🐛 fix determine gbk

server/down.go

@@ -99,18 +99,20 @@ func init() {
 
 func Text(c *gin.Context, link string) {
 	res, err := client.R().Get(link)
+	if err != nil {
+		ErrorResp(c, err, 500)
+		return
+	}
 	text := res.String()
-	if utils.IsGBK(res.Body()) {
+	t := utils.GetStrCoding(res.Body())
+	log.Debugf("text type: %s", t)
+	if t != utils.UTF8 {
 		body, err := utils.GbkToUtf8(res.Body())
 		if err != nil {
-			ErrorResp(c,err,500)
+			ErrorResp(c, err, 500)
 			return
 		}
 		text = string(body)
 	}
-	if err != nil {
-		ErrorResp(c,err,500)
-		return
-	}
-	c.String(200,text)
+	c.String(200, text)
 }

utils/code.go

@@ -0,0 +1,59 @@
+package utils
+
+import (
+	"bytes"
+	"golang.org/x/text/encoding/simplifiedchinese"
+	"golang.org/x/text/transform"
+	"io/ioutil"
+	"unicode/utf8"
+)
+
+func IsGBK(data []byte) bool {
+	length := len(data)
+	var i = 0
+	for i < length {
+		if data[i] <= 0x7f {
+			//编码0~127,只有一个字节的编码，兼容ASCII码
+			i++
+			continue
+		} else {
+			//大于127的使用双字节编码，落在gbk编码范围内的字符
+			if data[i] >= 0x81 &&
+				data[i] <= 0xfe &&
+				data[i+1] >= 0x40 &&
+				data[i+1] <= 0xfe &&
+				data[i+1] != 0xf7 {
+				i += 2
+				continue
+			} else {
+				return false
+			}
+		}
+	}
+	return true
+}
+
+const (
+	GBK     string = "GBK"
+	UTF8    string = "UTF8"
+	UNKNOWN string = "UNKNOWN"
+)
+
+func GetStrCoding(data []byte) string {
+	if utf8.Valid(data) {
+		return UTF8
+	} else if IsGBK(data) {
+		return GBK
+	} else {
+		return UNKNOWN
+	}
+}
+
+func GbkToUtf8(s []byte) ([]byte, error) {
+	reader := transform.NewReader(bytes.NewReader(s), simplifiedchinese.GBK.NewDecoder())
+	d, e := ioutil.ReadAll(reader)
+	if e != nil {
+		return nil, e
+	}
+	return d, nil
+}

utils/file.go

@@ -1,12 +1,9 @@
 package utils
 
 import (
-	"bytes"
 	"encoding/json"
 	"github.com/Xhofe/alist/conf"
 	log "github.com/sirupsen/logrus"
-	"golang.org/x/text/encoding/simplifiedchinese"
-	"golang.org/x/text/transform"
 	"io/ioutil"
 	"os"
 	"path/filepath"
@@ -92,37 +89,3 @@ func ParsePath(path string) string {
 	return path
 }
 
-func IsGBK(data []byte) bool {
-	length := len(data)
-	var i int = 0
-	for i < length {
-		//fmt.Printf("for %x\n", data[i])
-		if data[i] <= 0xff {
-			//编码小于等于127,只有一个字节的编码，兼容ASCII吗
-			i++
-			continue
-		} else {
-			//大于127的使用双字节编码
-			if  data[i] >= 0x81 &&
-				data[i] <= 0xfe &&
-				data[i + 1] >= 0x40 &&
-				data[i + 1] <= 0xfe &&
-				data[i + 1] != 0xf7 {
-				i += 2
-				continue
-			} else {
-				return false
-			}
-		}
-	}
-	return true
-}
-
-func GbkToUtf8(s []byte) ([]byte, error) {
-	reader := transform.NewReader(bytes.NewReader(s), simplifiedchinese.GBK.NewDecoder())
-	d, e := ioutil.ReadAll(reader)
-	if e != nil {
-		return nil, e
-	}
-	return d, nil
-}