123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556 |
- package decoder
- import (
- "bytes"
- "encoding/json"
- "io"
- "strconv"
- "unsafe"
- "github.com/goccy/go-json/internal/errors"
- )
- const (
- initBufSize = 512
- )
- type Stream struct {
- buf []byte
- bufSize int64
- length int64
- r io.Reader
- offset int64
- cursor int64
- filledBuffer bool
- allRead bool
- UseNumber bool
- DisallowUnknownFields bool
- Option *Option
- }
- func NewStream(r io.Reader) *Stream {
- return &Stream{
- r: r,
- bufSize: initBufSize,
- buf: make([]byte, initBufSize),
- Option: &Option{},
- }
- }
- func (s *Stream) TotalOffset() int64 {
- return s.totalOffset()
- }
- func (s *Stream) Buffered() io.Reader {
- buflen := int64(len(s.buf))
- for i := s.cursor; i < buflen; i++ {
- if s.buf[i] == nul {
- return bytes.NewReader(s.buf[s.cursor:i])
- }
- }
- return bytes.NewReader(s.buf[s.cursor:])
- }
- func (s *Stream) PrepareForDecode() error {
- for {
- switch s.char() {
- case ' ', '\t', '\r', '\n':
- s.cursor++
- continue
- case ',', ':':
- s.cursor++
- return nil
- case nul:
- if s.read() {
- continue
- }
- return io.EOF
- }
- break
- }
- return nil
- }
- func (s *Stream) totalOffset() int64 {
- return s.offset + s.cursor
- }
- func (s *Stream) char() byte {
- return s.buf[s.cursor]
- }
- func (s *Stream) equalChar(c byte) bool {
- cur := s.buf[s.cursor]
- if cur == nul {
- s.read()
- cur = s.buf[s.cursor]
- }
- return cur == c
- }
- func (s *Stream) stat() ([]byte, int64, unsafe.Pointer) {
- return s.buf, s.cursor, (*sliceHeader)(unsafe.Pointer(&s.buf)).data
- }
- func (s *Stream) bufptr() unsafe.Pointer {
- return (*sliceHeader)(unsafe.Pointer(&s.buf)).data
- }
- func (s *Stream) statForRetry() ([]byte, int64, unsafe.Pointer) {
- s.cursor-- // for retry ( because caller progress cursor position in each loop )
- return s.buf, s.cursor, (*sliceHeader)(unsafe.Pointer(&s.buf)).data
- }
- func (s *Stream) Reset() {
- s.reset()
- s.bufSize = int64(len(s.buf))
- }
- func (s *Stream) More() bool {
- for {
- switch s.char() {
- case ' ', '\n', '\r', '\t':
- s.cursor++
- continue
- case '}', ']':
- return false
- case nul:
- if s.read() {
- continue
- }
- return false
- }
- break
- }
- return true
- }
- func (s *Stream) Token() (interface{}, error) {
- for {
- c := s.char()
- switch c {
- case ' ', '\n', '\r', '\t':
- s.cursor++
- case '{', '[', ']', '}':
- s.cursor++
- return json.Delim(c), nil
- case ',', ':':
- s.cursor++
- case '-', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
- bytes := floatBytes(s)
- str := *(*string)(unsafe.Pointer(&bytes))
- if s.UseNumber {
- return json.Number(str), nil
- }
- f64, err := strconv.ParseFloat(str, 64)
- if err != nil {
- return nil, err
- }
- return f64, nil
- case '"':
- bytes, err := stringBytes(s)
- if err != nil {
- return nil, err
- }
- return string(bytes), nil
- case 't':
- if err := trueBytes(s); err != nil {
- return nil, err
- }
- return true, nil
- case 'f':
- if err := falseBytes(s); err != nil {
- return nil, err
- }
- return false, nil
- case 'n':
- if err := nullBytes(s); err != nil {
- return nil, err
- }
- return nil, nil
- case nul:
- if s.read() {
- continue
- }
- goto END
- default:
- return nil, errors.ErrInvalidCharacter(s.char(), "token", s.totalOffset())
- }
- }
- END:
- return nil, io.EOF
- }
- func (s *Stream) reset() {
- s.offset += s.cursor
- s.buf = s.buf[s.cursor:]
- s.length -= s.cursor
- s.cursor = 0
- }
- func (s *Stream) readBuf() []byte {
- if s.filledBuffer {
- s.bufSize *= 2
- remainBuf := s.buf
- s.buf = make([]byte, s.bufSize)
- copy(s.buf, remainBuf)
- }
- remainLen := s.length - s.cursor
- remainNotNulCharNum := int64(0)
- for i := int64(0); i < remainLen; i++ {
- if s.buf[s.cursor+i] == nul {
- break
- }
- remainNotNulCharNum++
- }
- s.length = s.cursor + remainNotNulCharNum
- return s.buf[s.cursor+remainNotNulCharNum:]
- }
- func (s *Stream) read() bool {
- if s.allRead {
- return false
- }
- buf := s.readBuf()
- last := len(buf) - 1
- buf[last] = nul
- n, err := s.r.Read(buf[:last])
- s.length += int64(n)
- if n == last {
- s.filledBuffer = true
- } else {
- s.filledBuffer = false
- }
- if err == io.EOF {
- s.allRead = true
- } else if err != nil {
- return false
- }
- return true
- }
- func (s *Stream) skipWhiteSpace() byte {
- p := s.bufptr()
- LOOP:
- c := char(p, s.cursor)
- switch c {
- case ' ', '\n', '\t', '\r':
- s.cursor++
- goto LOOP
- case nul:
- if s.read() {
- p = s.bufptr()
- goto LOOP
- }
- }
- return c
- }
- func (s *Stream) skipObject(depth int64) error {
- braceCount := 1
- _, cursor, p := s.stat()
- for {
- switch char(p, cursor) {
- case '{':
- braceCount++
- depth++
- if depth > maxDecodeNestingDepth {
- return errors.ErrExceededMaxDepth(s.char(), s.cursor)
- }
- case '}':
- braceCount--
- depth--
- if braceCount == 0 {
- s.cursor = cursor + 1
- return nil
- }
- case '[':
- depth++
- if depth > maxDecodeNestingDepth {
- return errors.ErrExceededMaxDepth(s.char(), s.cursor)
- }
- case ']':
- depth--
- case '"':
- for {
- cursor++
- switch char(p, cursor) {
- case '\\':
- cursor++
- if char(p, cursor) == nul {
- s.cursor = cursor
- if s.read() {
- _, cursor, p = s.stat()
- continue
- }
- return errors.ErrUnexpectedEndOfJSON("string of object", cursor)
- }
- case '"':
- goto SWITCH_OUT
- case nul:
- s.cursor = cursor
- if s.read() {
- _, cursor, p = s.statForRetry()
- continue
- }
- return errors.ErrUnexpectedEndOfJSON("string of object", cursor)
- }
- }
- case nul:
- s.cursor = cursor
- if s.read() {
- _, cursor, p = s.stat()
- continue
- }
- return errors.ErrUnexpectedEndOfJSON("object of object", cursor)
- }
- SWITCH_OUT:
- cursor++
- }
- }
- func (s *Stream) skipArray(depth int64) error {
- bracketCount := 1
- _, cursor, p := s.stat()
- for {
- switch char(p, cursor) {
- case '[':
- bracketCount++
- depth++
- if depth > maxDecodeNestingDepth {
- return errors.ErrExceededMaxDepth(s.char(), s.cursor)
- }
- case ']':
- bracketCount--
- depth--
- if bracketCount == 0 {
- s.cursor = cursor + 1
- return nil
- }
- case '{':
- depth++
- if depth > maxDecodeNestingDepth {
- return errors.ErrExceededMaxDepth(s.char(), s.cursor)
- }
- case '}':
- depth--
- case '"':
- for {
- cursor++
- switch char(p, cursor) {
- case '\\':
- cursor++
- if char(p, cursor) == nul {
- s.cursor = cursor
- if s.read() {
- _, cursor, p = s.stat()
- continue
- }
- return errors.ErrUnexpectedEndOfJSON("string of object", cursor)
- }
- case '"':
- goto SWITCH_OUT
- case nul:
- s.cursor = cursor
- if s.read() {
- _, cursor, p = s.statForRetry()
- continue
- }
- return errors.ErrUnexpectedEndOfJSON("string of object", cursor)
- }
- }
- case nul:
- s.cursor = cursor
- if s.read() {
- _, cursor, p = s.stat()
- continue
- }
- return errors.ErrUnexpectedEndOfJSON("array of object", cursor)
- }
- SWITCH_OUT:
- cursor++
- }
- }
- func (s *Stream) skipValue(depth int64) error {
- _, cursor, p := s.stat()
- for {
- switch char(p, cursor) {
- case ' ', '\n', '\t', '\r':
- cursor++
- continue
- case nul:
- s.cursor = cursor
- if s.read() {
- _, cursor, p = s.stat()
- continue
- }
- return errors.ErrUnexpectedEndOfJSON("value of object", s.totalOffset())
- case '{':
- s.cursor = cursor + 1
- return s.skipObject(depth + 1)
- case '[':
- s.cursor = cursor + 1
- return s.skipArray(depth + 1)
- case '"':
- for {
- cursor++
- switch char(p, cursor) {
- case '\\':
- cursor++
- if char(p, cursor) == nul {
- s.cursor = cursor
- if s.read() {
- _, cursor, p = s.stat()
- continue
- }
- return errors.ErrUnexpectedEndOfJSON("value of string", s.totalOffset())
- }
- case '"':
- s.cursor = cursor + 1
- return nil
- case nul:
- s.cursor = cursor
- if s.read() {
- _, cursor, p = s.statForRetry()
- continue
- }
- return errors.ErrUnexpectedEndOfJSON("value of string", s.totalOffset())
- }
- }
- case '-', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
- for {
- cursor++
- c := char(p, cursor)
- if floatTable[c] {
- continue
- } else if c == nul {
- if s.read() {
- _, cursor, p = s.stat()
- continue
- }
- }
- s.cursor = cursor
- return nil
- }
- case 't':
- s.cursor = cursor
- if err := trueBytes(s); err != nil {
- return err
- }
- return nil
- case 'f':
- s.cursor = cursor
- if err := falseBytes(s); err != nil {
- return err
- }
- return nil
- case 'n':
- s.cursor = cursor
- if err := nullBytes(s); err != nil {
- return err
- }
- return nil
- }
- cursor++
- }
- }
- func nullBytes(s *Stream) error {
- // current cursor's character is 'n'
- s.cursor++
- if s.char() != 'u' {
- if err := retryReadNull(s); err != nil {
- return err
- }
- }
- s.cursor++
- if s.char() != 'l' {
- if err := retryReadNull(s); err != nil {
- return err
- }
- }
- s.cursor++
- if s.char() != 'l' {
- if err := retryReadNull(s); err != nil {
- return err
- }
- }
- s.cursor++
- return nil
- }
- func retryReadNull(s *Stream) error {
- if s.char() == nul && s.read() {
- return nil
- }
- return errors.ErrInvalidCharacter(s.char(), "null", s.totalOffset())
- }
- func trueBytes(s *Stream) error {
- // current cursor's character is 't'
- s.cursor++
- if s.char() != 'r' {
- if err := retryReadTrue(s); err != nil {
- return err
- }
- }
- s.cursor++
- if s.char() != 'u' {
- if err := retryReadTrue(s); err != nil {
- return err
- }
- }
- s.cursor++
- if s.char() != 'e' {
- if err := retryReadTrue(s); err != nil {
- return err
- }
- }
- s.cursor++
- return nil
- }
- func retryReadTrue(s *Stream) error {
- if s.char() == nul && s.read() {
- return nil
- }
- return errors.ErrInvalidCharacter(s.char(), "bool(true)", s.totalOffset())
- }
- func falseBytes(s *Stream) error {
- // current cursor's character is 'f'
- s.cursor++
- if s.char() != 'a' {
- if err := retryReadFalse(s); err != nil {
- return err
- }
- }
- s.cursor++
- if s.char() != 'l' {
- if err := retryReadFalse(s); err != nil {
- return err
- }
- }
- s.cursor++
- if s.char() != 's' {
- if err := retryReadFalse(s); err != nil {
- return err
- }
- }
- s.cursor++
- if s.char() != 'e' {
- if err := retryReadFalse(s); err != nil {
- return err
- }
- }
- s.cursor++
- return nil
- }
- func retryReadFalse(s *Stream) error {
- if s.char() == nul && s.read() {
- return nil
- }
- return errors.ErrInvalidCharacter(s.char(), "bool(false)", s.totalOffset())
- }
|