Merge pull request #135 from goccy/feature/merge-stream

Optimize streaming decoder by BCE

decode_stream.go

@@ -3,6 +3,7 @@ package json
 import (
 	"bytes"
 	"io"
+	"unsafe"
 )
 
 const (
@@ -43,14 +44,14 @@ func (s *stream) totalOffset() int64 {
 	return s.offset + s.cursor
 }
 
-func (s *stream) prevChar() byte {
-	return s.buf[s.cursor-1]
-}
-
 func (s *stream) char() byte {
 	return s.buf[s.cursor]
 }
 
+func (s *stream) stat() ([]byte, int64, unsafe.Pointer) {
+	return s.buf, s.cursor, (*sliceHeader)(unsafe.Pointer(&s.buf)).data
+}
+
 func (s *stream) reset() {
 	s.offset += s.cursor
 	s.buf = s.buf[s.cursor:]
@@ -96,107 +97,171 @@ LOOP:
 	}
 }
 
-func (s *stream) skipValue() error {
-	s.skipWhiteSpace()
-	braceCount := 0
-	bracketCount := 0
-	start := s.cursor
+func (s *stream) skipObject() error {
+	braceCount := 1
+	_, cursor, p := s.stat()
 	for {
-		switch s.char() {
-		case nul:
-			if s.read() {
-				continue
-			}
-			if start == s.cursor {
-				return errUnexpectedEndOfJSON("value of object", s.totalOffset())
-			}
-			if braceCount == 0 && bracketCount == 0 {
-				return nil
-			}
-			return errUnexpectedEndOfJSON("value of object", s.totalOffset())
+		switch char(p, cursor) {
 		case '{':
 			braceCount++
-		case '[':
-			bracketCount++
 		case '}':
 			braceCount--
-			if braceCount == -1 && bracketCount == 0 {
-				return nil
-			}
-		case ']':
-			bracketCount--
-			if braceCount == 0 && bracketCount == -1 {
-				return nil
-			}
-		case ',':
-			if bracketCount == 0 && braceCount == 0 {
+			if braceCount == 0 {
+				s.cursor = cursor + 1
 				return nil
 			}
 		case '"':
 			for {
-				s.cursor++
-				c := s.char()
-				if c == nul {
-					if !s.read() {
-						return errUnexpectedEndOfJSON("value of string", s.totalOffset())
-					}
-					c = s.char()
-				}
-				if c != '"' {
+				cursor++
+				switch char(p, cursor) {
+				case '"':
+					if char(p, cursor-1) == '\\' {
 						continue
 					}
-				if s.prevChar() == '\\' {
+					goto SWITCH_OUT
+				case nul:
+					s.cursor = cursor
+					if s.read() {
+						s.cursor-- // for retry current character
+						_, cursor, p = s.stat()
 						continue
 					}
-				if bracketCount == 0 && braceCount == 0 {
-					s.cursor++
+					return errUnexpectedEndOfJSON("string of object", cursor)
+				}
+			}
+		case nul:
+			s.cursor = cursor
+			if s.read() {
+				_, cursor, p = s.stat()
+				continue
+			}
+			return errUnexpectedEndOfJSON("object of object", cursor)
+		}
+	SWITCH_OUT:
+		cursor++
+	}
+}
+
+func (s *stream) skipArray() error {
+	bracketCount := 1
+	_, cursor, p := s.stat()
+	for {
+		switch char(p, cursor) {
+		case '[':
+			bracketCount++
+		case ']':
+			bracketCount--
+			if bracketCount == 0 {
+				s.cursor = cursor + 1
 				return nil
 			}
-				break
+		case '"':
+			for {
+				cursor++
+				switch char(p, cursor) {
+				case '"':
+					if char(p, cursor-1) == '\\' {
+						continue
+					}
+					goto SWITCH_OUT
+				case nul:
+					s.cursor = cursor
+					if s.read() {
+						s.cursor-- // for retry current character
+						_, cursor, p = s.stat()
+						continue
+					}
+					return errUnexpectedEndOfJSON("string of object", cursor)
+				}
+			}
+		case nul:
+			s.cursor = cursor
+			if s.read() {
+				_, cursor, p = s.stat()
+				continue
+			}
+			return errUnexpectedEndOfJSON("array of object", cursor)
+		}
+	SWITCH_OUT:
+		cursor++
+	}
+}
+
+func (s *stream) skipValue() 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 errUnexpectedEndOfJSON("value of object", s.totalOffset())
+		case '{':
+			s.cursor = cursor + 1
+			return s.skipObject()
+		case '[':
+			s.cursor = cursor + 1
+			return s.skipArray()
+		case '"':
+			for {
+				cursor++
+				switch char(p, cursor) {
+				case '"':
+					if char(p, cursor-1) == '\\' {
+						continue
+					}
+					s.cursor = cursor + 1
+					return nil
+				case nul:
+					s.cursor = cursor
+					if s.read() {
+						s.cursor-- // for retry current character
+						_, cursor, p = s.stat()
+						continue
+					}
+					return errUnexpectedEndOfJSON("value of string", s.totalOffset())
+				}
 			}
 		case '-', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
 			for {
-				s.cursor++
-				c := s.char()
+				cursor++
+				c := char(p, cursor)
 				if floatTable[c] {
 					continue
 				} else if c == nul {
 					if s.read() {
 						s.cursor-- // for retry current character
+						_, cursor, p = s.stat()
 						continue
 					}
 				}
-				break
-			}
-			if bracketCount == 0 && braceCount == 0 {
+				s.cursor = cursor
 				return nil
 			}
-			continue
 		case 't':
+			s.cursor = cursor
 			if err := trueBytes(s); err != nil {
 				return err
 			}
-			if bracketCount == 0 && braceCount == 0 {
 			return nil
-			}
-			continue
 		case 'f':
+			s.cursor = cursor
 			if err := falseBytes(s); err != nil {
 				return err
 			}
-			if bracketCount == 0 && braceCount == 0 {
 			return nil
-			}
-			continue
 		case 'n':
+			s.cursor = cursor
 			if err := nullBytes(s); err != nil {
 				return err
 			}
-			if bracketCount == 0 && braceCount == 0 {
 			return nil
 		}
-			continue
-		}
-		s.cursor++
+		cursor++
 	}
 }

decode_string.go

@@ -162,27 +162,34 @@ func appendCoerceInvalidUTF8(b []byte, s []byte) []byte {
 }
 
 func stringBytes(s *stream) ([]byte, error) {
-	s.cursor++
-	start := s.cursor
+	buf, cursor, p := s.stat()
+
+	cursor++ // skip double quote char
+	start := cursor
 	for {
-		switch s.char() {
+		switch char(p, cursor) {
 		case '\\':
+			s.cursor = cursor
 			if err := decodeEscapeString(s); err != nil {
 				return nil, err
 			}
+			buf, cursor, p = s.stat()
 		case '"':
-			literal := s.buf[start:s.cursor]
+			literal := buf[start:cursor]
 			// TODO: this flow is so slow sequence.
 			// literal = appendCoerceInvalidUTF8(make([]byte, 0, len(literal)), literal)
-			s.cursor++
+			cursor++
+			s.cursor = cursor
 			return literal, nil
 		case nul:
+			s.cursor = cursor
 			if s.read() {
+				buf, cursor, p = s.stat()
 				continue
 			}
 			goto ERROR
 		}
-		s.cursor++
+		cursor++
 	}
 ERROR:
 	return nil, errUnexpectedEndOfJSON("string", s.totalOffset())

decode_struct.go

@@ -288,25 +288,31 @@ func decodeKeyByBitmapUint8Stream(d *structDecoder, s *stream) (*structFieldSet,
 		field  *structFieldSet
 		curBit uint8 = math.MaxUint8
 	)
+	buf, cursor, p := s.stat()
 	for {
-		switch s.char() {
+		switch char(p, cursor) {
 		case ' ', '\n', '\t', '\r':
-			s.cursor++
+			cursor++
 		case nul:
+			s.cursor = cursor
 			if s.read() {
+				buf, cursor, p = s.stat()
 				continue
 			}
 			return nil, "", errNotAtBeginningOfValue(s.totalOffset())
 		case '"':
-			s.cursor++
+			cursor++
 		FIRST_CHAR:
-			start := s.cursor
-			switch s.char() {
+			start := cursor
+			switch char(p, cursor) {
 			case '"':
-				s.cursor++
+				cursor++
+				s.cursor = cursor
 				return field, "", nil
 			case nul:
+				s.cursor = cursor
 				if s.read() {
+					buf, cursor, p = s.stat()
 					goto FIRST_CHAR
 				}
 				return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
@@ -314,20 +320,23 @@ func decodeKeyByBitmapUint8Stream(d *structDecoder, s *stream) (*structFieldSet,
 			keyIdx := 0
 			bitmap := d.keyBitmapUint8
 			for {
-				c := s.char()
+				c := char(p, cursor)
 				switch c {
 				case '"':
 					fieldSetIndex := bits.TrailingZeros8(curBit)
 					field = d.sortedFieldSets[fieldSetIndex]
-					keyLen := s.cursor - start
-					s.cursor++
+					keyLen := cursor - start
+					cursor++
+					s.cursor = cursor
 					if keyLen < field.keyLen {
 						// early match
 						return nil, field.key, nil
 					}
 					return field, field.key, nil
 				case nul:
+					s.cursor = cursor
 					if s.read() {
+						buf, cursor, p = s.stat()
 						continue
 					}
 					return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
@@ -335,30 +344,35 @@ func decodeKeyByBitmapUint8Stream(d *structDecoder, s *stream) (*structFieldSet,
 					curBit &= bitmap[keyIdx][largeToSmallTable[c]]
 					if curBit == 0 {
 						for {
-							s.cursor++
-							switch s.char() {
+							cursor++
+							switch char(p, cursor) {
 							case '"':
-								b := s.buf[start:s.cursor]
+								b := buf[start:cursor]
 								key := *(*string)(unsafe.Pointer(&b))
-								s.cursor++
+								cursor++
+								s.cursor = cursor
 								return field, key, nil
 							case '\\':
-								s.cursor++
-								if s.char() == nul {
+								cursor++
+								if char(p, cursor) == nul {
+									s.cursor = cursor
 									if !s.read() {
 										return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
 									}
+									buf, cursor, p = s.stat()
 								}
 							case nul:
+								s.cursor = cursor
 								if !s.read() {
 									return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
 								}
+								buf, cursor, p = s.stat()
 							}
 						}
 					}
 					keyIdx++
 				}
-				s.cursor++
+				cursor++
 			}
 		default:
 			return nil, "", errNotAtBeginningOfValue(s.totalOffset())
@@ -371,25 +385,31 @@ func decodeKeyByBitmapUint16Stream(d *structDecoder, s *stream) (*structFieldSet
 		field  *structFieldSet
 		curBit uint16 = math.MaxUint16
 	)
+	buf, cursor, p := s.stat()
 	for {
-		switch s.char() {
+		switch char(p, cursor) {
 		case ' ', '\n', '\t', '\r':
-			s.cursor++
+			cursor++
 		case nul:
+			s.cursor = cursor
 			if s.read() {
+				buf, cursor, p = s.stat()
 				continue
 			}
 			return nil, "", errNotAtBeginningOfValue(s.totalOffset())
 		case '"':
-			s.cursor++
+			cursor++
 		FIRST_CHAR:
-			start := s.cursor
-			switch s.char() {
+			start := cursor
+			switch char(p, cursor) {
 			case '"':
-				s.cursor++
+				cursor++
+				s.cursor = cursor
 				return field, "", nil
 			case nul:
+				s.cursor = cursor
 				if s.read() {
+					buf, cursor, p = s.stat()
 					goto FIRST_CHAR
 				}
 				return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
@@ -397,20 +417,23 @@ func decodeKeyByBitmapUint16Stream(d *structDecoder, s *stream) (*structFieldSet
 			keyIdx := 0
 			bitmap := d.keyBitmapUint16
 			for {
-				c := s.char()
+				c := char(p, cursor)
 				switch c {
 				case '"':
 					fieldSetIndex := bits.TrailingZeros16(curBit)
 					field = d.sortedFieldSets[fieldSetIndex]
-					keyLen := s.cursor - start
-					s.cursor++
+					keyLen := cursor - start
+					cursor++
+					s.cursor = cursor
 					if keyLen < field.keyLen {
 						// early match
 						return nil, field.key, nil
 					}
 					return field, field.key, nil
 				case nul:
+					s.cursor = cursor
 					if s.read() {
+						buf, cursor, p = s.stat()
 						continue
 					}
 					return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
@@ -418,30 +441,35 @@ func decodeKeyByBitmapUint16Stream(d *structDecoder, s *stream) (*structFieldSet
 					curBit &= bitmap[keyIdx][largeToSmallTable[c]]
 					if curBit == 0 {
 						for {
-							s.cursor++
-							switch s.char() {
+							cursor++
+							switch char(p, cursor) {
 							case '"':
-								b := s.buf[start:s.cursor]
+								b := buf[start:cursor]
 								key := *(*string)(unsafe.Pointer(&b))
-								s.cursor++
+								cursor++
+								s.cursor = cursor
 								return field, key, nil
 							case '\\':
-								s.cursor++
-								if s.char() == nul {
+								cursor++
+								if char(p, cursor) == nul {
+									s.cursor = cursor
 									if !s.read() {
 										return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
 									}
+									buf, cursor, p = s.stat()
 								}
 							case nul:
+								s.cursor = cursor
 								if !s.read() {
 									return nil, "", errUnexpectedEndOfJSON("string", s.totalOffset())
 								}
+								buf, cursor, p = s.stat()
 							}
 						}
 					}
 					keyIdx++
 				}
-				s.cursor++
+				cursor++
 			}
 		default:
 			return nil, "", errNotAtBeginningOfValue(s.totalOffset())