Merge pull request #721 from ttys3/fixup-isBinary

fixup func isBinary to handle CJK runes correctly

Former-commit-id: 08c8613759eed7d4790c9fea99273f60e88531a0 [formerly 82af4df6a155423555865a4d16a62c74befcefd1] [formerly 20751c4cd90807c51145c5284573509f26e5c934 [formerly 22bbad84fb]]
Former-commit-id: 4561311363fd52b51838e71408ff270c2f376c97 [formerly e6452a14dd39fa07dc8d77a2b47b7a03b58c0029]
Former-commit-id: aca07de59fcdb2f079b839572cb431c792719158

files/file.go

@@ -164,7 +164,7 @@ func (i *FileInfo) detectType(modify, saveContent bool) error {
 	case strings.HasPrefix(mimetype, "image"):
 		i.Type = "image"
 		return nil
-	case isBinary(string(buffer[:n])) || i.Size > 10*1024*1024: // 10 MB
+	case isBinary(buffer[:n], n) || i.Size > 10*1024*1024: // 10 MB
 		i.Type = "blob"
 		return nil
 	default:

files/utils.go

@@ -1,12 +1,46 @@
 package files
 
-func isBinary(content string) bool {
+import (
-	for _, b := range content {
+	"unicode/utf8"
-		// 65533 is the unknown char
+)
+
+func isBinary(content []byte, n int) bool {
+	maybeStr := string(content)
+	runeCnt := utf8.RuneCount(content)
+	runeIndex := 0
+	gotRuneErrCnt := 0
+	firstRuneErrIndex := -1
+
+	for _, b := range maybeStr {
 		// 8 and below are control chars (e.g. backspace, null, eof, etc)
-		if b <= 8 || b == 65533 {
+		if b <= 8 {
 			return true
 		}
+
+		// 0xFFFD(65533) is  the "error" Rune or "Unicode replacement character"
+		// see https://golang.org/pkg/unicode/utf8/#pkg-constants
+		if b == 0xFFFD {
+			//if it is not the last (utf8.UTFMax - x) rune
+			if runeCnt > utf8.UTFMax && runeIndex < runeCnt-utf8.UTFMax {
+				return true
+			} else {
+				//else it is the last (utf8.UTFMax - x) rune
+				//there maybe Vxxx, VVxx, VVVx, thus, we may got max 3 0xFFFD rune (asume V is the byte we got)
+				//for Chinese, it can only be Vxx, VVx, we may got max 2 0xFFFD rune
+				gotRuneErrCnt++
+
+				//mark the first time
+				if firstRuneErrIndex == -1 {
+					firstRuneErrIndex = runeIndex
+				}
+			}
+		}
+		runeIndex++
+	}
+
+	//if last (utf8.UTFMax - x ) rune has the "error" Rune, but not all
+	if firstRuneErrIndex != -1 && gotRuneErrCnt != runeCnt-firstRuneErrIndex {
+		return true
 	}
 	return false
 }