使用encoding库做字符集转换

g/encoding/gcharset/gcharset.go

@@ -9,24 +9,48 @@
 // Package gcharset provides converting string to requested character encoding.
 //
 // 字符集转换方法,
-// 使用mahonia实现的字符集转换方法，支持的字符集包括常见的utf8/UTF-16/UTF-16LE/macintosh/big5/gbk/gb18030,支持的全量字符集可以参考mahonia包
 package gcharset
 
 import (
 	"bytes"
 	"errors"
 	"fmt"
+	"github.com/gogf/gf/g/container/gmap"
 	"github.com/gogf/gf/third/github.com/axgle/mahonia"
+	"github.com/gogf/gf/third/golang.org/x/text/encoding"
+	"github.com/gogf/gf/third/golang.org/x/text/encoding/japanese"
+	"github.com/gogf/gf/third/golang.org/x/text/encoding/korean"
 	"github.com/gogf/gf/third/golang.org/x/text/encoding/simplifiedchinese"
+	"github.com/gogf/gf/third/golang.org/x/text/encoding/traditionalchinese"
 	"github.com/gogf/gf/third/golang.org/x/text/transform"
 	"io/ioutil"
 	"strings"
 )
 
+var encodingMap *gmap.Map
+
+func init() {
+	encodingMap = gmap.New()
+	encodingMap.Sets(
+		map[interface{}]interface{}{
+			"GBK":       simplifiedchinese.GBK,
+			"GB18030":   simplifiedchinese.GB18030,
+			"HZGB2312":  simplifiedchinese.HZGB2312,
+			"GB2312":    simplifiedchinese.HZGB2312,
+			"EUCJP":     japanese.EUCJP,
+			"ISO2022JP": japanese.ISO2022JP,
+			"SHIFTJIS":  japanese.ShiftJIS,
+			"EUCKR":     korean.EUCKR,
+			"BIG5":      traditionalchinese.Big5,
+		})
+}
+
 // 2个字符集之间的转换
 func Convert(dstCharset string, srcCharset string, src string) (dst string, err error) {
+	srcCharsetUpper := strings.ToUpper(srcCharset)
+	dstCharsetUpper := strings.ToUpper(dstCharset)
 
-	if strings.EqualFold(srcCharset, dstCharset) {
+	if srcCharsetUpper == dstCharsetUpper {
 		return src, nil
 	}
 
@@ -34,72 +58,48 @@ func Convert(dstCharset string, srcCharset string, src string) (dst string, err
 	d := new(mahonia.Charset)
 	srctmp := src
 
-	switch {
-	case strings.EqualFold("GBK", srcCharset):
-		tmp, err := ioutil.ReadAll(transform.NewReader(bytes.NewReader([]byte(src)), simplifiedchinese.GBK.NewDecoder()))
-		if err != nil {
-			return "", fmt.Errorf("gbk to utf8 failed. %v", err)
-		}
-		srctmp = string(tmp)
-	case strings.EqualFold("GB18030", srcCharset):
-		tmp, err := ioutil.ReadAll(transform.NewReader(bytes.NewReader([]byte(src)), simplifiedchinese.GB18030.NewDecoder()))
-		if err != nil {
-			return "", fmt.Errorf("GB18030 to utf8 failed. %v", err)
-		}
-		srctmp = string(tmp)
-	case strings.EqualFold("GB2312", srcCharset) || strings.EqualFold("HZGB2312", srcCharset):
-		tmp, err := ioutil.ReadAll(transform.NewReader(bytes.NewReader([]byte(src)), simplifiedchinese.HZGB2312.NewDecoder()))
-		if err != nil {
-			return "", fmt.Errorf("GB2312 to utf8 failed. %v", err)
-		}
-		srctmp = string(tmp)
-	case strings.EqualFold("UTF-8", srcCharset):
-	default:
-		s = mahonia.GetCharset(srcCharset)
-		if s == nil {
-			return "", errors.New(fmt.Sprintf("not support charset:%s", srcCharset))
-		}
+	if srcCharset != "UTF-8" {
+		enc := encodingMap.Get(srcCharset)
+		if enc != nil {
+			tmp, err := ioutil.ReadAll(transform.NewReader(bytes.NewReader([]byte(src)), enc.(encoding.Encoding).NewDecoder()))
+			if err != nil {
+				return "", fmt.Errorf("%s to utf8 failed. %v", srcCharset, err)
+			}
+			srctmp = string(tmp)
+		} else {
+			s = mahonia.GetCharset(srcCharsetUpper)
+			if s == nil {
+				return "", errors.New(fmt.Sprintf("not support charset:%s", srcCharset))
+			}
 
-		if s.Name != "UTF-8" {
-			srctmp = s.NewDecoder().ConvertString(srctmp)
+			if s.Name != "UTF-8" {
+				srctmp = s.NewDecoder().ConvertString(srctmp)
+			}
 		}
 	}
 
 	dst = srctmp
 
-	switch {
-	case strings.EqualFold("GBK", dstCharset):
-		tmp, err := ioutil.ReadAll(transform.NewReader(bytes.NewReader([]byte(srctmp)), simplifiedchinese.GBK.NewEncoder()))
-		if err != nil {
-			return "", fmt.Errorf("utf to gbk failed. %v", err)
-		}
-		dst = string(tmp)
-	case strings.EqualFold("GB18030", dstCharset):
-		tmp, err := ioutil.ReadAll(transform.NewReader(bytes.NewReader([]byte(srctmp)), simplifiedchinese.GB18030.NewEncoder()))
-		if err != nil {
-			return "", fmt.Errorf("utf8 to gb18030 failed. %v", err)
-		}
-		dst = string(tmp)
-	case strings.EqualFold("GB2312", dstCharset) || strings.EqualFold("HZGB2312", dstCharset):
-		tmp, err := ioutil.ReadAll(transform.NewReader(bytes.NewReader([]byte(srctmp)), simplifiedchinese.HZGB2312.NewEncoder()))
-		if err != nil {
-			return "", fmt.Errorf("utf8 to gb2312 failed. %v", err)
-		}
-		dst = string(tmp)
-	case strings.EqualFold("UTF-8", dstCharset):
-	default:
-		d = mahonia.GetCharset(dstCharset)
-		if d == nil {
-			return "", errors.New(fmt.Sprintf("not support charset:%s", dstCharset))
-		}
+	if dstCharset != "UTF-8" {
+		enc := encodingMap.Get(dstCharset)
+		if enc != nil {
+			tmp, err := ioutil.ReadAll(transform.NewReader(bytes.NewReader([]byte(srctmp)), enc.(encoding.Encoding).NewEncoder()))
+			if err != nil {
+				return "", fmt.Errorf("utf to %s failed. %v", dstCharset, err)
+			}
+			dst = string(tmp)
+		} else {
+			d = mahonia.GetCharset(dstCharsetUpper)
+			if d == nil {
+				return "", errors.New(fmt.Sprintf("not support charset:%s", dstCharset))
+			}
 
-		dst = srctmp
-		if d.Name != "UTF-8" {
-			dst = d.NewEncoder().ConvertString(dst)
+			dst = srctmp
+			if d.Name != "UTF-8" {
+				dst = d.NewEncoder().ConvertString(dst)
+			}
 		}
-
 	}
-
 	return dst, nil
 }