This pull request introduces significant improvements to the handling of
the `Replace` and `Save` operations for multiple database drivers,
especially for MSSQL and PostgreSQL. The changes ensure that these
operations now auto-detect primary keys when conflict columns are not
explicitly provided, improving usability and aligning behavior across
drivers. Additionally, the pull request updates related tests to reflect
these enhancements and includes some minor documentation and code
cleanup.
**Key changes:**
### Enhanced Replace/Save Logic for Database Drivers
* **MSSQL Driver:**
- `Replace` and `Save` operations now auto-detect primary keys if
`OnConflict` is not specified, using the `MERGE` statement for upsert
functionality. If no primary key is found in the data, a detailed error
is returned.
[[1]](diffhunk://#diff-87815aa559a927e2de09bd05148f9841dfc06a1b5f3ecc5e3d5fcb80323a87f8L23-R61)
[[2]](diffhunk://#diff-87815aa559a927e2de09bd05148f9841dfc06a1b5f3ecc5e3d5fcb80323a87f8L43-L59)
- Updated tests to verify that `Replace` correctly updates or inserts
records, and that missing conflict columns are properly handled.
[[1]](diffhunk://#diff-bdbde9d7d6ee14c795343767b414740c4396f4dd3e97788b1f9d4e615405a42dL141-R151)
[[2]](diffhunk://#diff-26338e93e473300b1313936eb0f6826546473793442f24715fa294b595f7a805L2661-R2707)
* **PostgreSQL Driver:**
- Similar to MSSQL, `Replace` and `Save` now auto-detect primary keys
for conflict resolution if `OnConflict` is not set, and treat `Replace`
as a `Save` operation.
- Adjusted tests to ensure `Save` and `Replace` work as expected,
including verifying data replacement and insertion.
[[1]](diffhunk://#diff-c22703c37ebb6836c332f7cd2ada570577ba4564fe39886db02f7c2d0e7a2048L93-R93)
[[2]](diffhunk://#diff-c22703c37ebb6836c332f7cd2ada570577ba4564fe39886db02f7c2d0e7a2048R102)
[[3]](diffhunk://#diff-c22703c37ebb6836c332f7cd2ada570577ba4564fe39886db02f7c2d0e7a2048L110-R130)
* **DM Driver:**
- Improved conflict detection: now checks that at least one primary key
exists in the provided data when `OnConflict` is not specified, and
provides clearer error messages.
- Refactored to use the core method for primary key detection and
removed redundant code.
### Minor Improvements and Documentation
* Added clarifying comments to `DoInsert` methods for ClickHouse, DM,
MSSQL, Oracle, and PostgreSQL drivers, specifying that the input list
must have at least one validated record.
[[1]](diffhunk://#diff-f2e003895041ed3c52b91bb8c270696adc3528d77c39d2f7137af3396267444cR19)
[[2]](diffhunk://#diff-f51b30e3f0b0f1284b905385a89992efd0de2fe9ff8c5a4062344dfab17d428eR23)
[[3]](diffhunk://#diff-87815aa559a927e2de09bd05148f9841dfc06a1b5f3ecc5e3d5fcb80323a87f8L23-R61)
[[4]](diffhunk://#diff-f61dac3fcfd5df4a3936cd8743499c8c0fc45f4f5d0f5398ed84a0cb1603202cR24)
[[5]](diffhunk://#diff-c1dfed79aaa3a432057d2bd74d270e4b4094ebcf72984f1161d4972bea009410R16-R72)
* Minor code and comment cleanups, including improved formatting and
error handling.
[[1]](diffhunk://#diff-f61dac3fcfd5df4a3936cd8743499c8c0fc45f4f5d0f5398ed84a0cb1603202cR37)
[[2]](diffhunk://#diff-f61dac3fcfd5df4a3936cd8743499c8c0fc45f4f5d0f5398ed84a0cb1603202cL96-R98)
[[3]](diffhunk://#diff-f61dac3fcfd5df4a3936cd8743499c8c0fc45f4f5d0f5398ed84a0cb1603202cL106-L116)
[[4]](diffhunk://#diff-a17b44c76aaac53d1f164a2bb9440a5531659f4355e7ccfabdadff8dc8633c09L170-R171)
[[5]](diffhunk://#diff-56189fa9ae1df51716b50d34d7fe56bfe67a330e8ac2c6b0de7b958db6817ed5R83-R98)
### Workflow and Documentation Updates
* Updated example Docker commands in the CI workflow for consistency and
clarity.
[[1]](diffhunk://#diff-a1a3cb9bdeb5541d148091d973cf266aa3b317e6415a86630e816cbe27cf8b9cL57-R57)
[[2]](diffhunk://#diff-a1a3cb9bdeb5541d148091d973cf266aa3b317e6415a86630e816cbe27cf8b9cL78-R78)
[[3]](diffhunk://#diff-a1a3cb9bdeb5541d148091d973cf266aa3b317e6415a86630e816cbe27cf8b9cL92-R92)
[[4]](diffhunk://#diff-a1a3cb9bdeb5541d148091d973cf266aa3b317e6415a86630e816cbe27cf8b9cL106-R106)
[[5]](diffhunk://#diff-a1a3cb9bdeb5541d148091d973cf266aa3b317e6415a86630e816cbe27cf8b9cL153-R153)
[[6]](diffhunk://#diff-a1a3cb9bdeb5541d148091d973cf266aa3b317e6415a86630e816cbe27cf8b9cL164-R164)
* Removed outdated note about `Replace` support from the SQLite driver
documentation.
These changes improve the consistency, reliability, and developer
experience when performing upsert operations across different database
backends.
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Lance Add <1196661499@qq.com>
This pull request significantly improves PostgreSQL array type handling
and conversion in the `pgsql` driver, providing more accurate type
mapping and conversion logic, especially for array types. It introduces
comprehensive documentation, refactors conversion logic to use the `pq`
package for array types, and adds extensive unit tests to ensure
correctness and error handling. Additionally, minor enhancements and
clarifications are made to upsert formatting and table field queries.
### PostgreSQL Array Type Handling and Conversion
* Refactored `CheckLocalTypeForField` and `ConvertValueForLocal` methods
in `contrib/drivers/pgsql/pgsql_convert.go` to accurately map PostgreSQL
array types (such as `_int2`, `_int4`, `_int8`, `_float4`, `_float8`,
`_bool`, `_varchar`, `_text`, `_char`, `_bpchar`, `_numeric`,
`_decimal`, `_money`, `_bytea`) to their corresponding Go types, using
the `pq` package for conversion. Added detailed documentation and
mapping tables for supported types.
[[1]](diffhunk://#diff-a3b1e68bfa29fbcfda7c703bbe875fa82e958f6c3ad942ef82193a9dd8ad67e2R46-R63)
[[2]](diffhunk://#diff-a3b1e68bfa29fbcfda7c703bbe875fa82e958f6c3ad942ef82193a9dd8ad67e2L56-R103)
[[3]](diffhunk://#diff-a3b1e68bfa29fbcfda7c703bbe875fa82e958f6c3ad942ef82193a9dd8ad67e2R112-R209)
* Added comprehensive unit tests in
`contrib/drivers/pgsql/pgsql_z_unit_convert_test.go` to verify type
mapping and conversion for all supported array types, including error
cases for invalid input.
### Utility and API Improvements
* Added a new `Bools()` method to the `gvar.Var` type in
`container/gvar/gvar_slice.go` for converting values to `[]bool`, with
corresponding unit tests in `container/gvar/gvar_z_unit_slice_test.go`.
[[1]](diffhunk://#diff-32e887e540e0170f785508d105cb794e4d54d854b53b6950973c80022973c490R11-R15)
[[2]](diffhunk://#diff-01453eca4d4b3e35d07ca105cb924c6441d0cd9df6cbcc337a89832c8d53057fR24-R41)
### SQL Formatting and Documentation
* Improved documentation and formatting in the upsert logic of
`contrib/drivers/pgsql/pgsql_format_upsert.go` to clarify the use of
`EXCLUDED` in PostgreSQL's `ON CONFLICT DO UPDATE`.
* Enhanced readability of the table field query in
`contrib/drivers/pgsql/pgsql_table_fields.go` by reformatting SQL and
clarifying field extraction.
---------
Co-authored-by: hailaz <739476267@qq.com>
Co-authored-by: houseme <housemecn@gmail.com>
This pull request introduces significant improvements to the DM database
driver, especially around insert operations, and refines documentation
and tests to reflect these changes. The main focus is enabling support
for "replace" and "insert ignore" operations using DM's `MERGE`
statement, improving type reporting for table fields, and updating
documentation for clarity and accuracy.
### DM Driver Insert Operations
* Added support for `Replace` and `InsertIgnore` operations in the DM
driver by internally mapping them to DM's `MERGE` statement. This
enables upsert and insert-ignore behavior for DM databases, improving
compatibility with other drivers.
* Implemented helper methods (`doMergeInsert`, `doInsertIgnore`, and
`getPrimaryKeys`) to generate correct `MERGE` SQL statements and
automatically detect primary keys when needed.
[[1]](diffhunk://#diff-f51b30e3f0b0f1284b905385a89992efd0de2fe9ff8c5a4062344dfab17d428eL31-R94)
[[2]](diffhunk://#diff-f51b30e3f0b0f1284b905385a89992efd0de2fe9ff8c5a4062344dfab17d428eL115-R212)
* Updated the logic for building update values and SQL generation to
ensure correct behavior for both upsert and insert-ignore cases.
[[1]](diffhunk://#diff-f51b30e3f0b0f1284b905385a89992efd0de2fe9ff8c5a4062344dfab17d428eL61-R109)
[[2]](diffhunk://#diff-f51b30e3f0b0f1284b905385a89992efd0de2fe9ff8c5a4062344dfab17d428eL89-R132)
[[3]](diffhunk://#diff-f51b30e3f0b0f1284b905385a89992efd0de2fe9ff8c5a4062344dfab17d428eL100-R144)
[[4]](diffhunk://#diff-f51b30e3f0b0f1284b905385a89992efd0de2fe9ff8c5a4062344dfab17d428eL115-R212)
### Table Field Type Reporting
* Improved the DM driver's `TableFields` method to report column types
with length/precision (e.g., `VARCHAR(128)` instead of just `VARCHAR`),
aligning with expectations and other drivers.
[[1]](diffhunk://#diff-40a365112421ae1967bd960f8acefcc91ddb8180865b78bc49cd090fbf4883daL26-R26)
[[2]](diffhunk://#diff-40a365112421ae1967bd960f8acefcc91ddb8180865b78bc49cd090fbf4883daR88-R105)
* Updated related unit tests to expect the new type format for DM table
fields.
### Documentation Updates
* Removed outdated or redundant documentation in both English and
Chinese driver README files, and clarified supported features and
limitations for DM and other drivers.
[[1]](diffhunk://#diff-d49f5bc3a34b11a6ccb82cc54675b06a7dea5f0a943ae91c4ca0d28bd5003299L1)
[[2]](diffhunk://#diff-d49f5bc3a34b11a6ccb82cc54675b06a7dea5f0a943ae91c4ca0d28bd5003299L47-R46)
[[3]](diffhunk://#diff-d49f5bc3a34b11a6ccb82cc54675b06a7dea5f0a943ae91c4ca0d28bd5003299L119-L122)
[[4]](diffhunk://#diff-05411a14e9c7ca235f7f436bfde732853aa93b364361fe80d65ac768f4e4d613L1-L126)
### Test Suite Enhancements
* Refactored and restored unit tests for DM driver insert operations,
including tests for `Save`, `Insert`, and the new `InsertIgnore`
functionality to ensure correct behavior and compatibility.
[[1]](diffhunk://#diff-2b1a59b8b2adaa1ca3074629374ab122929e4d4fbb4cc794b8e1db60ebf8d4c2L143-L245)
[[2]](diffhunk://#diff-2b1a59b8b2adaa1ca3074629374ab122929e4d4fbb4cc794b8e1db60ebf8d4c2R512-R632)
* Minor adjustments to DM test initialization for improved clarity.
### Core Insert Logic Minor Refactoring
* Minor variable renaming for clarity in the core insert logic
(`gdb_core.go`), improving code readability.
[[1]](diffhunk://#diff-b1bbe5e3995261813e4e0ac6ffee8a37c236eaa2759f2bd82e211711695a70bcL449-R452)
[[2]](diffhunk://#diff-b1bbe5e3995261813e4e0ac6ffee8a37c236eaa2759f2bd82e211711695a70bcL466-R474)
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Enhance the Chinese documentation by adding a new README file and
updating existing database driver instructions with the latest `go get`
commands. Additionally, provide Chinese explanations for the `gf`
command documentation.
fix https://github.com/gogf/gf/issues/4533
When using JOIN queries in MySQL with the `Group()` method, column names
in GROUP BY clauses become ambiguous if multiple tables contain columns
with the same name (commonly `id`). This results in MySQL errors like
"Column 'id' in group statement is ambiguous".
**Example Issue:**
```go
model := t.Ctx(ctx).Fields("t_inf_job.*, t_inf_job_attr.*").
LeftJoin("t_inf_job_attr", "t_inf_job.id = t_inf_job_attr.job_id").
Where(t.Columns().Deleted, 0)
// This would fail with "Column 'id' in group statement is ambiguous"
err = model.Group(t.Columns().Id).Scan(&jobs)
```
### **Key Changes**
1. **Modified function signature**: `Group(groupBy ...string)` →
`Group(groupBy ...any)` to support Raw SQL expressions
2. **Auto-prefixing logic**: When JOINs are detected (by checking for "
JOIN " in the tables string), unqualified column names are automatically
prefixed with the primary table name
3. **Preserved existing behavior**: Already qualified columns
(containing ".") and Raw expressions are handled as before
4. **Added comprehensive test**: `Test_Model_Group_WithJoin` verifies
the fix works correctly with JOIN queries
---------
Co-authored-by: hailaz <739476267@qq.com>
Code example：
``` go
var res *BasicInfo
err := g.DB().Model("basic_info").
FieldsPrefix("basic_info", basicInfoColumns).
Where("id", 35813305356386305).Scan(&res)
if err != nil {
panic(err)
}
g.Dump(res)
```
SQL generated before modification ：
``` sql
SELECT basic_info.id,basic_info.full_name,basic_info.contact FROM `basic_info` WHERE (`id`=35813305356386305) AND `delete_time` IS NULL LIMIT 1
```
SQL generated after modification：
``` sql
SELECT `basic_info`.`id`,`basic_info`.`full_name`,`basic_info`.`contact` FROM `basic_info` WHERE (`id`=35813305356386305) AND `delete_time` IS NULL LIMIT 1
```
---------
Co-authored-by: hailaz <739476267@qq.com>
It is wrote with glist.List's and list.List's source codes and improve
to support T type.
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: hailaz <739476267@qq.com>
Remove the t array of wrapper array. Now it's a real one. Other normal
array will base on it.
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Fixed an issue where `gcfg config apollo` failed to retrieve
configurations for multiple namespaces; fixed an issue where `watch
apollo change` resulted in missing configurations.
---------
Co-authored-by: DAWN <xiongchao@cdfsunrise.com>
Co-authored-by: hailaz <739476267@qq.com>
优化run命令使得只在文件有写入事件时才触发reload:
gf run 的文件监控逻辑之前会对所有文件系统事件做出响应，包括非内容修改的事件（如文件access
time变化），这会导致开发过程中不必要的频繁重载。本次修改在文件监控回调中增加了event.IsWrite()的判断，确保只有在文件内容被实际写入时才触发重载逻辑，优化了开发体验。
fix(net/gclient): fix content-type 'application/json;charset=utf-8' can
not match `application/json`
---------
Co-authored-by: houseme <housemecn@gmail.com>
