re
/
go-json
forked from mirror/go-json
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge pull request #237 from goccy/feature/fix-escaped-char-field 

Fix decoding fields containing escaped characters
This commit is contained in:
Masaaki Goshima 2021-06-03 15:18:48 +09:00 committed by GitHub
parent fee2d1fc68 917f71bbcf
commit f1ec7120aa
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 260 additions and 102 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

302
decode_struct.go
View File

@ -6,6 +6,8 @@ import (
"math/bits"
"sort"
"strings"
"unicode"
"unicode/utf16"
"unsafe"
)
@ -136,9 +138,54 @@ func (d *structDecoder) tryOptimize() {
}
}
// decode from '\uXXXX'
func decodeKeyCharByUnicodeRune(buf []byte, cursor int64) ([]byte, int64) {
const defaultOffset = 4
const surrogateOffset = 6
r := unicodeToRune(buf[cursor : cursor+defaultOffset])
if utf16.IsSurrogate(r) {
cursor += defaultOffset
if cursor+surrogateOffset >= int64(len(buf)) || buf[cursor] != '\\' || buf[cursor+1] != 'u' {
return []byte(string(unicode.ReplacementChar)), cursor + defaultOffset - 1
}
cursor += 2
r2 := unicodeToRune(buf[cursor : cursor+defaultOffset])
if r := utf16.DecodeRune(r, r2); r != unicode.ReplacementChar {
return []byte(string(r)), cursor + defaultOffset - 1
}
}
return []byte(string(r)), cursor + defaultOffset - 1
}
func decodeKeyCharByEscapedChar(buf []byte, cursor int64) ([]byte, int64) {
c := buf[cursor]
cursor++
switch c {
case '"':
return []byte{'"'}, cursor
case '\\':
return []byte{'\\'}, cursor
case '/':
return []byte{'/'}, cursor
case 'b':
return []byte{'\b'}, cursor
case 'f':
return []byte{'\f'}, cursor
case 'n':
return []byte{'\n'}, cursor
case 'r':
return []byte{'\r'}, cursor
case 't':
return []byte{'\t'}, cursor
case 'u':
return decodeKeyCharByUnicodeRune(buf, cursor)
}
return nil, cursor
}
func decodeKeyByBitmapUint8(d *structDecoder, buf []byte, cursor int64) (int64, *structFieldSet, error) {
var (
field *structFieldSet
curBit uint8 = math.MaxUint8
)
b := (*sliceHeader)(unsafe.Pointer(&buf)).data
@ -152,7 +199,7 @@ func decodeKeyByBitmapUint8(d *structDecoder, buf []byte, cursor int64) (int64,
switch c {
case '"':
cursor++
return cursor, field, nil
return cursor, nil, nil
case nul:
return 0, nil, errUnexpectedEndOfJSON("string", cursor)
}
@ -164,7 +211,7 @@ func decodeKeyByBitmapUint8(d *structDecoder, buf []byte, cursor int64) (int64,
switch c {
case '"':
fieldSetIndex := bits.TrailingZeros8(curBit)
field = d.sortedFieldSets[fieldSetIndex]
field := d.sortedFieldSets[fieldSetIndex]
keyLen := cursor - start
cursor++
if keyLen < field.keyLen {
@ -174,24 +221,21 @@ func decodeKeyByBitmapUint8(d *structDecoder, buf []byte, cursor int64) (int64,
return cursor, field, nil
case nul:
return 0, nil, errUnexpectedEndOfJSON("string", cursor)
case '\\':
cursor++
chars, nextCursor := decodeKeyCharByEscapedChar(buf, cursor)
for _, c := range chars {
curBit &= bitmap[keyIdx][largeToSmallTable[c]]
if curBit == 0 {
return decodeKeyNotFound(b, cursor)
}
keyIdx++
}
cursor = nextCursor
default:
curBit &= bitmap[keyIdx][largeToSmallTable[c]]
if curBit == 0 {
for {
cursor++
switch char(b, cursor) {
case '"':
cursor++
return cursor, field, nil
case '\\':
cursor++
if char(b, cursor) == nul {
return 0, nil, errUnexpectedEndOfJSON("string", cursor)
}
case nul:
return 0, nil, errUnexpectedEndOfJSON("string", cursor)
}
}
return decodeKeyNotFound(b, cursor)
}
keyIdx++
}
@ -205,7 +249,6 @@ func decodeKeyByBitmapUint8(d *structDecoder, buf []byte, cursor int64) (int64,
func decodeKeyByBitmapUint16(d *structDecoder, buf []byte, cursor int64) (int64, *structFieldSet, error) {
var (
field *structFieldSet
curBit uint16 = math.MaxUint16
)
b := (*sliceHeader)(unsafe.Pointer(&buf)).data
@ -219,7 +262,7 @@ func decodeKeyByBitmapUint16(d *structDecoder, buf []byte, cursor int64) (int64,
switch c {
case '"':
cursor++
return cursor, field, nil
return cursor, nil, nil
case nul:
return 0, nil, errUnexpectedEndOfJSON("string", cursor)
}
@ -231,7 +274,7 @@ func decodeKeyByBitmapUint16(d *structDecoder, buf []byte, cursor int64) (int64,
switch c {
case '"':
fieldSetIndex := bits.TrailingZeros16(curBit)
field = d.sortedFieldSets[fieldSetIndex]
field := d.sortedFieldSets[fieldSetIndex]
keyLen := cursor - start
cursor++
if keyLen < field.keyLen {
@ -241,24 +284,21 @@ func decodeKeyByBitmapUint16(d *structDecoder, buf []byte, cursor int64) (int64,
return cursor, field, nil
case nul:
return 0, nil, errUnexpectedEndOfJSON("string", cursor)
case '\\':
cursor++
chars, nextCursor := decodeKeyCharByEscapedChar(buf, cursor)
for _, c := range chars {
curBit &= bitmap[keyIdx][largeToSmallTable[c]]
if curBit == 0 {
return decodeKeyNotFound(b, cursor)
}
keyIdx++
}
cursor = nextCursor
default:
curBit &= bitmap[keyIdx][largeToSmallTable[c]]
if curBit == 0 {
for {
cursor++
switch char(b, cursor) {
case '"':
cursor++
return cursor, field, nil
case '\\':
cursor++
if char(b, cursor) == nul {
return 0, nil, errUnexpectedEndOfJSON("string", cursor)
}
case nul:
return 0, nil, errUnexpectedEndOfJSON("string", cursor)
}
}
return decodeKeyNotFound(b, cursor)
}
keyIdx++
}
@ -270,6 +310,24 @@ func decodeKeyByBitmapUint16(d *structDecoder, buf []byte, cursor int64) (int64,
}
}
func decodeKeyNotFound(b unsafe.Pointer, cursor int64) (int64, *structFieldSet, error) {
for {
cursor++
switch char(b, cursor) {
case '"':
cursor++
return cursor, nil, nil
case '\\':
cursor++
if char(b, cursor) == nul {
return 0, nil, errUnexpectedEndOfJSON("string", cursor)
}
case nul:
return 0, nil, errUnexpectedEndOfJSON("string", cursor)
}
}
}
func decodeKey(d *structDecoder, buf []byte, cursor int64) (int64, *structFieldSet, error) {
key, c, err := d.stringDecoder.decodeByte(buf, cursor)
if err != nil {
@ -286,10 +344,9 @@ func decodeKey(d *structDecoder, buf []byte, cursor int64) (int64, *structFieldS
func decodeKeyByBitmapUint8Stream(d *structDecoder, s *stream) (*structFieldSet, string, error) {
var (
field *structFieldSet
curBit uint8 = math.MaxUint8
)
buf, cursor, p := s.stat()
_, cursor, p := s.stat()
for {
switch char(p, cursor) {
case ' ', '\n', '\t', '\r':
@ -297,7 +354,7 @@ func decodeKeyByBitmapUint8Stream(d *structDecoder, s *stream) (*structFieldSet,
case nul:
s.cursor = cursor
if s.read() {
buf, cursor, p = s.stat()
_, cursor, p = s.stat()
continue
}
return nil, "", errNotAtBeginningOfValue(s.totalOffset())
@ -309,11 +366,11 @@ func decodeKeyByBitmapUint8Stream(d *structDecoder, s *stream) (*structFieldSet,
case '"':
cursor++
s.cursor = cursor
return field, "", nil
return nil, "", nil
case nul:
s.cursor = cursor
if s.read() {
buf, cursor, p = s.stat()
_, cursor, p = s.stat()
goto FIRST_CHAR
}
return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
@ -325,7 +382,7 @@ func decodeKeyByBitmapUint8Stream(d *structDecoder, s *stream) (*structFieldSet,
switch c {
case '"':
fieldSetIndex := bits.TrailingZeros8(curBit)
field = d.sortedFieldSets[fieldSetIndex]
field := d.sortedFieldSets[fieldSetIndex]
keyLen := cursor - start
cursor++
s.cursor = cursor
@ -337,39 +394,30 @@ func decodeKeyByBitmapUint8Stream(d *structDecoder, s *stream) (*structFieldSet,
case nul:
s.cursor = cursor
if s.read() {
buf, cursor, p = s.stat()
_, cursor, p = s.stat()
continue
}
return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
case '\\':
s.cursor = cursor + 1 // skip '\' char
chars, err := decodeKeyCharByEscapeCharStream(s)
if err != nil {
return nil, "", err
}
cursor = s.cursor
for _, c := range chars {
curBit &= bitmap[keyIdx][largeToSmallTable[c]]
if curBit == 0 {
s.cursor = cursor
return decodeKeyNotFoundStream(s, start)
}
keyIdx++
}
default:
curBit &= bitmap[keyIdx][largeToSmallTable[c]]
if curBit == 0 {
for {
cursor++
switch char(p, cursor) {
case '"':
b := buf[start:cursor]
key := *(*string)(unsafe.Pointer(&b))
cursor++
s.cursor = cursor
return field, key, nil
case '\\':
cursor++
if char(p, cursor) == nul {
s.cursor = cursor
if !s.read() {
return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
}
buf, cursor, p = s.statForRetry()
}
case nul:
s.cursor = cursor
if !s.read() {
return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
}
buf, cursor, p = s.statForRetry()
}
}
return decodeKeyNotFoundStream(s, start)
}
keyIdx++
}
@ -383,10 +431,9 @@ func decodeKeyByBitmapUint8Stream(d *structDecoder, s *stream) (*structFieldSet,
func decodeKeyByBitmapUint16Stream(d *structDecoder, s *stream) (*structFieldSet, string, error) {
var (
field *structFieldSet
curBit uint16 = math.MaxUint16
)
buf, cursor, p := s.stat()
_, cursor, p := s.stat()
for {
switch char(p, cursor) {
case ' ', '\n', '\t', '\r':
@ -394,7 +441,7 @@ func decodeKeyByBitmapUint16Stream(d *structDecoder, s *stream) (*structFieldSet
case nul:
s.cursor = cursor
if s.read() {
buf, cursor, p = s.stat()
_, cursor, p = s.stat()
continue
}
return nil, "", errNotAtBeginningOfValue(s.totalOffset())
@ -406,11 +453,11 @@ func decodeKeyByBitmapUint16Stream(d *structDecoder, s *stream) (*structFieldSet
case '"':
cursor++
s.cursor = cursor
return field, "", nil
return nil, "", nil
case nul:
s.cursor = cursor
if s.read() {
buf, cursor, p = s.stat()
_, cursor, p = s.stat()
goto FIRST_CHAR
}
return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
@ -422,7 +469,7 @@ func decodeKeyByBitmapUint16Stream(d *structDecoder, s *stream) (*structFieldSet
switch c {
case '"':
fieldSetIndex := bits.TrailingZeros16(curBit)
field = d.sortedFieldSets[fieldSetIndex]
field := d.sortedFieldSets[fieldSetIndex]
keyLen := cursor - start
cursor++
s.cursor = cursor
@ -434,13 +481,107 @@ func decodeKeyByBitmapUint16Stream(d *structDecoder, s *stream) (*structFieldSet
case nul:
s.cursor = cursor
if s.read() {
buf, cursor, p = s.stat()
_, cursor, p = s.stat()
continue
}
return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
case '\\':
s.cursor = cursor + 1 // skip '\' char
chars, err := decodeKeyCharByEscapeCharStream(s)
if err != nil {
return nil, "", err
}
cursor = s.cursor
for _, c := range chars {
curBit &= bitmap[keyIdx][largeToSmallTable[c]]
if curBit == 0 {
s.cursor = cursor
return decodeKeyNotFoundStream(s, start)
}
keyIdx++
}
default:
curBit &= bitmap[keyIdx][largeToSmallTable[c]]
if curBit == 0 {
s.cursor = cursor
return decodeKeyNotFoundStream(s, start)
}
keyIdx++
}
cursor++
}
default:
return nil, "", errNotAtBeginningOfValue(s.totalOffset())
}
}
}
// decode from '\uXXXX'
func decodeKeyCharByUnicodeRuneStream(s *stream) ([]byte, error) {
const defaultOffset = 4
const surrogateOffset = 6
if s.cursor+defaultOffset >= s.length {
if !s.read() {
return nil, errInvalidCharacter(s.char(), "escaped unicode char", s.totalOffset())
}
}
r := unicodeToRune(s.buf[s.cursor : s.cursor+defaultOffset])
if utf16.IsSurrogate(r) {
s.cursor += defaultOffset
if s.cursor+surrogateOffset >= s.length {
s.read()
}
if s.cursor+surrogateOffset >= s.length || s.buf[s.cursor] != '\\' || s.buf[s.cursor+1] != 'u' {
s.cursor += defaultOffset - 1
return []byte(string(unicode.ReplacementChar)), nil
}
r2 := unicodeToRune(s.buf[s.cursor+defaultOffset+2 : s.cursor+surrogateOffset])
if r := utf16.DecodeRune(r, r2); r != unicode.ReplacementChar {
s.cursor += defaultOffset - 1
return []byte(string(r)), nil
}
}
s.cursor += defaultOffset - 1
return []byte(string(r)), nil
}
func decodeKeyCharByEscapeCharStream(s *stream) ([]byte, error) {
c := s.buf[s.cursor]
s.cursor++
RETRY:
switch c {
case '"':
return []byte{'"'}, nil
case '\\':
return []byte{'\\'}, nil
case '/':
return []byte{'/'}, nil
case 'b':
return []byte{'\b'}, nil
case 'f':
return []byte{'\f'}, nil
case 'n':
return []byte{'\n'}, nil
case 'r':
return []byte{'\r'}, nil
case 't':
return []byte{'\t'}, nil
case 'u':
return decodeKeyCharByUnicodeRuneStream(s)
case nul:
if !s.read() {
return nil, errInvalidCharacter(s.char(), "escaped char", s.totalOffset())
}
goto RETRY
default:
return nil, errUnexpectedEndOfJSON("struct field", s.totalOffset())
}
}
func decodeKeyNotFoundStream(s *stream, start int64) (*structFieldSet, string, error) {
buf, cursor, p := s.stat()
for {
cursor++
switch char(p, cursor) {
@ -449,7 +590,7 @@ func decodeKeyByBitmapUint16Stream(d *structDecoder, s *stream) (*structFieldSet
key := *(*string)(unsafe.Pointer(&b))
cursor++
s.cursor = cursor
return field, key, nil
return nil, key, nil
case '\\':
cursor++
if char(p, cursor) == nul {
@ -467,15 +608,6 @@ func decodeKeyByBitmapUint16Stream(d *structDecoder, s *stream) (*structFieldSet
buf, cursor, p = s.statForRetry()
}
}
}
keyIdx++
}
cursor++
}
default:
return nil, "", errNotAtBeginningOfValue(s.totalOffset())
}
}
}
func decodeKeyStream(d *structDecoder, s *stream) (*structFieldSet, string, error) {

26
decode_test.go
View File

@ -3594,3 +3594,29 @@ func TestIssue218(t *testing.T) {
})
}
}
func TestDecodeEscapedCharField(t *testing.T) {
b := []byte(`{"\u6D88\u606F":"\u6D88\u606F"}`)
t.Run("unmarshal", func(t *testing.T) {
v := struct {
Msg string `json:"消息"`
}{}
if err := json.Unmarshal(b, &v); err != nil {
t.Fatal(err)
}
if !bytes.Equal([]byte(v.Msg), []byte("消息")) {
t.Fatal("failed to decode unicode char")
}
})
t.Run("stream", func(t *testing.T) {
v := struct {
Msg string `json:"消息"`
}{}
if err := json.NewDecoder(bytes.NewBuffer(b)).Decode(&v); err != nil {
t.Fatal(err)
}
if !bytes.Equal([]byte(v.Msg), []byte("消息")) {
t.Fatal("failed to decode unicode char")
}
})
}