Merge pull request #164 from goccy/feature/support-coerce-invalid-utf8

Support to coerce invalid utf8 char without performance down

decode_string.go

@@ -156,20 +156,13 @@ RETRY:
 	return nil
 }
 
-//nolint:deadcode,unused
-func appendCoerceInvalidUTF8(b []byte, s []byte) []byte {
-	c := [4]byte{}
-
-	for _, r := range string(s) {
-		b = append(b, c[:utf8.EncodeRune(c[:], r)]...)
-	}
-
-	return b
-}
+var (
+	runeErrBytes    = []byte(string(utf8.RuneError))
+	runeErrBytesLen = int64(len(runeErrBytes))
+)
 
 func stringBytes(s *stream) ([]byte, error) {
-	buf, cursor, p := s.stat()
-
+	_, cursor, p := s.stat()
 	cursor++ // skip double quote char
 	start := cursor
 	for {
@@ -179,21 +172,60 @@ func stringBytes(s *stream) ([]byte, error) {
 			if err := decodeEscapeString(s); err != nil {
 				return nil, err
 			}
-			buf, cursor, p = s.stat()
+			cursor = s.cursor
 		case '"':
-			literal := buf[start:cursor]
-			// TODO: this flow is so slow sequence.
-			// literal = appendCoerceInvalidUTF8(make([]byte, 0, len(literal)), literal)
+			literal := s.buf[start:cursor]
 			cursor++
 			s.cursor = cursor
 			return literal, nil
+		case
+			// 0x00 is nul, 0x5c is '\\', 0x22 is '"' .
+			0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F, // 0x00-0x0F
+			0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 0x18, 0x19, 0x1A, 0x1B, 0x1C, 0x1D, 0x1E, 0x1F, // 0x10-0x1F
+			0x20, 0x21 /*0x22,*/, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F, // 0x20-0x2F
+			0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3A, 0x3B, 0x3C, 0x3D, 0x3E, 0x3F, // 0x30-0x3F
+			0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F, // 0x40-0x4F
+			0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5A, 0x5B /*0x5C,*/, 0x5D, 0x5E, 0x5F, // 0x50-0x5F
+			0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F, // 0x60-0x6F
+			0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F: // 0x70-0x7F
+			// character is ASCII. skip to next char
+		case
+			0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8A, 0x8B, 0x8C, 0x8D, 0x8E, 0x8F, // 0x80-0x8F
+			0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0x9A, 0x9B, 0x9C, 0x9D, 0x9E, 0x9F, // 0x90-0x9F
+			0xA0, 0xA1, 0xA2, 0xA3, 0xA4, 0xA5, 0xA6, 0xA7, 0xA8, 0xA9, 0xAA, 0xAB, 0xAC, 0xAD, 0xAE, 0xAF, // 0xA0-0xAF
+			0xB0, 0xB1, 0xB2, 0xB3, 0xB4, 0xB5, 0xB6, 0xB7, 0xB8, 0xB9, 0xBA, 0xBB, 0xBC, 0xBD, 0xBE, 0xBF, // 0xB0-0xBF
+			0xC0, 0xC1, // 0xC0-0xC1
+			0xF5, 0xF6, 0xF7, 0xF8, 0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE, 0xFF: // 0xF5-0xFE
+			// character is invalid
+			s.buf = append(append(append([]byte{}, s.buf[:cursor]...), runeErrBytes...), s.buf[cursor+1:]...)
+			_, _, p = s.stat()
+			cursor += runeErrBytesLen
+			continue
 		case nul:
 			s.cursor = cursor
 			if s.read() {
-				buf, cursor, p = s.stat()
+				_, cursor, p = s.stat()
 				continue
 			}
 			goto ERROR
+		case 0xEF:
+			// RuneError is {0xEF, 0xBF, 0xBD}
+			if s.buf[cursor+1] == 0xBF && s.buf[cursor+2] == 0xBD {
+				// found RuneError: skip
+				cursor += 2
+				break
+			}
+			fallthrough
+		default:
+			// multi bytes character
+			r, _ := utf8.DecodeRune(s.buf[cursor:])
+			b := []byte(string(r))
+			if r == utf8.RuneError {
+				s.buf = append(append(append([]byte{}, s.buf[:cursor]...), b...), s.buf[cursor+1:]...)
+				_, _, p = s.stat()
+			}
+			cursor += int64(len(b))
+			continue
 		}
 		cursor++
 	}

decode_test.go

@@ -1277,29 +1277,27 @@ var unmarshalTests = []unmarshalTest{
 		ptr: new(map[string]json.Number),
 		err: fmt.Errorf(`strconv.ParseFloat: parsing "invalid": invalid syntax`),
 	},
-	/*
-		// invalid UTF-8 is coerced to valid UTF-8.
-			{
-				in:  "\"hello\xffworld\"", // 148
-				ptr: new(string),
-				out: "hello\ufffdworld",
-			},
-			{
-				in:  "\"hello\xc2\xc2world\"", // 149
-				ptr: new(string),
-				out: "hello\ufffd\ufffdworld",
-			},
-			{
-				in:  "\"hello\xc2\xffworld\"", // 150
-				ptr: new(string),
-				out: "hello\ufffd\ufffdworld",
-			},
-			{
-				in:  "\"hello\xed\xa0\x80\xed\xb0\x80world\"", // 151
-				ptr: new(string),
-				out: "hello\ufffd\ufffd\ufffd\ufffd\ufffd\ufffdworld",
-			},
-	*/
+	// invalid UTF-8 is coerced to valid UTF-8.
+	{
+		in:  "\"hello\xffworld\"", // 148
+		ptr: new(string),
+		out: "hello\ufffdworld",
+	},
+	{
+		in:  "\"hello\xc2\xc2world\"", // 149
+		ptr: new(string),
+		out: "hello\ufffd\ufffdworld",
+	},
+	{
+		in:  "\"hello\xc2\xffworld\"", // 150
+		ptr: new(string),
+		out: "hello\ufffd\ufffdworld",
+	},
+	{
+		in:  "\"hello\xed\xa0\x80\xed\xb0\x80world\"", // 151
+		ptr: new(string),
+		out: "hello\ufffd\ufffd\ufffd\ufffd\ufffd\ufffdworld",
+	},
 }
 
 type All struct {