAny() map[string]any {
-	m.mu.RLock()
-	var node *gListMapNode
-	var data map[string]any
-	if m.list != nil {
-		data = make(map[string]any, len(m.data))
-		m.list.IteratorAsc(func(e *glist.Element) bool {
-			node = e.Value.(*gListMapNode)
-			data[gconv.String(node.key)] = node.value
-			return true
-		})
-	}
-	m.mu.RUnlock()
-	return data
+	m.lazyInit()
+	return m.ListKVMap.MapStrAny()
 }

 // FilterEmpty deletes all key-value pair of which the value is empty.
 func (m *ListMap) FilterEmpty() {
-	m.mu.Lock()
-	if m.list != nil {
-		var (
-			keys = make([]any, 0)
-			node *gListMapNode
-		)
-		m.list.IteratorAsc(func(e *glist.Element) bool {
-			node = e.Value.(*gListMapNode)
-			if empty.IsEmpty(node.value) {
-				keys = append(keys, node.key)
-			}
-			return true
-		})
-		if len(keys) > 0 {
-			for _, key := range keys {
-				if e, ok := m.data[key]; ok {
-					delete(m.data, key)
-					m.list.Remove(e)
-				}
-			}
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.ListKVMap.FilterEmpty()
 }

 // Set sets key-value to the map.
 func (m *ListMap) Set(key any, value any) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = make(map[any]*glist.Element)
-		m.list = glist.New()
-	}
-	if e, ok := m.data[key]; !ok {
-		m.data[key] = m.list.PushBack(&gListMapNode{key, value})
-	} else {
-		e.Value = &gListMapNode{key, value}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.ListKVMap.Set(key, value)
 }

 // Sets batch sets key-values to the map.
 func (m *ListMap) Sets(data map[any]any) {
-	m.mu.Lock()
-	if m.data == nil {
-		m.data = make(map[any]*glist.Element)
-		m.list = glist.New()
-	}
-	for key, value := range data {
-		if e, ok := m.data[key]; !ok {
-			m.data[key] = m.list.PushBack(&gListMapNode{key, value})
-		} else {
-			e.Value = &gListMapNode{key, value}
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.ListKVMap.Sets(data)
 }

 // Search searches the map with given `key`.
 // Second return parameter `found` is true if key was found, otherwise false.
 func (m *ListMap) Search(key any) (value any, found bool) {
-	m.mu.RLock()
-	if m.data != nil {
-		if e, ok := m.data[key]; ok {
-			value = e.Value.(*gListMapNode).value
-			found = ok
-		}
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.ListKVMap.Search(key)
 }

 // Get returns the value by given `key`.
 func (m *ListMap) Get(key any) (value any) {
-	m.mu.RLock()
-	if m.data != nil {
-		if e, ok := m.data[key]; ok {
-			value = e.Value.(*gListMapNode).value
-		}
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.ListKVMap.Get(key)
 }

 // Pop retrieves and deletes an item from the map.
 func (m *ListMap) Pop() (key, value any) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	for k, e := range m.data {
-		value = e.Value.(*gListMapNode).value
-		delete(m.data, k)
-		m.list.Remove(e)
-		return k, value
-	}
-	return
+	m.lazyInit()
+	return m.ListKVMap.Pop()
 }

 // Pops retrieves and deletes `size` items from the map.
 // It returns all items if size == -1.
 func (m *ListMap) Pops(size int) map[any]any {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if size > len(m.data) || size == -1 {
-		size = len(m.data)
-	}
-	if size == 0 {
-		return nil
-	}
-	index := 0
-	newMap := make(map[any]any, size)
-	for k, e := range m.data {
-		value := e.Value.(*gListMapNode).value
-		delete(m.data, k)
-		m.list.Remove(e)
-		newMap[k] = value
-		index++
-		if index == size {
-			break
-		}
-	}
-	return newMap
-}
-
-// doSetWithLockCheck checks whether value of the key exists with mutex.Lock,
-// if not exists, set value to the map with given `key`,
-// or else just return the existing value.
-//
-// When setting value, if `value` is type of `func() interface {}`,
-// it will be executed with mutex.Lock of the map,
-// and its return value will be set to the map with `key`.
-//
-// It returns value with given `key`.
-func (m *ListMap) doSetWithLockCheck(key any, value any) any {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[any]*glist.Element)
-		m.list = glist.New()
-	}
-	if e, ok := m.data[key]; ok {
-		return e.Value.(*gListMapNode).value
-	}
-	if f, ok := value.(func() any); ok {
-		value = f()
-	}
-	if value != nil {
-		m.data[key] = m.list.PushBack(&gListMapNode{key, value})
-	}
-	return value
+	m.lazyInit()
+	return m.ListKVMap.Pops(size)
 }

 // GetOrSet returns the value by key,
 // or sets value with given `value` if it does not exist and then returns this value.
 func (m *ListMap) GetOrSet(key any, value any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, value)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.ListKVMap.GetOrSet(key, value)
 }

 // GetOrSetFunc returns the value by key,
 // or sets value with returned value of callback function `f` if it does not exist
 // and then returns this value.
 func (m *ListMap) GetOrSetFunc(key any, f func() any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f())
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.ListKVMap.GetOrSetFunc(key, f)
 }

 // GetOrSetFuncLock returns the value by key,
@ -333,55 +169,50 @@ func (m *ListMap) GetOrSetFunc(key any, f func() any) any {
 // GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`
 // with mutex.Lock of the map.
 func (m *ListMap) GetOrSetFuncLock(key any, f func() any) any {
-	if v, ok := m.Search(key); !ok {
-		return m.doSetWithLockCheck(key, f)
-	} else {
-		return v
-	}
+	m.lazyInit()
+	return m.ListKVMap.GetOrSetFuncLock(key, f)
 }

 // GetVar returns a Var with the value by given `key`.
 // The returned Var is un-concurrent safe.
 func (m *ListMap) GetVar(key any) *gvar.Var {
-	return gvar.New(m.Get(key))
+	m.lazyInit()
+	return m.ListKVMap.GetVar(key)
 }

 // GetVarOrSet returns a Var with result from GetVarOrSet.
 // The returned Var is un-concurrent safe.
 func (m *ListMap) GetVarOrSet(key any, value any) *gvar.Var {
-	return gvar.New(m.GetOrSet(key, value))
+	m.lazyInit()
+	return m.ListKVMap.GetVarOrSet(key, value)
 }

 // GetVarOrSetFunc returns a Var with result from GetOrSetFunc.
 // The returned Var is un-concurrent safe.
 func (m *ListMap) GetVarOrSetFunc(key any, f func() any) *gvar.Var {
-	return gvar.New(m.GetOrSetFunc(key, f))
+	m.lazyInit()
+	return m.ListKVMap.GetVarOrSetFunc(key, f)
 }

 // GetVarOrSetFuncLock returns a Var with result from GetOrSetFuncLock.
 // The returned Var is un-concurrent safe.
 func (m *ListMap) GetVarOrSetFuncLock(key any, f func() any) *gvar.Var {
-	return gvar.New(m.GetOrSetFuncLock(key, f))
+	m.lazyInit()
+	return m.ListKVMap.GetVarOrSetFuncLock(key, f)
 }

 // SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *ListMap) SetIfNotExist(key any, value any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, value)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.ListKVMap.SetIfNotExist(key, value)
 }

 // SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
 // It returns false if `key` exists, and `value` would be ignored.
 func (m *ListMap) SetIfNotExistFunc(key any, f func() any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f())
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.ListKVMap.SetIfNotExistFunc(key, f)
 }

 // SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
@ -390,100 +221,52 @@ func (m *ListMap) SetIfNotExistFunc(key any, f func() any) bool {
 // SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
 // it executes function `f` with mutex.Lock of the map.
 func (m *ListMap) SetIfNotExistFuncLock(key any, f func() any) bool {
-	if !m.Contains(key) {
-		m.doSetWithLockCheck(key, f)
-		return true
-	}
-	return false
+	m.lazyInit()
+	return m.ListKVMap.SetIfNotExistFuncLock(key, f)
 }

 // Remove deletes value from map by given `key`, and return this deleted value.
 func (m *ListMap) Remove(key any) (value any) {
-	m.mu.Lock()
-	if m.data != nil {
-		if e, ok := m.data[key]; ok {
-			value = e.Value.(*gListMapNode).value
-			delete(m.data, key)
-			m.list.Remove(e)
-		}
-	}
-	m.mu.Unlock()
-	return
+	m.lazyInit()
+	return m.ListKVMap.Remove(key)
 }

 // Removes batch deletes values of the map by keys.
 func (m *ListMap) Removes(keys []any) {
-	m.mu.Lock()
-	if m.data != nil {
-		for _, key := range keys {
-			if e, ok := m.data[key]; ok {
-				delete(m.data, key)
-				m.list.Remove(e)
-			}
-		}
-	}
-	m.mu.Unlock()
+	m.lazyInit()
+	m.ListKVMap.Removes(keys)
 }

 // Keys returns all keys of the map as a slice in ascending order.
 func (m *ListMap) Keys() []any {
-	m.mu.RLock()
-	var (
-		keys  = make([]any, m.list.Len())
-		index = 0
-	)
-	if m.list != nil {
-		m.list.IteratorAsc(func(e *glist.Element) bool {
-			keys[index] = e.Value.(*gListMapNode).key
-			index++
-			return true
-		})
-	}
-	m.mu.RUnlock()
-	return keys
+	m.lazyInit()
+	return m.ListKVMap.Keys()
 }

 // Values returns all values of the map as a slice.
 func (m *ListMap) Values() []any {
-	m.mu.RLock()
-	var (
-		values = make([]any, m.list.Len())
-		index  = 0
-	)
-	if m.list != nil {
-		m.list.IteratorAsc(func(e *glist.Element) bool {
-			values[index] = e.Value.(*gListMapNode).value
-			index++
-			return true
-		})
-	}
-	m.mu.RUnlock()
-	return values
+	m.lazyInit()
+	return m.ListKVMap.Values()
 }

 // Contains checks whether a key exists.
 // It returns true if the `key` exists, or else false.
 func (m *ListMap) Contains(key any) (ok bool) {
-	m.mu.RLock()
-	if m.data != nil {
-		_, ok = m.data[key]
-	}
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.ListKVMap.Contains(key)
 }

 // Size returns the size of the map.
 func (m *ListMap) Size() (size int) {
-	m.mu.RLock()
-	size = len(m.data)
-	m.mu.RUnlock()
-	return
+	m.lazyInit()
+	return m.ListKVMap.Size()
 }

 // IsEmpty checks whether the map is empty.
 // It returns true if map is empty, or else false.
 func (m *ListMap) IsEmpty() bool {
-	return m.Size() == 0
+	m.lazyInit()
+	return m.ListKVMap.IsEmpty()
 }

 // Flip exchanges key-value of the map to value-key.
@ -498,90 +281,35 @@ func (m *ListMap) Flip() {
 // Merge merges two link maps.
 // The `other` map will be merged into the map `m`.
 func (m *ListMap) Merge(other *ListMap) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[any]*glist.Element)
-		m.list = glist.New()
-	}
-	if other != m {
-		other.mu.RLock()
-		defer other.mu.RUnlock()
-	}
-	var node *gListMapNode
-	other.list.IteratorAsc(func(e *glist.Element) bool {
-		node = e.Value.(*gListMapNode)
-		if e, ok := m.data[node.key]; !ok {
-			m.data[node.key] = m.list.PushBack(&gListMapNode{node.key, node.value})
-		} else {
-			e.Value = &gListMapNode{node.key, node.value}
-		}
-		return true
-	})
+	m.lazyInit()
+	other.lazyInit()
+	m.ListKVMap.Merge(other.ListKVMap)
 }

 // String returns the map as a string.
 func (m *ListMap) String() string {
-	if m == nil {
-		return ""
-	}
-	b, _ := m.MarshalJSON()
-	return string(b)
+	m.lazyInit()
+	return m.ListKVMap.String()
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (m ListMap) MarshalJSON() (jsonBytes []byte, err error) {
-	if m.data == nil {
-		return []byte("null"), nil
-	}
-	buffer := bytes.NewBuffer(nil)
-	buffer.WriteByte('{')
-	m.Iterator(func(key, value any) bool {
-		valueBytes, valueJSONErr := json.Marshal(value)
-		if valueJSONErr != nil {
-			err = valueJSONErr
-			return false
-		}
-		if buffer.Len() > 1 {
-			buffer.WriteByte(',')
-		}
-		fmt.Fprintf(buffer, `"%v":%s`, key, valueBytes)
-		return true
-	})
-	buffer.WriteByte('}')
-	return buffer.Bytes(), nil
+	return m.ListKVMap.MarshalJSON()
 }

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (m *ListMap) UnmarshalJSON(b []byte) error {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[any]*glist.Element)
-		m.list = glist.New()
-	}
-	var data map[string]any
-	if err := json.UnmarshalUseNumber(b, &data); err != nil {
-		return err
-	}
-	for key, value := range data {
-		if e, ok := m.data[key]; !ok {
-			m.data[key] = m.list.PushBack(&gListMapNode{key, value})
-		} else {
-			e.Value = &gListMapNode{key, value}
-		}
-	}
-	return nil
+	m.lazyInit()
+	return m.ListKVMap.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for map.
 func (m *ListMap) UnmarshalValue(value any) (err error) {
+	m.lazyInit()
+
 	m.mu.Lock()
 	defer m.mu.Unlock()
-	if m.data == nil {
-		m.data = make(map[any]*glist.Element)
-		m.list = glist.New()
-	}
+
 	for k, v := range gconv.Map(value) {
 		if e, ok := m.data[k]; !ok {
 			m.data[k] = m.list.PushBack(&gListMapNode{k, v})
@ -597,16 +325,8 @@ func (m *ListMap) DeepCopy() any {
 	if m == nil {
 		return nil
 	}
-	m.mu.RLock()
-	defer m.mu.RUnlock()
-	data := make(map[any]any, len(m.data))
-	if m.list != nil {
-		var node *gListMapNode
-		m.list.IteratorAsc(func(e *glist.Element) bool {
-			node = e.Value.(*gListMapNode)
-			data[node.key] = deepcopy.Copy(node.value)
-			return true
-		})
+	m.lazyInit()
+	return &ListMap{
+		ListKVMap: m.ListKVMap.DeepCopy().(*ListKVMap[any, any]),
 	}
-	return NewListMapFrom(data, m.mu.IsSafe())
 }
--- a/container/gmap/gmap_tree_k_v_map.go
+++ b/container/gmap/gmap_tree_k_v_map.go
@ -0,0 +1,32 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+//go:build go1.24
+
+package gmap
+
+import (
+	"github.com/gogf/gf/v2/container/gtree"
+)
+
+// TreeKVMap based on red-black tree, alias of RedBlackKVTree.
+type TreeKVMap[K comparable, V any] = gtree.RedBlackKVTree[K, V]
+
+// NewTreeKVMap instantiates a tree map with the custom comparator.
+// The parameter `safe` is used to specify whether using tree in concurrent-safety,
+// which is false in default.
+func NewTreeKVMap[K comparable, V any](comparator func(v1, v2 K) int, safe ...bool) *TreeKVMap[K, V] {
+	return gtree.NewRedBlackKVTree[K, V](comparator, safe...)
+}
+
+// NewTreeKVMapFrom instantiates a tree map with the custom comparator and `data` map.
+// Note that, the param `data` map will be set as the underlying data map(no deep copy),
+// there might be some concurrent-safe issues when changing the map outside.
+// The parameter `safe` is used to specify whether using tree in concurrent-safety,
+// which is false in default.
+func NewTreeKVMapFrom[K comparable, V any](comparator func(v1, v2 K) int, data map[K]V, safe ...bool) *TreeKVMap[K, V] {
+	return gtree.NewRedBlackKVTreeFrom(comparator, data, safe...)
+}
--- a/container/gmap/gmap_z_unit_hash_any_any_test.go
+++ b/container/gmap/gmap_z_unit_hash_any_any_test.go
@ -443,3 +443,49 @@ func Test_AnyAnyMap_Diff(t *testing.T) {
 		t.Assert(updatedKeys, []any{3})
 	})
 }
+
+func Test_AnyAnyMap_DoSetWithLockCheck_FuncValue(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		m := gmap.NewAnyAnyMap(true)
+
+		// Test GetOrSetFuncLock with function value
+		// Function should be executed and its return value should be set
+		callCount := 0
+		result := m.GetOrSetFuncLock(1, func() any {
+			callCount++
+			return "value1"
+		})
+		t.Assert(result, "value1")
+		t.Assert(callCount, 1)
+		t.Assert(m.Get(1), "value1")
+
+		// Test GetOrSetFuncLock again with same key
+		// Function should NOT be called since key exists
+		result = m.GetOrSetFuncLock(1, func() any {
+			callCount++
+			return "value2"
+		})
+		t.Assert(result, "value1")
+		t.Assert(callCount, 1) // Should still be 1, function not called
+
+		// Test SetIfNotExistFuncLock with function value
+		callCount = 0
+		ok := m.SetIfNotExistFuncLock(2, func() any {
+			callCount++
+			return "value2"
+		})
+		t.Assert(ok, true)
+		t.Assert(callCount, 1)
+		t.Assert(m.Get(2), "value2")
+
+		// Test SetIfNotExistFuncLock again with same key
+		// Function should NOT be called since key exists
+		ok = m.SetIfNotExistFuncLock(2, func() any {
+			callCount++
+			return "value3"
+		})
+		t.Assert(ok, false)
+		t.Assert(callCount, 1)       // Should still be 1, function not called
+		t.Assert(m.Get(2), "value2") // Value should not change
+	})
+}
--- a/container/gmap/gmap_z_unit_hash_str_any_test.go
+++ b/container/gmap/gmap_z_unit_hash_str_any_test.go
@ -96,6 +96,42 @@ func Test_StrAnyMap_Set_Fun(t *testing.T) {

 		t.Assert(m.SetIfNotExistFuncLock("b", getAny), false)
 		t.Assert(m.SetIfNotExistFuncLock("d", getAny), true)
+
+		type T struct {
+			A int
+		}
+
+		av := m.GetOrSetFunc("s1", func() any {
+			return &T{
+				A: 1,
+			}
+		})
+		ta, ok := av.(*T)
+		t.Assert(ok, true)
+		t.Assert(ta.A, 1)
+
+		av = m.GetOrSetFunc("s1", func() any {
+			return &T{
+				A: 2,
+			}
+		})
+		ta, ok = av.(*T)
+		t.Assert(ok, true)
+		t.Assert(ta.A, 1)
+
+		av = m.GetOrSet("s1", &T{
+			A: 3,
+		})
+		ta, ok = av.(*T)
+		t.Assert(ok, true)
+		t.Assert(ta.A, 1)
+
+		av = m.GetOrSet("s2", &T{
+			A: 4,
+		})
+		ta, ok = av.(*T)
+		t.Assert(ok, true)
+		t.Assert(ta.A, 4)
 	})
 }

--- a/container/gmap/gmap_z_unit_k_v_map_test.go
+++ b/container/gmap/gmap_z_unit_k_v_map_test.go
--- a/container/gmap/gmap_z_unit_list_k_v_map_race_test.go
+++ b/container/gmap/gmap_z_unit_list_k_v_map_race_test.go
@ -0,0 +1,326 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package gmap_test
+
+import (
+	"sync"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/gogf/gf/v2/container/gmap"
+	"github.com/gogf/gf/v2/test/gtest"
+)
+
+// Test_ListKVMap_GetOrSetFuncLock_Race tests the atomicity of GetOrSetFuncLock.
+// This test ensures that the callback function is only executed once even under
+// high concurrency, which verifies that the function holds the lock during the
+// entire check-and-set operation.
+func Test_ListKVMap_GetOrSetFuncLock_Race(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		m := gmap.NewListKVMap[string, int](true)
+		key := "counter"
+		callCount := int32(0)
+		goroutines := 100
+
+		var wg sync.WaitGroup
+		wg.Add(goroutines)
+
+		// Start multiple goroutines trying to set the same key
+		for i := 0; i < goroutines; i++ {
+			go func() {
+				defer wg.Done()
+				m.GetOrSetFuncLock(key, func() int {
+					// Increment call count atomically
+					atomic.AddInt32(&callCount, 1)
+					// Simulate some work
+					time.Sleep(time.Microsecond)
+					return 100
+				})
+			}()
+		}
+
+		wg.Wait()
+
+		// The callback should only be called once because of proper locking
+		t.Assert(atomic.LoadInt32(&callCount), 1)
+		t.Assert(m.Get(key), 100)
+		t.Assert(m.Size(), 1)
+	})
+}
+
+// Test_ListKVMap_SetIfNotExistFuncLock_Race tests the atomicity of SetIfNotExistFuncLock.
+// This test ensures that only one goroutine can successfully set the value and
+// execute the callback function, even under high concurrency.
+func Test_ListKVMap_SetIfNotExistFuncLock_Race(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		m := gmap.NewListKVMap[string, int](true)
+		key := "counter"
+		callCount := int32(0)
+		successCount := int32(0)
+		goroutines := 100
+
+		var wg sync.WaitGroup
+		wg.Add(goroutines)
+
+		// Start multiple goroutines trying to set the same key
+		for i := 0; i < goroutines; i++ {
+			go func() {
+				defer wg.Done()
+				success := m.SetIfNotExistFuncLock(key, func() int {
+					// Increment call count atomically
+					atomic.AddInt32(&callCount, 1)
+					// Simulate some work
+					time.Sleep(time.Microsecond)
+					return 200
+				})
+				if success {
+					atomic.AddInt32(&successCount, 1)
+				}
+			}()
+		}
+
+		wg.Wait()
+
+		// The callback should only be called once
+		t.Assert(atomic.LoadInt32(&callCount), 1)
+		// Only one goroutine should succeed
+		t.Assert(atomic.LoadInt32(&successCount), 1)
+		t.Assert(m.Get(key), 200)
+		t.Assert(m.Size(), 1)
+	})
+}
+
+// Test_ListKVMap_GetOrSetFuncLock_MultipleKeys tests GetOrSetFuncLock with different keys.
+// This ensures that operations on different keys don't interfere with each other.
+func Test_ListKVMap_GetOrSetFuncLock_MultipleKeys(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		m := gmap.NewListKVMap[string, int](true)
+		keys := []string{"key1", "key2", "key3", "key4", "key5"}
+		callCounts := make([]int32, len(keys))
+		goroutines := 20
+
+		var wg sync.WaitGroup
+
+		// For each key, start multiple goroutines
+		for i, key := range keys {
+			keyIndex := i
+			for j := 0; j < goroutines; j++ {
+				wg.Add(1)
+				go func(idx int, k string) {
+					defer wg.Done()
+					m.GetOrSetFuncLock(k, func() int {
+						atomic.AddInt32(&callCounts[idx], 1)
+						time.Sleep(time.Microsecond)
+						return (idx + 1) * 100
+					})
+				}(keyIndex, key)
+			}
+		}
+
+		wg.Wait()
+
+		// Each key's callback should only be called once
+		for _, count := range callCounts {
+			t.Assert(atomic.LoadInt32(&count), 1)
+		}
+
+		// Verify all keys are set correctly
+		for i, key := range keys {
+			t.Assert(m.Get(key), (i+1)*100)
+		}
+		t.Assert(m.Size(), len(keys))
+	})
+}
+
+// Test_ListKVMap_SetIfNotExistFuncLock_MultipleKeys tests SetIfNotExistFuncLock with different keys.
+func Test_ListKVMap_SetIfNotExistFuncLock_MultipleKeys(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		m := gmap.NewListKVMap[int, string](true)
+		keys := []int{1, 2, 3, 4, 5}
+		callCounts := make([]int32, len(keys))
+		successCounts := make([]int32, len(keys))
+		goroutines := 20
+
+		var wg sync.WaitGroup
+
+		// For each key, start multiple goroutines
+		for i, key := range keys {
+			keyIndex := i
+			for j := 0; j < goroutines; j++ {
+				wg.Add(1)
+				go func(idx int, k int) {
+					defer wg.Done()
+					success := m.SetIfNotExistFuncLock(k, func() string {
+						atomic.AddInt32(&callCounts[idx], 1)
+						time.Sleep(time.Microsecond)
+						return gtest.DataContent()
+					})
+					if success {
+						atomic.AddInt32(&successCounts[idx], 1)
+					}
+				}(keyIndex, key)
+			}
+		}
+
+		wg.Wait()
+
+		// Each key's callback should only be called once
+		for _, count := range callCounts {
+			t.Assert(atomic.LoadInt32(&count), 1)
+		}
+
+		// Each key should have exactly one successful set
+		for _, count := range successCounts {
+			t.Assert(atomic.LoadInt32(&count), 1)
+		}
+
+		t.Assert(m.Size(), len(keys))
+	})
+}
+
+// Test_ListKVMap_GetOrSetFuncLock_NilValue tests that nil values are handled correctly.
+func Test_ListKVMap_GetOrSetFuncLock_NilValue(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		m := gmap.NewListKVMap[string, *int](true)
+		key := "nilKey"
+		callCount := int32(0)
+
+		var wg sync.WaitGroup
+		goroutines := 50
+		wg.Add(goroutines)
+
+		for i := 0; i < goroutines; i++ {
+			go func() {
+				defer wg.Done()
+				m.GetOrSetFuncLock(key, func() *int {
+					atomic.AddInt32(&callCount, 1)
+					return nil
+				})
+			}()
+		}
+
+		wg.Wait()
+
+		// Callback should be called once
+		t.Assert(atomic.LoadInt32(&callCount), 1)
+		// Typed nil pointer (*int)(nil) is stored because any(value) != nil for typed nil
+		// This is a Go language feature: typed nil is not the same as interface nil
+		t.Assert(m.Contains(key), true)
+		t.Assert(m.Get(key), (*int)(nil))
+		t.Assert(m.Size(), 1)
+	})
+}
+
+// Test_ListKVMap_SetIfNotExistFuncLock_NilValue tests that nil values are handled correctly.
+func Test_ListKVMap_SetIfNotExistFuncLock_NilValue(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		m := gmap.NewListKVMap[string, *string](true)
+		key := "nilKey"
+		callCount := int32(0)
+		successCount := int32(0)
+
+		var wg sync.WaitGroup
+		goroutines := 50
+		wg.Add(goroutines)
+
+		for i := 0; i < goroutines; i++ {
+			go func() {
+				defer wg.Done()
+				success := m.SetIfNotExistFuncLock(key, func() *string {
+					atomic.AddInt32(&callCount, 1)
+					return nil
+				})
+				if success {
+					atomic.AddInt32(&successCount, 1)
+				}
+			}()
+		}
+
+		wg.Wait()
+
+		// Callback should be called once
+		t.Assert(atomic.LoadInt32(&callCount), 1)
+		// Should report success once
+		t.Assert(atomic.LoadInt32(&successCount), 1)
+		// Typed nil pointer (*string)(nil) is stored because any(value) != nil for typed nil
+		t.Assert(m.Contains(key), true)
+		t.Assert(m.Get(key), (*string)(nil))
+		t.Assert(m.Size(), 1)
+	})
+}
+
+// Test_ListKVMap_GetOrSetFuncLock_ExistingKey tests behavior when key already exists.
+func Test_ListKVMap_GetOrSetFuncLock_ExistingKey(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		m := gmap.NewListKVMap[string, int](true)
+		key := "existing"
+		m.Set(key, 999)
+
+		callCount := int32(0)
+		goroutines := 50
+
+		var wg sync.WaitGroup
+		wg.Add(goroutines)
+
+		for i := 0; i < goroutines; i++ {
+			go func() {
+				defer wg.Done()
+				val := m.GetOrSetFuncLock(key, func() int {
+					atomic.AddInt32(&callCount, 1)
+					return 123
+				})
+				// Should always get the existing value
+				t.Assert(val, 999)
+			}()
+		}
+
+		wg.Wait()
+
+		// Callback should never be called since key exists
+		t.Assert(atomic.LoadInt32(&callCount), 0)
+		t.Assert(m.Get(key), 999)
+	})
+}
+
+// Test_ListKVMap_SetIfNotExistFuncLock_ExistingKey tests behavior when key already exists.
+func Test_ListKVMap_SetIfNotExistFuncLock_ExistingKey(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		m := gmap.NewListKVMap[string, int](true)
+		key := "existing"
+		m.Set(key, 888)
+
+		callCount := int32(0)
+		successCount := int32(0)
+		goroutines := 50
+
+		var wg sync.WaitGroup
+		wg.Add(goroutines)
+
+		for i := 0; i < goroutines; i++ {
+			go func() {
+				defer wg.Done()
+				success := m.SetIfNotExistFuncLock(key, func() int {
+					atomic.AddInt32(&callCount, 1)
+					return 456
+				})
+				if success {
+					atomic.AddInt32(&successCount, 1)
+				}
+			}()
+		}
+
+		wg.Wait()
+
+		// Callback should never be called since key exists
+		t.Assert(atomic.LoadInt32(&callCount), 0)
+		// No goroutine should succeed
+		t.Assert(atomic.LoadInt32(&successCount), 0)
+		// Original value should remain
+		t.Assert(m.Get(key), 888)
+	})
+}
--- a/container/gmap/gmap_z_unit_list_k_v_map_test.go
+++ b/container/gmap/gmap_z_unit_list_k_v_map_test.go
--- a/container/gpool/gpool.go
+++ b/container/gpool/gpool.go
@ -8,41 +8,19 @@
 package gpool

 import (
-	"context"
 	"time"
-
-	"github.com/gogf/gf/v2/container/glist"
-	"github.com/gogf/gf/v2/container/gtype"
-	"github.com/gogf/gf/v2/errors/gcode"
-	"github.com/gogf/gf/v2/errors/gerror"
-	"github.com/gogf/gf/v2/os/gtime"
-	"github.com/gogf/gf/v2/os/gtimer"
 )

 // Pool is an Object-Reusable Pool.
 type Pool struct {
-	list    *glist.List         // Available/idle items list.
-	closed  *gtype.Bool         // Whether the pool is closed.
-	TTL     time.Duration       // Time To Live for pool items.
-	NewFunc func() (any, error) // Callback function to create pool item.
-	// ExpireFunc is the function for expired items destruction.
-	// This function needs to be defined when the pool items
-	// need to perform additional destruction operations.
-	// Eg: net.Conn, os.File, etc.
-	ExpireFunc func(any)
-}
-
-// Pool item.
-type poolItem struct {
-	value    any   // Item value.
-	expireAt int64 // Expire timestamp in milliseconds.
+	*TPool[any]
 }

 // NewFunc Creation function for object.
-type NewFunc func() (any, error)
+type NewFunc = TPoolNewFunc[any]

 // ExpireFunc Destruction function for object.
-type ExpireFunc func(any)
+type ExpireFunc = TPoolExpireFunc[any]

 // New creates and returns a new object pool.
 // To ensure execution efficiency, the expiration time cannot be modified once it is set.
@ -52,134 +30,40 @@ type ExpireFunc func(any)
 // ttl < 0 : immediate expired after use;
 // ttl > 0 : timeout expired;
 func New(ttl time.Duration, newFunc NewFunc, expireFunc ...ExpireFunc) *Pool {
-	r := &Pool{
-		list:    glist.New(true),
-		closed:  gtype.NewBool(),
-		TTL:     ttl,
-		NewFunc: newFunc,
+	return &Pool{
+		TPool: NewTPool(ttl, newFunc, expireFunc...),
 	}
-	if len(expireFunc) > 0 {
-		r.ExpireFunc = expireFunc[0]
-	}
-	gtimer.AddSingleton(context.Background(), time.Second, r.checkExpireItems)
-	return r
 }

 // Put puts an item to pool.
 func (p *Pool) Put(value any) error {
-	if p.closed.Val() {
-		return gerror.NewCode(gcode.CodeInvalidOperation, "pool is closed")
-	}
-	item := &poolItem{
-		value: value,
-	}
-	if p.TTL == 0 {
-		item.expireAt = 0
-	} else {
-		// As for Golang version < 1.13, there's no method Milliseconds for time.Duration.
-		// So we need calculate the milliseconds using its nanoseconds value.
-		item.expireAt = gtime.TimestampMilli() + p.TTL.Nanoseconds()/1000000
-	}
-	p.list.PushBack(item)
-	return nil
+	return p.TPool.Put(value)
 }

 // MustPut puts an item to pool, it panics if any error occurs.
 func (p *Pool) MustPut(value any) {
-	if err := p.Put(value); err != nil {
-		panic(err)
-	}
+	p.TPool.MustPut(value)
 }

 // Clear clears pool, which means it will remove all items from pool.
 func (p *Pool) Clear() {
-	if p.ExpireFunc != nil {
-		for {
-			if r := p.list.PopFront(); r != nil {
-				p.ExpireFunc(r.(*poolItem).value)
-			} else {
-				break
-			}
-		}
-	} else {
-		p.list.RemoveAll()
-	}
+	p.TPool.Clear()
 }

 // Get picks and returns an item from pool. If the pool is empty and NewFunc is defined,
 // it creates and returns one from NewFunc.
 func (p *Pool) Get() (any, error) {
-	for !p.closed.Val() {
-		if r := p.list.PopFront(); r != nil {
-			f := r.(*poolItem)
-			if f.expireAt == 0 || f.expireAt > gtime.TimestampMilli() {
-				return f.value, nil
-			} else if p.ExpireFunc != nil {
-				// TODO: move expire function calling asynchronously out from `Get` operation.
-				p.ExpireFunc(f.value)
-			}
-		} else {
-			break
-		}
-	}
-	if p.NewFunc != nil {
-		return p.NewFunc()
-	}
-	return nil, gerror.NewCode(gcode.CodeInvalidOperation, "pool is empty")
+	return p.TPool.Get()
 }

 // Size returns the count of available items of pool.
 func (p *Pool) Size() int {
-	return p.list.Len()
+	return p.TPool.Size()
 }

 // Close closes the pool. If `p` has ExpireFunc,
 // then it automatically closes all items using this function before it's closed.
 // Commonly you do not need to call this function manually.
 func (p *Pool) Close() {
-	p.closed.Set(true)
-}
-
-// checkExpire removes expired items from pool in every second.
-func (p *Pool) checkExpireItems(ctx context.Context) {
-	if p.closed.Val() {
-		// If p has ExpireFunc,
-		// then it must close all items using this function.
-		if p.ExpireFunc != nil {
-			for {
-				if r := p.list.PopFront(); r != nil {
-					p.ExpireFunc(r.(*poolItem).value)
-				} else {
-					break
-				}
-			}
-		}
-		gtimer.Exit()
-	}
-	// All items do not expire.
-	if p.TTL == 0 {
-		return
-	}
-	// The latest item expire timestamp in milliseconds.
-	var latestExpire int64 = -1
-	// Retrieve the current timestamp in milliseconds, it expires the items
-	// by comparing with this timestamp. It is not accurate comparison for
-	// every item expired, but high performance.
-	var timestampMilli = gtime.TimestampMilli()
-	for latestExpire <= timestampMilli {
-		if r := p.list.PopFront(); r != nil {
-			item := r.(*poolItem)
-			latestExpire = item.expireAt
-			// TODO improve the auto-expiration mechanism of the pool.
-			if item.expireAt > timestampMilli {
-				p.list.PushFront(item)
-				break
-			}
-			if p.ExpireFunc != nil {
-				p.ExpireFunc(item.value)
-			}
-		} else {
-			break
-		}
-	}
+	p.TPool.Close()
 }
--- a/container/gpool/gpool_t.go
+++ b/container/gpool/gpool_t.go
@ -0,0 +1,183 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package gpool
+
+import (
+	"context"
+	"time"
+
+	"github.com/gogf/gf/v2/container/glist"
+	"github.com/gogf/gf/v2/container/gtype"
+	"github.com/gogf/gf/v2/errors/gcode"
+	"github.com/gogf/gf/v2/errors/gerror"
+	"github.com/gogf/gf/v2/os/gtime"
+	"github.com/gogf/gf/v2/os/gtimer"
+)
+
+// TPool is an Object-Reusable Pool.
+type TPool[T any] struct {
+	list    *glist.TList[*tPoolItem[T]] // Available/idle items list.
+	closed  *gtype.Bool                 // Whether the pool is closed.
+	TTL     time.Duration               // Time To Live for pool items.
+	NewFunc func() (T, error)           // Callback function to create pool item.
+	// ExpireFunc is the function for expired items destruction.
+	// This function needs to be defined when the pool items
+	// need to perform additional destruction operations.
+	// Eg: net.Conn, os.File, etc.
+	ExpireFunc func(T)
+}
+
+// TPool item.
+type tPoolItem[T any] struct {
+	value    T     // Item value.
+	expireAt int64 // Expire timestamp in milliseconds.
+}
+
+// TPoolNewFunc Creation function for object.
+type TPoolNewFunc[T any] func() (T, error)
+
+// TPoolExpireFunc Destruction function for object.
+type TPoolExpireFunc[T any] func(T)
+
+// NewTPool creates and returns a new object pool.
+// To ensure execution efficiency, the expiration time cannot be modified once it is set.
+//
+// Note the expiration logic:
+// ttl = 0 : not expired;
+// ttl < 0 : immediate expired after use;
+// ttl > 0 : timeout expired;
+func NewTPool[T any](ttl time.Duration, newFunc TPoolNewFunc[T], expireFunc ...TPoolExpireFunc[T]) *TPool[T] {
+	r := &TPool[T]{
+		list:    glist.NewT[*tPoolItem[T]](true),
+		closed:  gtype.NewBool(),
+		TTL:     ttl,
+		NewFunc: newFunc,
+	}
+	if len(expireFunc) > 0 {
+		r.ExpireFunc = expireFunc[0]
+	}
+	gtimer.AddSingleton(context.Background(), time.Second, r.checkExpireItems)
+	return r
+}
+
+// Put puts an item to pool.
+func (p *TPool[T]) Put(value T) error {
+	if p.closed.Val() {
+		return gerror.NewCode(gcode.CodeInvalidOperation, "pool is closed")
+	}
+	item := &tPoolItem[T]{
+		value: value,
+	}
+	if p.TTL == 0 {
+		item.expireAt = 0
+	} else {
+		// As for Golang version < 1.13, there's no method Milliseconds for time.Duration.
+		// So we need calculate the milliseconds using its nanoseconds value.
+		item.expireAt = gtime.TimestampMilli() + p.TTL.Nanoseconds()/1000000
+	}
+	p.list.PushBack(item)
+	return nil
+}
+
+// MustPut puts an item to pool, it panics if any error occurs.
+func (p *TPool[T]) MustPut(value T) {
+	if err := p.Put(value); err != nil {
+		panic(err)
+	}
+}
+
+// Clear clears pool, which means it will remove all items from pool.
+func (p *TPool[T]) Clear() {
+	if p.ExpireFunc != nil {
+		for {
+			if r := p.list.PopFront(); r != nil {
+				p.ExpireFunc(r.value)
+			} else {
+				break
+			}
+		}
+	} else {
+		p.list.RemoveAll()
+	}
+}
+
+// Get picks and returns an item from pool. If the pool is empty and NewFunc is defined,
+// it creates and returns one from NewFunc.
+func (p *TPool[T]) Get() (value T, err error) {
+	for !p.closed.Val() {
+		if f := p.list.PopFront(); f != nil {
+			if f.expireAt == 0 || f.expireAt > gtime.TimestampMilli() {
+				return f.value, nil
+			} else if p.ExpireFunc != nil {
+				// TODO: move expire function calling asynchronously out from `Get` operation.
+				p.ExpireFunc(f.value)
+			}
+		} else {
+			break
+		}
+	}
+	if p.NewFunc != nil {
+		return p.NewFunc()
+	}
+	err = gerror.NewCode(gcode.CodeInvalidOperation, "pool is empty")
+	return
+}
+
+// Size returns the count of available items of pool.
+func (p *TPool[T]) Size() int {
+	return p.list.Len()
+}
+
+// Close closes the pool. If `p` has ExpireFunc,
+// then it automatically closes all items using this function before it's closed.
+// Commonly you do not need to call this function manually.
+func (p *TPool[T]) Close() {
+	p.closed.Set(true)
+}
+
+// checkExpire removes expired items from pool in every second.
+func (p *TPool[T]) checkExpireItems(ctx context.Context) {
+	if p.closed.Val() {
+		// If p has ExpireFunc,
+		// then it must close all items using this function.
+		if p.ExpireFunc != nil {
+			for {
+				if r := p.list.PopFront(); r != nil {
+					p.ExpireFunc(r.value)
+				} else {
+					break
+				}
+			}
+		}
+		gtimer.Exit()
+	}
+	// All items do not expire.
+	if p.TTL == 0 {
+		return
+	}
+	// The latest item expire timestamp in milliseconds.
+	var latestExpire int64 = -1
+	// Retrieve the current timestamp in milliseconds, it expires the items
+	// by comparing with this timestamp. It is not accurate comparison for
+	// every item expired, but high performance.
+	var timestampMilli = gtime.TimestampMilli()
+	for latestExpire <= timestampMilli {
+		if item := p.list.PopFront(); item != nil {
+			latestExpire = item.expireAt
+			// TODO improve the auto-expiration mechanism of the pool.
+			if item.expireAt > timestampMilli {
+				p.list.PushFront(item)
+				break
+			}
+			if p.ExpireFunc != nil {
+				p.ExpireFunc(item.value)
+			}
+		} else {
+			break
+		}
+	}
+}
--- a/container/gpool/gpool_z_unit_generic_test.go
+++ b/container/gpool/gpool_z_unit_generic_test.go
@ -0,0 +1,112 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package gpool_test
+
+import (
+	"testing"
+	"time"
+
+	"github.com/gogf/gf/v2/container/gpool"
+	"github.com/gogf/gf/v2/container/gtype"
+	"github.com/gogf/gf/v2/frame/g"
+	"github.com/gogf/gf/v2/test/gtest"
+)
+
+func Test_TPool_Int(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		// Create a pool for int
+		var (
+			newFunc = func() (int, error) {
+				return 100, nil
+			}
+			expireVal  = gtype.NewInt(0)
+			expireFunc = func(i int) {
+				expireVal.Set(i)
+			}
+		)
+
+		// TTL = 0, no expiration by time
+		p := gpool.NewTPool(0, newFunc, expireFunc)
+
+		// Test Put and Get
+		p.Put(1)
+		p.Put(2)
+		t.Assert(p.Size(), 2)
+
+		v, err := p.Get()
+		t.AssertNil(err)
+		t.AssertIN(v, g.Slice{1, 2})
+
+		v, err = p.Get()
+		t.AssertNil(err)
+		t.AssertIN(v, g.Slice{1, 2})
+
+		t.Assert(p.Size(), 0)
+
+		// Test NewFunc when empty
+		v, err = p.Get()
+		t.AssertNil(err)
+		t.Assert(v, 100)
+
+		// Test Clear and ExpireFunc
+		p.Put(50)
+		t.Assert(p.Size(), 1)
+		p.Clear()
+		t.Assert(p.Size(), 0)
+		t.Assert(expireVal.Val(), 50)
+
+		// Test Close
+		p.Put(60)
+		p.Close()
+		// Close should trigger expire for existing items?
+		// Looking at implementation: Close() sets closed=true.
+		// It does NOT automatically clear items unless checkExpireItems runs or we call Clear?
+		// Wait, checkExpireItems checks closed.Val(). If closed, it clears items.
+		// But checkExpireItems runs in a separate goroutine every second.
+		// So we might need to wait or trigger it.
+		// Actually, let's check the implementation of Close again.
+		/*
+			func (p *TPool[T]) Close() {
+				p.closed.Set(true)
+			}
+		*/
+		// And checkExpireItems:
+		/*
+			func (p *TPool[T]) checkExpireItems(ctx context.Context) {
+				if p.closed.Val() {
+					// ... clears items ...
+					gtimer.Exit()
+				}
+				// ...
+			}
+		*/
+		// So it relies on the timer to clean up.
+	})
+}
+
+func Test_TPool_Struct(t *testing.T) {
+	type User struct {
+		Id   int
+		Name string
+	}
+
+	gtest.C(t, func(t *gtest.T) {
+		p := gpool.NewTPool[User](time.Hour, nil)
+		u1 := User{Id: 1, Name: "john"}
+		p.Put(u1)
+
+		v, err := p.Get()
+		t.AssertNil(err)
+		t.Assert(v, u1)
+
+		// Test empty with no NewFunc
+		v, err = p.Get()
+		t.AssertNE(err, nil)
+		t.Assert(err.Error(), "pool is empty")
+		t.Assert(v, User{}) // Zero value
+	})
+}
--- a/container/gpool/gpool_z_unit_test.go
+++ b/container/gpool/gpool_z_unit_test.go
@ -77,7 +77,7 @@ func Test_Gpool(t *testing.T) {
 		t.Assert(err2, errors.New("pool is empty"))
 		t.Assert(v2, nil)
 		// test close expireFunc
-		for index := 0; index < 10; index++ {
+		for index := range 10 {
 			p2.Put(index)
 		}
 		t.Assert(p2.Size(), 10)
--- a/container/gqueue/gqueue.go
+++ b/container/gqueue/gqueue.go
@ -17,20 +17,9 @@
 // 4. Blocking when reading data from queue;
 package gqueue

-import (
-	"math"
-
-	"github.com/gogf/gf/v2/container/glist"
-	"github.com/gogf/gf/v2/container/gtype"
-)
-
 // Queue is a concurrent-safe queue built on doubly linked list and channel.
 type Queue struct {
-	limit  int           // Limit for queue size.
-	list   *glist.List   // Underlying list structure for data maintaining.
-	closed *gtype.Bool   // Whether queue is closed.
-	events chan struct{} // Events for data writing.
-	C      chan any      // Underlying channel for data reading.
+	*TQueue[any]
 }

 const (
@ -42,74 +31,35 @@ const (
 // Optional parameter `limit` is used to limit the size of the queue, which is unlimited in default.
 // When `limit` is given, the queue will be static and high performance which is comparable with stdlib channel.
 func New(limit ...int) *Queue {
-	q := &Queue{
-		closed: gtype.NewBool(),
+	return &Queue{
+		TQueue: NewTQueue[any](limit...),
 	}
-	if len(limit) > 0 && limit[0] > 0 {
-		q.limit = limit[0]
-		q.C = make(chan any, limit[0])
-	} else {
-		q.list = glist.New(true)
-		q.events = make(chan struct{}, math.MaxInt32)
-		q.C = make(chan any, defaultQueueSize)
-		go q.asyncLoopFromListToChannel()
-	}
-	return q
 }

 // Push pushes the data `v` into the queue.
 // Note that it would panic if Push is called after the queue is closed.
 func (q *Queue) Push(v any) {
-	if q.limit > 0 {
-		q.C <- v
-	} else {
-		q.list.PushBack(v)
-		if len(q.events) < defaultQueueSize {
-			q.events <- struct{}{}
-		}
-	}
+	q.TQueue.Push(v)
 }

 // Pop pops an item from the queue in FIFO way.
 // Note that it would return nil immediately if Pop is called after the queue is closed.
 func (q *Queue) Pop() any {
-	return <-q.C
+	return q.TQueue.Pop()
 }

 // Close closes the queue.
 // Notice: It would notify all goroutines return immediately,
 // which are being blocked reading using Pop method.
 func (q *Queue) Close() {
-	if !q.closed.Cas(false, true) {
-		return
-	}
-	if q.events != nil {
-		close(q.events)
-	}
-	if q.limit > 0 {
-		close(q.C)
-	} else {
-		for range defaultBatchSize {
-			q.Pop()
-		}
-	}
+	q.TQueue.Close()
 }

 // Len returns the length of the queue.
 // Note that the result might not be accurate if using unlimited queue size as there's an
 // asynchronous channel reading the list constantly.
 func (q *Queue) Len() (length int64) {
-	bufferedSize := int64(len(q.C))
-	if q.limit > 0 {
-		return bufferedSize
-	}
-	// If the queue is unlimited and the buffered size is exactly the default size,
-	// it means there might be some data in the list not synchronized to channel yet.
-	// So we need to add 1 to the buffered size to make the result more accurate.
-	if bufferedSize == defaultQueueSize {
-		bufferedSize++
-	}
-	return int64(q.list.Size()) + bufferedSize
+	return q.TQueue.Len()
 }

 // Size is alias of Len.
@ -118,34 +68,3 @@ func (q *Queue) Len() (length int64) {
 func (q *Queue) Size() int64 {
 	return q.Len()
 }
-
-// asyncLoopFromListToChannel starts an asynchronous goroutine,
-// which handles the data synchronization from list `q.list` to channel `q.C`.
-func (q *Queue) asyncLoopFromListToChannel() {
-	defer func() {
-		if q.closed.Val() {
-			_ = recover()
-		}
-	}()
-	for !q.closed.Val() {
-		<-q.events
-		for !q.closed.Val() {
-			if bufferLength := q.list.Len(); bufferLength > 0 {
-				// When q.C is closed, it will panic here, especially q.C is being blocked for writing.
-				// If any error occurs here, it will be caught by recover and be ignored.
-				for range bufferLength {
-					q.C <- q.list.PopFront()
-				}
-			} else {
-				break
-			}
-		}
-		// Clear q.events to remain just one event to do the next synchronization check.
-		for i := 0; i < len(q.events)-1; i++ {
-			<-q.events
-		}
-	}
-	// It should be here to close `q.C` if `q` is unlimited size.
-	// It's the sender's responsibility to close channel when it should be closed.
-	close(q.C)
-}
--- a/container/gqueue/gqueue_t.go
+++ b/container/gqueue/gqueue_t.go
@ -0,0 +1,134 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+package gqueue
+
+import (
+	"math"
+
+	"github.com/gogf/gf/v2/container/glist"
+	"github.com/gogf/gf/v2/container/gtype"
+)
+
+// TQueue is a concurrent-safe queue built on doubly linked list and channel.
+type TQueue[T any] struct {
+	limit  int             // Limit for queue size.
+	list   *glist.TList[T] // Underlying list structure for data maintaining.
+	closed *gtype.Bool     // Whether queue is closed.
+	events chan struct{}   // Events for data writing.
+	C      chan T          // Underlying channel for data reading.
+}
+
+// NewTQueue returns an empty queue object.
+// Optional parameter `limit` is used to limit the size of the queue, which is unlimited in default.
+// When `limit` is given, the queue will be static and high performance which is comparable with stdlib channel.
+func NewTQueue[T any](limit ...int) *TQueue[T] {
+	q := &TQueue[T]{
+		closed: gtype.NewBool(),
+	}
+	if len(limit) > 0 && limit[0] > 0 {
+		q.limit = limit[0]
+		q.C = make(chan T, limit[0])
+	} else {
+		q.list = glist.NewT[T](true)
+		q.events = make(chan struct{}, math.MaxInt32)
+		q.C = make(chan T, defaultQueueSize)
+		go q.asyncLoopFromListToChannel()
+	}
+	return q
+}
+
+// Push pushes the data `v` into the queue.
+// Note that it would panic if Push is called after the queue is closed.
+func (q *TQueue[T]) Push(v T) {
+	if q.limit > 0 {
+		q.C <- v
+	} else {
+		q.list.PushBack(v)
+		if len(q.events) < defaultQueueSize {
+			q.events <- struct{}{}
+		}
+	}
+}
+
+// Pop pops an item from the queue in FIFO way.
+// Note that it would return nil immediately if Pop is called after the queue is closed.
+func (q *TQueue[T]) Pop() T {
+	return <-q.C
+}
+
+// Close closes the queue.
+// Notice: It would notify all goroutines return immediately,
+// which are being blocked reading using Pop method.
+func (q *TQueue[T]) Close() {
+	if !q.closed.Cas(false, true) {
+		return
+	}
+	if q.events != nil {
+		close(q.events)
+	}
+	if q.limit > 0 {
+		close(q.C)
+	} else {
+		for range defaultBatchSize {
+			q.Pop()
+		}
+	}
+}
+
+// Len returns the length of the queue.
+// Note that the result might not be accurate if using unlimited queue size as there's an
+// asynchronous channel reading the list constantly.
+func (q *TQueue[T]) Len() (length int64) {
+	bufferedSize := int64(len(q.C))
+	if q.limit > 0 {
+		return bufferedSize
+	}
+	// If the queue is unlimited and the buffered size is exactly the default size,
+	// it means there might be some data in the list not synchronized to channel yet.
+	// So we need to add 1 to the buffered size to make the result more accurate.
+	if bufferedSize == defaultQueueSize {
+		bufferedSize++
+	}
+	return int64(q.list.Size()) + bufferedSize
+}
+
+// Size is alias of Len.
+//
+// Deprecated: use Len instead.
+func (q *TQueue[T]) Size() int64 {
+	return q.Len()
+}
+
+// asyncLoopFromListToChannel starts an asynchronous goroutine,
+// which handles the data synchronization from list `q.list` to channel `q.C`.
+func (q *TQueue[T]) asyncLoopFromListToChannel() {
+	defer func() {
+		if q.closed.Val() {
+			_ = recover()
+		}
+	}()
+	for !q.closed.Val() {
+		<-q.events
+		for !q.closed.Val() {
+			if bufferLength := q.list.Len(); bufferLength > 0 {
+				// When q.C is closed, it will panic here, especially q.C is being blocked for writing.
+				// If any error occurs here, it will be caught by recover and be ignored.
+				for range bufferLength {
+					q.C <- q.list.PopFront()
+				}
+			} else {
+				break
+			}
+		}
+		// Clear q.events to remain just one event to do the next synchronization check.
+		for i := 0; i < len(q.events)-1; i++ {
+			<-q.events
+		}
+	}
+	// It should be here to close `q.C` if `q` is unlimited size.
+	// It's the sender's responsibility to close channel when it should be closed.
+	close(q.C)
+}
--- a/container/gqueue/gqueue_z_unit_test.go
+++ b/container/gqueue/gqueue_z_unit_test.go
@ -128,3 +128,218 @@ func TestIssue4376(t *testing.T) {
 		t.Log(gq.Len(), len(cq))
 	})
 }
+
+// Test static queue (with limit) close operation
+func TestQueue_StaticClose(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New(10)
+		defer func() {
+			if err := recover(); err == nil {
+				t.Log("Close succeeded")
+			}
+		}()
+		q.Push(1)
+		q.Push(2)
+		q.Close()
+		// After closing, Pop should return nil
+		v := q.Pop()
+		t.Assert(v, nil)
+	})
+}
+
+// Test Size() method (deprecated alias of Len)
+func TestQueue_Size(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New(20)
+		for i := range 10 {
+			q.Push(i)
+		}
+		t.Assert(q.Size(), 10)
+		t.Assert(q.Len(), 10)
+		q.Close()
+	})
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New()
+		for i := range 15 {
+			q.Push(i)
+		}
+		time.Sleep(10 * time.Millisecond)
+		t.Assert(q.Size(), q.Len())
+		q.Close()
+	})
+}
+
+// Test TQueue directly with generic type
+func TestTQueue_Generic(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		// Test with custom type
+		q := gqueue.NewTQueue[string]()
+		defer q.Close()
+		q.Push("hello")
+		q.Push("world")
+		t.Assert(q.Pop(), "hello")
+		t.Assert(q.Pop(), "world")
+	})
+}
+
+// Test TQueue Size method directly
+func TestTQueue_Size(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.NewTQueue[int]()
+		defer q.Close()
+		for i := range 10 {
+			q.Push(i)
+		}
+		time.Sleep(10 * time.Millisecond)
+		// Size is an alias of Len for TQueue
+		t.Assert(q.Size(), q.Len())
+	})
+}
+
+// Test TQueue with static limit
+func TestTQueue_StaticLimit(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.NewTQueue[int](5)
+		defer q.Close()
+		for i := range 5 {
+			q.Push(i)
+		}
+		t.Assert(q.Len(), 5)
+		for i := range 5 {
+			t.Assert(q.Pop(), i)
+		}
+		t.Assert(q.Len(), 0)
+	})
+}
+
+// Test queue with large data push/pop
+func TestQueue_LargeDataScale(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New()
+		defer q.Close()
+		n := 5000
+		for i := range n {
+			q.Push(i)
+		}
+		time.Sleep(50 * time.Millisecond)
+		// Pop should retrieve all items in order
+		for i := range n {
+			v := q.Pop()
+			t.Assert(v, i)
+		}
+	})
+}
+
+// Test double close (idempotent close)
+func TestQueue_DoubleClose(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New()
+		q.Push(1)
+		q.Close()
+		// Second close should not panic
+		q.Close()
+		t.Assert(q.Pop(), nil)
+	})
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New(10)
+		q.Push(1)
+		q.Close()
+		// Second close should not panic for static queue
+		q.Close()
+		// Pop from closed static queue returns the buffered value
+		v := q.Pop()
+		t.Assert(v, 1)
+	})
+}
+
+// Test concurrent push and pop
+func TestQueue_ConcurrentPushPop(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New()
+		defer q.Close()
+		// Producer goroutine
+		go func() {
+			for i := range 100 {
+				q.Push(i)
+			}
+			time.Sleep(50 * time.Millisecond)
+			q.Close()
+		}()
+		// Consumer
+		count := 0
+		for {
+			v := q.Pop()
+			if v == nil {
+				break
+			}
+			count++
+		}
+		t.AssertGE(count, 1)
+	})
+}
+
+// Test Pop on empty queue returns nil when closed
+func TestQueue_PopEmptyClosed(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New()
+		q.Close()
+		v := q.Pop()
+		t.Assert(v, nil)
+	})
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New(10)
+		q.Close()
+		v := q.Pop()
+		t.Assert(v, nil)
+	})
+}
+
+// Test Len with dynamic queue at capacity boundary
+func TestQueue_LenAtBoundary(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New()
+		defer q.Close()
+		// Push exactly defaultQueueSize items to test boundary condition
+		for i := range 10000 {
+			q.Push(i)
+		}
+		time.Sleep(50 * time.Millisecond)
+		len := q.Len()
+		t.AssertGE(len, 0)
+	})
+}
+
+// Test Close on dynamic queue with pending asyncLoopFromListToChannel
+func TestQueue_CloseWithAsyncLoop(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New()
+		// Push some data to activate asyncLoopFromListToChannel
+		for i := range 100 {
+			q.Push(i)
+		}
+		// Immediately close
+		q.Close()
+		// Pop should return values until exhausted, then nil
+		for {
+			v := q.Pop()
+			if v == nil {
+				break
+			}
+		}
+		t.Assert(q.Pop(), nil)
+	})
+}
+
+// Test static queue edge case with zero limit (should create unlimited queue)
+func TestQueue_ZeroLimitCreatesUnlimited(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		q := gqueue.New(0)
+		defer q.Close()
+		for i := range 100 {
+			q.Push(i)
+		}
+		time.Sleep(10 * time.Millisecond)
+		len := q.Len()
+		t.Assert(len, 100)
+	})
+}
--- a/container/gring/gring.go
+++ b/container/gring/gring.go
@ -9,27 +9,11 @@
 // Deprecated.
 package gring

-import (
-	"container/ring"
-
-	"github.com/gogf/gf/v2/container/gtype"
-	"github.com/gogf/gf/v2/internal/rwmutex"
-)
-
 // Ring is a struct of ring structure.
 //
 // Deprecated.
 type Ring struct {
-	mu    *rwmutex.RWMutex
-	ring  *ring.Ring  // Underlying ring.
-	len   *gtype.Int  // Length(already used size).
-	cap   *gtype.Int  // Capability(>=len).
-	dirty *gtype.Bool // Dirty, which means the len and cap should be recalculated. It's marked dirty when the size of ring changes.
-}
-
-// internalRingItem stores the ring element value.
-type internalRingItem struct {
-	Value any
+	*TRing[any]
 }

 // New creates and returns a Ring structure of `cap` elements.
@ -39,108 +23,53 @@ type internalRingItem struct {
 // Deprecated.
 func New(cap int, safe ...bool) *Ring {
 	return &Ring{
-		mu:    rwmutex.New(safe...),
-		ring:  ring.New(cap),
-		len:   gtype.NewInt(),
-		cap:   gtype.NewInt(cap),
-		dirty: gtype.NewBool(),
+		TRing: NewTRing[any](cap, safe...),
 	}
 }

 // Val returns the item's value of current position.
 func (r *Ring) Val() any {
-	var value any
-	r.mu.RLock()
-	if r.ring.Value != nil {
-		value = r.ring.Value.(internalRingItem).Value
-	}
-	r.mu.RUnlock()
-	return value
+	return r.TRing.Val()
 }

 // Len returns the size of ring.
 func (r *Ring) Len() int {
-	r.checkAndUpdateLenAndCap()
-	return r.len.Val()
+	return r.TRing.Len()
 }

 // Cap returns the capacity of ring.
 func (r *Ring) Cap() int {
-	r.checkAndUpdateLenAndCap()
-	return r.cap.Val()
-}
-
-// Checks and updates the len and cap of ring when ring is dirty.
-func (r *Ring) checkAndUpdateLenAndCap() {
-	if !r.dirty.Val() {
-		return
-	}
-	r.mu.RLock()
-	defer r.mu.RUnlock()
-	totalLen := 0
-	emptyLen := 0
-	if r.ring != nil {
-		if r.ring.Value == nil {
-			emptyLen++
-		}
-		totalLen++
-		for p := r.ring.Next(); p != r.ring; p = p.Next() {
-			if p.Value == nil {
-				emptyLen++
-			}
-			totalLen++
-		}
-	}
-	r.cap.Set(totalLen)
-	r.len.Set(totalLen - emptyLen)
-	r.dirty.Set(false)
+	return r.TRing.Cap()
 }

 // Set sets value to the item of current position.
 func (r *Ring) Set(value any) *Ring {
-	r.mu.Lock()
-	if r.ring.Value == nil {
-		r.len.Add(1)
-	}
-	r.ring.Value = internalRingItem{Value: value}
-	r.mu.Unlock()
+	r.TRing.Set(value)
 	return r
 }

 // Put sets `value` to current item of ring and moves position to next item.
 func (r *Ring) Put(value any) *Ring {
-	r.mu.Lock()
-	if r.ring.Value == nil {
-		r.len.Add(1)
-	}
-	r.ring.Value = internalRingItem{Value: value}
-	r.ring = r.ring.Next()
-	r.mu.Unlock()
+	r.TRing.Put(value)
 	return r
 }

 // Move moves n % r.Len() elements backward (n < 0) or forward (n >= 0)
 // in the ring and returns that ring element. r must not be empty.
 func (r *Ring) Move(n int) *Ring {
-	r.mu.Lock()
-	r.ring = r.ring.Move(n)
-	r.mu.Unlock()
+	r.TRing.Move(n)
 	return r
 }

 // Prev returns the previous ring element. r must not be empty.
 func (r *Ring) Prev() *Ring {
-	r.mu.Lock()
-	r.ring = r.ring.Prev()
-	r.mu.Unlock()
+	r.TRing.Prev()
 	return r
 }

 // Next returns the next ring element. r must not be empty.
 func (r *Ring) Next() *Ring {
-	r.mu.Lock()
-	r.ring = r.ring.Next()
-	r.mu.Unlock()
+	r.TRing.Next()
 	return r
 }

@ -160,13 +89,7 @@ func (r *Ring) Next() *Ring {
 // after r. The result points to the element following the
 // last element of s after insertion.
 func (r *Ring) Link(s *Ring) *Ring {
-	r.mu.Lock()
-	s.mu.Lock()
-	r.ring.Link(s.ring)
-	s.mu.Unlock()
-	r.mu.Unlock()
-	r.dirty.Set(true)
-	s.dirty.Set(true)
+	r.TRing.Link(s.TRing)
 	return r
 }

@ -174,78 +97,31 @@ func (r *Ring) Link(s *Ring) *Ring {
 // at r.Next(). If n % r.Len() == 0, r remains unchanged.
 // The result is the removed sub-ring. r must not be empty.
 func (r *Ring) Unlink(n int) *Ring {
-	r.mu.Lock()
-	resultRing := r.ring.Unlink(n)
-	r.dirty.Set(true)
-	r.mu.Unlock()
-	resultGRing := New(resultRing.Len())
-	resultGRing.ring = resultRing
-	resultGRing.dirty.Set(true)
-	return resultGRing
+	return &Ring{
+		TRing: r.TRing.Unlink(n),
+	}
 }

 // RLockIteratorNext iterates and locks reading forward
 // with given callback function `f` within RWMutex.RLock.
 // If `f` returns true, then it continues iterating; or false to stop.
 func (r *Ring) RLockIteratorNext(f func(value any) bool) {
-	r.mu.RLock()
-	defer r.mu.RUnlock()
-	if r.ring.Value != nil && !f(r.ring.Value.(internalRingItem).Value) {
-		return
-	}
-	for p := r.ring.Next(); p != r.ring; p = p.Next() {
-		if p.Value == nil || !f(p.Value.(internalRingItem).Value) {
-			break
-		}
-	}
+	r.TRing.RLockIteratorNext(f)
 }

-// RLockIteratorPrev iterates and locks writing backward
+// RLockIteratorPrev iterates and locks reading backward
 // with given callback function `f` within RWMutex.RLock.
 // If `f` returns true, then it continues iterating; or false to stop.
 func (r *Ring) RLockIteratorPrev(f func(value any) bool) {
-	r.mu.RLock()
-	defer r.mu.RUnlock()
-	if r.ring.Value != nil && !f(r.ring.Value.(internalRingItem).Value) {
-		return
-	}
-	for p := r.ring.Prev(); p != r.ring; p = p.Prev() {
-		if p.Value == nil || !f(p.Value.(internalRingItem).Value) {
-			break
-		}
-	}
+	r.TRing.RLockIteratorPrev(f)
 }

 // SliceNext returns a copy of all item values as slice forward from current position.
 func (r *Ring) SliceNext() []any {
-	s := make([]any, 0)
-	r.mu.RLock()
-	if r.ring.Value != nil {
-		s = append(s, r.ring.Value.(internalRingItem).Value)
-	}
-	for p := r.ring.Next(); p != r.ring; p = p.Next() {
-		if p.Value == nil {
-			break
-		}
-		s = append(s, p.Value.(internalRingItem).Value)
-	}
-	r.mu.RUnlock()
-	return s
+	return r.TRing.SliceNext()
 }

 // SlicePrev returns a copy of all item values as slice backward from current position.
 func (r *Ring) SlicePrev() []any {
-	s := make([]any, 0)
-	r.mu.RLock()
-	if r.ring.Value != nil {
-		s = append(s, r.ring.Value.(internalRingItem).Value)
-	}
-	for p := r.ring.Prev(); p != r.ring; p = p.Prev() {
-		if p.Value == nil {
-			break
-		}
-		s = append(s, p.Value.(internalRingItem).Value)
-	}
-	r.mu.RUnlock()
-	return s
+	return r.TRing.SlicePrev()
 }
--- a/container/gring/gring_t.go
+++ b/container/gring/gring_t.go
@ -0,0 +1,244 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package gring
+
+import (
+	"container/ring"
+
+	"github.com/gogf/gf/v2/container/gtype"
+	"github.com/gogf/gf/v2/internal/rwmutex"
+)
+
+// TRing is a struct of ring structure.
+type TRing[T any] struct {
+	mu    *rwmutex.RWMutex
+	ring  *ring.Ring  // Underlying ring.
+	len   *gtype.Int  // Length(already used size).
+	cap   *gtype.Int  // Capability(>=len).
+	dirty *gtype.Bool // Dirty, which means the len and cap should be recalculated. It's marked dirty when the size of ring changes.
+}
+
+// internalTRingItem[T] stores the ring element value.
+type internalTRingItem[T any] struct {
+	Value T
+}
+
+// NewTRing creates and returns a Ring structure of `cap` elements.
+// The optional parameter `safe` specifies whether using this structure in concurrent safety,
+// which is false in default.
+func NewTRing[T any](cap int, safe ...bool) *TRing[T] {
+	return &TRing[T]{
+		mu:    rwmutex.New(safe...),
+		ring:  ring.New(cap),
+		len:   gtype.NewInt(),
+		cap:   gtype.NewInt(cap),
+		dirty: gtype.NewBool(),
+	}
+}
+
+// Val returns the item's value of current position.
+func (r *TRing[T]) Val() T {
+	var value T
+	r.mu.RLock()
+	if r.ring.Value != nil {
+		value = r.ring.Value.(internalTRingItem[T]).Value
+	}
+	r.mu.RUnlock()
+	return value
+}
+
+// Len returns the size of ring.
+func (r *TRing[T]) Len() int {
+	r.checkAndUpdateLenAndCap()
+	return r.len.Val()
+}
+
+// Cap returns the capacity of ring.
+func (r *TRing[T]) Cap() int {
+	r.checkAndUpdateLenAndCap()
+	return r.cap.Val()
+}
+
+// Checks and updates the len and cap of ring when ring is dirty.
+func (r *TRing[T]) checkAndUpdateLenAndCap() {
+	if !r.dirty.Val() {
+		return
+	}
+	r.mu.RLock()
+	defer r.mu.RUnlock()
+	totalLen := 0
+	emptyLen := 0
+	if r.ring != nil {
+		if r.ring.Value == nil {
+			emptyLen++
+		}
+		totalLen++
+		for p := r.ring.Next(); p != r.ring; p = p.Next() {
+			if p.Value == nil {
+				emptyLen++
+			}
+			totalLen++
+		}
+	}
+	r.cap.Set(totalLen)
+	r.len.Set(totalLen - emptyLen)
+	r.dirty.Set(false)
+}
+
+// Set sets value to the item of current position.
+func (r *TRing[T]) Set(value T) *TRing[T] {
+	r.mu.Lock()
+	if r.ring.Value == nil {
+		r.len.Add(1)
+	}
+	r.ring.Value = internalTRingItem[T]{Value: value}
+	r.mu.Unlock()
+	return r
+}
+
+// Put sets `value` to current item of ring and moves position to next item.
+func (r *TRing[T]) Put(value T) *TRing[T] {
+	r.mu.Lock()
+	if r.ring.Value == nil {
+		r.len.Add(1)
+	}
+	r.ring.Value = internalTRingItem[T]{Value: value}
+	r.ring = r.ring.Next()
+	r.mu.Unlock()
+	return r
+}
+
+// Move moves n % r.Len() elements backward (n < 0) or forward (n >= 0)
+// in the ring and returns that ring element. r must not be empty.
+func (r *TRing[T]) Move(n int) *TRing[T] {
+	r.mu.Lock()
+	r.ring = r.ring.Move(n)
+	r.mu.Unlock()
+	return r
+}
+
+// Prev returns the previous ring element. r must not be empty.
+func (r *TRing[T]) Prev() *TRing[T] {
+	r.mu.Lock()
+	r.ring = r.ring.Prev()
+	r.mu.Unlock()
+	return r
+}
+
+// Next returns the next ring element. r must not be empty.
+func (r *TRing[T]) Next() *TRing[T] {
+	r.mu.Lock()
+	r.ring = r.ring.Next()
+	r.mu.Unlock()
+	return r
+}
+
+// Link connects ring r with ring s such that r.Next()
+// becomes s and returns the original value for r.Next().
+// r must not be empty.
+//
+// If r and s point to the same ring, linking
+// them removes the elements between r and s from the ring.
+// The removed elements form a sub-ring and the result is a
+// reference to that sub-ring (if no elements were removed,
+// the result is still the original value for r.Next(),
+// and not nil).
+//
+// If r and s point to different rings, linking
+// them creates a single ring with the elements of s inserted
+// after r. The result points to the element following the
+// last element of s after insertion.
+func (r *TRing[T]) Link(s *TRing[T]) *TRing[T] {
+	r.mu.Lock()
+	s.mu.Lock()
+	r.ring.Link(s.ring)
+	s.mu.Unlock()
+	r.mu.Unlock()
+	r.dirty.Set(true)
+	s.dirty.Set(true)
+	return r
+}
+
+// Unlink removes n % r.Len() elements from the ring r, starting
+// at r.Next(). If n % r.Len() == 0, r remains unchanged.
+// The result is the removed sub-ring. r must not be empty.
+func (r *TRing[T]) Unlink(n int) *TRing[T] {
+	r.mu.Lock()
+	resultRing := r.ring.Unlink(n)
+	r.dirty.Set(true)
+	r.mu.Unlock()
+	resultGRing := NewTRing[T](resultRing.Len())
+	resultGRing.ring = resultRing
+	resultGRing.dirty.Set(true)
+	return resultGRing
+}
+
+// RLockIteratorNext iterates and locks reading forward
+// with given callback function `f` within RWMutex.RLock.
+// If `f` returns true, then it continues iterating; or false to stop.
+func (r *TRing[T]) RLockIteratorNext(f func(value T) bool) {
+	r.mu.RLock()
+	defer r.mu.RUnlock()
+	if r.ring.Value != nil && !f(r.ring.Value.(internalTRingItem[T]).Value) {
+		return
+	}
+	for p := r.ring.Next(); p != r.ring; p = p.Next() {
+		if p.Value == nil || !f(p.Value.(internalTRingItem[T]).Value) {
+			break
+		}
+	}
+}
+
+// RLockIteratorPrev iterates and locks reading backward
+// with given callback function `f` within RWMutex.RLock.
+// If `f` returns true, then it continues iterating; or false to stop.
+func (r *TRing[T]) RLockIteratorPrev(f func(value T) bool) {
+	r.mu.RLock()
+	defer r.mu.RUnlock()
+	if r.ring.Value != nil && !f(r.ring.Value.(internalTRingItem[T]).Value) {
+		return
+	}
+	for p := r.ring.Prev(); p != r.ring; p = p.Prev() {
+		if p.Value == nil || !f(p.Value.(internalTRingItem[T]).Value) {
+			break
+		}
+	}
+}
+
+// SliceNext returns a copy of all item values as slice forward from current position.
+func (r *TRing[T]) SliceNext() []T {
+	s := make([]T, 0, r.Len())
+	r.mu.RLock()
+	if r.ring.Value != nil {
+		s = append(s, r.ring.Value.(internalTRingItem[T]).Value)
+	}
+	for p := r.ring.Next(); p != r.ring; p = p.Next() {
+		if p.Value == nil {
+			break
+		}
+		s = append(s, p.Value.(internalTRingItem[T]).Value)
+	}
+	r.mu.RUnlock()
+	return s
+}
+
+// SlicePrev returns a copy of all item values as slice backward from current position.
+func (r *TRing[T]) SlicePrev() []T {
+	s := make([]T, 0, r.Len())
+	r.mu.RLock()
+	if r.ring.Value != nil {
+		s = append(s, r.ring.Value.(internalTRingItem[T]).Value)
+	}
+	for p := r.ring.Prev(); p != r.ring; p = p.Prev() {
+		if p.Value == nil {
+			break
+		}
+		s = append(s, p.Value.(internalTRingItem[T]).Value)
+	}
+	r.mu.RUnlock()
+	return s
+}
--- a/container/gring/gring_z_unit_test.go
+++ b/container/gring/gring_z_unit_test.go
@ -148,14 +148,11 @@ func Test_Issue1394(t *testing.T) {
 		for i := 0; i < 10; i++ {
 			gRing.Put(i)
 		}
-		t.Logf("the length:%d", gRing.Len())
 		gRingResult := gRing.Unlink(6)
 		for i := 0; i < 10; i++ {
 			t.Log(gRing.Val())
 			gRing = gRing.Next()
 		}
-		t.Logf("the ring length:%d", gRing.Len())
-		t.Logf("the result length:%d", gRingResult.Len())

 		// stdring
 		stdRing := ring.New(10)
@ -163,14 +160,11 @@ func Test_Issue1394(t *testing.T) {
 			stdRing.Value = i
 			stdRing = stdRing.Next()
 		}
-		t.Logf("the length:%d", stdRing.Len())
 		stdRingResult := stdRing.Unlink(6)
 		for i := 0; i < 10; i++ {
 			t.Log(stdRing.Value)
 			stdRing = stdRing.Next()
 		}
-		t.Logf("the ring length:%d", stdRing.Len())
-		t.Logf("the result length:%d", stdRingResult.Len())

 		// Assertion.
 		t.Assert(gRing.Len(), stdRing.Len())
--- a/container/gset/gset_any_set.go
+++ b/container/gset/gset_any_set.go
@ -8,18 +8,15 @@
 package gset

 import (
-	"bytes"
+	"sync"

-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
-	"github.com/gogf/gf/v2/text/gstr"
 	"github.com/gogf/gf/v2/util/gconv"
 )

 // Set is consisted of any items.
 type Set struct {
-	mu   rwmutex.RWMutex
-	data map[any]struct{}
+	*TSet[any]
+	once sync.Once
 }

 // New create and returns a new set, which contains un-repeated items.
@ -33,44 +30,38 @@ func New(safe ...bool) *Set {
 // Also see New.
 func NewSet(safe ...bool) *Set {
 	return &Set{
-		data: make(map[any]struct{}),
-		mu:   rwmutex.Create(safe...),
+		TSet: NewTSet[any](safe...),
 	}
 }

 // NewFrom returns a new set from `items`.
 // Parameter `items` can be either a variable of any type, or a slice.
 func NewFrom(items any, safe ...bool) *Set {
-	m := make(map[any]struct{})
-	for _, v := range gconv.Interfaces(items) {
-		m[v] = struct{}{}
-	}
 	return &Set{
-		data: m,
-		mu:   rwmutex.Create(safe...),
+		TSet: NewTSetFrom[any](gconv.Interfaces(items), safe...),
 	}
 }

+// lazyInit lazily initializes the set.
+func (a *Set) lazyInit() {
+	a.once.Do(func() {
+		if a.TSet == nil {
+			a.TSet = NewTSet[any]()
+		}
+	})
+}
+
 // Iterator iterates the set readonly with given callback function `f`,
 // if `f` returns true then continue iterating; or false to stop.
 func (set *Set) Iterator(f func(v any) bool) {
-	for _, k := range set.Slice() {
-		if !f(k) {
-			break
-		}
-	}
+	set.lazyInit()
+	set.TSet.Iterator(f)
 }

 // Add adds one or multiple items to the set.
 func (set *Set) Add(items ...any) {
-	set.mu.Lock()
-	if set.data == nil {
-		set.data = make(map[any]struct{})
-	}
-	for _, v := range items {
-		set.data[v] = struct{}{}
-	}
-	set.mu.Unlock()
+	set.lazyInit()
+	set.TSet.Add(items...)
 }

 // AddIfNotExist checks whether item exists in the set,
@ -79,21 +70,8 @@ func (set *Set) Add(items ...any) {
 //
 // Note that, if `item` is nil, it does nothing and returns false.
 func (set *Set) AddIfNotExist(item any) bool {
-	if item == nil {
-		return false
-	}
-	if !set.Contains(item) {
-		set.mu.Lock()
-		defer set.mu.Unlock()
-		if set.data == nil {
-			set.data = make(map[any]struct{})
-		}
-		if _, ok := set.data[item]; !ok {
-			set.data[item] = struct{}{}
-			return true
-		}
-	}
-	return false
+	set.lazyInit()
+	return set.TSet.AddIfNotExist(item)
 }

 // AddIfNotExistFunc checks whether item exists in the set,
@ -103,23 +81,8 @@ func (set *Set) AddIfNotExist(item any) bool {
 // Note that, if `item` is nil, it does nothing and returns false. The function `f`
 // is executed without writing lock.
 func (set *Set) AddIfNotExistFunc(item any, f func() bool) bool {
-	if item == nil {
-		return false
-	}
-	if !set.Contains(item) {
-		if f() {
-			set.mu.Lock()
-			defer set.mu.Unlock()
-			if set.data == nil {
-				set.data = make(map[any]struct{})
-			}
-			if _, ok := set.data[item]; !ok {
-				set.data[item] = struct{}{}
-				return true
-			}
-		}
-	}
-	return false
+	set.lazyInit()
+	return set.TSet.AddIfNotExistFunc(item, f)
 }

 // AddIfNotExistFuncLock checks whether item exists in the set,
@ -129,95 +92,44 @@ func (set *Set) AddIfNotExistFunc(item any, f func() bool) bool {
 // Note that, if `item` is nil, it does nothing and returns false. The function `f`
 // is executed within writing lock.
 func (set *Set) AddIfNotExistFuncLock(item any, f func() bool) bool {
-	if item == nil {
-		return false
-	}
-	if !set.Contains(item) {
-		set.mu.Lock()
-		defer set.mu.Unlock()
-		if set.data == nil {
-			set.data = make(map[any]struct{})
-		}
-		if f() {
-			if _, ok := set.data[item]; !ok {
-				set.data[item] = struct{}{}
-				return true
-			}
-		}
-	}
-	return false
+	set.lazyInit()
+	return set.TSet.AddIfNotExistFuncLock(item, f)
 }

 // Contains checks whether the set contains `item`.
 func (set *Set) Contains(item any) bool {
-	var ok bool
-	set.mu.RLock()
-	if set.data != nil {
-		_, ok = set.data[item]
-	}
-	set.mu.RUnlock()
-	return ok
+	set.lazyInit()
+	return set.TSet.Contains(item)
 }

 // Remove deletes `item` from set.
 func (set *Set) Remove(item any) {
-	set.mu.Lock()
-	if set.data != nil {
-		delete(set.data, item)
-	}
-	set.mu.Unlock()
+	set.lazyInit()
+	set.TSet.Remove(item)
 }

 // Size returns the size of the set.
 func (set *Set) Size() int {
-	set.mu.RLock()
-	l := len(set.data)
-	set.mu.RUnlock()
-	return l
+	set.lazyInit()
+	return set.TSet.Size()
 }

 // Clear deletes all items of the set.
 func (set *Set) Clear() {
-	set.mu.Lock()
-	set.data = make(map[any]struct{})
-	set.mu.Unlock()
+	set.lazyInit()
+	set.TSet.Clear()
 }

 // Slice returns all items of the set as slice.
 func (set *Set) Slice() []any {
-	set.mu.RLock()
-	var (
-		i   = 0
-		ret = make([]any, len(set.data))
-	)
-	for item := range set.data {
-		ret[i] = item
-		i++
-	}
-	set.mu.RUnlock()
-	return ret
+	set.lazyInit()
+	return set.TSet.Slice()
 }

 // Join joins items with a string `glue`.
 func (set *Set) Join(glue string) string {
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	if len(set.data) == 0 {
-		return ""
-	}
-	var (
-		l      = len(set.data)
-		i      = 0
-		buffer = bytes.NewBuffer(nil)
-	)
-	for k := range set.data {
-		buffer.WriteString(gconv.String(k))
-		if i != l-1 {
-			buffer.WriteString(glue)
-		}
-		i++
-	}
-	return buffer.String()
+	set.lazyInit()
+	return set.TSet.Join(glue)
 }

 // String returns items as a string, which implements like json.Marshal does.
@ -225,63 +137,27 @@ func (set *Set) String() string {
 	if set == nil {
 		return ""
 	}
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	var (
-		s      string
-		l      = len(set.data)
-		i      = 0
-		buffer = bytes.NewBuffer(nil)
-	)
-	buffer.WriteByte('[')
-	for k := range set.data {
-		s = gconv.String(k)
-		if gstr.IsNumeric(s) {
-			buffer.WriteString(s)
-		} else {
-			buffer.WriteString(`"` + gstr.QuoteMeta(s, `"\`) + `"`)
-		}
-		if i != l-1 {
-			buffer.WriteByte(',')
-		}
-		i++
-	}
-	buffer.WriteByte(']')
-	return buffer.String()
+	set.lazyInit()
+	return set.TSet.String()
 }

 // LockFunc locks writing with callback function `f`.
 func (set *Set) LockFunc(f func(m map[any]struct{})) {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	f(set.data)
+	set.lazyInit()
+	set.TSet.LockFunc(f)
 }

 // RLockFunc locks reading with callback function `f`.
 func (set *Set) RLockFunc(f func(m map[any]struct{})) {
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	f(set.data)
+	set.lazyInit()
+	set.TSet.RLockFunc(f)
 }

 // Equal checks whether the two sets equal.
 func (set *Set) Equal(other *Set) bool {
-	if set == other {
-		return true
-	}
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	if len(set.data) != len(other.data) {
-		return false
-	}
-	for key := range set.data {
-		if _, ok := other.data[key]; !ok {
-			return false
-		}
-	}
-	return true
+	set.lazyInit()
+	other.lazyInit()
+	return set.TSet.Equal(other.TSet)
 }

 // IsSubsetOf checks whether the current set is a sub-set of `other`.
@ -289,85 +165,40 @@ func (set *Set) IsSubsetOf(other *Set) bool {
 	if set == other {
 		return true
 	}
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	for key := range set.data {
-		if _, ok := other.data[key]; !ok {
-			return false
-		}
-	}
-	return true
+
+	set.lazyInit()
+	other.lazyInit()
+
+	return set.TSet.IsSubsetOf(other.TSet)
 }

 // Union returns a new set which is the union of `set` and `others`.
 // Which means, all the items in `newSet` are in `set` or in `others`.
 func (set *Set) Union(others ...*Set) (newSet *Set) {
-	newSet = NewSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for _, other := range others {
-		if set != other {
-			other.mu.RLock()
-		}
-		for k, v := range set.data {
-			newSet.data[k] = v
-		}
-		if set != other {
-			for k, v := range other.data {
-				newSet.data[k] = v
-			}
-		}
-		if set != other {
-			other.mu.RUnlock()
-		}
-	}
+	set.lazyInit()

-	return
+	return &Set{
+		TSet: set.TSet.Union(set.toTSetSlice(others)...),
+	}
 }

 // Diff returns a new set which is the difference set from `set` to `others`.
 // Which means, all the items in `newSet` are in `set` but not in `others`.
 func (set *Set) Diff(others ...*Set) (newSet *Set) {
-	newSet = NewSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for _, other := range others {
-		if set == other {
-			continue
-		}
-		other.mu.RLock()
-		for k, v := range set.data {
-			if _, ok := other.data[k]; !ok {
-				newSet.data[k] = v
-			}
-		}
-		other.mu.RUnlock()
+	set.lazyInit()
+
+	return &Set{
+		TSet: set.TSet.Diff(set.toTSetSlice(others)...),
 	}
-	return
 }

 // Intersect returns a new set which is the intersection from `set` to `others`.
 // Which means, all the items in `newSet` are in `set` and also in `others`.
 func (set *Set) Intersect(others ...*Set) (newSet *Set) {
-	newSet = NewSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for _, other := range others {
-		if set != other {
-			other.mu.RLock()
-		}
-		for k, v := range set.data {
-			if _, ok := other.data[k]; ok {
-				newSet.data[k] = v
-			}
-		}
-		if set != other {
-			other.mu.RUnlock()
-		}
+	set.lazyInit()
+	return &Set{
+		TSet: set.TSet.Intersect(set.toTSetSlice(others)...),
 	}
-	return
 }

 // Complement returns a new set which is the complement from `set` to `full`.
@ -376,36 +207,22 @@ func (set *Set) Intersect(others ...*Set) (newSet *Set) {
 // It returns the difference between `full` and `set`
 // if the given set `full` is not the full set of `set`.
 func (set *Set) Complement(full *Set) (newSet *Set) {
-	newSet = NewSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	if set != full {
-		full.mu.RLock()
-		defer full.mu.RUnlock()
-	}
-	for k, v := range full.data {
-		if _, ok := set.data[k]; !ok {
-			newSet.data[k] = v
+	set.lazyInit()
+	if full == nil {
+		return &Set{
+			TSet: NewTSet[any](true),
 		}
 	}
-	return
+	full.lazyInit()
+	return &Set{
+		TSet: set.TSet.Complement(full.TSet),
+	}
 }

 // Merge adds items from `others` sets into `set`.
 func (set *Set) Merge(others ...*Set) *Set {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	for _, other := range others {
-		if set != other {
-			other.mu.RLock()
-		}
-		for k, v := range other.data {
-			set.data[k] = v
-		}
-		if set != other {
-			other.mu.RUnlock()
-		}
-	}
+	set.lazyInit()
+	set.TSet.Merge(set.toTSetSlice(others)...)
 	return set
 }

@ -413,101 +230,46 @@ func (set *Set) Merge(others ...*Set) *Set {
 // Note: The items should be converted to int type,
 // or you'd get a result that you unexpected.
 func (set *Set) Sum() (sum int) {
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for k := range set.data {
-		sum += gconv.Int(k)
-	}
-	return
+	set.lazyInit()
+	return set.TSet.Sum()
 }

 // Pop randomly pops an item from set.
 func (set *Set) Pop() any {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	for k := range set.data {
-		delete(set.data, k)
-		return k
-	}
-	return nil
+	set.lazyInit()
+	return set.TSet.Pop()
 }

 // Pops randomly pops `size` items from set.
 // It returns all items if size == -1.
 func (set *Set) Pops(size int) []any {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	if size > len(set.data) || size == -1 {
-		size = len(set.data)
-	}
-	if size <= 0 {
-		return nil
-	}
-	index := 0
-	array := make([]any, size)
-	for k := range set.data {
-		delete(set.data, k)
-		array[index] = k
-		index++
-		if index == size {
-			break
-		}
-	}
-	return array
+	set.lazyInit()
+	return set.TSet.Pops(size)
 }

 // Walk applies a user supplied function `f` to every item of set.
 func (set *Set) Walk(f func(item any) any) *Set {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	m := make(map[any]struct{}, len(set.data))
-	for k, v := range set.data {
-		m[f(k)] = v
-	}
-	set.data = m
+	set.lazyInit()
+	set.TSet.Walk(f)
 	return set
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (set Set) MarshalJSON() ([]byte, error) {
-	return json.Marshal(set.Slice())
+	set.lazyInit()
+	return set.TSet.MarshalJSON()
 }

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (set *Set) UnmarshalJSON(b []byte) error {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	if set.data == nil {
-		set.data = make(map[any]struct{})
-	}
-	var array []any
-	if err := json.UnmarshalUseNumber(b, &array); err != nil {
-		return err
-	}
-	for _, v := range array {
-		set.data[v] = struct{}{}
-	}
-	return nil
+	set.lazyInit()
+	return set.TSet.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for set.
 func (set *Set) UnmarshalValue(value any) (err error) {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	if set.data == nil {
-		set.data = make(map[any]struct{})
-	}
-	var array []any
-	switch value.(type) {
-	case string, []byte:
-		err = json.UnmarshalUseNumber(gconv.Bytes(value), &array)
-	default:
-		array = gconv.SliceAny(value)
-	}
-	for _, v := range array {
-		set.data[v] = struct{}{}
-	}
-	return
+	set.lazyInit()
+	return set.TSet.UnmarshalValue(value)
 }

 // DeepCopy implements interface for deep copy of current type.
@ -515,11 +277,21 @@ func (set *Set) DeepCopy() any {
 	if set == nil {
 		return nil
 	}
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	data := make([]any, 0)
-	for k := range set.data {
-		data = append(data, k)
+	set.lazyInit()
+	return &Set{
+		TSet: set.TSet.DeepCopy().(*TSet[any]),
 	}
-	return NewFrom(data, set.mu.IsSafe())
+}
+
+// toTSetSlice converts []*Set to []*TSet[any]
+func (set *Set) toTSetSlice(sets []*Set) (tSets []*TSet[any]) {
+	tSets = make([]*TSet[any], len(sets))
+	for i, v := range sets {
+		if v == nil {
+			continue
+		}
+		v.lazyInit()
+		tSets[i] = v.TSet
+	}
+	return
 }
--- a/container/gset/gset_int_set.go
+++ b/container/gset/gset_int_set.go
@ -8,17 +8,13 @@
 package gset

 import (
-	"bytes"
-
-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
-	"github.com/gogf/gf/v2/util/gconv"
+	"sync"
 )

 // IntSet is consisted of int items.
 type IntSet struct {
-	mu   rwmutex.RWMutex
-	data map[int]struct{}
+	*TSet[int]
+	once sync.Once
 }

 // NewIntSet create and returns a new set, which contains un-repeated items.
@ -26,43 +22,37 @@ type IntSet struct {
 // which is false in default.
 func NewIntSet(safe ...bool) *IntSet {
 	return &IntSet{
-		mu:   rwmutex.Create(safe...),
-		data: make(map[int]struct{}),
+		TSet: NewTSet[int](safe...),
 	}
 }

 // NewIntSetFrom returns a new set from `items`.
 func NewIntSetFrom(items []int, safe ...bool) *IntSet {
-	m := make(map[int]struct{})
-	for _, v := range items {
-		m[v] = struct{}{}
-	}
 	return &IntSet{
-		mu:   rwmutex.Create(safe...),
-		data: m,
+		TSet: NewTSetFrom(items, safe...),
 	}
 }

+// lazyInit lazily initializes the set.
+func (a *IntSet) lazyInit() {
+	a.once.Do(func() {
+		if a.TSet == nil {
+			a.TSet = NewTSet[int]()
+		}
+	})
+}
+
 // Iterator iterates the set readonly with given callback function `f`,
 // if `f` returns true then continue iterating; or false to stop.
 func (set *IntSet) Iterator(f func(v int) bool) {
-	for _, k := range set.Slice() {
-		if !f(k) {
-			break
-		}
-	}
+	set.lazyInit()
+	set.TSet.Iterator(f)
 }

 // Add adds one or multiple items to the set.
 func (set *IntSet) Add(item ...int) {
-	set.mu.Lock()
-	if set.data == nil {
-		set.data = make(map[int]struct{})
-	}
-	for _, v := range item {
-		set.data[v] = struct{}{}
-	}
-	set.mu.Unlock()
+	set.lazyInit()
+	set.TSet.Add(item...)
 }

 // AddIfNotExist checks whether item exists in the set,
@ -71,18 +61,8 @@ func (set *IntSet) Add(item ...int) {
 //
 // Note that, if `item` is nil, it does nothing and returns false.
 func (set *IntSet) AddIfNotExist(item int) bool {
-	if !set.Contains(item) {
-		set.mu.Lock()
-		defer set.mu.Unlock()
-		if set.data == nil {
-			set.data = make(map[int]struct{})
-		}
-		if _, ok := set.data[item]; !ok {
-			set.data[item] = struct{}{}
-			return true
-		}
-	}
-	return false
+	set.lazyInit()
+	return set.TSet.AddIfNotExist(item)
 }

 // AddIfNotExistFunc checks whether item exists in the set,
@ -91,20 +71,8 @@ func (set *IntSet) AddIfNotExist(item int) bool {
 //
 // Note that, the function `f` is executed without writing lock.
 func (set *IntSet) AddIfNotExistFunc(item int, f func() bool) bool {
-	if !set.Contains(item) {
-		if f() {
-			set.mu.Lock()
-			defer set.mu.Unlock()
-			if set.data == nil {
-				set.data = make(map[int]struct{})
-			}
-			if _, ok := set.data[item]; !ok {
-				set.data[item] = struct{}{}
-				return true
-			}
-		}
-	}
-	return false
+	set.lazyInit()
+	return set.TSet.AddIfNotExistFunc(item, f)
 }

 // AddIfNotExistFuncLock checks whether item exists in the set,
@ -113,92 +81,44 @@ func (set *IntSet) AddIfNotExistFunc(item int, f func() bool) bool {
 //
 // Note that, the function `f` is executed without writing lock.
 func (set *IntSet) AddIfNotExistFuncLock(item int, f func() bool) bool {
-	if !set.Contains(item) {
-		set.mu.Lock()
-		defer set.mu.Unlock()
-		if set.data == nil {
-			set.data = make(map[int]struct{})
-		}
-		if f() {
-			if _, ok := set.data[item]; !ok {
-				set.data[item] = struct{}{}
-				return true
-			}
-		}
-	}
-	return false
+	set.lazyInit()
+	return set.TSet.AddIfNotExistFuncLock(item, f)
 }

 // Contains checks whether the set contains `item`.
 func (set *IntSet) Contains(item int) bool {
-	var ok bool
-	set.mu.RLock()
-	if set.data != nil {
-		_, ok = set.data[item]
-	}
-	set.mu.RUnlock()
-	return ok
+	set.lazyInit()
+	return set.TSet.Contains(item)
 }

 // Remove deletes `item` from set.
 func (set *IntSet) Remove(item int) {
-	set.mu.Lock()
-	if set.data != nil {
-		delete(set.data, item)
-	}
-	set.mu.Unlock()
+	set.lazyInit()
+	set.TSet.Remove(item)
 }

 // Size returns the size of the set.
 func (set *IntSet) Size() int {
-	set.mu.RLock()
-	l := len(set.data)
-	set.mu.RUnlock()
-	return l
+	set.lazyInit()
+	return set.TSet.Size()
 }

 // Clear deletes all items of the set.
 func (set *IntSet) Clear() {
-	set.mu.Lock()
-	set.data = make(map[int]struct{})
-	set.mu.Unlock()
+	set.lazyInit()
+	set.TSet.Clear()
 }

 // Slice returns the an of items of the set as slice.
 func (set *IntSet) Slice() []int {
-	set.mu.RLock()
-	var (
-		i   = 0
-		ret = make([]int, len(set.data))
-	)
-	for k := range set.data {
-		ret[i] = k
-		i++
-	}
-	set.mu.RUnlock()
-	return ret
+	set.lazyInit()
+	return set.TSet.Slice()
 }

 // Join joins items with a string `glue`.
 func (set *IntSet) Join(glue string) string {
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	if len(set.data) == 0 {
-		return ""
-	}
-	var (
-		l      = len(set.data)
-		i      = 0
-		buffer = bytes.NewBuffer(nil)
-	)
-	for k := range set.data {
-		buffer.WriteString(gconv.String(k))
-		if i != l-1 {
-			buffer.WriteString(glue)
-		}
-		i++
-	}
-	return buffer.String()
+	set.lazyInit()
+	return set.TSet.Join(glue)
 }

 // String returns items as a string, which implements like json.Marshal does.
@ -206,41 +126,27 @@ func (set *IntSet) String() string {
 	if set == nil {
 		return ""
 	}
-	return "[" + set.Join(",") + "]"
+	set.lazyInit()
+	return set.TSet.String()
 }

 // LockFunc locks writing with callback function `f`.
 func (set *IntSet) LockFunc(f func(m map[int]struct{})) {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	f(set.data)
+	set.lazyInit()
+	set.TSet.LockFunc(f)
 }

 // RLockFunc locks reading with callback function `f`.
 func (set *IntSet) RLockFunc(f func(m map[int]struct{})) {
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	f(set.data)
+	set.lazyInit()
+	set.TSet.RLockFunc(f)
 }

 // Equal checks whether the two sets equal.
 func (set *IntSet) Equal(other *IntSet) bool {
-	if set == other {
-		return true
-	}
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	if len(set.data) != len(other.data) {
-		return false
-	}
-	for key := range set.data {
-		if _, ok := other.data[key]; !ok {
-			return false
-		}
-	}
-	return true
+	set.lazyInit()
+	other.lazyInit()
+	return set.TSet.Equal(other.TSet)
 }

 // IsSubsetOf checks whether the current set is a sub-set of `other`.
@ -248,85 +154,38 @@ func (set *IntSet) IsSubsetOf(other *IntSet) bool {
 	if set == other {
 		return true
 	}
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	for key := range set.data {
-		if _, ok := other.data[key]; !ok {
-			return false
-		}
-	}
-	return true
+
+	set.lazyInit()
+	other.lazyInit()
+
+	return set.TSet.IsSubsetOf(other.TSet)
 }

 // Union returns a new set which is the union of `set` and `other`.
 // Which means, all the items in `newSet` are in `set` or in `other`.
 func (set *IntSet) Union(others ...*IntSet) (newSet *IntSet) {
-	newSet = NewIntSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for _, other := range others {
-		if set != other {
-			other.mu.RLock()
-		}
-		for k, v := range set.data {
-			newSet.data[k] = v
-		}
-		if set != other {
-			for k, v := range other.data {
-				newSet.data[k] = v
-			}
-		}
-		if set != other {
-			other.mu.RUnlock()
-		}
+	set.lazyInit()
+	return &IntSet{
+		TSet: set.TSet.Union(set.toTSetSlice(others)...),
 	}
-
-	return
 }

 // Diff returns a new set which is the difference set from `set` to `other`.
 // Which means, all the items in `newSet` are in `set` but not in `other`.
 func (set *IntSet) Diff(others ...*IntSet) (newSet *IntSet) {
-	newSet = NewIntSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for _, other := range others {
-		if set == other {
-			continue
-		}
-		other.mu.RLock()
-		for k, v := range set.data {
-			if _, ok := other.data[k]; !ok {
-				newSet.data[k] = v
-			}
-		}
-		other.mu.RUnlock()
+	set.lazyInit()
+	return &IntSet{
+		TSet: set.TSet.Diff(set.toTSetSlice(others)...),
 	}
-	return
 }

 // Intersect returns a new set which is the intersection from `set` to `other`.
 // Which means, all the items in `newSet` are in `set` and also in `other`.
 func (set *IntSet) Intersect(others ...*IntSet) (newSet *IntSet) {
-	newSet = NewIntSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for _, other := range others {
-		if set != other {
-			other.mu.RLock()
-		}
-		for k, v := range set.data {
-			if _, ok := other.data[k]; ok {
-				newSet.data[k] = v
-			}
-		}
-		if set != other {
-			other.mu.RUnlock()
-		}
+	set.lazyInit()
+	return &IntSet{
+		TSet: set.TSet.Intersect(set.toTSetSlice(others)...),
 	}
-	return
 }

 // Complement returns a new set which is the complement from `set` to `full`.
@ -335,36 +194,22 @@ func (set *IntSet) Intersect(others ...*IntSet) (newSet *IntSet) {
 // It returns the difference between `full` and `set`
 // if the given set `full` is not the full set of `set`.
 func (set *IntSet) Complement(full *IntSet) (newSet *IntSet) {
-	newSet = NewIntSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	if set != full {
-		full.mu.RLock()
-		defer full.mu.RUnlock()
-	}
-	for k, v := range full.data {
-		if _, ok := set.data[k]; !ok {
-			newSet.data[k] = v
+	set.lazyInit()
+	if full == nil {
+		return &IntSet{
+			TSet: NewTSet[int](),
 		}
 	}
-	return
+	full.lazyInit()
+	return &IntSet{
+		TSet: set.TSet.Complement(full.TSet),
+	}
 }

 // Merge adds items from `others` sets into `set`.
 func (set *IntSet) Merge(others ...*IntSet) *IntSet {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	for _, other := range others {
-		if set != other {
-			other.mu.RLock()
-		}
-		for k, v := range other.data {
-			set.data[k] = v
-		}
-		if set != other {
-			other.mu.RUnlock()
-		}
-	}
+	set.lazyInit()
+	set.TSet.Merge(set.toTSetSlice(others)...)
 	return set
 }

@ -372,101 +217,46 @@ func (set *IntSet) Merge(others ...*IntSet) *IntSet {
 // Note: The items should be converted to int type,
 // or you'd get a result that you unexpected.
 func (set *IntSet) Sum() (sum int) {
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for k := range set.data {
-		sum += k
-	}
-	return
+	set.lazyInit()
+	return set.TSet.Sum()
 }

 // Pop randomly pops an item from set.
 func (set *IntSet) Pop() int {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	for k := range set.data {
-		delete(set.data, k)
-		return k
-	}
-	return 0
+	set.lazyInit()
+	return set.TSet.Pop()
 }

 // Pops randomly pops `size` items from set.
 // It returns all items if size == -1.
 func (set *IntSet) Pops(size int) []int {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	if size > len(set.data) || size == -1 {
-		size = len(set.data)
-	}
-	if size <= 0 {
-		return nil
-	}
-	index := 0
-	array := make([]int, size)
-	for k := range set.data {
-		delete(set.data, k)
-		array[index] = k
-		index++
-		if index == size {
-			break
-		}
-	}
-	return array
+	set.lazyInit()
+	return set.TSet.Pops(size)
 }

 // Walk applies a user supplied function `f` to every item of set.
 func (set *IntSet) Walk(f func(item int) int) *IntSet {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	m := make(map[int]struct{}, len(set.data))
-	for k, v := range set.data {
-		m[f(k)] = v
-	}
-	set.data = m
+	set.lazyInit()
+	set.TSet.Walk(f)
 	return set
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (set IntSet) MarshalJSON() ([]byte, error) {
-	return json.Marshal(set.Slice())
+	set.lazyInit()
+	return set.TSet.MarshalJSON()
 }

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (set *IntSet) UnmarshalJSON(b []byte) error {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	if set.data == nil {
-		set.data = make(map[int]struct{})
-	}
-	var array []int
-	if err := json.UnmarshalUseNumber(b, &array); err != nil {
-		return err
-	}
-	for _, v := range array {
-		set.data[v] = struct{}{}
-	}
-	return nil
+	set.lazyInit()
+	return set.TSet.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for set.
 func (set *IntSet) UnmarshalValue(value any) (err error) {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	if set.data == nil {
-		set.data = make(map[int]struct{})
-	}
-	var array []int
-	switch value.(type) {
-	case string, []byte:
-		err = json.UnmarshalUseNumber(gconv.Bytes(value), &array)
-	default:
-		array = gconv.SliceInt(value)
-	}
-	for _, v := range array {
-		set.data[v] = struct{}{}
-	}
-	return
+	set.lazyInit()
+	return set.TSet.UnmarshalValue(value)
 }

 // DeepCopy implements interface for deep copy of current type.
@ -474,15 +264,21 @@ func (set *IntSet) DeepCopy() any {
 	if set == nil {
 		return nil
 	}
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	var (
-		slice = make([]int, len(set.data))
-		index = 0
-	)
-	for k := range set.data {
-		slice[index] = k
-		index++
+	set.lazyInit()
+	return &IntSet{
+		TSet: set.TSet.DeepCopy().(*TSet[int]),
 	}
-	return NewIntSetFrom(slice, set.mu.IsSafe())
+}
+
+// toTSetSlice converts []*IntSet to []*TSet[int]
+func (set *IntSet) toTSetSlice(sets []*IntSet) (tSets []*TSet[int]) {
+	tSets = make([]*TSet[int], len(sets))
+	for i, v := range sets {
+		if v == nil {
+			continue
+		}
+		v.lazyInit()
+		tSets[i] = v.TSet
+	}
+	return
 }
--- a/container/gset/gset_str_set.go
+++ b/container/gset/gset_str_set.go
@ -8,19 +8,14 @@
 package gset

 import (
-	"bytes"
 	"strings"
-
-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
-	"github.com/gogf/gf/v2/text/gstr"
-	"github.com/gogf/gf/v2/util/gconv"
+	"sync"
 )

 // StrSet is consisted of string items.
 type StrSet struct {
-	mu   rwmutex.RWMutex
-	data map[string]struct{}
+	*TSet[string]
+	once sync.Once
 }

 // NewStrSet create and returns a new set, which contains un-repeated items.
@ -28,61 +23,45 @@ type StrSet struct {
 // which is false in default.
 func NewStrSet(safe ...bool) *StrSet {
 	return &StrSet{
-		mu:   rwmutex.Create(safe...),
-		data: make(map[string]struct{}),
+		TSet: NewTSet[string](safe...),
 	}
 }

 // NewStrSetFrom returns a new set from `items`.
 func NewStrSetFrom(items []string, safe ...bool) *StrSet {
-	m := make(map[string]struct{})
-	for _, v := range items {
-		m[v] = struct{}{}
-	}
 	return &StrSet{
-		mu:   rwmutex.Create(safe...),
-		data: m,
+		TSet: NewTSetFrom(items, safe...),
 	}
 }

+// lazyInit lazily initializes the set.
+func (a *StrSet) lazyInit() {
+	a.once.Do(func() {
+		if a.TSet == nil {
+			a.TSet = NewTSet[string]()
+		}
+	})
+}
+
 // Iterator iterates the set readonly with given callback function `f`,
 // if `f` returns true then continue iterating; or false to stop.
 func (set *StrSet) Iterator(f func(v string) bool) {
-	for _, k := range set.Slice() {
-		if !f(k) {
-			break
-		}
-	}
+	set.lazyInit()
+	set.TSet.Iterator(f)
 }

 // Add adds one or multiple items to the set.
 func (set *StrSet) Add(item ...string) {
-	set.mu.Lock()
-	if set.data == nil {
-		set.data = make(map[string]struct{})
-	}
-	for _, v := range item {
-		set.data[v] = struct{}{}
-	}
-	set.mu.Unlock()
+	set.lazyInit()
+	set.TSet.Add(item...)
 }

 // AddIfNotExist checks whether item exists in the set,
 // it adds the item to set and returns true if it does not exist in the set,
 // or else it does nothing and returns false.
 func (set *StrSet) AddIfNotExist(item string) bool {
-	if !set.Contains(item) {
-		set.mu.Lock()
-		defer set.mu.Unlock()
-		if set.data == nil {
-			set.data = make(map[string]struct{})
-		}
-		if _, ok := set.data[item]; !ok {
-			set.data[item] = struct{}{}
-			return true
-		}
-	}
-	return false
+	set.lazyInit()
+	return set.TSet.AddIfNotExist(item)
 }

 // AddIfNotExistFunc checks whether item exists in the set,
@ -91,20 +70,8 @@ func (set *StrSet) AddIfNotExist(item string) bool {
 //
 // Note that, the function `f` is executed without writing lock.
 func (set *StrSet) AddIfNotExistFunc(item string, f func() bool) bool {
-	if !set.Contains(item) {
-		if f() {
-			set.mu.Lock()
-			defer set.mu.Unlock()
-			if set.data == nil {
-				set.data = make(map[string]struct{})
-			}
-			if _, ok := set.data[item]; !ok {
-				set.data[item] = struct{}{}
-				return true
-			}
-		}
-	}
-	return false
+	set.lazyInit()
+	return set.TSet.AddIfNotExistFunc(item, f)
 }

 // AddIfNotExistFuncLock checks whether item exists in the set,
@ -113,36 +80,20 @@ func (set *StrSet) AddIfNotExistFunc(item string, f func() bool) bool {
 //
 // Note that, the function `f` is executed without writing lock.
 func (set *StrSet) AddIfNotExistFuncLock(item string, f func() bool) bool {
-	if !set.Contains(item) {
-		set.mu.Lock()
-		defer set.mu.Unlock()
-		if set.data == nil {
-			set.data = make(map[string]struct{})
-		}
-		if f() {
-			if _, ok := set.data[item]; !ok {
-				set.data[item] = struct{}{}
-				return true
-			}
-		}
-	}
-	return false
+	set.lazyInit()
+	return set.TSet.AddIfNotExistFuncLock(item, f)
 }

 // Contains checks whether the set contains `item`.
 func (set *StrSet) Contains(item string) bool {
-	var ok bool
-	set.mu.RLock()
-	if set.data != nil {
-		_, ok = set.data[item]
-	}
-	set.mu.RUnlock()
-	return ok
+	set.lazyInit()
+	return set.TSet.Contains(item)
 }

 // ContainsI checks whether a value exists in the set with case-insensitively.
 // Note that it internally iterates the whole set to do the comparison with case-insensitively.
 func (set *StrSet) ContainsI(item string) bool {
+	set.lazyInit()
 	set.mu.RLock()
 	defer set.mu.RUnlock()
 	for k := range set.data {
@ -155,64 +106,32 @@ func (set *StrSet) ContainsI(item string) bool {

 // Remove deletes `item` from set.
 func (set *StrSet) Remove(item string) {
-	set.mu.Lock()
-	if set.data != nil {
-		delete(set.data, item)
-	}
-	set.mu.Unlock()
+	set.lazyInit()
+	set.TSet.Remove(item)
 }

 // Size returns the size of the set.
 func (set *StrSet) Size() int {
-	set.mu.RLock()
-	l := len(set.data)
-	set.mu.RUnlock()
-	return l
+	set.lazyInit()
+	return set.TSet.Size()
 }

 // Clear deletes all items of the set.
 func (set *StrSet) Clear() {
-	set.mu.Lock()
-	set.data = make(map[string]struct{})
-	set.mu.Unlock()
+	set.lazyInit()
+	set.TSet.Clear()
 }

 // Slice returns the an of items of the set as slice.
 func (set *StrSet) Slice() []string {
-	set.mu.RLock()
-	var (
-		i   = 0
-		ret = make([]string, len(set.data))
-	)
-	for item := range set.data {
-		ret[i] = item
-		i++
-	}
-
-	set.mu.RUnlock()
-	return ret
+	set.lazyInit()
+	return set.TSet.Slice()
 }

 // Join joins items with a string `glue`.
 func (set *StrSet) Join(glue string) string {
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	if len(set.data) == 0 {
-		return ""
-	}
-	var (
-		l      = len(set.data)
-		i      = 0
-		buffer = bytes.NewBuffer(nil)
-	)
-	for k := range set.data {
-		buffer.WriteString(k)
-		if i != l-1 {
-			buffer.WriteString(glue)
-		}
-		i++
-	}
-	return buffer.String()
+	set.lazyInit()
+	return set.TSet.Join(glue)
 }

 // String returns items as a string, which implements like json.Marshal does.
@ -220,57 +139,27 @@ func (set *StrSet) String() string {
 	if set == nil {
 		return ""
 	}
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	var (
-		l      = len(set.data)
-		i      = 0
-		buffer = bytes.NewBuffer(nil)
-	)
-	buffer.WriteByte('[')
-	for k := range set.data {
-		buffer.WriteString(`"` + gstr.QuoteMeta(k, `"\`) + `"`)
-		if i != l-1 {
-			buffer.WriteByte(',')
-		}
-		i++
-	}
-	buffer.WriteByte(']')
-	return buffer.String()
+	set.lazyInit()
+	return set.TSet.String()
 }

 // LockFunc locks writing with callback function `f`.
 func (set *StrSet) LockFunc(f func(m map[string]struct{})) {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	f(set.data)
+	set.lazyInit()
+	set.TSet.LockFunc(f)
 }

 // RLockFunc locks reading with callback function `f`.
 func (set *StrSet) RLockFunc(f func(m map[string]struct{})) {
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	f(set.data)
+	set.lazyInit()
+	set.TSet.RLockFunc(f)
 }

 // Equal checks whether the two sets equal.
 func (set *StrSet) Equal(other *StrSet) bool {
-	if set == other {
-		return true
-	}
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	if len(set.data) != len(other.data) {
-		return false
-	}
-	for key := range set.data {
-		if _, ok := other.data[key]; !ok {
-			return false
-		}
-	}
-	return true
+	set.lazyInit()
+	other.lazyInit()
+	return set.TSet.Equal(other.TSet)
 }

 // IsSubsetOf checks whether the current set is a sub-set of `other`.
@ -278,85 +167,38 @@ func (set *StrSet) IsSubsetOf(other *StrSet) bool {
 	if set == other {
 		return true
 	}
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	other.mu.RLock()
-	defer other.mu.RUnlock()
-	for key := range set.data {
-		if _, ok := other.data[key]; !ok {
-			return false
-		}
-	}
-	return true
+
+	set.lazyInit()
+	other.lazyInit()
+
+	return set.TSet.IsSubsetOf(other.TSet)
 }

 // Union returns a new set which is the union of `set` and `other`.
 // Which means, all the items in `newSet` are in `set` or in `other`.
 func (set *StrSet) Union(others ...*StrSet) (newSet *StrSet) {
-	newSet = NewStrSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for _, other := range others {
-		if set != other {
-			other.mu.RLock()
-		}
-		for k, v := range set.data {
-			newSet.data[k] = v
-		}
-		if set != other {
-			for k, v := range other.data {
-				newSet.data[k] = v
-			}
-		}
-		if set != other {
-			other.mu.RUnlock()
-		}
+	set.lazyInit()
+	return &StrSet{
+		TSet: set.TSet.Union(set.toTSetSlice(others)...),
 	}
-
-	return
 }

 // Diff returns a new set which is the difference set from `set` to `other`.
 // Which means, all the items in `newSet` are in `set` but not in `other`.
 func (set *StrSet) Diff(others ...*StrSet) (newSet *StrSet) {
-	newSet = NewStrSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for _, other := range others {
-		if set == other {
-			continue
-		}
-		other.mu.RLock()
-		for k, v := range set.data {
-			if _, ok := other.data[k]; !ok {
-				newSet.data[k] = v
-			}
-		}
-		other.mu.RUnlock()
+	set.lazyInit()
+	return &StrSet{
+		TSet: set.TSet.Diff(set.toTSetSlice(others)...),
 	}
-	return
 }

 // Intersect returns a new set which is the intersection from `set` to `other`.
 // Which means, all the items in `newSet` are in `set` and also in `other`.
 func (set *StrSet) Intersect(others ...*StrSet) (newSet *StrSet) {
-	newSet = NewStrSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for _, other := range others {
-		if set != other {
-			other.mu.RLock()
-		}
-		for k, v := range set.data {
-			if _, ok := other.data[k]; ok {
-				newSet.data[k] = v
-			}
-		}
-		if set != other {
-			other.mu.RUnlock()
-		}
+	set.lazyInit()
+	return &StrSet{
+		TSet: set.TSet.Intersect(set.toTSetSlice(others)...),
 	}
-	return
 }

 // Complement returns a new set which is the complement from `set` to `full`.
@ -365,36 +207,22 @@ func (set *StrSet) Intersect(others ...*StrSet) (newSet *StrSet) {
 // It returns the difference between `full` and `set`
 // if the given set `full` is not the full set of `set`.
 func (set *StrSet) Complement(full *StrSet) (newSet *StrSet) {
-	newSet = NewStrSet()
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	if set != full {
-		full.mu.RLock()
-		defer full.mu.RUnlock()
-	}
-	for k, v := range full.data {
-		if _, ok := set.data[k]; !ok {
-			newSet.data[k] = v
+	set.lazyInit()
+	if full == nil {
+		return &StrSet{
+			TSet: NewTSet[string](),
 		}
 	}
-	return
+	full.lazyInit()
+	return &StrSet{
+		TSet: set.TSet.Complement(full.TSet),
+	}
 }

 // Merge adds items from `others` sets into `set`.
 func (set *StrSet) Merge(others ...*StrSet) *StrSet {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	for _, other := range others {
-		if set != other {
-			other.mu.RLock()
-		}
-		for k, v := range other.data {
-			set.data[k] = v
-		}
-		if set != other {
-			other.mu.RUnlock()
-		}
-	}
+	set.lazyInit()
+	set.TSet.Merge(set.toTSetSlice(others)...)
 	return set
 }

@ -402,101 +230,46 @@ func (set *StrSet) Merge(others ...*StrSet) *StrSet {
 // Note: The items should be converted to int type,
 // or you'd get a result that you unexpected.
 func (set *StrSet) Sum() (sum int) {
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	for k := range set.data {
-		sum += gconv.Int(k)
-	}
-	return
+	set.lazyInit()
+	return set.TSet.Sum()
 }

 // Pop randomly pops an item from set.
 func (set *StrSet) Pop() string {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	for k := range set.data {
-		delete(set.data, k)
-		return k
-	}
-	return ""
+	set.lazyInit()
+	return set.TSet.Pop()
 }

 // Pops randomly pops `size` items from set.
 // It returns all items if size == -1.
 func (set *StrSet) Pops(size int) []string {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	if size > len(set.data) || size == -1 {
-		size = len(set.data)
-	}
-	if size <= 0 {
-		return nil
-	}
-	index := 0
-	array := make([]string, size)
-	for k := range set.data {
-		delete(set.data, k)
-		array[index] = k
-		index++
-		if index == size {
-			break
-		}
-	}
-	return array
+	set.lazyInit()
+	return set.TSet.Pops(size)
 }

 // Walk applies a user supplied function `f` to every item of set.
 func (set *StrSet) Walk(f func(item string) string) *StrSet {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	m := make(map[string]struct{}, len(set.data))
-	for k, v := range set.data {
-		m[f(k)] = v
-	}
-	set.data = m
+	set.lazyInit()
+	set.TSet.Walk(f)
 	return set
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (set StrSet) MarshalJSON() ([]byte, error) {
-	return json.Marshal(set.Slice())
+	set.lazyInit()
+	return set.TSet.MarshalJSON()
 }

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (set *StrSet) UnmarshalJSON(b []byte) error {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	if set.data == nil {
-		set.data = make(map[string]struct{})
-	}
-	var array []string
-	if err := json.UnmarshalUseNumber(b, &array); err != nil {
-		return err
-	}
-	for _, v := range array {
-		set.data[v] = struct{}{}
-	}
-	return nil
+	set.lazyInit()
+	return set.TSet.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for set.
 func (set *StrSet) UnmarshalValue(value any) (err error) {
-	set.mu.Lock()
-	defer set.mu.Unlock()
-	if set.data == nil {
-		set.data = make(map[string]struct{})
-	}
-	var array []string
-	switch value.(type) {
-	case string, []byte:
-		err = json.UnmarshalUseNumber(gconv.Bytes(value), &array)
-	default:
-		array = gconv.SliceStr(value)
-	}
-	for _, v := range array {
-		set.data[v] = struct{}{}
-	}
-	return
+	set.lazyInit()
+	return set.TSet.UnmarshalValue(value)
 }

 // DeepCopy implements interface for deep copy of current type.
@ -504,15 +277,21 @@ func (set *StrSet) DeepCopy() any {
 	if set == nil {
 		return nil
 	}
-	set.mu.RLock()
-	defer set.mu.RUnlock()
-	var (
-		slice = make([]string, len(set.data))
-		index = 0
-	)
-	for k := range set.data {
-		slice[index] = k
-		index++
+	set.lazyInit()
+	return &StrSet{
+		TSet: set.TSet.DeepCopy().(*TSet[string]),
 	}
-	return NewStrSetFrom(slice, set.mu.IsSafe())
+}
+
+// toTSetSlice converts []*StrSet to []*TSet[string]
+func (set *StrSet) toTSetSlice(sets []*StrSet) (tSets []*TSet[string]) {
+	tSets = make([]*TSet[string], len(sets))
+	for i, v := range sets {
+		if v == nil {
+			continue
+		}
+		v.lazyInit()
+		tSets[i] = v.TSet
+	}
+	return
 }
--- a/container/gset/gset_t_set.go
+++ b/container/gset/gset_t_set.go
@ -0,0 +1,531 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package gset
+
+import (
+	"bytes"
+
+	"github.com/gogf/gf/v2/internal/json"
+	"github.com/gogf/gf/v2/internal/rwmutex"
+	"github.com/gogf/gf/v2/text/gstr"
+	"github.com/gogf/gf/v2/util/gconv"
+)
+
+// TSet[T] is consisted of any items.
+type TSet[T comparable] struct {
+	mu   rwmutex.RWMutex
+	data map[T]struct{}
+}
+
+// NewTSet creates and returns a new set, which contains un-repeated items.
+// Also see New.
+func NewTSet[T comparable](safe ...bool) *TSet[T] {
+	return &TSet[T]{
+		data: make(map[T]struct{}),
+		mu:   rwmutex.Create(safe...),
+	}
+}
+
+// NewTSetFrom returns a new set from `items`.
+// `items` - A slice of type T.
+func NewTSetFrom[T comparable](items []T, safe ...bool) *TSet[T] {
+	m := make(map[T]struct{})
+	for _, v := range items {
+		m[v] = struct{}{}
+	}
+	return &TSet[T]{
+		data: m,
+		mu:   rwmutex.Create(safe...),
+	}
+}
+
+// Iterator iterates the set readonly with given callback function `f`,
+// if `f` returns true then continue iterating; or false to stop.
+func (set *TSet[T]) Iterator(f func(v T) bool) {
+	for _, k := range set.Slice() {
+		if !f(k) {
+			break
+		}
+	}
+}
+
+// Add adds one or multiple items to the set.
+func (set *TSet[T]) Add(items ...T) {
+	set.mu.Lock()
+	if set.data == nil {
+		set.data = make(map[T]struct{})
+	}
+	for _, v := range items {
+		set.data[v] = struct{}{}
+	}
+	set.mu.Unlock()
+}
+
+// AddIfNotExist checks whether item exists in the set,
+// it adds the item to set and returns true if it does not exists in the set,
+// or else it does nothing and returns false.
+//
+// Note that, if `item` is nil, it does nothing and returns false.
+func (set *TSet[T]) AddIfNotExist(item T) bool {
+	if any(item) == nil {
+		return false
+	}
+	if !set.Contains(item) {
+		set.mu.Lock()
+		defer set.mu.Unlock()
+		if set.data == nil {
+			set.data = make(map[T]struct{})
+		}
+		if _, ok := set.data[item]; !ok {
+			set.data[item] = struct{}{}
+			return true
+		}
+	}
+	return false
+}
+
+// AddIfNotExistFunc checks whether item exists in the set,
+// it adds the item to set and returns true if it does not exist in the set and
+// function `f` returns true, or else it does nothing and returns false.
+//
+// Note that, if `item` is nil, it does nothing and returns false. The function `f`
+// is executed without writing lock.
+func (set *TSet[T]) AddIfNotExistFunc(item T, f func() bool) bool {
+	if any(item) == nil {
+		return false
+	}
+	if !set.Contains(item) {
+		if f() {
+			set.mu.Lock()
+			defer set.mu.Unlock()
+			if set.data == nil {
+				set.data = make(map[T]struct{})
+			}
+			if _, ok := set.data[item]; !ok {
+				set.data[item] = struct{}{}
+				return true
+			}
+		}
+	}
+	return false
+}
+
+// AddIfNotExistFuncLock checks whether item exists in the set,
+// it adds the item to set and returns true if it does not exists in the set and
+// function `f` returns true, or else it does nothing and returns false.
+//
+// Note that, if `item` is nil, it does nothing and returns false. The function `f`
+// is executed within writing lock.
+func (set *TSet[T]) AddIfNotExistFuncLock(item T, f func() bool) bool {
+	if any(item) == nil {
+		return false
+	}
+	if !set.Contains(item) {
+		set.mu.Lock()
+		defer set.mu.Unlock()
+		if set.data == nil {
+			set.data = make(map[T]struct{})
+		}
+		if f() {
+			if _, ok := set.data[item]; !ok {
+				set.data[item] = struct{}{}
+				return true
+			}
+		}
+	}
+	return false
+}
+
+// Contains checks whether the set contains `item`.
+func (set *TSet[T]) Contains(item T) bool {
+	var ok bool
+	set.mu.RLock()
+	if set.data != nil {
+		_, ok = set.data[item]
+	}
+	set.mu.RUnlock()
+	return ok
+}
+
+// Remove deletes `item` from set.
+func (set *TSet[T]) Remove(item T) {
+	set.mu.Lock()
+	if set.data != nil {
+		delete(set.data, item)
+	}
+	set.mu.Unlock()
+}
+
+// Size returns the size of the set.
+func (set *TSet[T]) Size() int {
+	set.mu.RLock()
+	l := len(set.data)
+	set.mu.RUnlock()
+	return l
+}
+
+// Clear deletes all items of the set.
+func (set *TSet[T]) Clear() {
+	set.mu.Lock()
+	set.data = make(map[T]struct{})
+	set.mu.Unlock()
+}
+
+// Slice returns all items of the set as slice.
+func (set *TSet[T]) Slice() []T {
+	set.mu.RLock()
+	var (
+		i   = 0
+		ret = make([]T, len(set.data))
+	)
+	for item := range set.data {
+		ret[i] = item
+		i++
+	}
+	set.mu.RUnlock()
+	return ret
+}
+
+// Join joins items with a string `glue`.
+func (set *TSet[T]) Join(glue string) string {
+	set.mu.RLock()
+	defer set.mu.RUnlock()
+	if len(set.data) == 0 {
+		return ""
+	}
+	var (
+		l      = len(set.data)
+		i      = 0
+		buffer = bytes.NewBuffer(nil)
+	)
+	for k := range set.data {
+		buffer.WriteString(gconv.String(k))
+		if i != l-1 {
+			buffer.WriteString(glue)
+		}
+		i++
+	}
+	return buffer.String()
+}
+
+// String returns items as a string, which implements like json.Marshal does.
+func (set *TSet[T]) String() string {
+	if set == nil {
+		return ""
+	}
+	set.mu.RLock()
+	defer set.mu.RUnlock()
+	var (
+		s      string
+		l      = len(set.data)
+		i      = 0
+		buffer = bytes.NewBuffer(nil)
+	)
+	buffer.WriteByte('[')
+	for k := range set.data {
+		s = gconv.String(k)
+		if gstr.IsNumeric(s) {
+			buffer.WriteString(s)
+		} else {
+			buffer.WriteString(`"` + gstr.QuoteMeta(s, `"\`) + `"`)
+		}
+		if i != l-1 {
+			buffer.WriteByte(',')
+		}
+		i++
+	}
+	buffer.WriteByte(']')
+	return buffer.String()
+}
+
+// LockFunc locks writing with callback function `f`.
+func (set *TSet[T]) LockFunc(f func(m map[T]struct{})) {
+	set.mu.Lock()
+	defer set.mu.Unlock()
+	f(set.data)
+}
+
+// RLockFunc locks reading with callback function `f`.
+func (set *TSet[T]) RLockFunc(f func(m map[T]struct{})) {
+	set.mu.RLock()
+	defer set.mu.RUnlock()
+	f(set.data)
+}
+
+// Equal checks whether the two sets equal.
+func (set *TSet[T]) Equal(other *TSet[T]) bool {
+	if set == other {
+		return true
+	}
+	set.mu.RLock()
+	defer set.mu.RUnlock()
+	other.mu.RLock()
+	defer other.mu.RUnlock()
+	if len(set.data) != len(other.data) {
+		return false
+	}
+	for key := range set.data {
+		if _, ok := other.data[key]; !ok {
+			return false
+		}
+	}
+	return true
+}
+
+// IsSubsetOf checks whether the current set is a sub-set of `other`.
+func (set *TSet[T]) IsSubsetOf(other *TSet[T]) bool {
+	if set == other {
+		return true
+	}
+	set.mu.RLock()
+	defer set.mu.RUnlock()
+	other.mu.RLock()
+	defer other.mu.RUnlock()
+	for key := range set.data {
+		if _, ok := other.data[key]; !ok {
+			return false
+		}
+	}
+	return true
+}
+
+// Union returns a new set which is the union of `set` and `others`.
+// Which means, all the items in `newSet` are in `set` or in `others`.
+func (set *TSet[T]) Union(others ...*TSet[T]) (newSet *TSet[T]) {
+	newSet = NewTSet[T]()
+	set.mu.RLock()
+	defer set.mu.RUnlock()
+	for _, other := range others {
+		if other == nil {
+			continue
+		}
+		if set != other {
+			other.mu.RLock()
+		}
+		for k, v := range set.data {
+			newSet.data[k] = v
+		}
+		if set != other {
+			for k, v := range other.data {
+				newSet.data[k] = v
+			}
+		}
+		if set != other {
+			other.mu.RUnlock()
+		}
+	}
+
+	return
+}
+
+// Diff returns a new set which is the difference set from `set` to `others`.
+// Which means, all the items in `newSet` are in `set` but not in `others`.
+func (set *TSet[T]) Diff(others ...*TSet[T]) (newSet *TSet[T]) {
+	newSet = NewTSet[T]()
+	set.mu.RLock()
+	defer set.mu.RUnlock()
+	for _, other := range others {
+		if other == nil {
+			continue
+		}
+		if set == other {
+			continue
+		}
+		other.mu.RLock()
+		for k, v := range set.data {
+			if _, ok := other.data[k]; !ok {
+				newSet.data[k] = v
+			}
+		}
+		other.mu.RUnlock()
+	}
+	return
+}
+
+// Intersect returns a new set which is the intersection from `set` to `others`.
+// Which means, all the items in `newSet` are in `set` and also in `others`.
+func (set *TSet[T]) Intersect(others ...*TSet[T]) (newSet *TSet[T]) {
+	newSet = NewTSet[T]()
+	set.mu.RLock()
+	defer set.mu.RUnlock()
+	for _, other := range others {
+		if other == nil {
+			continue
+		}
+		if set != other {
+			other.mu.RLock()
+		}
+		for k, v := range set.data {
+			if _, ok := other.data[k]; ok {
+				newSet.data[k] = v
+			}
+		}
+		if set != other {
+			other.mu.RUnlock()
+		}
+	}
+	return
+}
+
+// Complement returns a new set which is the complement from `set` to `full`.
+// Which means, all the items in `newSet` are in `full` and not in `set`.
+//
+// It returns the difference between `full` and `set`
+// if the given set `full` is not the full set of `set`.
+func (set *TSet[T]) Complement(full *TSet[T]) (newSet *TSet[T]) {
+	newSet = NewTSet[T]()
+	set.mu.RLock()
+	defer set.mu.RUnlock()
+	if set != full {
+		full.mu.RLock()
+		defer full.mu.RUnlock()
+	}
+	for k, v := range full.data {
+		if _, ok := set.data[k]; !ok {
+			newSet.data[k] = v
+		}
+	}
+	return
+}
+
+// Merge adds items from `others` sets into `set`.
+func (set *TSet[T]) Merge(others ...*TSet[T]) *TSet[T] {
+	set.mu.Lock()
+	defer set.mu.Unlock()
+	for _, other := range others {
+		if other == nil {
+			continue
+		}
+		if set != other {
+			other.mu.RLock()
+		}
+		for k, v := range other.data {
+			set.data[k] = v
+		}
+		if set != other {
+			other.mu.RUnlock()
+		}
+	}
+	return set
+}
+
+// Sum sums items.
+// Note: The items should be converted to int type,
+// or you'd get a result that you unexpected.
+func (set *TSet[T]) Sum() (sum int) {
+	set.mu.RLock()
+	defer set.mu.RUnlock()
+	for k := range set.data {
+		sum += gconv.Int(k)
+	}
+	return
+}
+
+// Pop randomly pops an item from set.
+func (set *TSet[T]) Pop() (item T) {
+	set.mu.Lock()
+	defer set.mu.Unlock()
+	for k := range set.data {
+		delete(set.data, k)
+		return k
+	}
+	return
+}
+
+// Pops randomly pops `size` items from set.
+// It returns all items if size == -1.
+func (set *TSet[T]) Pops(size int) []T {
+	set.mu.Lock()
+	defer set.mu.Unlock()
+	if size > len(set.data) || size == -1 {
+		size = len(set.data)
+	}
+	if size <= 0 {
+		return nil
+	}
+	index := 0
+	array := make([]T, size)
+	for k := range set.data {
+		delete(set.data, k)
+		array[index] = k
+		index++
+		if index == size {
+			break
+		}
+	}
+	return array
+}
+
+// Walk applies a user supplied function `f` to every item of set.
+func (set *TSet[T]) Walk(f func(item T) T) *TSet[T] {
+	set.mu.Lock()
+	defer set.mu.Unlock()
+	m := make(map[T]struct{}, len(set.data))
+	for k, v := range set.data {
+		m[f(k)] = v
+	}
+	set.data = m
+	return set
+}
+
+// MarshalJSON implements the interface MarshalJSON for json.Marshal.
+func (set TSet[T]) MarshalJSON() ([]byte, error) {
+	return json.Marshal(set.Slice())
+}
+
+// UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
+func (set *TSet[T]) UnmarshalJSON(b []byte) error {
+	set.mu.Lock()
+	defer set.mu.Unlock()
+	if set.data == nil {
+		set.data = make(map[T]struct{})
+	}
+	var array []T
+	if err := json.UnmarshalUseNumber(b, &array); err != nil {
+		return err
+	}
+	for _, v := range array {
+		set.data[v] = struct{}{}
+	}
+	return nil
+}
+
+// UnmarshalValue is an interface implement which sets any type of value for set.
+func (set *TSet[T]) UnmarshalValue(value any) (err error) {
+	set.mu.Lock()
+	defer set.mu.Unlock()
+	if set.data == nil {
+		set.data = make(map[T]struct{})
+	}
+	var array []T
+	switch value.(type) {
+	case string, []byte:
+		err = json.UnmarshalUseNumber(gconv.Bytes(value), &array)
+	default:
+		if err = gconv.Scan(value, &array); err != nil {
+			return
+		}
+	}
+	for _, v := range array {
+		set.data[v] = struct{}{}
+	}
+	return
+}
+
+// DeepCopy implements interface for deep copy of current type.
+func (set *TSet[T]) DeepCopy() any {
+	if set == nil {
+		return nil
+	}
+	set.mu.RLock()
+	defer set.mu.RUnlock()
+	data := make([]T, 0)
+	for k := range set.data {
+		data = append(data, k)
+	}
+	return NewTSetFrom[T](data, set.mu.IsSafe())
+}
--- a/container/gset/gset_z_unit_any_test.go
+++ b/container/gset/gset_z_unit_any_test.go
@ -187,6 +187,19 @@ func TestSet_Union(t *testing.T) {
 		t.Assert(s3.Contains(3), true)
 		t.Assert(s3.Contains(4), true)
 	})
+
+	// Test with nil element in slice
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewSet()
+		s2 := gset.NewSet()
+		s1.Add(1, 2)
+		s2.Add(3, 4)
+		s3 := s1.Union(s2, nil)
+		t.Assert(s3.Contains(1), true)
+		t.Assert(s3.Contains(2), true)
+		t.Assert(s3.Contains(3), true)
+		t.Assert(s3.Contains(4), true)
+	})
 }

 func TestSet_Diff(t *testing.T) {
@ -236,6 +249,14 @@ func TestSet_Complement(t *testing.T) {
 		t.Assert(s3.Contains(4), true)
 		t.Assert(s3.Contains(5), true)
 	})
+
+	// Test with nil full set
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewSet()
+		s1.Add(1, 2, 3)
+		s3 := s1.Complement(nil)
+		t.Assert(s3.Size(), 0)
+	})
 }

 func TestNewFrom(t *testing.T) {
--- a/container/gset/gset_z_unit_int_test.go
+++ b/container/gset/gset_z_unit_int_test.go
@ -167,6 +167,19 @@ func TestIntSet_Union(t *testing.T) {
 		t.Assert(s3.Contains(3), true)
 		t.Assert(s3.Contains(4), true)
 	})
+
+	// Test with nil element in slice
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewIntSet()
+		s2 := gset.NewIntSet()
+		s1.Add(1, 2)
+		s2.Add(3, 4)
+		s3 := s1.Union(s2, nil)
+		t.Assert(s3.Contains(1), true)
+		t.Assert(s3.Contains(2), true)
+		t.Assert(s3.Contains(3), true)
+		t.Assert(s3.Contains(4), true)
+	})
 }

 func TestIntSet_Diff(t *testing.T) {
@ -216,6 +229,14 @@ func TestIntSet_Complement(t *testing.T) {
 		t.Assert(s3.Contains(4), true)
 		t.Assert(s3.Contains(5), true)
 	})
+
+	// Test with nil full set
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewIntSet()
+		s1.Add(1, 2, 3)
+		s3 := s1.Complement(nil)
+		t.Assert(s3.Size(), 0)
+	})
 }

 func TestIntSet_Size(t *testing.T) {
--- a/container/gset/gset_z_unit_str_test.go
+++ b/container/gset/gset_z_unit_str_test.go
@ -178,6 +178,19 @@ func TestStrSet_Union(t *testing.T) {
 		t.Assert(s3.Contains("3"), true)
 		t.Assert(s3.Contains("4"), true)
 	})
+
+	// Test with nil element in slice
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewStrSet()
+		s2 := gset.NewStrSet()
+		s1.Add("1", "2")
+		s2.Add("3", "4")
+		s3 := s1.Union(s2, nil)
+		t.Assert(s3.Contains("1"), true)
+		t.Assert(s3.Contains("2"), true)
+		t.Assert(s3.Contains("3"), true)
+		t.Assert(s3.Contains("4"), true)
+	})
 }

 func TestStrSet_Diff(t *testing.T) {
@ -227,6 +240,14 @@ func TestStrSet_Complement(t *testing.T) {
 		t.Assert(s3.Contains("4"), true)
 		t.Assert(s3.Contains("5"), true)
 	})
+
+	// Test with nil full set
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewStrSet()
+		s1.Add("1", "2", "3")
+		s3 := s1.Complement(nil)
+		t.Assert(s3.Size(), 0)
+	})
 }

 func TestNewIntSetFrom(t *testing.T) {
--- a/container/gset/gset_z_unit_t_set_test.go
+++ b/container/gset/gset_z_unit_t_set_test.go
@ -0,0 +1,593 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package gset_test
+
+import (
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/gogf/gf/v2/container/gset"
+	"github.com/gogf/gf/v2/internal/json"
+	"github.com/gogf/gf/v2/test/gtest"
+)
+
+func TestTSet_New(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		s.Add(1, 1, 2)
+		s.Add([]int{3, 4}...)
+		t.Assert(s.Size(), 4)
+		t.AssertIN(1, s.Slice())
+		t.AssertIN(2, s.Slice())
+		t.AssertIN(3, s.Slice())
+		t.AssertIN(4, s.Slice())
+		t.AssertNI(0, s.Slice())
+		t.Assert(s.Contains(4), true)
+		t.Assert(s.Contains(5), false)
+		s.Remove(1)
+		t.Assert(s.Size(), 3)
+		s.Clear()
+		t.Assert(s.Size(), 0)
+	})
+}
+
+func TestTSet_NewFrom(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3}, true)
+		t.Assert(s.Size(), 3)
+		t.Assert(s.Contains(1), true)
+		t.Assert(s.Contains(2), true)
+		t.Assert(s.Contains(3), true)
+		t.Assert(s.Contains(4), false)
+	})
+}
+
+func TestTSet_Add_NilData(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		var s gset.TSet[int]
+		s.Add(1, 2, 3)
+		t.Assert(s.Size(), 3)
+		t.Assert(s.Contains(1), true)
+	})
+}
+
+func TestTSet_AddIfNotExist(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int](true)
+		s.Add(1)
+		t.Assert(s.Contains(1), true)
+		t.Assert(s.AddIfNotExist(1), false)
+		t.Assert(s.AddIfNotExist(2), true)
+		t.Assert(s.Contains(2), true)
+		t.Assert(s.AddIfNotExist(2), false)
+	})
+
+	// Test with pointer type to test nil check
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[*int](true)
+
+		val := 1
+		ptr := &val
+		t.Assert(s.AddIfNotExist(ptr), true)
+		t.Assert(s.AddIfNotExist(ptr), false)
+	})
+
+	// Test nil data map initialization
+	gtest.C(t, func(t *gtest.T) {
+		var s gset.TSet[int]
+		t.Assert(s.AddIfNotExist(1), true)
+		t.Assert(s.Size(), 1)
+	})
+}
+
+func TestTSet_AddIfNotExistFunc(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int](true)
+		s.Add(1)
+		t.Assert(s.Contains(1), true)
+		t.Assert(s.Contains(2), false)
+		t.Assert(s.AddIfNotExistFunc(2, func() bool { return false }), false)
+		t.Assert(s.Contains(2), false)
+		t.Assert(s.AddIfNotExistFunc(2, func() bool { return true }), true)
+		t.Assert(s.Contains(2), true)
+		t.Assert(s.AddIfNotExistFunc(2, func() bool { return true }), false)
+		t.Assert(s.Contains(2), true)
+	})
+
+	// Test concurrent scenario
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int](true)
+		wg := sync.WaitGroup{}
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+			r := s.AddIfNotExistFunc(1, func() bool {
+				time.Sleep(100 * time.Millisecond)
+				return true
+			})
+			t.Assert(r, false)
+		}()
+		s.Add(1)
+		wg.Wait()
+	})
+
+	// Test nil data map initialization
+	gtest.C(t, func(t *gtest.T) {
+		var s gset.TSet[int]
+		t.Assert(s.AddIfNotExistFunc(1, func() bool { return true }), true)
+		t.Assert(s.Size(), 1)
+	})
+}
+
+func TestTSet_AddIfNotExistFuncLock(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int](true)
+		s.Add(1)
+		t.Assert(s.Contains(1), true)
+		t.Assert(s.Contains(2), false)
+		t.Assert(s.AddIfNotExistFuncLock(2, func() bool { return false }), false)
+		t.Assert(s.Contains(2), false)
+		t.Assert(s.AddIfNotExistFuncLock(2, func() bool { return true }), true)
+		t.Assert(s.Contains(2), true)
+		t.Assert(s.AddIfNotExistFuncLock(2, func() bool { return true }), false)
+		t.Assert(s.Contains(2), true)
+	})
+
+	// Test nil data map initialization
+	gtest.C(t, func(t *gtest.T) {
+		var s gset.TSet[int]
+		t.Assert(s.AddIfNotExistFuncLock(1, func() bool { return true }), true)
+		t.Assert(s.Size(), 1)
+	})
+}
+
+func TestTSet_Iterator(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3, 4, 5}, true)
+		var sum int
+		s.Iterator(func(v int) bool {
+			sum += v
+			return true
+		})
+		t.Assert(sum, 15)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3, 4, 5}, true)
+		var count int
+		s.Iterator(func(v int) bool {
+			count++
+			return count < 3
+		})
+		t.Assert(count, 3)
+	})
+}
+
+func TestTSet_Join(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		t.Assert(s.Join(","), "")
+		s.Add(1, 2, 3)
+		result := s.Join(",")
+		t.Assert(len(result) > 0, true)
+	})
+}
+
+func TestTSet_String(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		var s *gset.TSet[int]
+		t.Assert(s.String(), "")
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		t.Assert(s.String(), "[]")
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		result := s.String()
+		t.Assert(len(result) > 2, true)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[string]([]string{"a", "b", "c"})
+		result := s.String()
+		t.Assert(len(result) > 2, true)
+	})
+}
+
+func TestTSet_Equal(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s2 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		t.Assert(s1.Equal(s2), true)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s2 := gset.NewTSetFrom[int]([]int{1, 2, 3, 4})
+		t.Assert(s1.Equal(s2), false)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		t.Assert(s1.Equal(s1), true)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s2 := gset.NewTSetFrom[int]([]int{1, 2, 4})
+		t.Assert(s1.Equal(s2), false)
+	})
+}
+
+func TestTSet_IsSubsetOf(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2})
+		s2 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		t.Assert(s1.IsSubsetOf(s2), true)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s2 := gset.NewTSetFrom[int]([]int{1, 2})
+		t.Assert(s1.IsSubsetOf(s2), false)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		t.Assert(s1.IsSubsetOf(s1), true)
+	})
+}
+
+func TestTSet_Union(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s2 := gset.NewTSetFrom[int]([]int{3, 4, 5})
+		s := s1.Union(s2)
+		t.Assert(s.Size(), 5)
+		t.Assert(s.Contains(1), true)
+		t.Assert(s.Contains(2), true)
+		t.Assert(s.Contains(3), true)
+		t.Assert(s.Contains(4), true)
+		t.Assert(s.Contains(5), true)
+	})
+
+	// Test with nil set - should skip it and copy s1 data
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		var s2 *gset.TSet[int]
+		s := s1.Union(s2)
+		// Since s2 is nil and skipped, newSet will be empty
+		// because the loop runs but nothing is copied when other is nil
+		t.Assert(s.Size(), 0)
+	})
+
+	// Test with self
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s := s1.Union(s1)
+		t.Assert(s.Size(), 3)
+	})
+}
+
+func TestTSet_Diff(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s2 := gset.NewTSetFrom[int]([]int{3, 4, 5})
+		s := s1.Diff(s2)
+		t.Assert(s.Size(), 2)
+		t.Assert(s.Contains(1), true)
+		t.Assert(s.Contains(2), true)
+		t.Assert(s.Contains(3), false)
+	})
+
+	// Test with nil set - should skip it
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		var s2 *gset.TSet[int]
+		s := s1.Diff(s2)
+		// Since s2 is nil and skipped, newSet will be empty
+		// because the loop runs but nothing is copied when other is nil
+		t.Assert(s.Size(), 0)
+	})
+
+	// Test with self
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s := s1.Diff(s1)
+		t.Assert(s.Size(), 0)
+	})
+}
+
+func TestTSet_Intersect(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s2 := gset.NewTSetFrom[int]([]int{3, 4, 5})
+		s := s1.Intersect(s2)
+		t.Assert(s.Size(), 1)
+		t.Assert(s.Contains(3), true)
+	})
+
+	// Test with nil set
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		var s2 *gset.TSet[int]
+		s := s1.Intersect(s2)
+		t.Assert(s.Size(), 0)
+	})
+
+	// Test with self
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s := s1.Intersect(s1)
+		t.Assert(s.Size(), 3)
+	})
+}
+
+func TestTSet_Complement(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s2 := gset.NewTSetFrom[int]([]int{1, 2, 3, 4, 5})
+		s := s1.Complement(s2)
+		t.Assert(s.Size(), 2)
+		t.Assert(s.Contains(4), true)
+		t.Assert(s.Contains(5), true)
+	})
+
+	// Test with self
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s := s1.Complement(s1)
+		t.Assert(s.Size(), 0)
+	})
+}
+
+func TestTSet_Merge(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s2 := gset.NewTSetFrom[int]([]int{3, 4, 5})
+		s1.Merge(s2)
+		t.Assert(s1.Size(), 5)
+		t.Assert(s1.Contains(1), true)
+		t.Assert(s1.Contains(2), true)
+		t.Assert(s1.Contains(3), true)
+		t.Assert(s1.Contains(4), true)
+		t.Assert(s1.Contains(5), true)
+	})
+
+	// Test with nil set
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		var s2 *gset.TSet[int]
+		s1.Merge(s2)
+		t.Assert(s1.Size(), 3)
+	})
+
+	// Test with self
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		s1.Merge(s1)
+		t.Assert(s1.Size(), 3)
+	})
+}
+
+func TestTSet_Sum(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		t.Assert(s.Sum(), 6)
+	})
+}
+
+func TestTSet_Pop(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		item := s.Pop()
+		t.Assert(s.Size(), 2)
+		t.Assert(s.Contains(item), false)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		item := s.Pop()
+		t.Assert(item, 0)
+	})
+}
+
+func TestTSet_Pops(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3, 4, 5})
+		items := s.Pops(3)
+		t.Assert(len(items), 3)
+		t.Assert(s.Size(), 2)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		items := s.Pops(-1)
+		t.Assert(len(items), 3)
+		t.Assert(s.Size(), 0)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		items := s.Pops(0)
+		t.Assert(items, nil)
+		t.Assert(s.Size(), 3)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		items := s.Pops(10)
+		t.Assert(len(items), 3)
+		t.Assert(s.Size(), 0)
+	})
+}
+
+func TestTSet_Walk(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2})
+		s.Walk(func(item int) int {
+			return item + 10
+		})
+		t.Assert(s.Size(), 2)
+		t.Assert(s.Contains(11), true)
+		t.Assert(s.Contains(12), true)
+	})
+}
+
+func TestTSet_MarshalJSON(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3})
+		b, err := json.Marshal(s)
+		t.AssertNil(err)
+		t.Assert(len(b) > 0, true)
+	})
+}
+
+func TestTSet_UnmarshalJSON(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		b := []byte(`[1,2,3]`)
+		err := json.UnmarshalUseNumber(b, &s)
+		t.AssertNil(err)
+		t.Assert(s.Size(), 3)
+		t.Assert(s.Contains(1), true)
+		t.Assert(s.Contains(2), true)
+		t.Assert(s.Contains(3), true)
+	})
+
+	// Test with nil data map
+	gtest.C(t, func(t *gtest.T) {
+		var s gset.TSet[int]
+		b := []byte(`[1,2,3]`)
+		err := json.UnmarshalUseNumber(b, &s)
+		t.AssertNil(err)
+		t.Assert(s.Size(), 3)
+	})
+
+	// Test with invalid JSON
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		b := []byte(`{invalid}`)
+		err := json.UnmarshalUseNumber(b, &s)
+		t.AssertNE(err, nil)
+	})
+
+	// Test with empty array
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		b := []byte(`[]`)
+		err := json.UnmarshalUseNumber(b, &s)
+		t.AssertNil(err)
+		t.Assert(s.Size(), 0)
+	})
+}
+
+func TestTSet_UnmarshalValue(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		err := s.UnmarshalValue([]byte(`[1,2,3]`))
+		t.AssertNil(err)
+		t.Assert(s.Size(), 3)
+		t.Assert(s.Contains(1), true)
+		t.Assert(s.Contains(2), true)
+		t.Assert(s.Contains(3), true)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		err := s.UnmarshalValue(`[1,2,3]`)
+		t.AssertNil(err)
+		t.Assert(s.Size(), 3)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		err := s.UnmarshalValue([]int{1, 2, 3})
+		t.AssertNil(err)
+		t.Assert(s.Size(), 3)
+	})
+
+	// Test with nil data map
+	gtest.C(t, func(t *gtest.T) {
+		var s gset.TSet[int]
+		err := s.UnmarshalValue([]int{1, 2, 3})
+		t.AssertNil(err)
+		t.Assert(s.Size(), 3)
+	})
+
+	// Test error case with invalid JSON
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		err := s.UnmarshalValue([]byte(`{invalid}`))
+		t.AssertNE(err, nil)
+	})
+
+	// Test with empty array for string/bytes case
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		err := s.UnmarshalValue([]byte(`[]`))
+		t.AssertNil(err)
+		t.Assert(s.Size(), 0)
+	})
+
+	// Test with empty slice for default case
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSet[int]()
+		err := s.UnmarshalValue([]int{})
+		t.AssertNil(err)
+		t.Assert(s.Size(), 0)
+	})
+}
+
+func TestTSet_DeepCopy(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s1 := gset.NewTSetFrom[int]([]int{1, 2, 3}, true)
+		s2 := s1.DeepCopy().(*gset.TSet[int])
+		t.Assert(s1.Size(), s2.Size())
+		t.Assert(s1.Contains(1), s2.Contains(1))
+		t.Assert(s1.Contains(2), s2.Contains(2))
+		t.Assert(s1.Contains(3), s2.Contains(3))
+
+		s1.Add(4)
+		t.Assert(s1.Size(), 4)
+		t.Assert(s2.Size(), 3)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		var s1 *gset.TSet[int]
+		s2 := s1.DeepCopy()
+		t.Assert(s2, nil)
+	})
+}
+
+func TestTSet_LockFunc(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3}, true)
+		s.LockFunc(func(m map[int]struct{}) {
+			m[4] = struct{}{}
+		})
+		t.Assert(s.Size(), 4)
+		t.Assert(s.Contains(4), true)
+	})
+}
+
+func TestTSet_RLockFunc(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		s := gset.NewTSetFrom[int]([]int{1, 2, 3}, true)
+		var sum int
+		s.RLockFunc(func(m map[int]struct{}) {
+			for k := range m {
+				sum += k
+			}
+		})
+		t.Assert(sum, 6)
+	})
+}
--- a/container/gtree/gtree.go
+++ b/container/gtree/gtree.go
@ -162,12 +162,12 @@ type iTree interface {
 	IteratorDescFrom(key any, match bool, f func(key, value any) bool)
 }

-// iteratorFromGetIndex returns the index of the key in the keys slice.
+// iteratorFromGetIndexT returns the index of the key in the keys slice.
 //
 // The parameter `match` specifies whether starting iterating only if the `key` is fully matched,
 // or else using index searching iterating.
 // If `isIterator` is true, iterator is available; or else not.
-func iteratorFromGetIndex(key any, keys []any, match bool) (index int, canIterator bool) {
+func iteratorFromGetIndexT[T comparable](key T, keys []T, match bool) (index int, canIterator bool) {
 	if match {
 		for i, k := range keys {
 			if k == key {
@ -176,10 +176,19 @@ func iteratorFromGetIndex(key any, keys []any, match bool) (index int, canIterat
 			}
 		}
 	} else {
-		if i, ok := key.(int); ok {
+		if i, ok := any(key).(int); ok {
 			canIterator = true
 			index = i
 		}
 	}
 	return
 }
+
+// iteratorFromGetIndex returns the index of the key in the keys slice.
+//
+// The parameter `match` specifies whether starting iterating only if the `key` is fully matched,
+// or else using index searching iterating.
+// If `isIterator` is true, iterator is available; or else not.
+func iteratorFromGetIndex(key any, keys []any, match bool) (index int, canIterator bool) {
+	return iteratorFromGetIndexT(key, keys, match)
+}
--- a/container/gtree/gtree_avltree.go
+++ b/container/gtree/gtree_avltree.go
@ -7,31 +7,22 @@
 package gtree

 import (
-	"fmt"
-
-	"github.com/emirpasic/gods/trees/avltree"
+	"sync"

 	"github.com/gogf/gf/v2/container/gvar"
-	"github.com/gogf/gf/v2/internal/rwmutex"
-	"github.com/gogf/gf/v2/text/gstr"
-	"github.com/gogf/gf/v2/util/gconv"
+	"github.com/gogf/gf/v2/util/gutil"
 )

 var _ iTree = (*AVLTree)(nil)

 // AVLTree holds elements of the AVL tree.
 type AVLTree struct {
-	mu         rwmutex.RWMutex
-	root       *AVLTreeNode
-	comparator func(v1, v2 any) int
-	tree       *avltree.Tree
+	*AVLKVTree[any, any]
+	once sync.Once
 }

 // AVLTreeNode is a single element within the tree.
-type AVLTreeNode struct {
-	Key   any
-	Value any
-}
+type AVLTreeNode = AVLKVTreeNode[any, any]

 // NewAVLTree instantiates an AVL tree with the custom key comparator.
 //
@ -39,9 +30,7 @@ type AVLTreeNode struct {
 // which is false in default.
 func NewAVLTree(comparator func(v1, v2 any) int, safe ...bool) *AVLTree {
 	return &AVLTree{
-		mu:         rwmutex.Create(safe...),
-		comparator: comparator,
-		tree:       avltree.NewWith(comparator),
+		AVLKVTree: NewAVLKVTree[any, any](comparator, safe...),
 	}
 }

@ -49,58 +38,55 @@ func NewAVLTree(comparator func(v1, v2 any) int, safe ...bool) *AVLTree {
 //
 // The parameter `safe` is used to specify whether using tree in concurrent-safety, which is false in default.
 func NewAVLTreeFrom(comparator func(v1, v2 any) int, data map[any]any, safe ...bool) *AVLTree {
-	tree := NewAVLTree(comparator, safe...)
-	for k, v := range data {
-		tree.doSet(k, v)
+	return &AVLTree{
+		AVLKVTree: NewAVLKVTreeFrom(comparator, data, safe...),
 	}
-	return tree
+}
+
+// lazyInit lazily initializes the tree.
+func (tree *AVLTree) lazyInit() {
+	tree.once.Do(func() {
+		if tree.AVLKVTree == nil {
+			tree.AVLKVTree = NewAVLKVTree[any, any](gutil.ComparatorTStr, false)
+		}
+	})
 }

 // Clone clones and returns a new tree from current tree.
 func (tree *AVLTree) Clone() *AVLTree {
-	newTree := NewAVLTree(tree.comparator, tree.mu.IsSafe())
-	newTree.Sets(tree.Map())
-	return newTree
+	if tree == nil {
+		return nil
+	}
+	tree.lazyInit()
+	return &AVLTree{
+		AVLKVTree: tree.AVLKVTree.Clone(),
+	}
 }

 // Set sets key-value pair into the tree.
 func (tree *AVLTree) Set(key any, value any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	tree.doSet(key, value)
+	tree.lazyInit()
+	tree.AVLKVTree.Set(key, value)
 }

 // Sets batch sets key-values to the tree.
 func (tree *AVLTree) Sets(data map[any]any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	for key, value := range data {
-		tree.doSet(key, value)
-	}
+	tree.lazyInit()
+	tree.AVLKVTree.Sets(data)
 }

 // SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
 // It returns false if `key` exists, and such setting key-value pair operation would be ignored.
 func (tree *AVLTree) SetIfNotExist(key any, value any) bool {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if _, ok := tree.doGet(key); !ok {
-		tree.doSet(key, value)
-		return true
-	}
-	return false
+	tree.lazyInit()
+	return tree.AVLKVTree.SetIfNotExist(key, value)
 }

 // SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
 // It returns false if `key` exists, and such setting key-value pair operation would be ignored.
 func (tree *AVLTree) SetIfNotExistFunc(key any, f func() any) bool {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if _, ok := tree.doGet(key); !ok {
-		tree.doSet(key, f())
-		return true
-	}
-	return false
+	tree.lazyInit()
+	return tree.AVLKVTree.SetIfNotExistFunc(key, f)
 }

 // SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
@ -109,13 +95,8 @@ func (tree *AVLTree) SetIfNotExistFunc(key any, f func() any) bool {
 // SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
 // it executes function `f` within mutex lock.
 func (tree *AVLTree) SetIfNotExistFuncLock(key any, f func() any) bool {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if _, ok := tree.doGet(key); !ok {
-		tree.doSet(key, f)
-		return true
-	}
-	return false
+	tree.lazyInit()
+	return tree.AVLKVTree.SetIfNotExistFuncLock(key, f)
 }

 // Get searches the `key` in the tree and returns its associated `value` or nil if key is not found in tree.
@ -123,32 +104,22 @@ func (tree *AVLTree) SetIfNotExistFuncLock(key any, f func() any) bool {
 // Note that, the `nil` value from Get function cannot be used to determine key existence, please use Contains function
 // to do so.
 func (tree *AVLTree) Get(key any) (value any) {
-	value, _ = tree.Search(key)
-	return
+	tree.lazyInit()
+	return tree.AVLKVTree.Get(key)
 }

 // GetOrSet returns its `value` of `key`, or sets value with given `value` if it does not exist and then returns
 // this value.
 func (tree *AVLTree) GetOrSet(key any, value any) any {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if v, ok := tree.doGet(key); !ok {
-		return tree.doSet(key, value)
-	} else {
-		return v
-	}
+	tree.lazyInit()
+	return tree.AVLKVTree.GetOrSet(key, value)
 }

 // GetOrSetFunc returns its `value` of `key`, or sets value with returned value of callback function `f` if it does not
 // exist and then returns this value.
 func (tree *AVLTree) GetOrSetFunc(key any, f func() any) any {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if v, ok := tree.doGet(key); !ok {
-		return tree.doSet(key, f())
-	} else {
-		return v
-	}
+	tree.lazyInit()
+	return tree.AVLKVTree.GetOrSetFunc(key, f)
 }

 // GetOrSetFuncLock returns its `value` of `key`, or sets value with returned value of callback function `f` if it does
@ -156,13 +127,8 @@ func (tree *AVLTree) GetOrSetFunc(key any, f func() any) any {
 //
 // GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f` within mutex lock.
 func (tree *AVLTree) GetOrSetFuncLock(key any, f func() any) any {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if v, ok := tree.doGet(key); !ok {
-		return tree.doSet(key, f)
-	} else {
-		return v
-	}
+	tree.lazyInit()
+	return tree.AVLKVTree.GetOrSetFuncLock(key, f)
 }

 // GetVar returns a gvar.Var with the value by given `key`.
@ -170,7 +136,8 @@ func (tree *AVLTree) GetOrSetFuncLock(key any, f func() any) any {
 //
 // Also see function Get.
 func (tree *AVLTree) GetVar(key any) *gvar.Var {
-	return gvar.New(tree.Get(key))
+	tree.lazyInit()
+	return tree.AVLKVTree.GetVar(key)
 }

 // GetVarOrSet returns a gvar.Var with result from GetVarOrSet.
@ -178,7 +145,8 @@ func (tree *AVLTree) GetVar(key any) *gvar.Var {
 //
 // Also see function GetOrSet.
 func (tree *AVLTree) GetVarOrSet(key any, value any) *gvar.Var {
-	return gvar.New(tree.GetOrSet(key, value))
+	tree.lazyInit()
+	return tree.AVLKVTree.GetVarOrSet(key, value)
 }

 // GetVarOrSetFunc returns a gvar.Var with result from GetOrSetFunc.
@ -186,7 +154,8 @@ func (tree *AVLTree) GetVarOrSet(key any, value any) *gvar.Var {
 //
 // Also see function GetOrSetFunc.
 func (tree *AVLTree) GetVarOrSetFunc(key any, f func() any) *gvar.Var {
-	return gvar.New(tree.GetOrSetFunc(key, f))
+	tree.lazyInit()
+	return tree.AVLKVTree.GetVarOrSetFunc(key, f)
 }

 // GetVarOrSetFuncLock returns a gvar.Var with result from GetOrSetFuncLock.
@ -194,127 +163,100 @@ func (tree *AVLTree) GetVarOrSetFunc(key any, f func() any) *gvar.Var {
 //
 // Also see function GetOrSetFuncLock.
 func (tree *AVLTree) GetVarOrSetFuncLock(key any, f func() any) *gvar.Var {
-	return gvar.New(tree.GetOrSetFuncLock(key, f))
+	tree.lazyInit()
+	return tree.AVLKVTree.GetVarOrSetFuncLock(key, f)
 }

 // Search searches the tree with given `key`.
 // Second return parameter `found` is true if key was found, otherwise false.
 func (tree *AVLTree) Search(key any) (value any, found bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	if node, found := tree.doGet(key); found {
-		return node, true
-	}
-	return nil, false
+	tree.lazyInit()
+	return tree.AVLKVTree.Search(key)
 }

 // Contains checks and returns whether given `key` exists in the tree.
 func (tree *AVLTree) Contains(key any) bool {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	_, ok := tree.doGet(key)
-	return ok
+	tree.lazyInit()
+	return tree.AVLKVTree.Contains(key)
 }

 // Size returns number of nodes in the tree.
 func (tree *AVLTree) Size() int {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Size()
+	tree.lazyInit()
+	return tree.AVLKVTree.Size()
 }

 // IsEmpty returns true if the tree does not contain any nodes.
 func (tree *AVLTree) IsEmpty() bool {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Size() == 0
+	tree.lazyInit()
+	return tree.AVLKVTree.IsEmpty()
 }

 // Remove removes the node from the tree by `key`, and returns its associated value of `key`.
 // The given `key` should adhere to the comparator's type assertion, otherwise method panics.
 func (tree *AVLTree) Remove(key any) (value any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	return tree.doRemove(key)
+	tree.lazyInit()
+	return tree.AVLKVTree.Remove(key)
 }

 // Removes batch deletes key-value pairs from the tree by `keys`.
 func (tree *AVLTree) Removes(keys []any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	for _, key := range keys {
-		tree.doRemove(key)
-	}
+	tree.lazyInit()
+	tree.AVLKVTree.Removes(keys)
 }

 // Clear removes all nodes from the tree.
 func (tree *AVLTree) Clear() {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	tree.tree.Clear()
+	tree.lazyInit()
+	tree.AVLKVTree.Clear()
 }

 // Keys returns all keys from the tree in order by its comparator.
 func (tree *AVLTree) Keys() []any {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Keys()
+	tree.lazyInit()
+	return tree.AVLKVTree.Keys()
 }

 // Values returns all values from the true in order by its comparator based on the key.
 func (tree *AVLTree) Values() []any {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Values()
+	tree.lazyInit()
+	return tree.AVLKVTree.Values()
 }

 // Replace clears the data of the tree and sets the nodes by given `data`.
 func (tree *AVLTree) Replace(data map[any]any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	tree.tree.Clear()
-	for k, v := range data {
-		tree.doSet(k, v)
-	}
+	tree.lazyInit()
+	tree.AVLKVTree.Replace(data)
 }

 // Print prints the tree to stdout.
 func (tree *AVLTree) Print() {
-	fmt.Println(tree.String())
+	tree.lazyInit()
+	tree.AVLKVTree.Print()
 }

 // String returns a string representation of container.
 func (tree *AVLTree) String() string {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return gstr.Replace(tree.tree.String(), "AVLTree\n", "")
+	tree.lazyInit()
+	return tree.AVLKVTree.String()
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (tree *AVLTree) MarshalJSON() (jsonBytes []byte, err error) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.MarshalJSON()
+	tree.lazyInit()
+	return tree.AVLKVTree.MarshalJSON()
 }

 // Map returns all key-value pairs as map.
 func (tree *AVLTree) Map() map[any]any {
-	m := make(map[any]any, tree.Size())
-	tree.IteratorAsc(func(key, value any) bool {
-		m[key] = value
-		return true
-	})
-	return m
+	tree.lazyInit()
+	return tree.AVLKVTree.Map()
 }

 // MapStrAny returns all key-value items as map[string]any.
 func (tree *AVLTree) MapStrAny() map[string]any {
-	m := make(map[string]any, tree.Size())
-	tree.IteratorAsc(func(key, value any) bool {
-		m[gconv.String(key)] = value
-		return true
-	})
-	return m
+	tree.lazyInit()
+	return tree.AVLKVTree.MapStrAny()
 }

 // Iterator is alias of IteratorAsc.
@ -334,18 +276,8 @@ func (tree *AVLTree) IteratorFrom(key any, match bool, f func(key, value any) bo
 // IteratorAsc iterates the tree readonly in ascending order with given callback function `f`.
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *AVLTree) IteratorAsc(f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var (
-		ok bool
-		it = tree.tree.Iterator()
-	)
-	for it.Begin(); it.Next(); {
-		index, value := it.Key(), it.Value()
-		if ok = f(index, value); !ok {
-			break
-		}
-	}
+	tree.lazyInit()
+	tree.AVLKVTree.IteratorAsc(f)
 }

 // IteratorAscFrom iterates the tree readonly in ascending order with given callback function `f`.
@ -355,34 +287,16 @@ func (tree *AVLTree) IteratorAsc(f func(key, value any) bool) {
 // searching iterating.
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *AVLTree) IteratorAscFrom(key any, match bool, f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var keys = tree.tree.Keys()
-	index, canIterator := iteratorFromGetIndex(key, keys, match)
-	if !canIterator {
-		return
-	}
-	for ; index < len(keys); index++ {
-		f(keys[index], tree.Get(keys[index]))
-	}
+	tree.lazyInit()
+	tree.AVLKVTree.IteratorAscFrom(key, match, f)
 }

 // IteratorDesc iterates the tree readonly in descending order with given callback function `f`.
 //
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *AVLTree) IteratorDesc(f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var (
-		ok bool
-		it = tree.tree.Iterator()
-	)
-	for it.End(); it.Prev(); {
-		index, value := it.Key(), it.Value()
-		if ok = f(index, value); !ok {
-			break
-		}
-	}
+	tree.lazyInit()
+	tree.AVLKVTree.IteratorDesc(f)
 }

 // IteratorDescFrom iterates the tree readonly in descending order with given callback function `f`.
@ -392,44 +306,20 @@ func (tree *AVLTree) IteratorDesc(f func(key, value any) bool) {
 // searching iterating.
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *AVLTree) IteratorDescFrom(key any, match bool, f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var keys = tree.tree.Keys()
-	index, canIterator := iteratorFromGetIndex(key, keys, match)
-	if !canIterator {
-		return
-	}
-	for ; index >= 0; index-- {
-		f(keys[index], tree.Get(keys[index]))
-	}
+	tree.lazyInit()
+	tree.AVLKVTree.IteratorDescFrom(key, match, f)
 }

 // Left returns the minimum element corresponding to the comparator of the tree or nil if the tree is empty.
 func (tree *AVLTree) Left() *AVLTreeNode {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	node := tree.tree.Left()
-	if node == nil {
-		return nil
-	}
-	return &AVLTreeNode{
-		Key:   node.Key,
-		Value: node.Value,
-	}
+	tree.lazyInit()
+	return tree.AVLKVTree.Left()
 }

 // Right returns the maximum element corresponding to the comparator of the tree or nil if the tree is empty.
 func (tree *AVLTree) Right() *AVLTreeNode {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	node := tree.tree.Right()
-	if node == nil {
-		return nil
-	}
-	return &AVLTreeNode{
-		Key:   node.Key,
-		Value: node.Value,
-	}
+	tree.lazyInit()
+	return tree.AVLKVTree.Right()
 }

 // Floor Finds floor node of the input key, returns the floor node or nil if no floor node is found.
@ -441,16 +331,8 @@ func (tree *AVLTree) Right() *AVLTreeNode {
 //
 // Key should adhere to the comparator's type assertion, otherwise method panics.
 func (tree *AVLTree) Floor(key any) (floor *AVLTreeNode, found bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	node, ok := tree.tree.Floor(key)
-	if !ok {
-		return nil, false
-	}
-	return &AVLTreeNode{
-		Key:   node.Key,
-		Value: node.Value,
-	}, true
+	tree.lazyInit()
+	return tree.AVLKVTree.Floor(key)
 }

 // Ceiling finds ceiling node of the input key, returns the ceiling node or nil if no ceiling node is found.
@ -462,16 +344,8 @@ func (tree *AVLTree) Floor(key any) (floor *AVLTreeNode, found bool) {
 //
 // Key should adhere to the comparator's type assertion, otherwise method panics.
 func (tree *AVLTree) Ceiling(key any) (ceiling *AVLTreeNode, found bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	node, ok := tree.tree.Ceiling(key)
-	if !ok {
-		return nil, false
-	}
-	return &AVLTreeNode{
-		Key:   node.Key,
-		Value: node.Value,
-	}, true
+	tree.lazyInit()
+	return tree.AVLKVTree.Ceiling(key)
 }

 // Flip exchanges key-value of the tree to value-key.
@ -480,6 +354,8 @@ func (tree *AVLTree) Ceiling(key any) (ceiling *AVLTreeNode, found bool) {
 //
 // If the type of value is different with key, you pass the new `comparator`.
 func (tree *AVLTree) Flip(comparator ...func(v1, v2 any) int) {
+	tree.lazyInit()
+
 	var t = new(AVLTree)
 	if len(comparator) > 0 {
 		t = NewAVLTree(comparator[0], tree.mu.IsSafe())
@ -493,32 +369,3 @@ func (tree *AVLTree) Flip(comparator ...func(v1, v2 any) int) {
 	tree.Clear()
 	tree.Sets(t.Map())
 }
-
-// doSet inserts key-value pair node into the tree without lock.
-// If `key` already exists, then its value is updated with the new value.
-// If `value` is type of <func() any>, it will be executed and its return value will be set to the map with `key`.
-//
-// It returns value with given `key`.
-func (tree *AVLTree) doSet(key, value any) any {
-	if f, ok := value.(func() any); ok {
-		value = f()
-	}
-	if value == nil {
-		return value
-	}
-	tree.tree.Put(key, value)
-	return value
-}
-
-// doGet retrieves and returns the value of given key from tree without lock.
-func (tree *AVLTree) doGet(key any) (value any, found bool) {
-	return tree.tree.Get(key)
-}
-
-// doRemove removes key from tree and returns its associated value without lock.
-// Note that, the given `key` should adhere to the comparator's type assertion, otherwise method panics.
-func (tree *AVLTree) doRemove(key any) (value any) {
-	value, _ = tree.tree.Get(key)
-	tree.tree.Remove(key)
-	return
-}
--- a/container/gtree/gtree_btree.go
+++ b/container/gtree/gtree_btree.go
@ -7,31 +7,22 @@
 package gtree

 import (
-	"fmt"
-
-	"github.com/emirpasic/gods/trees/btree"
+	"sync"

 	"github.com/gogf/gf/v2/container/gvar"
-	"github.com/gogf/gf/v2/internal/rwmutex"
-	"github.com/gogf/gf/v2/text/gstr"
-	"github.com/gogf/gf/v2/util/gconv"
+	"github.com/gogf/gf/v2/util/gutil"
 )

 var _ iTree = (*BTree)(nil)

 // BTree holds elements of the B-tree.
 type BTree struct {
-	mu         rwmutex.RWMutex
-	comparator func(v1, v2 any) int
-	m          int // order (maximum number of children)
-	tree       *btree.Tree
+	*BKVTree[any, any]
+	once sync.Once
 }

 // BTreeEntry represents the key-value pair contained within nodes.
-type BTreeEntry struct {
-	Key   any
-	Value any
-}
+type BTreeEntry = BKVTreeEntry[any, any]

 // NewBTree instantiates a B-tree with `m` (maximum number of children) and a custom key comparator.
 // The parameter `safe` is used to specify whether using tree in concurrent-safety,
@ -39,10 +30,7 @@ type BTreeEntry struct {
 // Note that the `m` must be greater or equal than 3, or else it panics.
 func NewBTree(m int, comparator func(v1, v2 any) int, safe ...bool) *BTree {
 	return &BTree{
-		mu:         rwmutex.Create(safe...),
-		m:          m,
-		comparator: comparator,
-		tree:       btree.NewWith(m, comparator),
+		BKVTree: NewBKVTree[any, any](m, comparator, safe...),
 	}
 }

@ -50,58 +38,55 @@ func NewBTree(m int, comparator func(v1, v2 any) int, safe ...bool) *BTree {
 // The parameter `safe` is used to specify whether using tree in concurrent-safety,
 // which is false in default.
 func NewBTreeFrom(m int, comparator func(v1, v2 any) int, data map[any]any, safe ...bool) *BTree {
-	tree := NewBTree(m, comparator, safe...)
-	for k, v := range data {
-		tree.doSet(k, v)
+	return &BTree{
+		BKVTree: NewBKVTreeFrom(m, comparator, data, safe...),
 	}
-	return tree
+}
+
+// lazyInit lazily initializes the tree.
+func (tree *BTree) lazyInit() {
+	tree.once.Do(func() {
+		if tree.BKVTree == nil {
+			tree.BKVTree = NewBKVTree[any, any](3, gutil.ComparatorTStr, false)
+		}
+	})
 }

 // Clone clones and returns a new tree from current tree.
 func (tree *BTree) Clone() *BTree {
-	newTree := NewBTree(tree.m, tree.comparator, tree.mu.IsSafe())
-	newTree.Sets(tree.Map())
-	return newTree
+	if tree == nil {
+		return nil
+	}
+	tree.lazyInit()
+	return &BTree{
+		BKVTree: tree.BKVTree.Clone(),
+	}
 }

 // Set sets key-value pair into the tree.
 func (tree *BTree) Set(key any, value any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	tree.doSet(key, value)
+	tree.lazyInit()
+	tree.BKVTree.Set(key, value)
 }

 // Sets batch sets key-values to the tree.
 func (tree *BTree) Sets(data map[any]any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	for k, v := range data {
-		tree.doSet(k, v)
-	}
+	tree.lazyInit()
+	tree.BKVTree.Sets(data)
 }

 // SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
 // It returns false if `key` exists, and such setting key-value pair operation would be ignored.
 func (tree *BTree) SetIfNotExist(key any, value any) bool {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if _, ok := tree.doGet(key); !ok {
-		tree.doSet(key, value)
-		return true
-	}
-	return false
+	tree.lazyInit()
+	return tree.BKVTree.SetIfNotExist(key, value)
 }

 // SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
 // It returns false if `key` exists, and such setting key-value pair operation would be ignored.
 func (tree *BTree) SetIfNotExistFunc(key any, f func() any) bool {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if _, ok := tree.doGet(key); !ok {
-		tree.doSet(key, f())
-		return true
-	}
-	return false
+	tree.lazyInit()
+	return tree.BKVTree.SetIfNotExistFunc(key, f)
 }

 // SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
@ -110,13 +95,8 @@ func (tree *BTree) SetIfNotExistFunc(key any, f func() any) bool {
 // SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
 // it executes function `f` within mutex lock.
 func (tree *BTree) SetIfNotExistFuncLock(key any, f func() any) bool {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if _, ok := tree.doGet(key); !ok {
-		tree.doSet(key, f)
-		return true
-	}
-	return false
+	tree.lazyInit()
+	return tree.BKVTree.SetIfNotExistFuncLock(key, f)
 }

 // Get searches the `key` in the tree and returns its associated `value` or nil if key is not found in tree.
@ -124,34 +104,22 @@ func (tree *BTree) SetIfNotExistFuncLock(key any, f func() any) bool {
 // Note that, the `nil` value from Get function cannot be used to determine key existence, please use Contains function
 // to do so.
 func (tree *BTree) Get(key any) (value any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	value, _ = tree.doGet(key)
-	return
+	tree.lazyInit()
+	return tree.BKVTree.Get(key)
 }

 // GetOrSet returns its `value` of `key`, or sets value with given `value` if it does not exist and then returns
 // this value.
 func (tree *BTree) GetOrSet(key any, value any) any {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if v, ok := tree.doGet(key); !ok {
-		return tree.doSet(key, value)
-	} else {
-		return v
-	}
+	tree.lazyInit()
+	return tree.BKVTree.GetOrSet(key, value)
 }

 // GetOrSetFunc returns its `value` of `key`, or sets value with returned value of callback function `f` if it does not
 // exist and then returns this value.
 func (tree *BTree) GetOrSetFunc(key any, f func() any) any {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if v, ok := tree.doGet(key); !ok {
-		return tree.doSet(key, f())
-	} else {
-		return v
-	}
+	tree.lazyInit()
+	return tree.BKVTree.GetOrSetFunc(key, f)
 }

 // GetOrSetFuncLock returns its `value` of `key`, or sets value with returned value of callback function `f` if it does
@ -159,13 +127,8 @@ func (tree *BTree) GetOrSetFunc(key any, f func() any) any {
 //
 // GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f` within mutex lock.
 func (tree *BTree) GetOrSetFuncLock(key any, f func() any) any {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if v, ok := tree.doGet(key); !ok {
-		return tree.doSet(key, f)
-	} else {
-		return v
-	}
+	tree.lazyInit()
+	return tree.BKVTree.GetOrSetFuncLock(key, f)
 }

 // GetVar returns a gvar.Var with the value by given `key`.
@ -173,7 +136,8 @@ func (tree *BTree) GetOrSetFuncLock(key any, f func() any) any {
 //
 // Also see function Get.
 func (tree *BTree) GetVar(key any) *gvar.Var {
-	return gvar.New(tree.Get(key))
+	tree.lazyInit()
+	return tree.BKVTree.GetVar(key)
 }

 // GetVarOrSet returns a gvar.Var with result from GetVarOrSet.
@ -181,7 +145,8 @@ func (tree *BTree) GetVar(key any) *gvar.Var {
 //
 // Also see function GetOrSet.
 func (tree *BTree) GetVarOrSet(key any, value any) *gvar.Var {
-	return gvar.New(tree.GetOrSet(key, value))
+	tree.lazyInit()
+	return tree.BKVTree.GetVarOrSet(key, value)
 }

 // GetVarOrSetFunc returns a gvar.Var with result from GetOrSetFunc.
@ -189,7 +154,8 @@ func (tree *BTree) GetVarOrSet(key any, value any) *gvar.Var {
 //
 // Also see function GetOrSetFunc.
 func (tree *BTree) GetVarOrSetFunc(key any, f func() any) *gvar.Var {
-	return gvar.New(tree.GetOrSetFunc(key, f))
+	tree.lazyInit()
+	return tree.BKVTree.GetVarOrSetFunc(key, f)
 }

 // GetVarOrSetFuncLock returns a gvar.Var with result from GetOrSetFuncLock.
@ -197,155 +163,123 @@ func (tree *BTree) GetVarOrSetFunc(key any, f func() any) *gvar.Var {
 //
 // Also see function GetOrSetFuncLock.
 func (tree *BTree) GetVarOrSetFuncLock(key any, f func() any) *gvar.Var {
-	return gvar.New(tree.GetOrSetFuncLock(key, f))
+	tree.lazyInit()
+	return tree.BKVTree.GetVarOrSetFuncLock(key, f)
 }

 // Search searches the tree with given `key`.
 // Second return parameter `found` is true if key was found, otherwise false.
 func (tree *BTree) Search(key any) (value any, found bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Get(key)
+	tree.lazyInit()
+	return tree.BKVTree.Search(key)
 }

 // Contains checks and returns whether given `key` exists in the tree.
 func (tree *BTree) Contains(key any) bool {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	_, ok := tree.doGet(key)
-	return ok
+	tree.lazyInit()
+	return tree.BKVTree.Contains(key)
 }

 // Size returns number of nodes in the tree.
 func (tree *BTree) Size() int {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Size()
+	tree.lazyInit()
+	return tree.BKVTree.Size()
 }

 // IsEmpty returns true if tree does not contain any nodes
 func (tree *BTree) IsEmpty() bool {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Size() == 0
+	tree.lazyInit()
+	return tree.BKVTree.IsEmpty()
 }

 // Remove removes the node from the tree by `key`, and returns its associated value of `key`.
 // The given `key` should adhere to the comparator's type assertion, otherwise method panics.
 func (tree *BTree) Remove(key any) (value any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	return tree.doRemove(key)
+	tree.lazyInit()
+	return tree.BKVTree.Remove(key)
 }

 // Removes batch deletes key-value pairs from the tree by `keys`.
 func (tree *BTree) Removes(keys []any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	for _, key := range keys {
-		tree.doRemove(key)
-	}
+	tree.lazyInit()
+	tree.BKVTree.Removes(keys)
 }

 // Clear removes all nodes from the tree.
 func (tree *BTree) Clear() {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	tree.tree.Clear()
+	tree.lazyInit()
+	tree.BKVTree.Clear()
 }

 // Keys returns all keys from the tree in order by its comparator.
 func (tree *BTree) Keys() []any {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Keys()
+	tree.lazyInit()
+	return tree.BKVTree.Keys()
 }

 // Values returns all values from the true in order by its comparator based on the key.
 func (tree *BTree) Values() []any {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Values()
+	tree.lazyInit()
+	return tree.BKVTree.Values()
 }

 // Replace clears the data of the tree and sets the nodes by given `data`.
 func (tree *BTree) Replace(data map[any]any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	tree.tree.Clear()
-	for k, v := range data {
-		tree.doSet(k, v)
-	}
+	tree.lazyInit()
+	tree.BKVTree.Replace(data)
 }

 // Map returns all key-value pairs as map.
 func (tree *BTree) Map() map[any]any {
-	m := make(map[any]any, tree.Size())
-	tree.IteratorAsc(func(key, value any) bool {
-		m[key] = value
-		return true
-	})
-	return m
+	tree.lazyInit()
+	return tree.BKVTree.Map()
 }

 // MapStrAny returns all key-value items as map[string]any.
 func (tree *BTree) MapStrAny() map[string]any {
-	m := make(map[string]any, tree.Size())
-	tree.IteratorAsc(func(key, value any) bool {
-		m[gconv.String(key)] = value
-		return true
-	})
-	return m
+	tree.lazyInit()
+	return tree.BKVTree.MapStrAny()
 }

 // Print prints the tree to stdout.
 func (tree *BTree) Print() {
-	fmt.Println(tree.String())
+	tree.lazyInit()
+	tree.BKVTree.Print()
 }

 // String returns a string representation of container (for debugging purposes)
 func (tree *BTree) String() string {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return gstr.Replace(tree.tree.String(), "BTree\n", "")
+	tree.lazyInit()
+	return tree.BKVTree.String()
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
 func (tree *BTree) MarshalJSON() (jsonBytes []byte, err error) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.MarshalJSON()
+	tree.lazyInit()
+	return tree.BKVTree.MarshalJSON()
 }

 // Iterator is alias of IteratorAsc.
 //
 // Also see IteratorAsc.
 func (tree *BTree) Iterator(f func(key, value any) bool) {
-	tree.IteratorAsc(f)
+	tree.lazyInit()
+	tree.BKVTree.Iterator(f)
 }

 // IteratorFrom is alias of IteratorAscFrom.
 //
 // Also see IteratorAscFrom.
 func (tree *BTree) IteratorFrom(key any, match bool, f func(key, value any) bool) {
-	tree.IteratorAscFrom(key, match, f)
+	tree.lazyInit()
+	tree.BKVTree.IteratorFrom(key, match, f)
 }

 // IteratorAsc iterates the tree readonly in ascending order with given callback function `f`.
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *BTree) IteratorAsc(f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var (
-		ok bool
-		it = tree.tree.Iterator()
-	)
-	for it.Begin(); it.Next(); {
-		index, value := it.Key(), it.Value()
-		if ok = f(index, value); !ok {
-			break
-		}
-	}
+	tree.lazyInit()
+	tree.BKVTree.IteratorAsc(f)
 }

 // IteratorAscFrom iterates the tree readonly in ascending order with given callback function `f`.
@ -355,34 +289,16 @@ func (tree *BTree) IteratorAsc(f func(key, value any) bool) {
 // searching iterating.
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *BTree) IteratorAscFrom(key any, match bool, f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var keys = tree.tree.Keys()
-	index, canIterator := iteratorFromGetIndex(key, keys, match)
-	if !canIterator {
-		return
-	}
-	for ; index < len(keys); index++ {
-		f(keys[index], tree.Get(keys[index]))
-	}
+	tree.lazyInit()
+	tree.BKVTree.IteratorAscFrom(key, match, f)
 }

 // IteratorDesc iterates the tree readonly in descending order with given callback function `f`.
 //
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *BTree) IteratorDesc(f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var (
-		ok bool
-		it = tree.tree.Iterator()
-	)
-	for it.End(); it.Prev(); {
-		index, value := it.Key(), it.Value()
-		if ok = f(index, value); !ok {
-			break
-		}
-	}
+	tree.lazyInit()
+	tree.BKVTree.IteratorDesc(f)
 }

 // IteratorDescFrom iterates the tree readonly in descending order with given callback function `f`.
@ -392,78 +308,24 @@ func (tree *BTree) IteratorDesc(f func(key, value any) bool) {
 // searching iterating.
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *BTree) IteratorDescFrom(key any, match bool, f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var keys = tree.tree.Keys()
-	index, canIterator := iteratorFromGetIndex(key, keys, match)
-	if !canIterator {
-		return
-	}
-	for ; index >= 0; index-- {
-		f(keys[index], tree.Get(keys[index]))
-	}
+	tree.lazyInit()
+	tree.BKVTree.IteratorDescFrom(key, match, f)
 }

 // Height returns the height of the tree.
 func (tree *BTree) Height() int {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Height()
+	tree.lazyInit()
+	return tree.BKVTree.Height()
 }

 // Left returns the minimum element corresponding to the comparator of the tree or nil if the tree is empty.
 func (tree *BTree) Left() *BTreeEntry {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	node := tree.tree.Left()
-	if node == nil || node.Entries == nil || len(node.Entries) == 0 {
-		return nil
-	}
-	return &BTreeEntry{
-		Key:   node.Entries[0].Key,
-		Value: node.Entries[0].Value,
-	}
+	tree.lazyInit()
+	return tree.BKVTree.Left()
 }

 // Right returns the maximum element corresponding to the comparator of the tree or nil if the tree is empty.
 func (tree *BTree) Right() *BTreeEntry {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	node := tree.tree.Right()
-	if node == nil || node.Entries == nil || len(node.Entries) == 0 {
-		return nil
-	}
-	return &BTreeEntry{
-		Key:   node.Entries[len(node.Entries)-1].Key,
-		Value: node.Entries[len(node.Entries)-1].Value,
-	}
-}
-
-// doSet inserts key-value pair node into the tree without lock.
-// If `key` already exists, then its value is updated with the new value.
-// If `value` is type of <func() any>, it will be executed and its return value will be set to the map with `key`.
-//
-// It returns value with given `key`.
-func (tree *BTree) doSet(key any, value any) any {
-	if f, ok := value.(func() any); ok {
-		value = f()
-	}
-	if value == nil {
-		return value
-	}
-	tree.tree.Put(key, value)
-	return value
-}
-
-// doGet get the value from the tree by key without lock.
-func (tree *BTree) doGet(key any) (value any, ok bool) {
-	return tree.tree.Get(key)
-}
-
-// doRemove removes key from tree and returns its associated value without lock.
-// Note that, the given `key` should adhere to the comparator's type assertion, otherwise method panics.
-func (tree *BTree) doRemove(key any) (value any) {
-	value, _ = tree.tree.Get(key)
-	tree.tree.Remove(key)
-	return
+	tree.lazyInit()
+	return tree.BKVTree.Right()
 }
--- a/container/gtree/gtree_k_v_avltree.go
+++ b/container/gtree/gtree_k_v_avltree.go
@ -0,0 +1,539 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package gtree
+
+import (
+	"fmt"
+
+	"github.com/emirpasic/gods/v2/trees/avltree"
+
+	"github.com/gogf/gf/v2/container/gvar"
+	"github.com/gogf/gf/v2/internal/json"
+	"github.com/gogf/gf/v2/internal/rwmutex"
+	"github.com/gogf/gf/v2/text/gstr"
+	"github.com/gogf/gf/v2/util/gconv"
+)
+
+// AVLKVTree holds elements of the AVL tree.
+type AVLKVTree[K comparable, V any] struct {
+	mu         rwmutex.RWMutex
+	comparator func(v1, v2 K) int
+	tree       *avltree.Tree[K, V]
+}
+
+// AVLKVTreeNode is a single element within the tree.
+type AVLKVTreeNode[K comparable, V any] struct {
+	Key   K
+	Value V
+}
+
+// NewAVLKVTree instantiates an AVL tree with the custom key comparator.
+//
+// The parameter `safe` is used to specify whether using tree in concurrent-safety,
+// which is false in default.
+func NewAVLKVTree[K comparable, V any](comparator func(v1, v2 K) int, safe ...bool) *AVLKVTree[K, V] {
+	return &AVLKVTree[K, V]{
+		mu:         rwmutex.Create(safe...),
+		comparator: comparator,
+		tree:       avltree.NewWith[K, V](comparator),
+	}
+}
+
+// NewAVLKVTreeFrom instantiates an AVL tree with the custom key comparator and data map.
+//
+// The parameter `safe` is used to specify whether using tree in concurrent-safety, which is false in default.
+func NewAVLKVTreeFrom[K comparable, V any](comparator func(v1, v2 K) int, data map[K]V, safe ...bool) *AVLKVTree[K, V] {
+	tree := NewAVLKVTree[K, V](comparator, safe...)
+	for k, v := range data {
+		tree.doSet(k, v)
+	}
+	return tree
+}
+
+// Clone clones and returns a new tree from current tree.
+func (tree *AVLKVTree[K, V]) Clone() *AVLKVTree[K, V] {
+	if tree == nil {
+		return nil
+	}
+	newTree := NewAVLKVTree[K, V](tree.comparator, tree.mu.IsSafe())
+	newTree.Sets(tree.Map())
+	return newTree
+}
+
+// Set sets key-value pair into the tree.
+func (tree *AVLKVTree[K, V]) Set(key K, value V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	tree.doSet(key, value)
+}
+
+// Sets batch sets key-values to the tree.
+func (tree *AVLKVTree[K, V]) Sets(data map[K]V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	for key, value := range data {
+		tree.doSet(key, value)
+	}
+}
+
+// SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
+// It returns false if `key` exists, and such setting key-value pair operation would be ignored.
+func (tree *AVLKVTree[K, V]) SetIfNotExist(key K, value V) bool {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if _, ok := tree.doGet(key); !ok {
+		tree.doSet(key, value)
+		return true
+	}
+	return false
+}
+
+// SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
+// It returns false if `key` exists, and such setting key-value pair operation would be ignored.
+func (tree *AVLKVTree[K, V]) SetIfNotExistFunc(key K, f func() V) bool {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if _, ok := tree.doGet(key); !ok {
+		tree.doSet(key, f())
+		return true
+	}
+	return false
+}
+
+// SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
+// It returns false if `key` exists, and such setting key-value pair operation would be ignored.
+//
+// SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
+// it executes function `f` within mutex lock.
+func (tree *AVLKVTree[K, V]) SetIfNotExistFuncLock(key K, f func() V) bool {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if _, ok := tree.doGet(key); !ok {
+		tree.doSet(key, f())
+		return true
+	}
+	return false
+}
+
+// Get searches the `key` in the tree and returns its associated `value` or nil if key is not found in tree.
+//
+// Note that, the `nil` value from Get function cannot be used to determine key existence, please use Contains function
+// to do so.
+func (tree *AVLKVTree[K, V]) Get(key K) (value V) {
+	value, _ = tree.Search(key)
+	return
+}
+
+// GetOrSet returns its `value` of `key`, or sets value with given `value` if it does not exist and then returns
+// this value.
+func (tree *AVLKVTree[K, V]) GetOrSet(key K, value V) V {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if v, ok := tree.doGet(key); !ok {
+		return tree.doSet(key, value)
+	} else {
+		return v
+	}
+}
+
+// GetOrSetFunc returns its `value` of `key`, or sets value with returned value of callback function `f` if it does not
+// exist and then returns this value.
+func (tree *AVLKVTree[K, V]) GetOrSetFunc(key K, f func() V) V {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if v, ok := tree.doGet(key); !ok {
+		return tree.doSet(key, f())
+	} else {
+		return v
+	}
+}
+
+// GetOrSetFuncLock returns its `value` of `key`, or sets value with returned value of callback function `f` if it does
+// not exist and then returns this value.
+//
+// GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f` within mutex lock.
+func (tree *AVLKVTree[K, V]) GetOrSetFuncLock(key K, f func() V) V {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if v, ok := tree.doGet(key); !ok {
+		return tree.doSet(key, f())
+	} else {
+		return v
+	}
+}
+
+// GetVar returns a gvar.Var with the value by given `key`.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function Get.
+func (tree *AVLKVTree[K, V]) GetVar(key K) *gvar.Var {
+	return gvar.New(tree.Get(key))
+}
+
+// GetVarOrSet returns a gvar.Var with result from GetVarOrSet.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function GetOrSet.
+func (tree *AVLKVTree[K, V]) GetVarOrSet(key K, value V) *gvar.Var {
+	return gvar.New(tree.GetOrSet(key, value))
+}
+
+// GetVarOrSetFunc returns a gvar.Var with result from GetOrSetFunc.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function GetOrSetFunc.
+func (tree *AVLKVTree[K, V]) GetVarOrSetFunc(key K, f func() V) *gvar.Var {
+	return gvar.New(tree.GetOrSetFunc(key, f))
+}
+
+// GetVarOrSetFuncLock returns a gvar.Var with result from GetOrSetFuncLock.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function GetOrSetFuncLock.
+func (tree *AVLKVTree[K, V]) GetVarOrSetFuncLock(key K, f func() V) *gvar.Var {
+	return gvar.New(tree.GetOrSetFuncLock(key, f))
+}
+
+// Search searches the tree with given `key`.
+// Second return parameter `found` is true if key was found, otherwise false.
+func (tree *AVLKVTree[K, V]) Search(key K) (value V, found bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	if node, found := tree.doGet(key); found {
+		return node, true
+	}
+	found = false
+	return
+}
+
+// Contains checks and returns whether given `key` exists in the tree.
+func (tree *AVLKVTree[K, V]) Contains(key K) bool {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	_, ok := tree.doGet(key)
+	return ok
+}
+
+// Size returns number of nodes in the tree.
+func (tree *AVLKVTree[K, V]) Size() int {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Size()
+}
+
+// IsEmpty returns true if the tree does not contain any nodes.
+func (tree *AVLKVTree[K, V]) IsEmpty() bool {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Size() == 0
+}
+
+// Remove removes the node from the tree by `key`, and returns its associated value of `key`.
+// The given `key` should adhere to the comparator's type assertion, otherwise method panics.
+func (tree *AVLKVTree[K, V]) Remove(key K) (value V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	return tree.doRemove(key)
+}
+
+// Removes batch deletes key-value pairs from the tree by `keys`.
+func (tree *AVLKVTree[K, V]) Removes(keys []K) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	for _, key := range keys {
+		tree.doRemove(key)
+	}
+}
+
+// Clear removes all nodes from the tree.
+func (tree *AVLKVTree[K, V]) Clear() {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	tree.tree.Clear()
+}
+
+// Keys returns all keys from the tree in order by its comparator.
+func (tree *AVLKVTree[K, V]) Keys() []K {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Keys()
+}
+
+// Values returns all values from the true in order by its comparator based on the key.
+func (tree *AVLKVTree[K, V]) Values() []V {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Values()
+}
+
+// Replace clears the data of the tree and sets the nodes by given `data`.
+func (tree *AVLKVTree[K, V]) Replace(data map[K]V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	tree.tree.Clear()
+	for k, v := range data {
+		tree.doSet(k, v)
+	}
+}
+
+// Print prints the tree to stdout.
+func (tree *AVLKVTree[K, V]) Print() {
+	fmt.Println(tree.String())
+}
+
+// String returns a string representation of container.
+func (tree *AVLKVTree[K, V]) String() string {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return gstr.Replace(tree.tree.String(), "AVLTree\n", "")
+}
+
+// MarshalJSON implements the interface MarshalJSON for json.Marshal.
+func (tree *AVLKVTree[K, V]) MarshalJSON() (jsonBytes []byte, err error) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+
+	elements := make(map[string]V)
+	it := tree.tree.Iterator()
+	for it.Next() {
+		elements[gconv.String(it.Key())] = it.Value()
+	}
+	return json.Marshal(&elements)
+}
+
+// Map returns all key-value pairs as map.
+func (tree *AVLKVTree[K, V]) Map() map[K]V {
+	m := make(map[K]V, tree.Size())
+	tree.IteratorAsc(func(key K, value V) bool {
+		m[key] = value
+		return true
+	})
+	return m
+}
+
+// MapStrAny returns all key-value items as map[string]any.
+func (tree *AVLKVTree[K, V]) MapStrAny() map[string]any {
+	m := make(map[string]any, tree.Size())
+	tree.IteratorAsc(func(key K, value V) bool {
+		m[gconv.String(key)] = value
+		return true
+	})
+	return m
+}
+
+// Iterator is alias of IteratorAsc.
+//
+// Also see IteratorAsc.
+func (tree *AVLKVTree[K, V]) Iterator(f func(key K, value V) bool) {
+	tree.IteratorAsc(f)
+}
+
+// IteratorFrom is alias of IteratorAscFrom.
+//
+// Also see IteratorAscFrom.
+func (tree *AVLKVTree[K, V]) IteratorFrom(key K, match bool, f func(key K, value V) bool) {
+	tree.IteratorAscFrom(key, match, f)
+}
+
+// IteratorAsc iterates the tree readonly in ascending order with given callback function `f`.
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *AVLKVTree[K, V]) IteratorAsc(f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var (
+		ok bool
+		it = tree.tree.Iterator()
+	)
+	for it.Begin(); it.Next(); {
+		index, value := it.Key(), it.Value()
+		if ok = f(index, value); !ok {
+			break
+		}
+	}
+}
+
+// IteratorAscFrom iterates the tree readonly in ascending order with given callback function `f`.
+//
+// The parameter `key` specifies the start entry for iterating.
+// The parameter `match` specifies whether starting iterating only if the `key` is fully matched, or else using index
+// searching iterating.
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *AVLKVTree[K, V]) IteratorAscFrom(key K, match bool, f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var keys = tree.tree.Keys()
+	index, canIterator := iteratorFromGetIndexT(key, keys, match)
+	if !canIterator {
+		return
+	}
+	for ; index < len(keys); index++ {
+		f(keys[index], tree.Get(keys[index]))
+	}
+}
+
+// IteratorDesc iterates the tree readonly in descending order with given callback function `f`.
+//
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *AVLKVTree[K, V]) IteratorDesc(f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var (
+		ok bool
+		it = tree.tree.Iterator()
+	)
+	for it.End(); it.Prev(); {
+		index, value := it.Key(), it.Value()
+		if ok = f(index, value); !ok {
+			break
+		}
+	}
+}
+
+// IteratorDescFrom iterates the tree readonly in descending order with given callback function `f`.
+//
+// The parameter `key` specifies the start entry for iterating.
+// The parameter `match` specifies whether starting iterating only if the `key` is fully matched, or else using index
+// searching iterating.
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *AVLKVTree[K, V]) IteratorDescFrom(key K, match bool, f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var keys = tree.tree.Keys()
+	index, canIterator := iteratorFromGetIndexT(key, keys, match)
+	if !canIterator {
+		return
+	}
+	for ; index >= 0; index-- {
+		f(keys[index], tree.Get(keys[index]))
+	}
+}
+
+// Left returns the minimum element corresponding to the comparator of the tree or nil if the tree is empty.
+func (tree *AVLKVTree[K, V]) Left() *AVLKVTreeNode[K, V] {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	node := tree.tree.Left()
+	if node == nil {
+		return nil
+	}
+	return &AVLKVTreeNode[K, V]{
+		Key:   node.Key,
+		Value: node.Value,
+	}
+}
+
+// Right returns the maximum element corresponding to the comparator of the tree or nil if the tree is empty.
+func (tree *AVLKVTree[K, V]) Right() *AVLKVTreeNode[K, V] {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	node := tree.tree.Right()
+	if node == nil {
+		return nil
+	}
+	return &AVLKVTreeNode[K, V]{
+		Key:   node.Key,
+		Value: node.Value,
+	}
+}
+
+// Floor Finds floor node of the input key, returns the floor node or nil if no floor node is found.
+// The second returned parameter `found` is true if floor was found, otherwise false.
+//
+// Floor node is defined as the largest node that is smaller than or equal to the given node.
+// A floor node may not be found, either because the tree is empty, or because
+// all nodes in the tree is larger than the given node.
+//
+// Key should adhere to the comparator's type assertion, otherwise method panics.
+func (tree *AVLKVTree[K, V]) Floor(key K) (floor *AVLKVTreeNode[K, V], found bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	node, ok := tree.tree.Floor(key)
+	if !ok {
+		return nil, false
+	}
+	return &AVLKVTreeNode[K, V]{
+		Key:   node.Key,
+		Value: node.Value,
+	}, true
+}
+
+// Ceiling finds ceiling node of the input key, returns the ceiling node or nil if no ceiling node is found.
+// The second return parameter `found` is true if ceiling was found, otherwise false.
+//
+// Ceiling node is defined as the smallest node that is larger than or equal to the given node.
+// A ceiling node may not be found, either because the tree is empty, or because
+// all nodes in the tree is smaller than the given node.
+//
+// Key should adhere to the comparator's type assertion, otherwise method panics.
+func (tree *AVLKVTree[K, V]) Ceiling(key K) (ceiling *AVLKVTreeNode[K, V], found bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	node, ok := tree.tree.Ceiling(key)
+	if !ok {
+		return nil, false
+	}
+	return &AVLKVTreeNode[K, V]{
+		Key:   node.Key,
+		Value: node.Value,
+	}, true
+}
+
+// Flip exchanges key-value of the tree to value-key.
+// Note that you should guarantee the value is the same type as key,
+// or else the comparator would panic.
+//
+// If the type of value is different with key, you pass the new `comparator`.
+func (tree *AVLKVTree[K, V]) Flip(comparator ...func(v1, v2 K) int) {
+	var t = new(AVLKVTree[K, V])
+	if len(comparator) > 0 {
+		t = NewAVLKVTree[K, V](comparator[0], tree.mu.IsSafe())
+	} else {
+		t = NewAVLKVTree[K, V](tree.comparator, tree.mu.IsSafe())
+	}
+	var (
+		newKey   K
+		newValue V
+	)
+	tree.IteratorAsc(func(key K, value V) bool {
+		if err := gconv.Scan(key, &newValue); err != nil {
+			panic(err)
+		}
+		if err := gconv.Scan(value, &newKey); err != nil {
+			panic(err)
+		}
+		t.doSet(newKey, newValue)
+		return true
+	})
+	tree.Clear()
+	tree.Sets(t.Map())
+}
+
+// doSet inserts key-value pair node into the tree without lock.
+// If `key` already exists, then its value is updated with the new value.
+// If `value` is type of <func() any>, it will be executed and its return value will be set to the map with `key`.
+//
+// It returns value with given `key`.
+func (tree *AVLKVTree[K, V]) doSet(key K, value V) V {
+	if any(value) == nil {
+		return value
+	}
+	tree.tree.Put(key, value)
+	return value
+}
+
+// doGet retrieves and returns the value of given key from tree without lock.
+func (tree *AVLKVTree[K, V]) doGet(key K) (value V, found bool) {
+	return tree.tree.Get(key)
+}
+
+// doRemove removes key from tree and returns its associated value without lock.
+// Note that, the given `key` should adhere to the comparator's type assertion, otherwise method panics.
+func (tree *AVLKVTree[K, V]) doRemove(key K) (value V) {
+	value, _ = tree.tree.Get(key)
+	tree.tree.Remove(key)
+	return
+}
--- a/container/gtree/gtree_k_v_btree.go
+++ b/container/gtree/gtree_k_v_btree.go
@ -0,0 +1,474 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package gtree
+
+import (
+	"fmt"
+
+	"github.com/emirpasic/gods/v2/trees/btree"
+
+	"github.com/gogf/gf/v2/container/gvar"
+	"github.com/gogf/gf/v2/internal/json"
+	"github.com/gogf/gf/v2/internal/rwmutex"
+	"github.com/gogf/gf/v2/text/gstr"
+	"github.com/gogf/gf/v2/util/gconv"
+)
+
+// BKVTree holds elements of the B-tree.
+type BKVTree[K comparable, V any] struct {
+	mu         rwmutex.RWMutex
+	comparator func(v1, v2 K) int
+	m          int // order (maximum number of children)
+	tree       *btree.Tree[K, V]
+}
+
+// BKVTreeEntry represents the key-value pair contained within nodes.
+type BKVTreeEntry[K comparable, V any] struct {
+	Key   K
+	Value V
+}
+
+// NewBKVTree instantiates a B-tree with `m` (maximum number of children) and a custom key comparator.
+// The parameter `safe` is used to specify whether using tree in concurrent-safety,
+// which is false in default.
+// Note that the `m` must be greater or equal than 3, or else it panics.
+func NewBKVTree[K comparable, V any](m int, comparator func(v1, v2 K) int, safe ...bool) *BKVTree[K, V] {
+	return &BKVTree[K, V]{
+		mu:         rwmutex.Create(safe...),
+		m:          m,
+		comparator: comparator,
+		tree:       btree.NewWith[K, V](m, comparator),
+	}
+}
+
+// NewBKVTreeFrom instantiates a B-tree with `m` (maximum number of children), a custom key comparator and data map.
+// The parameter `safe` is used to specify whether using tree in concurrent-safety,
+// which is false in default.
+func NewBKVTreeFrom[K comparable, V any](m int, comparator func(v1, v2 K) int, data map[K]V, safe ...bool) *BKVTree[K, V] {
+	tree := NewBKVTree[K, V](m, comparator, safe...)
+	for k, v := range data {
+		tree.doSet(k, v)
+	}
+	return tree
+}
+
+// Clone clones and returns a new tree from current tree.
+func (tree *BKVTree[K, V]) Clone() *BKVTree[K, V] {
+	if tree == nil {
+		return nil
+	}
+	newTree := NewBKVTree[K, V](tree.m, tree.comparator, tree.mu.IsSafe())
+	newTree.Sets(tree.Map())
+	return newTree
+}
+
+// Set sets key-value pair into the tree.
+func (tree *BKVTree[K, V]) Set(key K, value V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	tree.doSet(key, value)
+}
+
+// Sets batch sets key-values to the tree.
+func (tree *BKVTree[K, V]) Sets(data map[K]V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	for k, v := range data {
+		tree.doSet(k, v)
+	}
+}
+
+// SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
+// It returns false if `key` exists, and such setting key-value pair operation would be ignored.
+func (tree *BKVTree[K, V]) SetIfNotExist(key K, value V) bool {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if _, ok := tree.doGet(key); !ok {
+		tree.doSet(key, value)
+		return true
+	}
+	return false
+}
+
+// SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
+// It returns false if `key` exists, and such setting key-value pair operation would be ignored.
+func (tree *BKVTree[K, V]) SetIfNotExistFunc(key K, f func() V) bool {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if _, ok := tree.doGet(key); !ok {
+		tree.doSet(key, f())
+		return true
+	}
+	return false
+}
+
+// SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
+// It returns false if `key` exists, and such setting key-value pair operation would be ignored.
+//
+// SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
+// it executes function `f` within mutex lock.
+func (tree *BKVTree[K, V]) SetIfNotExistFuncLock(key K, f func() V) bool {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if _, ok := tree.doGet(key); !ok {
+		tree.doSet(key, f())
+		return true
+	}
+	return false
+}
+
+// Get searches the `key` in the tree and returns its associated `value` or nil if key is not found in tree.
+//
+// Note that, the `nil` value from Get function cannot be used to determine key existence, please use Contains function
+// to do so.
+func (tree *BKVTree[K, V]) Get(key K) (value V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	value, _ = tree.doGet(key)
+	return
+}
+
+// GetOrSet returns its `value` of `key`, or sets value with given `value` if it does not exist and then returns
+// this value.
+func (tree *BKVTree[K, V]) GetOrSet(key K, value V) V {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if v, ok := tree.doGet(key); !ok {
+		return tree.doSet(key, value)
+	} else {
+		return v
+	}
+}
+
+// GetOrSetFunc returns its `value` of `key`, or sets value with returned value of callback function `f` if it does not
+// exist and then returns this value.
+func (tree *BKVTree[K, V]) GetOrSetFunc(key K, f func() V) V {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if v, ok := tree.doGet(key); !ok {
+		return tree.doSet(key, f())
+	} else {
+		return v
+	}
+}
+
+// GetOrSetFuncLock returns its `value` of `key`, or sets value with returned value of callback function `f` if it does
+// not exist and then returns this value.
+//
+// GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f` within mutex lock.
+func (tree *BKVTree[K, V]) GetOrSetFuncLock(key K, f func() V) V {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if v, ok := tree.doGet(key); !ok {
+		return tree.doSet(key, f())
+	} else {
+		return v
+	}
+}
+
+// GetVar returns a gvar.Var with the value by given `key`.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function Get.
+func (tree *BKVTree[K, V]) GetVar(key K) *gvar.Var {
+	return gvar.New(tree.Get(key))
+}
+
+// GetVarOrSet returns a gvar.Var with result from GetVarOrSet.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function GetOrSet.
+func (tree *BKVTree[K, V]) GetVarOrSet(key K, value V) *gvar.Var {
+	return gvar.New(tree.GetOrSet(key, value))
+}
+
+// GetVarOrSetFunc returns a gvar.Var with result from GetOrSetFunc.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function GetOrSetFunc.
+func (tree *BKVTree[K, V]) GetVarOrSetFunc(key K, f func() V) *gvar.Var {
+	return gvar.New(tree.GetOrSetFunc(key, f))
+}
+
+// GetVarOrSetFuncLock returns a gvar.Var with result from GetOrSetFuncLock.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function GetOrSetFuncLock.
+func (tree *BKVTree[K, V]) GetVarOrSetFuncLock(key K, f func() V) *gvar.Var {
+	return gvar.New(tree.GetOrSetFuncLock(key, f))
+}
+
+// Search searches the tree with given `key`.
+// Second return parameter `found` is true if key was found, otherwise false.
+func (tree *BKVTree[K, V]) Search(key K) (value V, found bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Get(key)
+}
+
+// Contains checks and returns whether given `key` exists in the tree.
+func (tree *BKVTree[K, V]) Contains(key K) bool {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	_, ok := tree.doGet(key)
+	return ok
+}
+
+// Size returns number of nodes in the tree.
+func (tree *BKVTree[K, V]) Size() int {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Size()
+}
+
+// IsEmpty returns true if tree does not contain any nodes
+func (tree *BKVTree[K, V]) IsEmpty() bool {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Size() == 0
+}
+
+// Remove removes the node from the tree by `key`, and returns its associated value of `key`.
+// The given `key` should adhere to the comparator's type assertion, otherwise method panics.
+func (tree *BKVTree[K, V]) Remove(key K) (value V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	return tree.doRemove(key)
+}
+
+// Removes batch deletes key-value pairs from the tree by `keys`.
+func (tree *BKVTree[K, V]) Removes(keys []K) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	for _, key := range keys {
+		tree.doRemove(key)
+	}
+}
+
+// Clear removes all nodes from the tree.
+func (tree *BKVTree[K, V]) Clear() {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	tree.tree.Clear()
+}
+
+// Keys returns all keys from the tree in order by its comparator.
+func (tree *BKVTree[K, V]) Keys() []K {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Keys()
+}
+
+// Values returns all values from the true in order by its comparator based on the key.
+func (tree *BKVTree[K, V]) Values() []V {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Values()
+}
+
+// Replace clears the data of the tree and sets the nodes by given `data`.
+func (tree *BKVTree[K, V]) Replace(data map[K]V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	tree.tree.Clear()
+	for k, v := range data {
+		tree.doSet(k, v)
+	}
+}
+
+// Map returns all key-value pairs as map.
+func (tree *BKVTree[K, V]) Map() map[K]V {
+	m := make(map[K]V, tree.Size())
+	tree.IteratorAsc(func(key K, value V) bool {
+		m[key] = value
+		return true
+	})
+	return m
+}
+
+// MapStrAny returns all key-value items as map[string]any.
+func (tree *BKVTree[K, V]) MapStrAny() map[string]any {
+	m := make(map[string]any, tree.Size())
+	tree.IteratorAsc(func(key K, value V) bool {
+		m[gconv.String(key)] = value
+		return true
+	})
+	return m
+}
+
+// Print prints the tree to stdout.
+func (tree *BKVTree[K, V]) Print() {
+	fmt.Println(tree.String())
+}
+
+// String returns a string representation of container (for debugging purposes)
+func (tree *BKVTree[K, V]) String() string {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return gstr.Replace(tree.tree.String(), "BTree\n", "")
+}
+
+// MarshalJSON implements the interface MarshalJSON for json.Marshal.
+func (tree *BKVTree[K, V]) MarshalJSON() (jsonBytes []byte, err error) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+
+	elements := make(map[string]V)
+	it := tree.tree.Iterator()
+	for it.Next() {
+		elements[gconv.String(it.Key())] = it.Value()
+	}
+	return json.Marshal(&elements)
+}
+
+// Iterator is alias of IteratorAsc.
+//
+// Also see IteratorAsc.
+func (tree *BKVTree[K, V]) Iterator(f func(key K, value V) bool) {
+	tree.IteratorAsc(f)
+}
+
+// IteratorFrom is alias of IteratorAscFrom.
+//
+// Also see IteratorAscFrom.
+func (tree *BKVTree[K, V]) IteratorFrom(key K, match bool, f func(key K, value V) bool) {
+	tree.IteratorAscFrom(key, match, f)
+}
+
+// IteratorAsc iterates the tree readonly in ascending order with given callback function `f`.
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *BKVTree[K, V]) IteratorAsc(f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var (
+		ok bool
+		it = tree.tree.Iterator()
+	)
+	for it.Begin(); it.Next(); {
+		index, value := it.Key(), it.Value()
+		if ok = f(index, value); !ok {
+			break
+		}
+	}
+}
+
+// IteratorAscFrom iterates the tree readonly in ascending order with given callback function `f`.
+//
+// The parameter `key` specifies the start entry for iterating.
+// The parameter `match` specifies whether starting iterating only if the `key` is fully matched, or else using index
+// searching iterating.
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *BKVTree[K, V]) IteratorAscFrom(key K, match bool, f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var keys = tree.tree.Keys()
+	index, canIterator := iteratorFromGetIndexT(key, keys, match)
+	if !canIterator {
+		return
+	}
+	for ; index < len(keys); index++ {
+		f(keys[index], tree.Get(keys[index]))
+	}
+}
+
+// IteratorDesc iterates the tree readonly in descending order with given callback function `f`.
+//
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *BKVTree[K, V]) IteratorDesc(f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var (
+		ok bool
+		it = tree.tree.Iterator()
+	)
+	for it.End(); it.Prev(); {
+		index, value := it.Key(), it.Value()
+		if ok = f(index, value); !ok {
+			break
+		}
+	}
+}
+
+// IteratorDescFrom iterates the tree readonly in descending order with given callback function `f`.
+//
+// The parameter `key` specifies the start entry for iterating.
+// The parameter `match` specifies whether starting iterating only if the `key` is fully matched, or else using index
+// searching iterating.
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *BKVTree[K, V]) IteratorDescFrom(key K, match bool, f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var keys = tree.tree.Keys()
+	index, canIterator := iteratorFromGetIndexT(key, keys, match)
+	if !canIterator {
+		return
+	}
+	for ; index >= 0; index-- {
+		f(keys[index], tree.Get(keys[index]))
+	}
+}
+
+// Height returns the height of the tree.
+func (tree *BKVTree[K, V]) Height() int {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Height()
+}
+
+// Left returns the minimum element corresponding to the comparator of the tree or nil if the tree is empty.
+func (tree *BKVTree[K, V]) Left() *BKVTreeEntry[K, V] {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	node := tree.tree.Left()
+	if node == nil || node.Entries == nil || len(node.Entries) == 0 {
+		return nil
+	}
+	return &BKVTreeEntry[K, V]{
+		Key:   node.Entries[0].Key,
+		Value: node.Entries[0].Value,
+	}
+}
+
+// Right returns the maximum element corresponding to the comparator of the tree or nil if the tree is empty.
+func (tree *BKVTree[K, V]) Right() *BKVTreeEntry[K, V] {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	node := tree.tree.Right()
+	if node == nil || node.Entries == nil || len(node.Entries) == 0 {
+		return nil
+	}
+	return &BKVTreeEntry[K, V]{
+		Key:   node.Entries[len(node.Entries)-1].Key,
+		Value: node.Entries[len(node.Entries)-1].Value,
+	}
+}
+
+// doSet inserts key-value pair node into the tree without lock.
+// If `key` already exists, then its value is updated with the new value.
+// If `value` is type of <func() any>, it will be executed and its return value will be set to the map with `key`.
+//
+// It returns value with given `key`.
+func (tree *BKVTree[K, V]) doSet(key K, value V) V {
+	if any(value) == nil {
+		return value
+	}
+	tree.tree.Put(key, value)
+	return value
+}
+
+// doGet get the value from the tree by key without lock.
+func (tree *BKVTree[K, V]) doGet(key K) (value V, ok bool) {
+	return tree.tree.Get(key)
+}
+
+// doRemove removes key from tree and returns its associated value without lock.
+// Note that, the given `key` should adhere to the comparator's type assertion, otherwise method panics.
+func (tree *BKVTree[K, V]) doRemove(key K) (value V) {
+	value, _ = tree.tree.Get(key)
+	tree.tree.Remove(key)
+	return
+}
--- a/container/gtree/gtree_k_v_redblacktree.go
+++ b/container/gtree/gtree_k_v_redblacktree.go
@ -0,0 +1,613 @@
+// Copyright GoFrame Author(https://goframe.org). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package gtree
+
+import (
+	"fmt"
+
+	"github.com/emirpasic/gods/v2/trees/redblacktree"
+
+	"github.com/gogf/gf/v2/container/gvar"
+	"github.com/gogf/gf/v2/internal/json"
+	"github.com/gogf/gf/v2/internal/rwmutex"
+	"github.com/gogf/gf/v2/text/gstr"
+	"github.com/gogf/gf/v2/util/gconv"
+	"github.com/gogf/gf/v2/util/gutil"
+)
+
+// RedBlackKVTree holds elements of the red-black tree.
+type RedBlackKVTree[K comparable, V any] struct {
+	mu         rwmutex.RWMutex
+	comparator func(v1, v2 K) int
+	tree       *redblacktree.Tree[K, V]
+}
+
+// RedBlackKVTreeNode is a single element within the tree.
+type RedBlackKVTreeNode[K comparable, V any] struct {
+	Key   K
+	Value V
+}
+
+// NewRedBlackKVTree instantiates a red-black tree with the custom key comparator.
+// The parameter `safe` is used to specify whether using tree in concurrent-safety,
+// which is false in default.
+func NewRedBlackKVTree[K comparable, V any](comparator func(v1, v2 K) int, safe ...bool) *RedBlackKVTree[K, V] {
+	var tree RedBlackKVTree[K, V]
+	RedBlackKVTreeInit(&tree, comparator, safe...)
+	return &tree
+}
+
+// NewRedBlackKVTreeFrom instantiates a red-black tree with the custom key comparator and `data` map.
+// The parameter `safe` is used to specify whether using tree in concurrent-safety,
+// which is false in default.
+func NewRedBlackKVTreeFrom[K comparable, V any](comparator func(v1, v2 K) int, data map[K]V, safe ...bool) *RedBlackKVTree[K, V] {
+	var tree RedBlackKVTree[K, V]
+	RedBlackKVTreeInitFrom(&tree, comparator, data, safe...)
+	return &tree
+}
+
+// RedBlackKVTreeInit instantiates a red-black tree with the custom key comparator.
+// The parameter `safe` is used to specify whether using tree in concurrent-safety,
+// which is false in default.
+func RedBlackKVTreeInit[K comparable, V any](tree *RedBlackKVTree[K, V], comparator func(v1, v2 K) int, safe ...bool) {
+	if tree == nil {
+		return
+	}
+	tree.mu = rwmutex.Create(safe...)
+	tree.comparator = comparator
+	tree.tree = redblacktree.NewWith[K, V](comparator)
+}
+
+// RedBlackKVTreeInitFrom instantiates a red-black tree with the custom key comparator and `data` map.
+// The parameter `safe` is used to specify whether using tree in concurrent-safety,
+// which is false in default.
+func RedBlackKVTreeInitFrom[K comparable, V any](tree *RedBlackKVTree[K, V], comparator func(v1, v2 K) int, data map[K]V, safe ...bool) {
+	if tree == nil {
+		return
+	}
+	RedBlackKVTreeInit(tree, comparator, safe...)
+	for k, v := range data {
+		tree.doSet(k, v)
+	}
+}
+
+// SetComparator sets/changes the comparator for sorting.
+func (tree *RedBlackKVTree[K, V]) SetComparator(comparator func(a, b K) int) {
+	tree.comparator = comparator
+	if tree.tree == nil {
+		tree.tree = redblacktree.NewWith[K, V](comparator)
+	}
+	size := tree.tree.Size()
+	if size > 0 {
+		m := tree.Map()
+		tree.Sets(m)
+	}
+}
+
+// Clone clones and returns a new tree from current tree.
+func (tree *RedBlackKVTree[K, V]) Clone() *RedBlackKVTree[K, V] {
+	if tree == nil {
+		return nil
+	}
+	newTree := NewRedBlackKVTree[K, V](tree.comparator, tree.mu.IsSafe())
+	newTree.Sets(tree.Map())
+	return newTree
+}
+
+// Set sets key-value pair into the tree.
+func (tree *RedBlackKVTree[K, V]) Set(key K, value V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	tree.doSet(key, value)
+}
+
+// Sets batch sets key-values to the tree.
+func (tree *RedBlackKVTree[K, V]) Sets(data map[K]V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	for key, value := range data {
+		tree.doSet(key, value)
+	}
+}
+
+// SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
+// It returns false if `key` exists, and such setting key-value pair operation would be ignored.
+func (tree *RedBlackKVTree[K, V]) SetIfNotExist(key K, value V) bool {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if _, ok := tree.doGet(key); !ok {
+		tree.doSet(key, value)
+		return true
+	}
+	return false
+}
+
+// SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
+// It returns false if `key` exists, and such setting key-value pair operation would be ignored.
+func (tree *RedBlackKVTree[K, V]) SetIfNotExistFunc(key K, f func() V) bool {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if _, ok := tree.doGet(key); !ok {
+		tree.doSet(key, f())
+		return true
+	}
+	return false
+}
+
+// SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
+// It returns false if `key` exists, and such setting key-value pair operation would be ignored.
+//
+// SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
+// it executes function `f` within mutex lock.
+func (tree *RedBlackKVTree[K, V]) SetIfNotExistFuncLock(key K, f func() V) bool {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if _, ok := tree.doGet(key); !ok {
+		tree.doSet(key, f())
+		return true
+	}
+	return false
+}
+
+// Get searches the `key` in the tree and returns its associated `value` or nil if key is not found in tree.
+//
+// Note that, the `nil` value from Get function cannot be used to determine key existence, please use Contains function
+// to do so.
+func (tree *RedBlackKVTree[K, V]) Get(key K) (value V) {
+	value, _ = tree.Search(key)
+	return
+}
+
+// GetOrSet returns its `value` of `key`, or sets value with given `value` if it does not exist and then returns
+// this value.
+func (tree *RedBlackKVTree[K, V]) GetOrSet(key K, value V) V {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if v, ok := tree.doGet(key); !ok {
+		return tree.doSet(key, value)
+	} else {
+		return v
+	}
+}
+
+// GetOrSetFunc returns its `value` of `key`, or sets value with returned value of callback function `f` if it does not
+// exist and then returns this value.
+func (tree *RedBlackKVTree[K, V]) GetOrSetFunc(key K, f func() V) V {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if v, ok := tree.doGet(key); !ok {
+		return tree.doSet(key, f())
+	} else {
+		return v
+	}
+}
+
+// GetOrSetFuncLock returns its `value` of `key`, or sets value with returned value of callback function `f` if it does
+// not exist and then returns this value.
+//
+// GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`within mutex lock.
+func (tree *RedBlackKVTree[K, V]) GetOrSetFuncLock(key K, f func() V) V {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if v, ok := tree.doGet(key); !ok {
+		return tree.doSet(key, f())
+	} else {
+		return v
+	}
+}
+
+// GetVar returns a gvar.Var with the value by given `key`.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function Get.
+func (tree *RedBlackKVTree[K, V]) GetVar(key K) *gvar.Var {
+	return gvar.New(tree.Get(key))
+}
+
+// GetVarOrSet returns a gvar.Var with result from GetVarOrSet.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function GetOrSet.
+func (tree *RedBlackKVTree[K, V]) GetVarOrSet(key K, value V) *gvar.Var {
+	return gvar.New(tree.GetOrSet(key, value))
+}
+
+// GetVarOrSetFunc returns a gvar.Var with result from GetOrSetFunc.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function GetOrSetFunc.
+func (tree *RedBlackKVTree[K, V]) GetVarOrSetFunc(key K, f func() V) *gvar.Var {
+	return gvar.New(tree.GetOrSetFunc(key, f))
+}
+
+// GetVarOrSetFuncLock returns a gvar.Var with result from GetOrSetFuncLock.
+// Note that, the returned gvar.Var is un-concurrent safe.
+//
+// Also see function GetOrSetFuncLock.
+func (tree *RedBlackKVTree[K, V]) GetVarOrSetFuncLock(key K, f func() V) *gvar.Var {
+	return gvar.New(tree.GetOrSetFuncLock(key, f))
+}
+
+// Search searches the tree with given `key`.
+// Second return parameter `found` is true if key was found, otherwise false.
+func (tree *RedBlackKVTree[K, V]) Search(key K) (value V, found bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	if node, found := tree.doGet(key); found {
+		return node, true
+	}
+	found = false
+	return
+}
+
+// Contains checks and returns whether given `key` exists in the tree.
+func (tree *RedBlackKVTree[K, V]) Contains(key K) bool {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	_, ok := tree.doGet(key)
+	return ok
+}
+
+// Size returns number of nodes in the tree.
+func (tree *RedBlackKVTree[K, V]) Size() int {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Size()
+}
+
+// IsEmpty returns true if tree does not contain any nodes.
+func (tree *RedBlackKVTree[K, V]) IsEmpty() bool {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Size() == 0
+}
+
+// Remove removes the node from the tree by `key`, and returns its associated value of `key`.
+// The given `key` should adhere to the comparator's type assertion, otherwise method panics.
+func (tree *RedBlackKVTree[K, V]) Remove(key K) (value V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	return tree.doRemove(key)
+}
+
+// Removes batch deletes key-value pairs from the tree by `keys`.
+func (tree *RedBlackKVTree[K, V]) Removes(keys []K) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	for _, key := range keys {
+		tree.doRemove(key)
+	}
+}
+
+// Clear removes all nodes from the tree.
+func (tree *RedBlackKVTree[K, V]) Clear() {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	tree.tree.Clear()
+}
+
+// Keys returns all keys from the tree in order by its comparator.
+func (tree *RedBlackKVTree[K, V]) Keys() []K {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Keys()
+}
+
+// Values returns all values from the true in order by its comparator based on the key.
+func (tree *RedBlackKVTree[K, V]) Values() []V {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return tree.tree.Values()
+}
+
+// Replace clears the data of the tree and sets the nodes by given `data`.
+func (tree *RedBlackKVTree[K, V]) Replace(data map[K]V) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	tree.tree.Clear()
+	for k, v := range data {
+		tree.doSet(k, v)
+	}
+}
+
+// Print prints the tree to stdout.
+func (tree *RedBlackKVTree[K, V]) Print() {
+	fmt.Println(tree.String())
+}
+
+// String returns a string representation of container
+func (tree *RedBlackKVTree[K, V]) String() string {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	return gstr.Replace(tree.tree.String(), "RedBlackTree\n", "")
+}
+
+// MarshalJSON implements the interface MarshalJSON for json.Marshal.
+func (tree RedBlackKVTree[K, V]) MarshalJSON() (jsonBytes []byte, err error) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+
+	elements := make(map[string]V)
+	it := tree.tree.Iterator()
+	for it.Next() {
+		elements[gconv.String(it.Key())] = it.Value()
+	}
+	return json.Marshal(&elements)
+}
+
+// Map returns all key-value pairs as map.
+func (tree *RedBlackKVTree[K, V]) Map() map[K]V {
+	m := make(map[K]V, tree.Size())
+	tree.IteratorAsc(func(key K, value V) bool {
+		m[key] = value
+		return true
+	})
+	return m
+}
+
+// MapStrAny returns all key-value items as map[string]any.
+func (tree *RedBlackKVTree[K, V]) MapStrAny() map[string]any {
+	m := make(map[string]any, tree.Size())
+	tree.IteratorAsc(func(key K, value V) bool {
+		m[gconv.String(key)] = value
+		return true
+	})
+	return m
+}
+
+// Iterator is alias of IteratorAsc.
+//
+// Also see IteratorAsc.
+func (tree *RedBlackKVTree[K, V]) Iterator(f func(key K, value V) bool) {
+	tree.IteratorAsc(f)
+}
+
+// IteratorFrom is alias of IteratorAscFrom.
+//
+// Also see IteratorAscFrom.
+func (tree *RedBlackKVTree[K, V]) IteratorFrom(key K, match bool, f func(key K, value V) bool) {
+	tree.IteratorAscFrom(key, match, f)
+}
+
+// IteratorAsc iterates the tree readonly in ascending order with given callback function `f`.
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *RedBlackKVTree[K, V]) IteratorAsc(f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var (
+		ok bool
+		it = tree.tree.Iterator()
+	)
+	for it.Begin(); it.Next(); {
+		index, value := it.Key(), it.Value()
+		if ok = f(index, value); !ok {
+			break
+		}
+	}
+}
+
+// IteratorAscFrom iterates the tree readonly in ascending order with given callback function `f`.
+//
+// The parameter `key` specifies the start entry for iterating.
+// The parameter `match` specifies whether starting iterating only if the `key` is fully matched, or else using index
+// searching iterating.
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *RedBlackKVTree[K, V]) IteratorAscFrom(key K, match bool, f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var keys = tree.tree.Keys()
+	index, canIterator := iteratorFromGetIndexT(key, keys, match)
+	if !canIterator {
+		return
+	}
+	for ; index < len(keys); index++ {
+		f(keys[index], tree.Get(keys[index]))
+	}
+}
+
+// IteratorDesc iterates the tree readonly in descending order with given callback function `f`.
+//
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *RedBlackKVTree[K, V]) IteratorDesc(f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var (
+		ok bool
+		it = tree.tree.Iterator()
+	)
+	for it.End(); it.Prev(); {
+		index, value := it.Key(), it.Value()
+		if ok = f(index, value); !ok {
+			break
+		}
+	}
+}
+
+// IteratorDescFrom iterates the tree readonly in descending order with given callback function `f`.
+//
+// The parameter `key` specifies the start entry for iterating.
+// The parameter `match` specifies whether starting iterating only if the `key` is fully matched, or else using index
+// searching iterating.
+// If callback function `f` returns true, then it continues iterating; or false to stop.
+func (tree *RedBlackKVTree[K, V]) IteratorDescFrom(key K, match bool, f func(key K, value V) bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	var keys = tree.tree.Keys()
+	index, canIterator := iteratorFromGetIndexT(key, keys, match)
+	if !canIterator {
+		return
+	}
+	for ; index >= 0; index-- {
+		f(keys[index], tree.Get(keys[index]))
+	}
+}
+
+// Left returns the minimum element corresponding to the comparator of the tree or nil if the tree is empty.
+func (tree *RedBlackKVTree[K, V]) Left() *RedBlackKVTreeNode[K, V] {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	node := tree.tree.Left()
+	if node == nil {
+		return nil
+	}
+	return &RedBlackKVTreeNode[K, V]{
+		Key:   node.Key,
+		Value: node.Value,
+	}
+}
+
+// Right returns the maximum element corresponding to the comparator of the tree or nil if the tree is empty.
+func (tree *RedBlackKVTree[K, V]) Right() *RedBlackKVTreeNode[K, V] {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	node := tree.tree.Right()
+	if node == nil {
+		return nil
+	}
+	return &RedBlackKVTreeNode[K, V]{
+		Key:   node.Key,
+		Value: node.Value,
+	}
+}
+
+// Floor Finds floor node of the input key, returns the floor node or nil if no floor node is found.
+// The second returned parameter `found` is true if floor was found, otherwise false.
+//
+// Floor node is defined as the largest node that is smaller than or equal to the given node.
+// A floor node may not be found, either because the tree is empty, or because
+// all nodes in the tree is larger than the given node.
+//
+// Key should adhere to the comparator's type assertion, otherwise method panics.
+func (tree *RedBlackKVTree[K, V]) Floor(key K) (floor *RedBlackKVTreeNode[K, V], found bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	node, found := tree.tree.Floor(key)
+	if !found {
+		return nil, false
+	}
+	return &RedBlackKVTreeNode[K, V]{
+		Key:   node.Key,
+		Value: node.Value,
+	}, true
+}
+
+// Ceiling finds ceiling node of the input key, returns the ceiling node or nil if no ceiling node is found.
+// The second return parameter `found` is true if ceiling was found, otherwise false.
+//
+// Ceiling node is defined as the smallest node that is larger than or equal to the given node.
+// A ceiling node may not be found, either because the tree is empty, or because
+// all nodes in the tree is smaller than the given node.
+//
+// Key should adhere to the comparator's type assertion, otherwise method panics.
+func (tree *RedBlackKVTree[K, V]) Ceiling(key K) (ceiling *RedBlackKVTreeNode[K, V], found bool) {
+	tree.mu.RLock()
+	defer tree.mu.RUnlock()
+	node, found := tree.tree.Ceiling(key)
+	if !found {
+		return nil, false
+	}
+	return &RedBlackKVTreeNode[K, V]{
+		Key:   node.Key,
+		Value: node.Value,
+	}, true
+}
+
+// Flip exchanges key-value of the tree to value-key.
+// Note that you should guarantee the value is the same type as key,
+// or else the comparator would panic.
+//
+// If the type of value is different with key, you pass the new `comparator`.
+func (tree *RedBlackKVTree[K, V]) Flip(comparator ...func(v1, v2 K) int) {
+	var t = new(RedBlackKVTree[K, V])
+	if len(comparator) > 0 {
+		t = NewRedBlackKVTree[K, V](comparator[0], tree.mu.IsSafe())
+	} else {
+		t = NewRedBlackKVTree[K, V](tree.comparator, tree.mu.IsSafe())
+	}
+	var (
+		newKey   K
+		newValue V
+	)
+	tree.IteratorAsc(func(key K, value V) bool {
+		if err := gconv.Scan(key, &newValue); err != nil {
+			panic(err)
+		}
+		if err := gconv.Scan(value, &newKey); err != nil {
+			panic(err)
+		}
+		t.doSet(newKey, newValue)
+		return true
+	})
+	tree.Clear()
+	tree.Sets(t.Map())
+}
+
+// UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
+func (tree *RedBlackKVTree[K, V]) UnmarshalJSON(b []byte) (err error) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if tree.comparator == nil {
+		tree.comparator = gutil.ComparatorTStr[K]
+		tree.tree = redblacktree.NewWith[K, V](tree.comparator)
+	}
+	var data map[string]any
+	if err := json.UnmarshalUseNumber(b, &data); err != nil {
+		return err
+	}
+	var m = make(map[K]V)
+	if err = gconv.Scan(data, &m); err != nil {
+		return
+	}
+	for k, v := range m {
+		tree.doSet(k, v)
+	}
+	return nil
+}
+
+// UnmarshalValue is an interface implement which sets any type of value for map.
+func (tree *RedBlackKVTree[K, V]) UnmarshalValue(value any) (err error) {
+	tree.mu.Lock()
+	defer tree.mu.Unlock()
+	if tree.comparator == nil {
+		tree.comparator = gutil.ComparatorTStr[K]
+		tree.tree = redblacktree.NewWith[K, V](tree.comparator)
+	}
+	var m = make(map[K]V)
+	if err = gconv.Scan(value, &m); err != nil {
+		return
+	}
+	for k, v := range m {
+		tree.doSet(k, v)
+	}
+	return
+}
+
+// doSet inserts key-value pair node into the tree without lock.
+// If `key` already exists, then its value is updated with the new value.
+// If `value` is type of <func() any>, it will be executed and its return value will be set to the map with `key`.
+//
+// It returns value with given `key`.
+func (tree *RedBlackKVTree[K, V]) doSet(key K, value V) (ret V) {
+	if any(value) == nil {
+		return
+	}
+	tree.tree.Put(key, value)
+	return value
+}
+
+// doGet retrieves and returns the value of given key from tree without lock.
+func (tree *RedBlackKVTree[K, V]) doGet(key K) (value V, found bool) {
+	return tree.tree.Get(key)
+}
+
+// doRemove removes key from tree and returns its associated value without lock.
+// Note that, the given `key` should adhere to the comparator's type assertion, otherwise method panics.
+func (tree *RedBlackKVTree[K, V]) doRemove(key K) (value V) {
+	value, _ = tree.tree.Get(key)
+	tree.tree.Remove(key)
+	return
+}
--- a/container/gtree/gtree_redblacktree.go
+++ b/container/gtree/gtree_redblacktree.go
@ -7,15 +7,9 @@
 package gtree

 import (
-	"fmt"
-
-	"github.com/emirpasic/gods/trees/redblacktree"
+	"sync"

 	"github.com/gogf/gf/v2/container/gvar"
-	"github.com/gogf/gf/v2/internal/json"
-	"github.com/gogf/gf/v2/internal/rwmutex"
-	"github.com/gogf/gf/v2/text/gstr"
-	"github.com/gogf/gf/v2/util/gconv"
 	"github.com/gogf/gf/v2/util/gutil"
 )

@ -23,25 +17,19 @@ var _ iTree = (*RedBlackTree)(nil)

 // RedBlackTree holds elements of the red-black tree.
 type RedBlackTree struct {
-	mu         rwmutex.RWMutex
-	comparator func(v1, v2 any) int
-	tree       *redblacktree.Tree
+	*RedBlackKVTree[any, any]
+	once sync.Once
 }

 // RedBlackTreeNode is a single element within the tree.
-type RedBlackTreeNode struct {
-	Key   any
-	Value any
-}
+type RedBlackTreeNode = RedBlackKVTreeNode[any, any]

 // NewRedBlackTree instantiates a red-black tree with the custom key comparator.
 // The parameter `safe` is used to specify whether using tree in concurrent-safety,
 // which is false in default.
 func NewRedBlackTree(comparator func(v1, v2 any) int, safe ...bool) *RedBlackTree {
 	return &RedBlackTree{
-		mu:         rwmutex.Create(safe...),
-		comparator: comparator,
-		tree:       redblacktree.NewWith(comparator),
+		RedBlackKVTree: NewRedBlackKVTree[any, any](comparator, safe...),
 	}
 }

@ -49,71 +37,61 @@ func NewRedBlackTree(comparator func(v1, v2 any) int, safe ...bool) *RedBlackTre
 // The parameter `safe` is used to specify whether using tree in concurrent-safety,
 // which is false in default.
 func NewRedBlackTreeFrom(comparator func(v1, v2 any) int, data map[any]any, safe ...bool) *RedBlackTree {
-	tree := NewRedBlackTree(comparator, safe...)
-	for k, v := range data {
-		tree.doSet(k, v)
+	return &RedBlackTree{
+		RedBlackKVTree: NewRedBlackKVTreeFrom(comparator, data, safe...),
 	}
-	return tree
+}
+
+// lazyInit lazily initializes the tree.
+func (tree *RedBlackTree) lazyInit() {
+	tree.once.Do(func() {
+		if tree.RedBlackKVTree == nil {
+			tree.RedBlackKVTree = NewRedBlackKVTree[any, any](gutil.ComparatorTStr, false)
+		}
+	})
 }

 // SetComparator sets/changes the comparator for sorting.
 func (tree *RedBlackTree) SetComparator(comparator func(a, b any) int) {
-	tree.comparator = comparator
-	if tree.tree == nil {
-		tree.tree = redblacktree.NewWith(comparator)
-	}
-	size := tree.tree.Size()
-	if size > 0 {
-		m := tree.Map()
-		tree.Sets(m)
-	}
+	tree.lazyInit()
+	tree.RedBlackKVTree.SetComparator(comparator)
 }

 // Clone clones and returns a new tree from current tree.
 func (tree *RedBlackTree) Clone() *RedBlackTree {
-	newTree := NewRedBlackTree(tree.comparator, tree.mu.IsSafe())
-	newTree.Sets(tree.Map())
-	return newTree
+	if tree == nil {
+		return nil
+	}
+	tree.lazyInit()
+	return &RedBlackTree{
+		RedBlackKVTree: tree.RedBlackKVTree.Clone(),
+	}
 }

 // Set sets key-value pair into the tree.
 func (tree *RedBlackTree) Set(key any, value any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	tree.doSet(key, value)
+	tree.lazyInit()
+	tree.RedBlackKVTree.Set(key, value)
 }

 // Sets batch sets key-values to the tree.
 func (tree *RedBlackTree) Sets(data map[any]any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	for key, value := range data {
-		tree.doSet(key, value)
-	}
+	tree.lazyInit()
+	tree.RedBlackKVTree.Sets(data)
 }

 // SetIfNotExist sets `value` to the map if the `key` does not exist, and then returns true.
 // It returns false if `key` exists, and such setting key-value pair operation would be ignored.
 func (tree *RedBlackTree) SetIfNotExist(key any, value any) bool {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if _, ok := tree.doGet(key); !ok {
-		tree.doSet(key, value)
-		return true
-	}
-	return false
+	tree.lazyInit()
+	return tree.RedBlackKVTree.SetIfNotExist(key, value)
 }

 // SetIfNotExistFunc sets value with return value of callback function `f`, and then returns true.
 // It returns false if `key` exists, and such setting key-value pair operation would be ignored.
 func (tree *RedBlackTree) SetIfNotExistFunc(key any, f func() any) bool {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if _, ok := tree.doGet(key); !ok {
-		tree.doSet(key, f())
-		return true
-	}
-	return false
+	tree.lazyInit()
+	return tree.RedBlackKVTree.SetIfNotExistFunc(key, f)
 }

 // SetIfNotExistFuncLock sets value with return value of callback function `f`, and then returns true.
@ -122,13 +100,8 @@ func (tree *RedBlackTree) SetIfNotExistFunc(key any, f func() any) bool {
 // SetIfNotExistFuncLock differs with SetIfNotExistFunc function is that
 // it executes function `f` within mutex lock.
 func (tree *RedBlackTree) SetIfNotExistFuncLock(key any, f func() any) bool {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if _, ok := tree.doGet(key); !ok {
-		tree.doSet(key, f)
-		return true
-	}
-	return false
+	tree.lazyInit()
+	return tree.RedBlackKVTree.SetIfNotExistFuncLock(key, f)
 }

 // Get searches the `key` in the tree and returns its associated `value` or nil if key is not found in tree.
@ -136,32 +109,22 @@ func (tree *RedBlackTree) SetIfNotExistFuncLock(key any, f func() any) bool {
 // Note that, the `nil` value from Get function cannot be used to determine key existence, please use Contains function
 // to do so.
 func (tree *RedBlackTree) Get(key any) (value any) {
-	value, _ = tree.Search(key)
-	return
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Get(key)
 }

 // GetOrSet returns its `value` of `key`, or sets value with given `value` if it does not exist and then returns
 // this value.
 func (tree *RedBlackTree) GetOrSet(key any, value any) any {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if v, ok := tree.doGet(key); !ok {
-		return tree.doSet(key, value)
-	} else {
-		return v
-	}
+	tree.lazyInit()
+	return tree.RedBlackKVTree.GetOrSet(key, value)
 }

 // GetOrSetFunc returns its `value` of `key`, or sets value with returned value of callback function `f` if it does not
 // exist and then returns this value.
 func (tree *RedBlackTree) GetOrSetFunc(key any, f func() any) any {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if v, ok := tree.doGet(key); !ok {
-		return tree.doSet(key, f())
-	} else {
-		return v
-	}
+	tree.lazyInit()
+	return tree.RedBlackKVTree.GetOrSetFunc(key, f)
 }

 // GetOrSetFuncLock returns its `value` of `key`, or sets value with returned value of callback function `f` if it does
@ -169,13 +132,8 @@ func (tree *RedBlackTree) GetOrSetFunc(key any, f func() any) any {
 //
 // GetOrSetFuncLock differs with GetOrSetFunc function is that it executes function `f`within mutex lock.
 func (tree *RedBlackTree) GetOrSetFuncLock(key any, f func() any) any {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if v, ok := tree.doGet(key); !ok {
-		return tree.doSet(key, f)
-	} else {
-		return v
-	}
+	tree.lazyInit()
+	return tree.RedBlackKVTree.GetOrSetFuncLock(key, f)
 }

 // GetVar returns a gvar.Var with the value by given `key`.
@ -183,7 +141,8 @@ func (tree *RedBlackTree) GetOrSetFuncLock(key any, f func() any) any {
 //
 // Also see function Get.
 func (tree *RedBlackTree) GetVar(key any) *gvar.Var {
-	return gvar.New(tree.Get(key))
+	tree.lazyInit()
+	return tree.RedBlackKVTree.GetVar(key)
 }

 // GetVarOrSet returns a gvar.Var with result from GetVarOrSet.
@ -191,7 +150,8 @@ func (tree *RedBlackTree) GetVar(key any) *gvar.Var {
 //
 // Also see function GetOrSet.
 func (tree *RedBlackTree) GetVarOrSet(key any, value any) *gvar.Var {
-	return gvar.New(tree.GetOrSet(key, value))
+	tree.lazyInit()
+	return tree.RedBlackKVTree.GetVarOrSet(key, value)
 }

 // GetVarOrSetFunc returns a gvar.Var with result from GetOrSetFunc.
@ -199,7 +159,8 @@ func (tree *RedBlackTree) GetVarOrSet(key any, value any) *gvar.Var {
 //
 // Also see function GetOrSetFunc.
 func (tree *RedBlackTree) GetVarOrSetFunc(key any, f func() any) *gvar.Var {
-	return gvar.New(tree.GetOrSetFunc(key, f))
+	tree.lazyInit()
+	return tree.RedBlackKVTree.GetVarOrSetFunc(key, f)
 }

 // GetVarOrSetFuncLock returns a gvar.Var with result from GetOrSetFuncLock.
@ -207,158 +168,123 @@ func (tree *RedBlackTree) GetVarOrSetFunc(key any, f func() any) *gvar.Var {
 //
 // Also see function GetOrSetFuncLock.
 func (tree *RedBlackTree) GetVarOrSetFuncLock(key any, f func() any) *gvar.Var {
-	return gvar.New(tree.GetOrSetFuncLock(key, f))
+	tree.lazyInit()
+	return tree.RedBlackKVTree.GetVarOrSetFuncLock(key, f)
 }

 // Search searches the tree with given `key`.
 // Second return parameter `found` is true if key was found, otherwise false.
 func (tree *RedBlackTree) Search(key any) (value any, found bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	if node, found := tree.doGet(key); found {
-		return node, true
-	}
-	return nil, false
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Search(key)
 }

 // Contains checks and returns whether given `key` exists in the tree.
 func (tree *RedBlackTree) Contains(key any) bool {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	_, ok := tree.doGet(key)
-	return ok
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Contains(key)
 }

 // Size returns number of nodes in the tree.
 func (tree *RedBlackTree) Size() int {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Size()
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Size()
 }

 // IsEmpty returns true if tree does not contain any nodes.
 func (tree *RedBlackTree) IsEmpty() bool {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Size() == 0
+	tree.lazyInit()
+	return tree.RedBlackKVTree.IsEmpty()
 }

 // Remove removes the node from the tree by `key`, and returns its associated value of `key`.
 // The given `key` should adhere to the comparator's type assertion, otherwise method panics.
 func (tree *RedBlackTree) Remove(key any) (value any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	return tree.doRemove(key)
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Remove(key)
 }

 // Removes batch deletes key-value pairs from the tree by `keys`.
 func (tree *RedBlackTree) Removes(keys []any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	for _, key := range keys {
-		tree.doRemove(key)
-	}
+	tree.lazyInit()
+	tree.RedBlackKVTree.Removes(keys)
 }

 // Clear removes all nodes from the tree.
 func (tree *RedBlackTree) Clear() {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	tree.tree.Clear()
+	tree.lazyInit()
+	tree.RedBlackKVTree.Clear()
 }

 // Keys returns all keys from the tree in order by its comparator.
 func (tree *RedBlackTree) Keys() []any {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Keys()
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Keys()
 }

 // Values returns all values from the true in order by its comparator based on the key.
 func (tree *RedBlackTree) Values() []any {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.Values()
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Values()
 }

 // Replace clears the data of the tree and sets the nodes by given `data`.
 func (tree *RedBlackTree) Replace(data map[any]any) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	tree.tree.Clear()
-	for k, v := range data {
-		tree.doSet(k, v)
-	}
+	tree.lazyInit()
+	tree.RedBlackKVTree.Replace(data)
 }

 // Print prints the tree to stdout.
 func (tree *RedBlackTree) Print() {
-	fmt.Println(tree.String())
+	tree.lazyInit()
+	tree.RedBlackKVTree.Print()
 }

 // String returns a string representation of container
 func (tree *RedBlackTree) String() string {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return gstr.Replace(tree.tree.String(), "RedBlackTree\n", "")
+	tree.lazyInit()
+	return tree.RedBlackKVTree.String()
 }

 // MarshalJSON implements the interface MarshalJSON for json.Marshal.
-func (tree *RedBlackTree) MarshalJSON() (jsonBytes []byte, err error) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	return tree.tree.MarshalJSON()
+func (tree RedBlackTree) MarshalJSON() (jsonBytes []byte, err error) {
+	tree.lazyInit()
+	return tree.RedBlackKVTree.MarshalJSON()
 }

 // Map returns all key-value pairs as map.
 func (tree *RedBlackTree) Map() map[any]any {
-	m := make(map[any]any, tree.Size())
-	tree.IteratorAsc(func(key, value any) bool {
-		m[key] = value
-		return true
-	})
-	return m
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Map()
 }

 // MapStrAny returns all key-value items as map[string]any.
 func (tree *RedBlackTree) MapStrAny() map[string]any {
-	m := make(map[string]any, tree.Size())
-	tree.IteratorAsc(func(key, value any) bool {
-		m[gconv.String(key)] = value
-		return true
-	})
-	return m
+	tree.lazyInit()
+	return tree.RedBlackKVTree.MapStrAny()
 }

 // Iterator is alias of IteratorAsc.
 //
 // Also see IteratorAsc.
 func (tree *RedBlackTree) Iterator(f func(key, value any) bool) {
-	tree.IteratorAsc(f)
+	tree.lazyInit()
+	tree.RedBlackKVTree.Iterator(f)
 }

 // IteratorFrom is alias of IteratorAscFrom.
 //
 // Also see IteratorAscFrom.
 func (tree *RedBlackTree) IteratorFrom(key any, match bool, f func(key, value any) bool) {
-	tree.IteratorAscFrom(key, match, f)
+	tree.lazyInit()
+	tree.RedBlackKVTree.IteratorFrom(key, match, f)
 }

 // IteratorAsc iterates the tree readonly in ascending order with given callback function `f`.
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *RedBlackTree) IteratorAsc(f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var (
-		ok bool
-		it = tree.tree.Iterator()
-	)
-	for it.Begin(); it.Next(); {
-		index, value := it.Key(), it.Value()
-		if ok = f(index, value); !ok {
-			break
-		}
-	}
+	tree.lazyInit()
+	tree.RedBlackKVTree.IteratorAsc(f)
 }

 // IteratorAscFrom iterates the tree readonly in ascending order with given callback function `f`.
@ -368,34 +294,16 @@ func (tree *RedBlackTree) IteratorAsc(f func(key, value any) bool) {
 // searching iterating.
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *RedBlackTree) IteratorAscFrom(key any, match bool, f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var keys = tree.tree.Keys()
-	index, canIterator := iteratorFromGetIndex(key, keys, match)
-	if !canIterator {
-		return
-	}
-	for ; index < len(keys); index++ {
-		f(keys[index], tree.Get(keys[index]))
-	}
+	tree.lazyInit()
+	tree.RedBlackKVTree.IteratorAscFrom(key, match, f)
 }

 // IteratorDesc iterates the tree readonly in descending order with given callback function `f`.
 //
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *RedBlackTree) IteratorDesc(f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var (
-		ok bool
-		it = tree.tree.Iterator()
-	)
-	for it.End(); it.Prev(); {
-		index, value := it.Key(), it.Value()
-		if ok = f(index, value); !ok {
-			break
-		}
-	}
+	tree.lazyInit()
+	tree.RedBlackKVTree.IteratorDesc(f)
 }

 // IteratorDescFrom iterates the tree readonly in descending order with given callback function `f`.
@ -405,44 +313,20 @@ func (tree *RedBlackTree) IteratorDesc(f func(key, value any) bool) {
 // searching iterating.
 // If callback function `f` returns true, then it continues iterating; or false to stop.
 func (tree *RedBlackTree) IteratorDescFrom(key any, match bool, f func(key, value any) bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	var keys = tree.tree.Keys()
-	index, canIterator := iteratorFromGetIndex(key, keys, match)
-	if !canIterator {
-		return
-	}
-	for ; index >= 0; index-- {
-		f(keys[index], tree.Get(keys[index]))
-	}
+	tree.lazyInit()
+	tree.RedBlackKVTree.IteratorDescFrom(key, match, f)
 }

 // Left returns the minimum element corresponding to the comparator of the tree or nil if the tree is empty.
 func (tree *RedBlackTree) Left() *RedBlackTreeNode {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	node := tree.tree.Left()
-	if node == nil {
-		return nil
-	}
-	return &RedBlackTreeNode{
-		Key:   node.Key,
-		Value: node.Value,
-	}
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Left()
 }

 // Right returns the maximum element corresponding to the comparator of the tree or nil if the tree is empty.
 func (tree *RedBlackTree) Right() *RedBlackTreeNode {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	node := tree.tree.Right()
-	if node == nil {
-		return nil
-	}
-	return &RedBlackTreeNode{
-		Key:   node.Key,
-		Value: node.Value,
-	}
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Right()
 }

 // Floor Finds floor node of the input key, returns the floor node or nil if no floor node is found.
@ -454,16 +338,8 @@ func (tree *RedBlackTree) Right() *RedBlackTreeNode {
 //
 // Key should adhere to the comparator's type assertion, otherwise method panics.
 func (tree *RedBlackTree) Floor(key any) (floor *RedBlackTreeNode, found bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	node, found := tree.tree.Floor(key)
-	if !found {
-		return nil, false
-	}
-	return &RedBlackTreeNode{
-		Key:   node.Key,
-		Value: node.Value,
-	}, true
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Floor(key)
 }

 // Ceiling finds ceiling node of the input key, returns the ceiling node or nil if no ceiling node is found.
@ -475,16 +351,8 @@ func (tree *RedBlackTree) Floor(key any) (floor *RedBlackTreeNode, found bool) {
 //
 // Key should adhere to the comparator's type assertion, otherwise method panics.
 func (tree *RedBlackTree) Ceiling(key any) (ceiling *RedBlackTreeNode, found bool) {
-	tree.mu.RLock()
-	defer tree.mu.RUnlock()
-	node, found := tree.tree.Ceiling(key)
-	if !found {
-		return nil, false
-	}
-	return &RedBlackTreeNode{
-		Key:   node.Key,
-		Value: node.Value,
-	}, true
+	tree.lazyInit()
+	return tree.RedBlackKVTree.Ceiling(key)
 }

 // Flip exchanges key-value of the tree to value-key.
@ -493,6 +361,7 @@ func (tree *RedBlackTree) Ceiling(key any) (ceiling *RedBlackTreeNode, found boo
 //
 // If the type of value is different with key, you pass the new `comparator`.
 func (tree *RedBlackTree) Flip(comparator ...func(v1, v2 any) int) {
+	tree.lazyInit()
 	var t = new(RedBlackTree)
 	if len(comparator) > 0 {
 		t = NewRedBlackTree(comparator[0], tree.mu.IsSafe())
@ -509,61 +378,12 @@ func (tree *RedBlackTree) Flip(comparator ...func(v1, v2 any) int) {

 // UnmarshalJSON implements the interface UnmarshalJSON for json.Unmarshal.
 func (tree *RedBlackTree) UnmarshalJSON(b []byte) error {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if tree.comparator == nil {
-		tree.comparator = gutil.ComparatorString
-		tree.tree = redblacktree.NewWith(tree.comparator)
-	}
-	var data map[string]any
-	if err := json.UnmarshalUseNumber(b, &data); err != nil {
-		return err
-	}
-	for k, v := range data {
-		tree.doSet(k, v)
-	}
-	return nil
+	tree.lazyInit()
+	return tree.RedBlackKVTree.UnmarshalJSON(b)
 }

 // UnmarshalValue is an interface implement which sets any type of value for map.
 func (tree *RedBlackTree) UnmarshalValue(value any) (err error) {
-	tree.mu.Lock()
-	defer tree.mu.Unlock()
-	if tree.comparator == nil {
-		tree.comparator = gutil.ComparatorString
-		tree.tree = redblacktree.NewWith(tree.comparator)
-	}
-	for k, v := range gconv.Map(value) {
-		tree.doSet(k, v)
-	}
-	return
-}
-
-// doSet inserts key-value pair node into the tree without lock.
-// If `key` already exists, then its value is updated with the new value.
-// If `value` is type of <func() any>, it will be executed and its return value will be set to the map with `key`.
-//
-// It returns value with given `key`.
-func (tree *RedBlackTree) doSet(key, value any) any {
-	if f, ok := value.(func() any); ok {
-		value = f()
-	}
-	if value == nil {
-		return value
-	}
-	tree.tree.Put(key, value)
-	return value
-}
-
-// doGet retrieves and returns the value of given key from tree without lock.
-func (tree *RedBlackTree) doGet(key any) (value any, found bool) {
-	return tree.tree.Get(key)
-}
-
-// doRemove removes key from tree and returns its associated value without lock.
-// Note that, the given `key` should adhere to the comparator's type assertion, otherwise method panics.
-func (tree *RedBlackTree) doRemove(key any) (value any) {
-	value, _ = tree.tree.Get(key)
-	tree.tree.Remove(key)
-	return
+	tree.lazyInit()
+	return tree.RedBlackKVTree.UnmarshalValue(value)
 }
--- a/container/gvar/gvar_slice.go
+++ b/container/gvar/gvar_slice.go
@ -8,6 +8,11 @@ package gvar

 import "github.com/gogf/gf/v2/util/gconv"

+// Bools converts and returns `v` as []bool.
+func (v *Var) Bools() []bool {
+	return gconv.Bools(v.Val())
+}
+
 // Ints converts and returns `v` as []int.
 func (v *Var) Ints() []int {
 	return gconv.Ints(v.Val())
--- a/container/gvar/gvar_z_unit_slice_test.go
+++ b/container/gvar/gvar_z_unit_slice_test.go
@ -21,6 +21,24 @@ func TestVar_Ints(t *testing.T) {
 	})
 }

+func TestVar_Bools(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		var arr = []bool{true, false, true, false}
+		objOne := gvar.New(arr, true)
+		t.AssertEQ(objOne.Bools(), arr)
+	})
+	gtest.C(t, func(t *gtest.T) {
+		var arr = []int{1, 0, 1, 0}
+		objOne := gvar.New(arr, true)
+		t.AssertEQ(objOne.Bools(), []bool{true, false, true, false})
+	})
+	gtest.C(t, func(t *gtest.T) {
+		var arr = []string{"true", "false", "1", "0"}
+		objOne := gvar.New(arr, true)
+		t.AssertEQ(objOne.Bools(), []bool{true, false, true, false})
+	})
+}
+
 func TestVar_Uints(t *testing.T) {
 	gtest.C(t, func(t *gtest.T) {
 		var arr = []int{1, 2, 3, 4, 5}
--- a/contrib/config/apollo/go.mod
+++ b/contrib/config/apollo/go.mod
@ -4,13 +4,13 @@ go 1.23.0

 require (
 	github.com/apolloconfig/agollo/v4 v4.3.1
-	github.com/gogf/gf/v2 v2.9.5
+	github.com/gogf/gf/v2 v2.9.7
 )

 require (
 	github.com/BurntSushi/toml v1.5.0 // indirect
 	github.com/clbanning/mxj/v2 v2.7.0 // indirect
-	github.com/emirpasic/gods v1.18.1 // indirect
+	github.com/emirpasic/gods/v2 v2.0.0-alpha // indirect
 	github.com/fatih/color v1.18.0 // indirect
 	github.com/fsnotify/fsnotify v1.9.0 // indirect
 	github.com/go-logr/logr v1.4.3 // indirect
--- a/contrib/config/apollo/go.sum
+++ b/contrib/config/apollo/go.sum
@ -64,8 +64,8 @@ github.com/coreos/go-systemd/v22 v22.3.2/go.mod h1:Y58oyj3AT4RCenI/lSvhwexgC+NSV
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
-github.com/emirpasic/gods v1.18.1 h1:FXtiHYKDGKCW2KzwZKx0iC0PQmdlorYgdFG9jPXJ1Bc=
-github.com/emirpasic/gods v1.18.1/go.mod h1:8tpGGwCnJ5H4r6BWwaV6OrWmMoPhUl5jm/FMNAnJvWQ=
+github.com/emirpasic/gods/v2 v2.0.0-alpha h1:dwFlh8pBg1VMOXWGipNMRt8v96dKAIvBehtCt6OtunU=
+github.com/emirpasic/gods/v2 v2.0.0-alpha/go.mod h1:W0y4M2dtBB9U5z3YlghmpuUhiaZT2h6yoeE+C1sCp6A=
 github.com/envoyproxy/go-control-plane v0.9.0/go.mod h1:YTl/9mNaCwkRvm6d1a2C3ymFceY/DCBVvsKhRF0iEA4=
 github.com/envoyproxy/go-control-plane v0.9.1-0.20191026205805-5f8ba28d4473/go.mod h1:YTl/9mNaCwkRvm6d1a2C3ymFceY/DCBVvsKhRF0iEA4=
 github.com/envoyproxy/go-control-plane v0.9.4/go.mod h1:6rpuAdCZL397s3pYoYcLgu1mIlRU8Am5FuJP05cCM98=
--- a/contrib/config/consul/go.mod
+++ b/contrib/config/consul/go.mod
@ -3,7 +3,7 @@ module github.com/gogf/gf/contrib/config/consul/v2
 go 1.23.0

 require (
-	github.com/gogf/gf/v2 v2.9.5
+	github.com/gogf/gf/v2 v2.9.7
 	github.com/hashicorp/consul/api v1.24.0
 	github.com/hashicorp/go-cleanhttp v0.5.2
 )
@ -12,7 +12,7 @@ require (
 	github.com/BurntSushi/toml v1.5.0 // indirect
 	github.com/armon/go-metrics v0.4.1 // indirect
 	github.com/clbanning/mxj/v2 v2.7.0 // indirect
-	github.com/emirpasic/gods v1.18.1 // indirect
+	github.com/emirpasic/gods/v2 v2.0.0-alpha // indirect
 	github.com/fatih/color v1.18.0 // indirect
 	github.com/fsnotify/fsnotify v1.9.0 // indirect
 	github.com/go-logr/logr v1.4.3 // indirect
--- a/contrib/config/consul/go.sum
+++ b/contrib/config/consul/go.sum
@ -24,8 +24,8 @@ github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSs
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc h1:U9qPSI2PIWSS1VwoXQT9A3Wy9MM3WgvqSxFWenqJduM=
 github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
-github.com/emirpasic/gods v1.18.1 h1:FXtiHYKDGKCW2KzwZKx0iC0PQmdlorYgdFG9jPXJ1Bc=
-github.com/emirpasic/gods v1.18.1/go.mod h1:8tpGGwCnJ5H4r6BWwaV6OrWmMoPhUl5jm/FMNAnJvWQ=
+github.com/emirpasic/gods/v2 v2.0.0-alpha h1:dwFlh8pBg1VMOXWGipNMRt8v96dKAIvBehtCt6OtunU=
+github.com/emirpasic/gods/v2 v2.0.0-alpha/go.mod h1:W0y4M2dtBB9U5z3YlghmpuUhiaZT2h6yoeE+C1sCp6A=
 github.com/fatih/color v1.7.0/go.mod h1:Zm6kSWBoL9eyXnKyktHP6abPY2pDugNf5KwzbycvMj4=
 github.com/fatih/color v1.9.0/go.mod h1:eQcE1qtQxscV5RaZvpXrrb8Drkc3/DdQ+uUYCNjL+zU=
 github.com/fatih/color v1.13.0/go.mod h1:kLAiJbzzSOZDVNGyDpeOxJ47H46qBXwg5ILebYFFOfk=
--- a/contrib/config/kubecm/go.mod
+++ b/contrib/config/kubecm/go.mod
@ -3,7 +3,7 @@ module github.com/gogf/gf/contrib/config/kubecm/v2
 go 1.24.0

 require (
-	github.com/gogf/gf/v2 v2.9.5
+	github.com/gogf/gf/v2 v2.9.7
 	k8s.io/api v0.33.4
 	k8s.io/apimachinery v0.33.4
 	k8s.io/client-go v0.33.4
@ -14,7 +14,7 @@ require (
 	github.com/clbanning/mxj/v2 v2.7.0 // indirect
 	github.com/davecgh/go-spew v1.1.1 // indirect
 	github.com/emicklei/go-restful/v3 v3.11.0 // indirect
-	github.com/emirpasic/gods v1.18.1 // indirect
+	github.com/emirpasic/gods/v2 v2.0.0-alpha // indirect
 	github.com/fatih/color v1.18.0 // indirect
 	github.com/fsnotify/fsnotify v1.9.0 // indirect
 	github.com/fxamacker/cbor/v2 v2.7.0 // indirect
--- a/contrib/config/kubecm/go.sum
+++ b/contrib/config/kubecm/go.sum
@ -8,8 +8,8 @@ github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/emicklei/go-restful/v3 v3.11.0 h1:rAQeMHw1c7zTmncogyy8VvRZwtkmkZ4FxERmMY4rD+g=
 github.com/emicklei/go-restful/v3 v3.11.0/go.mod h1:6n3XBCmQQb25CM2LCACGz8ukIrRry+4bhvbpWn3mrbc=
-github.com/emirpasic/gods v1.18.1 h1:FXtiHYKDGKCW2KzwZKx0iC0PQmdlorYgdFG9jPXJ1Bc=
-github.com/emirpasic/gods v1.18.1/go.mod h1:8tpGGwCnJ5H4r6BWwaV6OrWmMoPhUl5jm/FMNAnJvWQ=
+github.com/emirpasic/gods/v2 v2.0.0-alpha h1:dwFlh8pBg1VMOXWGipNMRt8v96dKAIvBehtCt6OtunU=
+github.com/emirpasic/gods/v2 v2.0.0-alpha/go.mod h1:W0y4M2dtBB9U5z3YlghmpuUhiaZT2h6yoeE+C1sCp6A=
 github.com/fatih/color v1.18.0 h1:S8gINlzdQ840/4pfAwic/ZE0djQEH3wM94VfqLTZcOM=
 github.com/fatih/color v1.18.0/go.mod h1:4FelSpRwEGDpQ12mAdzqdOukCy4u8WUtOY6lkT/6HfU=
 github.com/fsnotify/fsnotify v1.9.0 h1:2Ml+OJNzbYCTzsxtv8vKSFD9PbJjmhYF14k/jKC7S9k=
--- a/contrib/config/nacos/go.mod
+++ b/contrib/config/nacos/go.mod
@ -3,7 +3,7 @@ module github.com/gogf/gf/contrib/config/nacos/v2
 go 1.23.0

 require (
-	github.com/gogf/gf/v2 v2.9.5
+	github.com/gogf/gf/v2 v2.9.7
 	github.com/nacos-group/nacos-sdk-go/v2 v2.3.3
 )

@ -35,7 +35,7 @@ require (
 	github.com/cespare/xxhash/v2 v2.3.0 // indirect
 	github.com/clbanning/mxj/v2 v2.7.0 // indirect
 	github.com/deckarep/golang-set v1.7.1 // indirect
-	github.com/emirpasic/gods v1.18.1 // indirect
+	github.com/emirpasic/gods/v2 v2.0.0-alpha // indirect
 	github.com/fatih/color v1.18.0 // indirect
 	github.com/fsnotify/fsnotify v1.9.0 // indirect
 	github.com/go-logr/logr v1.4.3 // indirect
--- a/contrib/config/nacos/go.sum
+++ b/contrib/config/nacos/go.sum
@ -127,8 +127,8 @@ github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/deckarep/golang-set v1.7.1 h1:SCQV0S6gTtp6itiFrTqI+pfmJ4LN85S1YzhDf9rTHJQ=
 github.com/deckarep/golang-set v1.7.1/go.mod h1:93vsz/8Wt4joVM7c2AVqh+YRMiUSc14yDtF28KmMOgQ=
-github.com/emirpasic/gods v1.18.1 h1:FXtiHYKDGKCW2KzwZKx0iC0PQmdlorYgdFG9jPXJ1Bc=
-github.com/emirpasic/gods v1.18.1/go.mod h1:8tpGGwCnJ5H4r6BWwaV6OrWmMoPhUl5jm/FMNAnJvWQ=
+github.com/emirpasic/gods/v2 v2.0.0-alpha h1:dwFlh8pBg1VMOXWGipNMRt8v96dKAIvBehtCt6OtunU=
+github.com/emirpasic/gods/v2 v2.0.0-alpha/go.mod h1:W0y4M2dtBB9U5z3YlghmpuUhiaZT2h6yoeE+C1sCp6A=
 github.com/envoyproxy/go-control-plane v0.9.0/go.mod h1:YTl/9mNaCwkRvm6d1a2C3ymFceY/DCBVvsKhRF0iEA4=
 github.com/envoyproxy/go-control-plane v0.9.1-0.20191026205805-5f8ba28d4473/go.mod h1:YTl/9mNaCwkRvm6d1a2C3ymFceY/DCBVvsKhRF0iEA4=
 github.com/envoyproxy/go-control-plane v0.9.4/go.mod h1:6rpuAdCZL397s3pYoYcLgu1mIlRU8Am5FuJP05cCM98=
--- a/contrib/config/polaris/go.mod
+++ b/contrib/config/polaris/go.mod
@ -3,7 +3,7 @@ module github.com/gogf/gf/contrib/config/polaris/v2
 go 1.23.0

 require (
-	github.com/gogf/gf/v2 v2.9.5
+	github.com/gogf/gf/v2 v2.9.7
 	github.com/polarismesh/polaris-go v1.6.1
 )

@ -13,7 +13,7 @@ require (
 	github.com/cespare/xxhash/v2 v2.1.2 // indirect
 	github.com/clbanning/mxj/v2 v2.7.0 // indirect
 	github.com/dlclark/regexp2 v1.7.0 // indirect
-	github.com/emirpasic/gods v1.18.1 // indirect
+	github.com/emirpasic/gods/v2 v2.0.0-alpha // indirect
 	github.com/fatih/color v1.18.0 // indirect
 	github.com/fsnotify/fsnotify v1.9.0 // indirect
 	github.com/go-logr/logr v1.4.3 // indirect
--- a/contrib/config/polaris/go.sum
+++ b/contrib/config/polaris/go.sum
@ -210,8 +210,8 @@ github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/dlclark/regexp2 v1.7.0 h1:7lJfhqlPssTb1WQx4yvTHN0uElPEv52sbaECrAQxjAo=
 github.com/dlclark/regexp2 v1.7.0/go.mod h1:DHkYz0B9wPfa6wondMfaivmHpzrQ3v9q8cnmRbL6yW8=
-github.com/emirpasic/gods v1.18.1 h1:FXtiHYKDGKCW2KzwZKx0iC0PQmdlorYgdFG9jPXJ1Bc=
-github.com/emirpasic/gods v1.18.1/go.mod h1:8tpGGwCnJ5H4r6BWwaV6OrWmMoPhUl5jm/FMNAnJvWQ=
+github.com/emirpasic/gods/v2 v2.0.0-alpha h1:dwFlh8pBg1VMOXWGipNMRt8v96dKAIvBehtCt6OtunU=
+github.com/emirpasic/gods/v2 v2.0.0-alpha/go.mod h1:W0y4M2dtBB9U5z3YlghmpuUhiaZT2h6yoeE+C1sCp6A=
 github.com/envoyproxy/go-control-plane v0.9.0/go.mod h1:YTl/9mNaCwkRvm6d1a2C3ymFceY/DCBVvsKhRF0iEA4=
 github.com/envoyproxy/go-control-plane v0.9.1-0.20191026205805-5f8ba28d4473/go.mod h1:YTl/9mNaCwkRvm6d1a2C3ymFceY/DCBVvsKhRF0iEA4=
 github.com/envoyproxy/go-control-plane v0.9.4/go.mod h1:6rpuAdCZL397s3pYoYcLgu1mIlRU8Am5FuJP05cCM98=
--- a/contrib/drivers/README.MD
+++ b/contrib/drivers/README.MD
@ -1,4 +1,3 @@
-[简体中文](README.zh_CN.MD)

 # Database drivers

@ -9,15 +8,20 @@ Powerful database drivers for package gdb.
 Let's take `mysql` for example.

 ```shell
-go get -u github.com/gogf/gf/contrib/drivers/mysql/v2
-# Easy to copy
-go get -u github.com/gogf/gf/contrib/drivers/clickhouse/v2
-go get -u github.com/gogf/gf/contrib/drivers/dm/v2
-go get -u github.com/gogf/gf/contrib/drivers/mssql/v2
-go get -u github.com/gogf/gf/contrib/drivers/oracle/v2
-go get -u github.com/gogf/gf/contrib/drivers/pgsql/v2
-go get -u github.com/gogf/gf/contrib/drivers/sqlite/v2
-go get -u github.com/gogf/gf/contrib/drivers/sqlitecgo/v2
+go get github.com/gogf/gf/contrib/drivers/mysql/v2@latest
+
+# Easy for copying:
+go get github.com/gogf/gf/contrib/drivers/clickhouse/v2@latest
+go get github.com/gogf/gf/contrib/drivers/dm/v2@latest
+go get github.com/gogf/gf/contrib/drivers/gaussdb/v2@latest
+go get github.com/gogf/gf/contrib/drivers/mariadb/v2@latest
+go get github.com/gogf/gf/contrib/drivers/mssql/v2@latest
+go get github.com/gogf/gf/contrib/drivers/oceanbase/v2@latest
+go get github.com/gogf/gf/contrib/drivers/oracle/v2@latest
+go get github.com/gogf/gf/contrib/drivers/pgsql/v2@latest
+go get github.com/gogf/gf/contrib/drivers/sqlite/v2@latest
+go get github.com/gogf/gf/contrib/drivers/sqlitecgo/v2@latest
+go get github.com/gogf/gf/contrib/drivers/tidb/v2@latest
 ```

 Choose and import the driver to your project:
@ -44,12 +48,36 @@ func main() {

 ## Supported Drivers

-### MySQL/MariaDB/TiDB
+### MySQL

 ```go
 import _ "github.com/gogf/gf/contrib/drivers/mysql/v2"
 ```

+### MariaDB
+
+```go
+import _ "github.com/gogf/gf/contrib/drivers/mariadb/v2"
+```
+
+### TiDB
+
+```go
+import _ "github.com/gogf/gf/contrib/drivers/tidb/v2"
+```
+
+### OceanBase
+
+```go
+import _ "github.com/gogf/gf/contrib/drivers/oceanbase/v2"
+```
+
+### GaussDB
+
+```go
+import _ "github.com/gogf/gf/contrib/drivers/gaussdb/v2"
+```
+
 ### SQLite

 ```go
@ -58,7 +86,7 @@ import _ "github.com/gogf/gf/contrib/drivers/sqlite/v2"

 #### cgo version

-When the target is a 32-bit Windows system, the cgo version needs to be used.
+When the target is a `32-bit` Windows system, the `cgo` version needs to be used.

 ```go
 import _ "github.com/gogf/gf/contrib/drivers/sqlitecgo/v2"
@ -70,10 +98,6 @@ import _ "github.com/gogf/gf/contrib/drivers/sqlitecgo/v2"
 import _ "github.com/gogf/gf/contrib/drivers/pgsql/v2"
 ```

-Note:
-
- It does not support `Replace` features.
-
 ### SQL Server

 ```go
@ -82,9 +106,10 @@ import _ "github.com/gogf/gf/contrib/drivers/mssql/v2"

 Note:

- It does not support `Replace` features.
+- `InsertIgnore` returns error if there is no primary key or unique index submitted with record.
 - It supports server version >= `SQL Server2005`
- It ONLY supports datetime2 and datetimeoffset types for auto handling created_at/updated_at/deleted_at columns, because datetime type does not support microseconds precision when column value is passed as string.
+- It ONLY supports `datetime2` and `datetimeoffset` types for auto handling created_at/updated_at/deleted_at columns, 
+  because datetime type does not support microseconds precision when column value is passed as string.

 ### Oracle

@ -94,8 +119,8 @@ import _ "github.com/gogf/gf/contrib/drivers/oracle/v2"

 Note:

- It does not support `Replace` features.
 - It does not support `LastInsertId`.
+- `InsertIgnore` returns error if there is no primary key or unique index submitted with record.

 ### ClickHouse

@ -105,7 +130,7 @@ import _ "github.com/gogf/gf/contrib/drivers/clickhouse/v2"

 Note:

- It does not support `InsertIgnore/InsertGetId` features.
+- It does not support `InsertIgnore/InsertAndGetId` features.
 - It does not support `Save/Replace` features.
 - It does not support `Transaction` feature.
 - It does not support `RowsAffected` feature.
@ -118,7 +143,7 @@ import _ "github.com/gogf/gf/contrib/drivers/dm/v2"

 Note:

- It does not support `Replace` features.
+- `InsertIgnore` returns error if there is no primary key or unique index submitted with record.

 ## Custom Drivers

--- a/contrib/drivers/README.zh_CN.MD
+++ b/contrib/drivers/README.zh_CN.MD
@ -1,124 +0,0 @@
-# 数据库驱动程序
-
-用于gdb包的数据库驱动程序。
-
-## 安装
-
-以 `mysql` 为例。
-
-```shell
-go get -u github.com/gogf/gf/contrib/drivers/mysql/v2
-# 方便复制
-go get -u github.com/gogf/gf/contrib/drivers/clickhouse/v2
-go get -u github.com/gogf/gf/contrib/drivers/dm/v2
-go get -u github.com/gogf/gf/contrib/drivers/mssql/v2
-go get -u github.com/gogf/gf/contrib/drivers/oracle/v2
-go get -u github.com/gogf/gf/contrib/drivers/pgsql/v2
-go get -u github.com/gogf/gf/contrib/drivers/sqlite/v2
-go get -u github.com/gogf/gf/contrib/drivers/sqlitecgo/v2
-```
-
-选择并将驱动程序导入到您的项目中：
-
-```go
-import _ "github.com/gogf/gf/contrib/drivers/mysql/v2"
-```
-
-通常在 `main.go` 的顶部导入：
-
-```go
-package main
-
-import (
-	_ "github.com/gogf/gf/contrib/drivers/mysql/v2"
-
-	// 其他导入的包。
-)
-
-func main() {
-	// 主要逻辑。
-}
-```
-
-## 支持的驱动程序
-
-### MySQL/MariaDB/TiDB
-
-```go
-import _ "github.com/gogf/gf/contrib/drivers/mysql/v2"
-```
-
-### SQLite
-
-```go
-import _ "github.com/gogf/gf/contrib/drivers/sqlite/v2"
-```
-
-#### cgo 版本
-
-32位Windows请使用cgo版本
-
-```go
-import _ "github.com/gogf/gf/contrib/drivers/sqlitecgo/v2"
-```
-
-### PostgreSQL
-
-```go
-import _ "github.com/gogf/gf/contrib/drivers/pgsql/v2"
-```
-
-注意：
-
- 不支持 `Replace` 功能。
-
-### SQL Server
-
-```go
-import _ "github.com/gogf/gf/contrib/drivers/mssql/v2"
-```
-
-注意：
-
- 不支持 `Replace` 功能。
- 仅支持服务器版本 >= `SQL Server2005`
- 仅支持 datetime2 和 datetimeoffset 类型来自动处理 created_at/updated_at/deleted_at 列，因为 datetime 类型在将列值作为字符串传递时不支持微秒精度。
-
-### Oracle
-
-```go
-import _ "github.com/gogf/gf/contrib/drivers/oracle/v2"
-```
-
-注意：
-
- 不支持 `Replace` 功能。
- 不支持 `LastInsertId`。
-
-### ClickHouse
-
-```go
-import _ "github.com/gogf/gf/contrib/drivers/clickhouse/v2"
-```
-
-注意：
-
- 不支持 `InsertIgnore/InsertGetId` 功能。
- 不支持 `Save/Replace` 功能。
- 不支持 `Transaction` 功能。
- 不支持 `RowsAffected` 功能。
-
-### DM
-
-```go
-import _ "github.com/gogf/gf/contrib/drivers/dm/v2"
-```
-
-注意：
-
- 不支持 `Replace` 功能。
-
-## 自定义驱动程序
-
-自定义驱动程序非常快速和简单，您可以参考当前驱动程序的源代码来进行开发。
-如果您有关于支持新驱动程序的PR（Pull Request），我们将非常感激地接受您的提交到当前仓库。
--- a/contrib/drivers/clickhouse/clickhouse_do_insert.go
+++ b/contrib/drivers/clickhouse/clickhouse_do_insert.go
@ -16,6 +16,7 @@ import (
 )

 // DoInsert inserts or updates data for given table.
+// The list parameter must contain at least one record, which was previously validated.
 func (d *Driver) DoInsert(
 	ctx context.Context, link gdb.Link, table string, list gdb.List, option gdb.DoInsertOption,
 ) (result sql.Result, err error) {
--- a/contrib/drivers/clickhouse/go.mod
+++ b/contrib/drivers/clickhouse/go.mod
@ -4,7 +4,7 @@ go 1.23.0

 require (
 	github.com/ClickHouse/clickhouse-go/v2 v2.0.15
-	github.com/gogf/gf/v2 v2.9.5
+	github.com/gogf/gf/v2 v2.9.7
 	github.com/google/uuid v1.6.0
 	github.com/shopspring/decimal v1.3.1
 )
@ -12,7 +12,7 @@ require (
 require (
 	github.com/BurntSushi/toml v1.5.0 // indirect
 	github.com/clbanning/mxj/v2 v2.7.0 // indirect
-	github.com/emirpasic/gods v1.18.1 // indirect
+	github.com/emirpasic/gods/v2 v2.0.0-alpha // indirect
 	github.com/fatih/color v1.18.0 // indirect
 	github.com/fsnotify/fsnotify v1.9.0 // indirect
 	github.com/go-logr/logr v1.4.3 // indirect
--- a/contrib/drivers/clickhouse/go.sum
+++ b/contrib/drivers/clickhouse/go.sum
@ -9,8 +9,8 @@ github.com/cloudflare/golz4 v0.0.0-20150217214814-ef862a3cdc58/go.mod h1:EOBUe0h
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
-github.com/emirpasic/gods v1.18.1 h1:FXtiHYKDGKCW2KzwZKx0iC0PQmdlorYgdFG9jPXJ1Bc=
-github.com/emirpasic/gods v1.18.1/go.mod h1:8tpGGwCnJ5H4r6BWwaV6OrWmMoPhUl5jm/FMNAnJvWQ=
+github.com/emirpasic/gods/v2 v2.0.0-alpha h1:dwFlh8pBg1VMOXWGipNMRt8v96dKAIvBehtCt6OtunU=
+github.com/emirpasic/gods/v2 v2.0.0-alpha/go.mod h1:W0y4M2dtBB9U5z3YlghmpuUhiaZT2h6yoeE+C1sCp6A=
 github.com/fatih/color v1.18.0 h1:S8gINlzdQ840/4pfAwic/ZE0djQEH3wM94VfqLTZcOM=
 github.com/fatih/color v1.18.0/go.mod h1:4FelSpRwEGDpQ12mAdzqdOukCy4u8WUtOY6lkT/6HfU=
 github.com/fsnotify/fsnotify v1.9.0 h1:2Ml+OJNzbYCTzsxtv8vKSFD9PbJjmhYF14k/jKC7S9k=
--- a/contrib/drivers/dm/dm.go
+++ b/contrib/drivers/dm/dm.go
@ -14,6 +14,7 @@ import (
 	"github.com/gogf/gf/v2/frame/g"
 )

+// Driver is the driver for dm database.
 type Driver struct {
 	*gdb.Core
 }
--- a/contrib/drivers/dm/dm_do_filter.go
+++ b/contrib/drivers/dm/dm_do_filter.go
@ -20,7 +20,7 @@ func (d *Driver) DoFilter(
 	ctx context.Context, link gdb.Link, sql string, args []any,
 ) (newSql string, newArgs []any, err error) {
 	// There should be no need to capitalize, because it has been done from field processing before
-	newSql, _ = gregex.ReplaceString(`["\n\t]`, "", sql)
+	newSql, _ = gregex.ReplaceString(`["]`, "", sql)
 	newSql = gstr.ReplaceI(gstr.ReplaceI(newSql, "GROUP_CONCAT", "LISTAGG"), "SEPARATOR", ",")

 	// TODO The current approach is too rough. We should deal with the GROUP_CONCAT function and the
--- a/contrib/drivers/dm/dm_do_insert.go
+++ b/contrib/drivers/dm/dm_do_insert.go
@ -20,6 +20,7 @@ import (
 )

 // DoInsert inserts or updates data for given table.
+// The list parameter must contain at least one record, which was previously validated.
 func (d *Driver) DoInsert(
 	ctx context.Context, link gdb.Link, table string, list gdb.List, option gdb.DoInsertOption,
 ) (result sql.Result, err error) {
@ -28,44 +29,92 @@ func (d *Driver) DoInsert(
 		return d.doSave(ctx, link, table, list, option)

 	case gdb.InsertOptionReplace:
-		// TODO:: Should be Supported
-		return nil, gerror.NewCode(
-			gcode.CodeNotSupported, `Replace operation is not supported by dm driver`,
-		)
-	}
+		// dm does not support REPLACE INTO syntax, use SAVE instead.
+		return d.doSave(ctx, link, table, list, option)

-	return d.Core.DoInsert(ctx, link, table, list, option)
+	case gdb.InsertOptionIgnore:
+		// dm does not support INSERT IGNORE syntax, use MERGE instead.
+		return d.doInsertIgnore(ctx, link, table, list, option)
+
+	default:
+		// DM database supports IDENTITY auto-increment columns natively.
+		// The driver automatically returns LastInsertId through sql.Result.
+		//
+		// Note: DM IDENTITY columns cannot accept explicit ID values unless
+		// IDENTITY_INSERT is enabled. When using tables with IDENTITY columns,
+		// avoid providing explicit ID values in the data.
+		return d.Core.DoInsert(ctx, link, table, list, option)
+	}
 }

 // doSave support upsert for dm
 func (d *Driver) doSave(ctx context.Context,
 	link gdb.Link, table string, list gdb.List, option gdb.DoInsertOption,
 ) (result sql.Result, err error) {
-	if len(option.OnConflict) == 0 {
-		return nil, gerror.NewCode(
-			gcode.CodeMissingParameter, `Please specify conflict columns`,
-		)
-	}
+	return d.doMergeInsert(ctx, link, table, list, option, true)
+}

-	if len(list) == 0 {
-		return nil, gerror.NewCode(
-			gcode.CodeInvalidRequest, `Save operation list is empty by oracle driver`,
-		)
+// doInsertIgnore implements INSERT IGNORE operation using MERGE statement for DM database.
+// It only inserts records when there's no conflict on primary/unique keys.
+func (d *Driver) doInsertIgnore(ctx context.Context,
+	link gdb.Link, table string, list gdb.List, option gdb.DoInsertOption,
+) (result sql.Result, err error) {
+	return d.doMergeInsert(ctx, link, table, list, option, false)
+}
+
+// doMergeInsert implements MERGE-based insert operations for DM database.
+// When withUpdate is true, it performs upsert (insert or update).
+// When withUpdate is false, it performs insert ignore (insert only when no conflict).
+func (d *Driver) doMergeInsert(
+	ctx context.Context,
+	link gdb.Link, table string, list gdb.List, option gdb.DoInsertOption, withUpdate bool,
+) (result sql.Result, err error) {
+	// If OnConflict is not specified, automatically get the primary key of the table
+	conflictKeys := option.OnConflict
+	if len(conflictKeys) == 0 {
+		primaryKeys, err := d.Core.GetPrimaryKeys(ctx, table)
+		if err != nil {
+			return nil, gerror.WrapCode(
+				gcode.CodeInternalError,
+				err,
+				`failed to get primary keys for table`,
+			)
+		}
+		foundPrimaryKey := false
+		for _, primaryKey := range primaryKeys {
+			for dataKey := range list[0] {
+				if strings.EqualFold(dataKey, primaryKey) {
+					foundPrimaryKey = true
+					break
+				}
+			}
+			if foundPrimaryKey {
+				break
+			}
+		}
+		if !foundPrimaryKey {
+			return nil, gerror.NewCodef(
+				gcode.CodeMissingParameter,
+				`Replace/Save/InsertIgnore operation requires conflict detection: `+
+					`either specify OnConflict() columns or ensure table '%s' has a primary key in the data`,
+				table,
+			)
+		}
+		// TODO consider composite primary keys.
+		conflictKeys = primaryKeys
 	}

 	var (
-		one          = list[0]
-		oneLen       = len(one)
-		charL, charR = d.GetChars()
-
-		conflictKeys   = option.OnConflict
+		one            = list[0]
+		oneLen         = len(one)
+		charL, charR   = d.GetChars()
 		conflictKeySet = gset.New(false)

-		// queryHolders:	Handle data with Holder that need to be upsert
-		// queryValues:		Handle data that need to be upsert
+		// queryHolders:	Handle data with Holder that need to be merged
+		// queryValues:		Handle data that need to be merged
 		// insertKeys:		Handle valid keys that need to be inserted
 		// insertValues:	Handle values that need to be inserted
-		// updateValues:	Handle values that need to be updated
+		// updateValues:	Handle values that need to be updated (only when withUpdate=true)
 		queryHolders = make([]string, oneLen)
 		queryValues  = make([]any, oneLen)
 		insertKeys   = make([]string, oneLen)
@ -86,9 +135,9 @@ func (d *Driver) doSave(ctx context.Context,
 		insertKeys[index] = keyWithChar
 		insertValues[index] = fmt.Sprintf("T2.%s", keyWithChar)

-		// filter conflict keys in updateValues.
-		// And the key is not a soft created field.
-		if !(conflictKeySet.Contains(key) || d.Core.IsSoftCreatedFieldName(key)) {
+		// Build updateValues only when withUpdate is true
+		// Filter conflict keys and soft created fields from updateValues
+		if withUpdate && !(conflictKeySet.Contains(key) || d.Core.IsSoftCreatedFieldName(key)) {
 			updateValues = append(
 				updateValues,
 				fmt.Sprintf(`T1.%s = T2.%s`, keyWithChar, keyWithChar),
@ -97,8 +146,10 @@ func (d *Driver) doSave(ctx context.Context,
 		index++
 	}

-	batchResult := new(gdb.SqlResult)
-	sqlStr := parseSqlForUpsert(table, queryHolders, insertKeys, insertValues, updateValues, conflictKeys)
+	var (
+		batchResult = new(gdb.SqlResult)
+		sqlStr      = parseSqlForMerge(table, queryHolders, insertKeys, insertValues, updateValues, conflictKeys)
+	)
 	r, err := d.DoExec(ctx, link, sqlStr, queryValues...)
 	if err != nil {
 		return r, err
@ -112,40 +163,40 @@ func (d *Driver) doSave(ctx context.Context,
 	return batchResult, nil
 }

-// parseSqlForUpsert
-// MERGE INTO {{table}} T1
-// USING ( SELECT {{queryHolders}} FROM DUAL T2
-// ON (T1.{{duplicateKey}} = T2.{{duplicateKey}} AND ...)
-// WHEN NOT MATCHED THEN
-// INSERT {{insertKeys}} VALUES {{insertValues}}
-// WHEN MATCHED THEN
-// UPDATE SET {{updateValues}}
-func parseSqlForUpsert(table string,
+// parseSqlForMerge generates MERGE statement for DM database.
+// When updateValues is empty, it only inserts (INSERT IGNORE behavior).
+// When updateValues is provided, it performs upsert (INSERT or UPDATE).
+// Examples:
+// - INSERT IGNORE: MERGE INTO table T1 USING (...) T2 ON (...) WHEN NOT MATCHED THEN INSERT(...) VALUES (...)
+// - UPSERT: MERGE INTO table T1 USING (...) T2 ON (...) WHEN NOT MATCHED THEN INSERT(...) VALUES (...) WHEN MATCHED THEN UPDATE SET ...
+func parseSqlForMerge(table string,
 	queryHolders, insertKeys, insertValues, updateValues, duplicateKey []string,
 ) (sqlStr string) {
 	var (
 		queryHolderStr  = strings.Join(queryHolders, ",")
 		insertKeyStr    = strings.Join(insertKeys, ",")
 		insertValueStr  = strings.Join(insertValues, ",")
-		updateValueStr  = strings.Join(updateValues, ",")
 		duplicateKeyStr string
-		pattern         = gstr.Trim(`MERGE INTO %s T1 USING (SELECT %s FROM DUAL) T2 ON (%s) WHEN NOT MATCHED THEN INSERT(%s) VALUES (%s) WHEN MATCHED THEN UPDATE SET %s;`)
 	)

+	// Build ON condition
 	for index, keys := range duplicateKey {
 		if index != 0 {
 			duplicateKeyStr += " AND "
 		}
-		duplicateTmp := fmt.Sprintf("T1.%s = T2.%s", keys, keys)
-		duplicateKeyStr += duplicateTmp
+		duplicateKeyStr += fmt.Sprintf("T1.%s = T2.%s", keys, keys)
 	}

-	return fmt.Sprintf(pattern,
-		table,
-		queryHolderStr,
-		duplicateKeyStr,
-		insertKeyStr,
-		insertValueStr,
-		updateValueStr,
-	)
+	// Build SQL based on whether UPDATE is needed
+	pattern := gstr.Trim(`MERGE INTO %s T1 USING (SELECT %s FROM DUAL) T2 ON (%s) WHEN NOT MATCHED THEN INSERT(%s) VALUES (%s)`)
+	if len(updateValues) > 0 {
+		// Upsert: INSERT or UPDATE
+		pattern += gstr.Trim(`WHEN MATCHED THEN UPDATE SET %s`)
+		return fmt.Sprintf(
+			pattern, table, queryHolderStr, duplicateKeyStr, insertKeyStr, insertValueStr,
+			strings.Join(updateValues, ","),
+		)
+	}
+	// Insert Ignore: INSERT only
+	return fmt.Sprintf(pattern, table, queryHolderStr, duplicateKeyStr, insertKeyStr, insertValueStr)
 }
--- a/contrib/drivers/dm/dm_table_fields.go
+++ b/contrib/drivers/dm/dm_table_fields.go
@ -11,12 +11,21 @@ import (
 	"fmt"
 	"strings"

+	"github.com/gogf/gf/v2/container/gmap"
 	"github.com/gogf/gf/v2/database/gdb"
 	"github.com/gogf/gf/v2/util/gutil"
 )

+// escapeSingleQuote escapes single quotes in the string to prevent SQL injection.
+// In SQL, single quotes are escaped by doubling them (two single quotes).
+func escapeSingleQuote(s string) string {
+	return strings.ReplaceAll(s, "'", "''")
+}
+
 const (
-	tableFieldsSqlTmp = `SELECT * FROM ALL_TAB_COLUMNS WHERE Table_Name= '%s' AND OWNER = '%s'`
+	tableFieldsSqlTmp         = `SELECT c.COLUMN_NAME, c.DATA_TYPE, c.DATA_LENGTH, c.DATA_DEFAULT, c.NULLABLE, cc.COMMENTS FROM ALL_TAB_COLUMNS c LEFT JOIN ALL_COL_COMMENTS cc ON c.COLUMN_NAME = cc.COLUMN_NAME AND c.TABLE_NAME = cc.TABLE_NAME AND c.OWNER = cc.OWNER WHERE c.TABLE_NAME = '%s' AND c.OWNER = '%s'`
+	tableFieldsPkSqlSchemaTmp = `SELECT COLS.COLUMN_NAME AS PRIMARY_KEY_COLUMN FROM USER_CONSTRAINTS CONS JOIN USER_CONS_COLUMNS COLS ON CONS.CONSTRAINT_NAME = COLS.CONSTRAINT_NAME WHERE CONS.TABLE_NAME = '%s' AND CONS.CONSTRAINT_TYPE = 'P'`
+	tableFieldsPkSqlDBATmp    = `SELECT COLS.COLUMN_NAME AS PRIMARY_KEY_COLUMN FROM DBA_CONSTRAINTS CONS JOIN DBA_CONS_COLUMNS COLS ON CONS.CONSTRAINT_NAME = COLS.CONSTRAINT_NAME WHERE CONS.TABLE_NAME = '%s' AND CONS.OWNER = '%s' AND CONS.CONSTRAINT_TYPE = 'P'`
 )

 // TableFields retrieves and returns the fields' information of specified table of current schema.
@ -24,8 +33,9 @@ func (d *Driver) TableFields(
 	ctx context.Context, table string, schema ...string,
 ) (fields map[string]*gdb.TableField, err error) {
 	var (
-		result gdb.Result
-		link   gdb.Link
+		result   gdb.Result
+		pkResult gdb.Result
+		link     gdb.Link
 		// When no schema is specified, the configuration item is returned by default
 		usedSchema = gutil.GetOrDefaultStr(d.GetSchema(), schema...)
 	)
@ -38,14 +48,35 @@ func (d *Driver) TableFields(
 		ctx, link,
 		fmt.Sprintf(
 			tableFieldsSqlTmp,
-			strings.ToUpper(table),
-			strings.ToUpper(d.GetSchema()),
+			escapeSingleQuote(strings.ToUpper(table)),
+			escapeSingleQuote(strings.ToUpper(d.GetSchema())),
 		),
 	)
 	if err != nil {
 		return nil, err
 	}
+	// Query the primary key field
+	pkResult, err = d.DoSelect(
+		ctx, link,
+		fmt.Sprintf(tableFieldsPkSqlSchemaTmp, escapeSingleQuote(strings.ToUpper(table))),
+	)
+	if err != nil {
+		return nil, err
+	}
+	if pkResult.IsEmpty() {
+		pkResult, err = d.DoSelect(
+			ctx, link,
+			fmt.Sprintf(tableFieldsPkSqlDBATmp, escapeSingleQuote(strings.ToUpper(table)), escapeSingleQuote(strings.ToUpper(d.GetSchema()))),
+		)
+		if err != nil {
+			return nil, err
+		}
+	}
 	fields = make(map[string]*gdb.TableField)
+	pkFields := gmap.NewStrStrMap()
+	for _, pk := range pkResult {
+		pkFields.Set(pk["PRIMARY_KEY_COLUMN"].String(), "PRI")
+	}
 	for i, m := range result {
 		// m[NULLABLE] returns "N" "Y"
 		// "N" means not null
@ -54,15 +85,29 @@ func (d *Driver) TableFields(
 		if m["NULLABLE"].String() != "N" {
 			nullable = true
 		}
+
+		// Build field type with length/precision
+		// For NUMBER(p,s): use DATA_PRECISION and DATA_SCALE
+		// For VARCHAR2/CHAR: use DATA_LENGTH
+		var (
+			fieldType  string
+			dataType   = m["DATA_TYPE"].String()
+			dataLength = m["DATA_LENGTH"].Int()
+		)
+		if dataLength > 0 {
+			fieldType = fmt.Sprintf("%s(%d)", dataType, dataLength)
+		} else {
+			fieldType = dataType
+		}
 		fields[m["COLUMN_NAME"].String()] = &gdb.TableField{
 			Index:   i,
 			Name:    m["COLUMN_NAME"].String(),
-			Type:    m["DATA_TYPE"].String(),
+			Type:    fieldType,
 			Null:    nullable,
 			Default: m["DATA_DEFAULT"].Val(),
-			// Key:     m["Key"].String(),
+			Key:     pkFields.Get(m["COLUMN_NAME"].String()),
 			// Extra:   m["Extra"].String(),
-			// Comment: m["Comment"].String(),
+			Comment: m["COMMENTS"].String(),
 		}
 	}
 	return fields, nil
--- a/contrib/drivers/dm/dm_z_unit_basic_test.go
+++ b/contrib/drivers/dm/dm_z_unit_basic_test.go
@ -7,7 +7,6 @@
 package dm_test

 import (
-	"database/sql"
 	"fmt"
 	"strings"
 	"testing"
@ -28,10 +27,7 @@ func Test_DB_Ping(t *testing.T) {
 }

 func TestTables(t *testing.T) {
-	tables := []string{"A_tables", "A_tables2"}
-	for _, v := range tables {
-		createInitTable(v)
-	}
+	tables := createInitTables(2)
 	gtest.C(t, func(t *gtest.T) {
 		result, err := db.Tables(ctx)
 		gtest.AssertNil(err)
@ -39,7 +35,7 @@ func TestTables(t *testing.T) {
 		for i := 0; i < len(tables); i++ {
 			find := false
 			for j := 0; j < len(result); j++ {
-				if strings.ToUpper(tables[i]) == result[j] {
+				if strings.ToUpper(tables[i]) == strings.ToUpper(result[j]) {
 					find = true
 					break
 				}
@ -52,7 +48,7 @@ func TestTables(t *testing.T) {
 		for i := 0; i < len(tables); i++ {
 			find := false
 			for j := 0; j < len(result); j++ {
-				if strings.ToUpper(tables[i]) == result[j] {
+				if strings.ToUpper(tables[i]) == strings.ToUpper(result[j]) {
 					find = true
 					break
 				}
@ -83,17 +79,14 @@ func TestTableFields(t *testing.T) {
 	createInitTable(tables)
 	gtest.C(t, func(t *gtest.T) {
 		var expect = map[string][]any{
-			"ID":           {"BIGINT", false},
-			"ACCOUNT_NAME": {"VARCHAR", false},
-			"PWD_RESET":    {"TINYINT", false},
-			"ATTR_INDEX":   {"INT", true},
-			"DELETED":      {"INT", false},
-			"CREATED_TIME": {"TIMESTAMP", false},
+			"ID":           {"BIGINT(8)", false},
+			"ACCOUNT_NAME": {"VARCHAR(128)", false},
+			"PWD_RESET":    {"TINYINT(1)", false},
+			"ATTR_INDEX":   {"INT(4)", true},
+			"DELETED":      {"INT(4)", false},
+			"CREATED_TIME": {"TIMESTAMP(8)", false},
 		}

-		_, err := dbErr.TableFields(ctx, "Fields")
-		gtest.AssertNE(err, nil)
-
 		res, err := db.TableFields(ctx, tables)
 		gtest.AssertNil(err)

@ -114,6 +107,14 @@ func TestTableFields(t *testing.T) {
 	})
 }

+func TestTableFields_WithWrongPassword(t *testing.T) {
+	gtest.C(t, func(t *gtest.T) {
+		// dbErr is configured with wrong password, so it should return an error
+		_, err := dbErr.TableFields(ctx, "Fields")
+		gtest.AssertNE(err, nil)
+	})
+}
+
 func Test_DB_Query(t *testing.T) {
 	tableName := "A_tables"
 	createInitTable(tableName)
@ -138,110 +139,6 @@ func Test_DB_Query(t *testing.T) {
 	})
 }

-func TestModelSave(t *testing.T) {
-	table := createTable()
-	defer dropTable(table)
-	gtest.C(t, func(t *gtest.T) {
-		type User struct {
-			Id          int
-			AccountName string
-			AttrIndex   int
-		}
-		var (
-			user   User
-			count  int
-			result sql.Result
-			err    error
-		)
-		db.SetDebug(true)
-
-		result, err = db.Model(table).Data(g.Map{
-			"id":          1,
-			"accountName": "ac1",
-			"attrIndex":   100,
-		}).OnConflict("id").Save()
-
-		t.AssertNil(err)
-		n, _ := result.RowsAffected()
-		t.Assert(n, 1)
-
-		err = db.Model(table).Scan(&user)
-		t.AssertNil(err)
-		t.Assert(user.Id, 1)
-		t.Assert(user.AccountName, "ac1")
-		t.Assert(user.AttrIndex, 100)
-
-		_, err = db.Model(table).Data(g.Map{
-			"id":          1,
-			"accountName": "ac2",
-			"attrIndex":   200,
-		}).OnConflict("id").Save()
-		t.AssertNil(err)
-
-		err = db.Model(table).Scan(&user)
-		t.AssertNil(err)
-		t.Assert(user.AccountName, "ac2")
-		t.Assert(user.AttrIndex, 200)
-
-		count, err = db.Model(table).Count()
-		t.AssertNil(err)
-		t.Assert(count, 1)
-	})
-}
-
-func TestModelInsert(t *testing.T) {
-	// g.Model.insert not lost default not null coloumn
-	table := "A_tables"
-	createInitTable(table)
-	gtest.C(t, func(t *gtest.T) {
-		i := 200
-		data := User{
-			ID:          int64(i),
-			AccountName: fmt.Sprintf(`A%dtwo`, i),
-			PwdReset:    0,
-			AttrIndex:   99,
-			CreatedTime: time.Now(),
-			UpdatedTime: time.Now(),
-		}
-		// _, err := db.Schema(TestDBName).Model(table).Data(data).Insert()
-		_, err := db.Model(table).Insert(&data)
-		gtest.AssertNil(err)
-	})
-
-	gtest.C(t, func(t *gtest.T) {
-		i := 201
-		data := User{
-			ID:          int64(i),
-			AccountName: fmt.Sprintf(`A%dtwoONE`, i),
-			PwdReset:    1,
-			CreatedTime: time.Now(),
-			AttrIndex:   98,
-			UpdatedTime: time.Now(),
-		}
-		// _, err := db.Schema(TestDBName).Model(table).Data(data).Insert()
-		_, err := db.Model(table).Data(&data).Insert()
-		gtest.AssertNil(err)
-	})
-}
-
-func TestDBInsert(t *testing.T) {
-	table := "A_tables"
-	createInitTable("A_tables")
-	gtest.C(t, func(t *gtest.T) {
-		i := 300
-		data := g.Map{
-			"ID":           i,
-			"ACCOUNT_NAME": fmt.Sprintf(`A%dthress`, i),
-			"PWD_RESET":    3,
-			"ATTR_INDEX":   98,
-			"CREATED_TIME": gtime.Now(),
-			"UPDATED_TIME": gtime.Now(),
-		}
-		_, err := db.Insert(ctx, table, &data)
-		gtest.AssertNil(err)
-	})
-}
-
 func Test_DB_Exec(t *testing.T) {
 	createInitTable("A_tables")
 	gtest.C(t, func(t *gtest.T) {
--- a/contrib/drivers/dm/dm_z_unit_feature_soft_time_test.go
+++ b/contrib/drivers/dm/dm_z_unit_feature_soft_time_test.go
--- a/contrib/drivers/dm/dm_z_unit_init_test.go
+++ b/contrib/drivers/dm/dm_z_unit_init_test.go
@ -63,11 +63,10 @@ func init() {
 		Weight:           1,
 		MaxIdleConnCount: 10,
 		MaxOpenConnCount: 10,
-		CreatedAt:        "created_time",
-		UpdatedAt:        "updated_time",
+		// CreatedAt:        "created_time",
+		// UpdatedAt:        "updated_time",
 	}

-	// todo
 	nodeLink := gdb.ConfigNode{
 		Type: TestDBType,
 		Name: TestDBName,
@ -111,6 +110,8 @@ func init() {
 	}

 	ctx = context.Background()
+
+	// db.SetDebug(true)
 }

 func dropTable(table string) {
@ -143,7 +144,7 @@ func createTable(table ...string) (name string) {
 	CREATE TABLE "%s"
 (
 "ID" BIGINT NOT NULL,
-"ACCOUNT_NAME" VARCHAR(128) DEFAULT '' NOT NULL,
+"ACCOUNT_NAME" VARCHAR(128) DEFAULT '' NOT NULL COMMENT 'Account Name',
 "PWD_RESET" TINYINT DEFAULT 0 NOT NULL,
 "ENABLED" INT DEFAULT 1 NOT NULL,
 "DELETED" INT DEFAULT 0 NOT NULL,
@ -156,7 +157,6 @@ NOT CLUSTER PRIMARY KEY("ID")) STORAGE(ON "MAIN", CLUSTERBTR) ;
 	`, name)); err != nil {
 		gtest.Fatal(err)
 	}
-
 	return
 }

@ -169,7 +169,7 @@ func createInitTable(table ...string) (name string) {
 			"account_name": fmt.Sprintf(`name_%d`, i),
 			"pwd_reset":    0,
 			"attr_index":   i,
-			"create_time":  gtime.Now().String(),
+			"created_time": gtime.Now(),
 		})
 	}
 	result, err := db.Schema(TestDBName).Insert(context.Background(), name, array.Slice())
@ -212,3 +212,41 @@ NOT CLUSTER PRIMARY KEY("ID")) STORAGE(ON "MAIN", CLUSTERBTR) ;

 	return name, nil
 }
+
+func createInitTables(len int) []string {
+	tables := make([]string, 0, len)
+	for range len {
+		tables = append(tables, createInitTable())
+	}
+	return tables
+}
+
+// createTableWithIdentity creates a table with IDENTITY column for LastInsertId testing
+func createTableWithIdentity(table ...string) (name string) {
+	if len(table) > 0 {
+		name = table[0]
+	} else {
+		name = fmt.Sprintf("random_%d", gtime.Timestamp())
+	}
+
+	dropTable(name)
+
+	if _, err := db.Exec(ctx, fmt.Sprintf(`
+	CREATE TABLE "%s"
+(
+"ID" BIGINT IDENTITY(1, 1) NOT NULL,
+"ACCOUNT_NAME" VARCHAR(128) DEFAULT '' NOT NULL COMMENT 'Account Name',
+"PWD_RESET" TINYINT DEFAULT 0 NOT NULL,
+"ENABLED" INT DEFAULT 1 NOT NULL,
+"DELETED" INT DEFAULT 0 NOT NULL,
+"ATTR_INDEX" INT DEFAULT 0 ,
+"CREATED_BY" VARCHAR(32) DEFAULT '' NOT NULL,
+"CREATED_TIME" TIMESTAMP(6) DEFAULT CURRENT_TIMESTAMP() NOT NULL,
+"UPDATED_BY" VARCHAR(32) DEFAULT '' NOT NULL,
+"UPDATED_TIME" TIMESTAMP(6) DEFAULT CURRENT_TIMESTAMP() NOT NULL,
+NOT CLUSTER PRIMARY KEY("ID")) STORAGE(ON "MAIN", CLUSTERBTR) ;
+	`, name)); err != nil {
+		gtest.Fatal(err)
+	}
+	return
+}
--- a/contrib/drivers/dm/dm_z_unit_issue_test.go
+++ b/contrib/drivers/dm/dm_z_unit_issue_test.go
@ -71,3 +71,93 @@ func Test_Issue2594(t *testing.T) {
 		t.Assert(h1, h2)
 	})
 }
+
+// Test_MultilineSQLStatement tests that multi-line SQL statements are properly supported.
+// This test verifies that newlines and tabs in SQL queries are preserved,
+// which is essential for readability and proper SQL statement handling.
+func Test_MultilineSQLStatement(t *testing.T) {
+	table := "A_tables"
+	createInitTable(table)
+	defer dropTable(table)
+
+	gtest.C(t, func(t *gtest.T) {
+		// Test multi-line SELECT statement with newlines and indentation
+		multilineSql := `
+		SELECT 
+			id,
+			account_name,
+			attr_index
+		FROM A_tables
+		WHERE id = ?
+		AND account_name = ?
+		`
+		result, err := db.GetAll(ctx, multilineSql, 1, "name_1")
+		t.AssertNil(err)
+		t.Assert(len(result), 1)
+		t.Assert(result[0]["ID"].Int(), 1)
+		t.Assert(result[0]["ACCOUNT_NAME"].String(), "name_1")
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		// Test multi-line SELECT with tabs
+		multilineSql := `SELECT
+			id,
+			account_name,
+			attr_index
+		FROM A_tables
+		WHERE id IN (?, ?)
+		ORDER BY id`
+		result, err := db.GetAll(ctx, multilineSql, 2, 3)
+		t.AssertNil(err)
+		t.Assert(len(result), 2)
+		t.Assert(result[0]["ID"].Int(), 2)
+		t.Assert(result[1]["ID"].Int(), 3)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		// Test that newlines in values don't cause issues
+		multilineSql := `
+		SELECT * 
+		FROM A_tables 
+		WHERE id = ?`
+		result, err := db.GetAll(ctx, multilineSql, 5)
+		t.AssertNil(err)
+		t.Assert(len(result), 1)
+		t.Assert(result[0]["ID"].Int(), 5)
+		t.Assert(result[0]["ACCOUNT_NAME"].String(), "name_5")
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		// Test multi-line INSERT with newlines
+		multilineSql := `
+		INSERT INTO A_tables
+		(ID, ACCOUNT_NAME, ATTR_INDEX, CREATED_TIME, UPDATED_TIME)
+		VALUES
+		(?, ?, ?, ?, ?)`
+		_, err := db.Exec(ctx, multilineSql, 1001, "multiline_insert_test", 100, gtime.Now(), gtime.Now())
+		t.AssertNil(err)
+
+		// Verify the insert worked
+		result, err := db.GetAll(ctx, "SELECT * FROM A_tables WHERE ID = ?", 1001)
+		t.AssertNil(err)
+		t.Assert(len(result), 1)
+		t.Assert(result[0]["ACCOUNT_NAME"].String(), "multiline_insert_test")
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		// Test multi-line UPDATE with newlines
+		multilineSql := `
+		UPDATE A_tables
+		SET account_name = ?,
+			attr_index = ?
+		WHERE id = ?`
+		_, err := db.Exec(ctx, multilineSql, "updated_multiline", 999, 1)
+		t.AssertNil(err)
+
+		// Verify the update worked
+		result, err := db.GetAll(ctx, "SELECT * FROM A_tables WHERE ID = ?", 1)
+		t.AssertNil(err)
+		t.Assert(len(result), 1)
+		t.Assert(result[0]["ACCOUNT_NAME"].String(), "updated_multiline")
+	})
+}
--- a/contrib/drivers/dm/dm_z_unit_model_test.go
+++ b/contrib/drivers/dm/dm_z_unit_model_test.go
@ -0,0 +1,185 @@
+// Copyright 2019 gf Author(https://github.com/gogf/gf). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package dm_test
+
+import (
+	"database/sql"
+	"fmt"
+	"testing"
+	"time"
+
+	"github.com/gogf/gf/v2/frame/g"
+	"github.com/gogf/gf/v2/os/gtime"
+	"github.com/gogf/gf/v2/test/gtest"
+)
+
+func Test_Model_Save(t *testing.T) {
+	table := createTableWithIdentity()
+	defer dropTable(table)
+	gtest.C(t, func(t *gtest.T) {
+		type User struct {
+			Id          int
+			AccountName string
+			AttrIndex   int
+		}
+		var (
+			user   User
+			count  int
+			result sql.Result
+			err    error
+		)
+
+		// First insert: let IDENTITY auto-generate ID - use Insert() instead of Save()
+		// because Save() requires a primary key in the data for conflict detection
+		result, err = db.Model(table).Data(g.Map{
+			"accountName": "ac1",
+			"attrIndex":   100,
+		}).Insert()
+
+		t.AssertNil(err)
+		n, _ := result.RowsAffected()
+		t.Assert(n, 1)
+
+		err = db.Model(table).Scan(&user)
+		t.AssertNil(err)
+		t.AssertGT(user.Id, 0) // ID should be auto-generated
+		t.Assert(user.AccountName, "ac1")
+		t.Assert(user.AttrIndex, 100)
+
+		// Second save: update the existing record using the generated ID
+		_, err = db.Model(table).Data(g.Map{
+			"id":          user.Id,
+			"accountName": "ac2",
+			"attrIndex":   200,
+		}).OnConflict("id").Save()
+		t.AssertNil(err)
+
+		err = db.Model(table).Scan(&user)
+		t.AssertNil(err)
+		t.Assert(user.AccountName, "ac2")
+		t.Assert(user.AttrIndex, 200)
+
+		_, err = db.Model(table).Data(g.Map{
+			"id":          user.Id,
+			"accountName": "ac2",
+			"attrIndex":   2000,
+		}).Save()
+		t.AssertNil(err)
+
+		err = db.Model(table).Scan(&user)
+		t.AssertNil(err)
+		t.Assert(user.AccountName, "ac2")
+		t.Assert(user.AttrIndex, 2000)
+
+		count, err = db.Model(table).Count()
+		t.AssertNil(err)
+		t.Assert(count, 1)
+	})
+}
+
+func Test_Model_Insert(t *testing.T) {
+	// g.Model.insert not lost default not null column
+	table := "A_tables"
+	createInitTable(table)
+	defer dropTable(table)
+
+	gtest.C(t, func(t *gtest.T) {
+		i := 200
+		data := User{
+			ID:          int64(i),
+			AccountName: fmt.Sprintf(`A%dtwo`, i),
+			PwdReset:    0,
+			AttrIndex:   99,
+			CreatedTime: time.Now(),
+			UpdatedTime: time.Now(),
+		}
+		result, err := db.Model(table).Insert(&data)
+		gtest.AssertNil(err)
+		n, err := result.RowsAffected()
+		gtest.AssertNil(err)
+		gtest.Assert(n, 1)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		i := 201
+		data := User{
+			ID:          int64(i),
+			AccountName: fmt.Sprintf(`A%dtwoONE`, i),
+			PwdReset:    1,
+			CreatedTime: time.Now(),
+			AttrIndex:   98,
+			UpdatedTime: time.Now(),
+		}
+		result, err := db.Model(table).Data(&data).Insert()
+		gtest.AssertNil(err)
+		n, err := result.RowsAffected()
+		gtest.AssertNil(err)
+		gtest.Assert(n, 1)
+	})
+}
+
+func Test_Model_InsertIgnore(t *testing.T) {
+	table := createInitTable()
+	defer dropTable(table)
+
+	// db.SetDebug(true)
+
+	gtest.C(t, func(t *gtest.T) {
+		data := g.Map{
+			"id":           1,
+			"account_name": fmt.Sprintf(`name_%d`, 777),
+			"pwd_reset":    0,
+			"attr_index":   777,
+			"created_time": gtime.Now(),
+		}
+		_, err := db.Model(table).Data(data).InsertIgnore()
+		t.AssertNil(err)
+
+		one, err := db.Model(table).WherePri(1).One()
+		t.AssertNil(err)
+		t.Assert(one["ACCOUNT_NAME"].String(), "name_1")
+
+		count, err := db.Model(table).Count()
+		t.AssertNil(err)
+		t.Assert(count, TableSize)
+	})
+
+	gtest.C(t, func(t *gtest.T) {
+		data := g.Map{
+			// "id":           1,
+			"account_name": fmt.Sprintf(`name_%d`, 777),
+			"pwd_reset":    0,
+			"attr_index":   777,
+			"created_time": gtime.Now(),
+		}
+		_, err := db.Model(table).Data(data).InsertIgnore()
+		t.AssertNE(err, nil)
+
+		count, err := db.Model(table).Count()
+		t.AssertNil(err)
+		t.Assert(count, TableSize)
+	})
+}
+
+func Test_Model_InsertAndGetId(t *testing.T) {
+	table := createTableWithIdentity()
+	defer dropTable(table)
+
+	gtest.C(t, func(t *gtest.T) {
+		data := g.Map{
+			// "id":          1,
+			"account_name": fmt.Sprintf(`name_%d`, 1),
+			"pwd_reset":    0,
+			"attr_index":   1,
+			"created_time": gtime.Now(),
+		}
+		lastId, err := db.Model(table).Data(data).InsertAndGetId()
+		t.AssertNil(err)
+		t.AssertGT(lastId, 0)
+	})
+
+}
--- a/contrib/drivers/dm/dm_z_unit_pr_test.go
+++ b/contrib/drivers/dm/dm_z_unit_pr_test.go
@ -0,0 +1,40 @@
+// Copyright 2019 gf Author(https://github.com/gogf/gf). All Rights Reserved.
+//
+// This Source Code Form is subject to the terms of the MIT License.
+// If a copy of the MIT was not distributed with this file,
+// You can obtain one at https://github.com/gogf/gf.
+
+package dm_test
+
+import (
+	"testing"
+
+	"github.com/gogf/gf/v2/test/gtest"
+)
+
+// PR #4157 WherePri
+func Test_WherePri_PR4157(t *testing.T) {
+	tableName := "A_tables"
+	createInitTable(tableName)
+	defer dropTable(tableName)
+	gtest.C(t, func(t *gtest.T) {
+		var resOne *User
+		err := db.Model(tableName).WherePri(1).Scan(&resOne)
+		t.AssertNil(err)
+		t.AssertNQ(resOne, nil)
+		t.AssertEQ(resOne.ID, int64(1))
+	})
+}
+
+// PR #4157 get table field comments
+func Test_TableFields_Comment_PR4157(t *testing.T) {
+	tableName := "A_tables"
+	schema := "SYSDBA"
+	createInitTable(tableName)
+	defer dropTable(tableName)
+	gtest.C(t, func(t *gtest.T) {
+		fields, err := db.Model().TableFields(tableName, schema)
+		t.AssertNil(err)
+		t.AssertEQ(fields["ACCOUNT_NAME"].Comment, "Account Name")
+	})
+}
--- a/contrib/drivers/dm/go.mod
+++ b/contrib/drivers/dm/go.mod
@ -6,13 +6,13 @@ replace github.com/gogf/gf/v2 => ../../../

 require (
 	gitee.com/chunanyong/dm v1.8.12
-	github.com/gogf/gf/v2 v2.9.5
+	github.com/gogf/gf/v2 v2.9.7
 )

 require (
 	github.com/BurntSushi/toml v1.5.0 // indirect
 	github.com/clbanning/mxj/v2 v2.7.0 // indirect
-	github.com/emirpasic/gods v1.18.1 // indirect
+	github.com/emirpasic/gods/v2 v2.0.0-alpha // indirect
 	github.com/fatih/color v1.18.0 // indirect
 	github.com/fsnotify/fsnotify v1.9.0 // indirect
 	github.com/go-logr/logr v1.4.3 // indirect
--- a/Show More
+++ b/Show More