diff options
Diffstat (limited to 'vendor/github.com/bytedance/sonic/internal/decoder')
7 files changed, 154 insertions, 121 deletions
diff --git a/vendor/github.com/bytedance/sonic/internal/decoder/asm_stubs_amd64_go121.go b/vendor/github.com/bytedance/sonic/internal/decoder/asm_stubs_amd64_go121.go index 5db5b5cd0..6adeac0cf 100644 --- a/vendor/github.com/bytedance/sonic/internal/decoder/asm_stubs_amd64_go121.go +++ b/vendor/github.com/bytedance/sonic/internal/decoder/asm_stubs_amd64_go121.go @@ -1,4 +1,4 @@ -// +build go1.21,!go1.22 +// +build go1.21,!go1.23 // Copyright 2023 CloudWeGo Authors // diff --git a/vendor/github.com/bytedance/sonic/internal/decoder/assembler_regabi_amd64.go b/vendor/github.com/bytedance/sonic/internal/decoder/assembler_regabi_amd64.go index 3d223e14e..0defb75a5 100644 --- a/vendor/github.com/bytedance/sonic/internal/decoder/assembler_regabi_amd64.go +++ b/vendor/github.com/bytedance/sonic/internal/decoder/assembler_regabi_amd64.go @@ -1,5 +1,4 @@ -//go:build go1.17 && !go1.22 -// +build go1.17,!go1.22 +// +build go1.17,!go1.23 /* * Copyright 2021 ByteDance Inc. @@ -420,9 +419,9 @@ func (self *_Assembler) call_go(fn obj.Addr) { } func (self *_Assembler) callc(fn obj.Addr) { - self.Emit("XCHGQ", _IP, _BP) + self.save(_IP) self.call(fn) - self.Emit("XCHGQ", _IP, _BP) + self.load(_IP) } func (self *_Assembler) call_c(fn obj.Addr) { @@ -1164,7 +1163,7 @@ var ( var ( _F_FieldMap_GetCaseInsensitive obj.Addr - _Empty_Slice = make([]byte, 0) + _Empty_Slice = []byte{} _Zero_Base = int64(uintptr(((*rt.GoSlice)(unsafe.Pointer(&_Empty_Slice))).Ptr)) ) @@ -1641,7 +1640,8 @@ func (self *_Assembler) _asm_OP_check_empty(p *_Instr) { self.Emit("CMPB", jit.Sib(_IP, _IC, 1, 0), jit.Imm(int64(rbracket))) // CMPB (IP)(IC), ']' self.Sjmp("JNE" , "_not_empty_array_{n}") // JNE _not_empty_array_{n} self.Emit("MOVQ", _AX, _IC) // MOVQ AX, IC - self.StorePtr(_Zero_Base, jit.Ptr(_VP, 0), _AX) // MOVQ $zerobase, (VP) + self.Emit("MOVQ", jit.Imm(_Zero_Base), _AX) + self.WritePtrAX(9, jit.Ptr(_VP, 0), false) self.Emit("PXOR", _X0, _X0) // PXOR X0, X0 self.Emit("MOVOU", _X0, jit.Ptr(_VP, 8)) // MOVOU X0, 8(VP) self.Xjmp("JMP" , p.vi()) // JMP {p.vi()} diff --git a/vendor/github.com/bytedance/sonic/internal/decoder/assembler_stkabi_amd64.go b/vendor/github.com/bytedance/sonic/internal/decoder/assembler_stkabi_amd64.go index 57a38b420..9e2acc23f 100644 --- a/vendor/github.com/bytedance/sonic/internal/decoder/assembler_stkabi_amd64.go +++ b/vendor/github.com/bytedance/sonic/internal/decoder/assembler_stkabi_amd64.go @@ -1651,7 +1651,8 @@ func (self *_Assembler) _asm_OP_check_empty(p *_Instr) { self.Emit("CMPB", jit.Sib(_IP, _IC, 1, 0), jit.Imm(int64(rbracket))) // CMPB (IP)(IC), ']' self.Sjmp("JNE" , "_not_empty_array_{n}") // JNE _not_empty_array_{n} self.Emit("MOVQ", _AX, _IC) // MOVQ AX, IC - self.StorePtr(_Zero_Base, jit.Ptr(_VP, 0), _AX) // MOVQ $zerobase, (VP) + self.Emit("MOVQ", jit.Imm(_Zero_Base), _AX) + self.WritePtrAX(9, jit.Ptr(_VP, 0), false) self.Emit("PXOR" , _X0, _X0) // PXOR X0, X0 self.Emit("MOVOU", _X0, jit.Ptr(_VP, 8)) // MOVOU X0, 8(VP) self.Xjmp("JMP" , p.vi()) // JMP {p.vi()} diff --git a/vendor/github.com/bytedance/sonic/internal/decoder/compiler.go b/vendor/github.com/bytedance/sonic/internal/decoder/compiler.go index e9e2b77fc..b350c0461 100644 --- a/vendor/github.com/bytedance/sonic/internal/decoder/compiler.go +++ b/vendor/github.com/bytedance/sonic/internal/decoder/compiler.go @@ -527,40 +527,47 @@ func (self *_Compiler) compile(vt reflect.Type) (ret _Program, err error) { return } -func (self *_Compiler) compileOne(p *_Program, sp int, vt reflect.Type) { - /* check for recursive nesting */ - ok := self.tab[vt] - if ok { - p.rtt(_OP_recurse, vt) - return - } - +func (self *_Compiler) checkMarshaler(p *_Program, vt reflect.Type) bool { pt := reflect.PtrTo(vt) /* check for `json.Unmarshaler` with pointer receiver */ if pt.Implements(jsonUnmarshalerType) { p.rtt(_OP_unmarshal_p, pt) - return + return true } /* check for `json.Unmarshaler` */ if vt.Implements(jsonUnmarshalerType) { p.add(_OP_lspace) self.compileUnmarshalJson(p, vt) - return + return true } /* check for `encoding.TextMarshaler` with pointer receiver */ if pt.Implements(encodingTextUnmarshalerType) { p.add(_OP_lspace) self.compileUnmarshalTextPtr(p, pt) - return + return true } /* check for `encoding.TextUnmarshaler` */ if vt.Implements(encodingTextUnmarshalerType) { p.add(_OP_lspace) self.compileUnmarshalText(p, vt) + return true + } + return false +} + +func (self *_Compiler) compileOne(p *_Program, sp int, vt reflect.Type) { + /* check for recursive nesting */ + ok := self.tab[vt] + if ok { + p.rtt(_OP_recurse, vt) + return + } + + if self.checkMarshaler(p, vt) { return } @@ -683,17 +690,9 @@ func (self *_Compiler) compilePtr(p *_Program, sp int, et reflect.Type) { /* dereference all the way down */ for et.Kind() == reflect.Ptr { - if et.Implements(jsonUnmarshalerType) { - p.rtt(_OP_unmarshal_p, et) + if self.checkMarshaler(p, et) { return } - - if et.Implements(encodingTextUnmarshalerType) { - p.add(_OP_lspace) - self.compileUnmarshalTextPtr(p, et) - return - } - et = et.Elem() p.rtt(_OP_deref, et) } @@ -706,7 +705,7 @@ func (self *_Compiler) compilePtr(p *_Program, sp int, et reflect.Type) { /* enter the recursion */ p.add(_OP_lspace) self.tab[et] = true - + /* not inline the pointer type * recursing the defined pointer type's elem will casue issue379. */ @@ -716,8 +715,12 @@ func (self *_Compiler) compilePtr(p *_Program, sp int, et reflect.Type) { j := p.pc() p.add(_OP_goto) + + // set val pointer as nil p.pin(i) p.add(_OP_nil_1) + + // nothing todo p.pin(j) } diff --git a/vendor/github.com/bytedance/sonic/internal/decoder/generic_regabi_amd64.go b/vendor/github.com/bytedance/sonic/internal/decoder/generic_regabi_amd64.go index 337af054c..c7514cb41 100644 --- a/vendor/github.com/bytedance/sonic/internal/decoder/generic_regabi_amd64.go +++ b/vendor/github.com/bytedance/sonic/internal/decoder/generic_regabi_amd64.go @@ -1,4 +1,4 @@ -// +build go1.17,!go1.22 +// +build go1.17,!go1.23 /* * Copyright 2021 ByteDance Inc. @@ -119,9 +119,9 @@ func (self *_ValueDecoder) call_go(fn obj.Addr) { } func (self *_ValueDecoder) callc(fn obj.Addr) { - self.Emit("XCHGQ", _IP, _BP) + self.save(_IP) self.call(fn) - self.Emit("XCHGQ", _IP, _BP) + self.load(_IP) } func (self *_ValueDecoder) call_c(fn obj.Addr) { diff --git a/vendor/github.com/bytedance/sonic/internal/decoder/generic_regabi_amd64_test.s b/vendor/github.com/bytedance/sonic/internal/decoder/generic_regabi_amd64_test.s index 1c46928de..b4b0de183 100644 --- a/vendor/github.com/bytedance/sonic/internal/decoder/generic_regabi_amd64_test.s +++ b/vendor/github.com/bytedance/sonic/internal/decoder/generic_regabi_amd64_test.s @@ -1,4 +1,4 @@ -// +build go1.17,!go1.22 +// +build go1.17,!go1.23 // // Copyright 2021 ByteDance Inc. diff --git a/vendor/github.com/bytedance/sonic/internal/decoder/stream.go b/vendor/github.com/bytedance/sonic/internal/decoder/stream.go index a3716435a..7eb8a6951 100644 --- a/vendor/github.com/bytedance/sonic/internal/decoder/stream.go +++ b/vendor/github.com/bytedance/sonic/internal/decoder/stream.go @@ -23,11 +23,12 @@ import ( `github.com/bytedance/sonic/internal/native` `github.com/bytedance/sonic/internal/native/types` + `github.com/bytedance/sonic/internal/rt` `github.com/bytedance/sonic/option` ) var ( - minLeftBufferShift uint = 1 + minLeftBufferShift uint = 1 ) // StreamDecoder is the decoder context object for streaming input. @@ -58,95 +59,71 @@ func NewStreamDecoder(r io.Reader) *StreamDecoder { // Either io error from underlying io.Reader (except io.EOF) // or syntax error from data will be recorded and stop subsequently decoding. func (self *StreamDecoder) Decode(val interface{}) (err error) { - if self.err != nil { - return self.err - } - - var buf = self.buf[self.scanp:] - var p = 0 - var recycle bool - if cap(buf) == 0 { - buf = bufPool.Get().([]byte) - recycle = true - } - - var first = true - var repeat = true - -read_more: - for { - l := len(buf) - realloc(&buf) - n, err := self.r.Read(buf[l:cap(buf)]) - buf = buf[:l+n] - if err != nil { - repeat = false - if err == io.EOF { - if len(buf) == 0 { - return err - } - break - } - self.err = err - return err - } - if n > 0 || first { - break - } - } - first = false - - l := len(buf) - if l > 0 { - self.Decoder.Reset(string(buf)) - - var x int - if ret := native.SkipOneFast(&self.s, &x); ret < 0 { - if repeat { - goto read_more + // read more data into buf + if self.More() { + // println(string(self.buf)) + var s = self.scanp + try_skip: + var e = len(self.buf) + // println("s:", s, "e:", e, "scanned:",self.scanned, "scanp:",self.scanp, self.buf) + var src = rt.Mem2Str(self.buf[s:e]) + // if len(src) > 5 { + // println(src[:5], src[len(src)-5:]) + // } else { + // println(src) + // } + // try skip + var x = 0; + if y := native.SkipOneFast(&src, &x); y < 0 { + if self.readMore() { + // println("more") + goto try_skip } else { - err = SyntaxError{x, self.s, types.ParsingError(-ret), ""} - self.err = err + // println("no more") + err = SyntaxError{e, self.s, types.ParsingError(-s), ""} + self.setErr(err) return } + } else { + s = y + s + e = x + s } - + + // println("decode: ", s, e) + // must copy string here for safety + self.Decoder.Reset(string(self.buf[s:e])) err = self.Decoder.Decode(val) if err != nil { - self.err = err + self.setErr(err) + return } - p = self.Decoder.Pos() - self.scanned += int64(p) - self.scanp = 0 - } - - if l > p { - // remain undecoded bytes, so copy them into self.buf - self.buf = append(self.buf[:0], buf[p:]...) - } else { - self.buf = nil - recycle = true - } + self.scanp = e + _, empty := self.scan() + if empty { + // println("recycle") + // no remain valid bytes, thus we just recycle buffer + mem := self.buf + self.buf = nil + bufPool.Put(mem[:0]) + } else { + // println("keep") + // remain undecoded bytes, move them onto head + n := copy(self.buf, self.buf[self.scanp:]) + self.buf = self.buf[:n] + } - if recycle { - buf = buf[:0] - bufPool.Put(buf) - } - return err -} + self.scanned += int64(self.scanp) + self.scanp = 0 + } -func (self StreamDecoder) repeatable(err error) bool { - if ee, ok := err.(SyntaxError); ok && - (ee.Code == types.ERR_EOF || (ee.Code == types.ERR_INVALID_CHAR && self.i >= len(self.s)-1)) { - return true - } - return false + return self.err } // InputOffset returns the input stream byte offset of the current decoder position. // The offset gives the location of the end of the most recently returned token and the beginning of the next token. func (self *StreamDecoder) InputOffset() int64 { + // println("input offset",self.scanned, self.scanp) return self.scanned + int64(self.scanp) } @@ -166,28 +143,72 @@ func (self *StreamDecoder) More() bool { return err == nil && c != ']' && c != '}' } +// More reports whether there is another element in the +// current array or object being parsed. +func (self *StreamDecoder) readMore() bool { + if self.err != nil { + return false + } + + var err error + var n int + for { + // Grow buffer if not large enough. + l := len(self.buf) + realloc(&self.buf) + + n, err = self.r.Read(self.buf[l:cap(self.buf)]) + self.buf = self.buf[: l+n] + + self.scanp = l + _, empty := self.scan() + if !empty { + return true + } + + // buffer has been scanned, now report any error + if err != nil { + self.setErr(err) + return false + } + } +} + +func (self *StreamDecoder) setErr(err error) { + self.err = err + mem := self.buf[:0] + self.buf = nil + bufPool.Put(mem) +} + func (self *StreamDecoder) peek() (byte, error) { var err error for { - for i := self.scanp; i < len(self.buf); i++ { - c := self.buf[i] - if isSpace(c) { - continue - } - self.scanp = i - return c, nil + c, empty := self.scan() + if !empty { + return byte(c), nil } // buffer has been scanned, now report any error if err != nil { - if err != io.EOF { - self.err = err - } + self.setErr(err) return 0, err } err = self.refill() } } +func (self *StreamDecoder) scan() (byte, bool) { + for i := self.scanp; i < len(self.buf); i++ { + c := self.buf[i] + if isSpace(c) { + continue + } + self.scanp = i + return c, false + } + return 0, true +} + func isSpace(c byte) bool { return types.SPACE_MASK & (1 << c) != 0 } @@ -212,17 +233,25 @@ func (self *StreamDecoder) refill() error { return err } -func realloc(buf *[]byte) { +func realloc(buf *[]byte) bool { l := uint(len(*buf)) c := uint(cap(*buf)) + if c == 0 { + // println("use pool!") + *buf = bufPool.Get().([]byte) + return true + } if c - l <= c >> minLeftBufferShift { + // println("realloc!") e := l+(l>>minLeftBufferShift) - if e < option.DefaultDecoderBufferSize { - e = option.DefaultDecoderBufferSize + if e <= c { + e = c*2 } tmp := make([]byte, l, e) copy(tmp, *buf) *buf = tmp + return true } + return false } |