|
| 1 | +package jsoniter |
| 2 | + |
| 3 | +import ( |
| 4 | + "fmt" |
| 5 | + "io" |
| 6 | +) |
| 7 | + |
| 8 | +// ValueType the type for JSON element |
| 9 | +type ValueType int |
| 10 | + |
| 11 | +const ( |
| 12 | + // InvalidValue invalid JSON element |
| 13 | + InvalidValue ValueType = iota |
| 14 | + // StringValue JSON element "string" |
| 15 | + StringValue |
| 16 | + // NumberValue JSON element 100 or 0.10 |
| 17 | + NumberValue |
| 18 | + // NilValue JSON element null |
| 19 | + NilValue |
| 20 | + // BoolValue JSON element true or false |
| 21 | + BoolValue |
| 22 | + // ArrayValue JSON element [] |
| 23 | + ArrayValue |
| 24 | + // ObjectValue JSON element {} |
| 25 | + ObjectValue |
| 26 | +) |
| 27 | +const uint32SafeToMultiply10 = uint32(0xffffffff)/10 - 1 |
| 28 | +const uint64SafeToMultiple10 = uint64(0xffffffffffffffff)/10 - 1 |
| 29 | +const maxFloat64 = 1<<53 - 1 |
| 30 | + |
| 31 | +var pow10 []uint64 |
| 32 | +var hexDigits []byte |
| 33 | +var valueTypes []ValueType |
| 34 | + |
| 35 | +func init() { |
| 36 | + pow10 = []uint64{1, 10, 100, 1000, 10000, 100000, 1000000} |
| 37 | + hexDigits = make([]byte, 256) |
| 38 | + for i := 0; i < len(hexDigits); i++ { |
| 39 | + hexDigits[i] = 255 |
| 40 | + } |
| 41 | + for i := '0'; i <= '9'; i++ { |
| 42 | + hexDigits[i] = byte(i - '0') |
| 43 | + } |
| 44 | + for i := 'a'; i <= 'f'; i++ { |
| 45 | + hexDigits[i] = byte((i - 'a') + 10) |
| 46 | + } |
| 47 | + for i := 'A'; i <= 'F'; i++ { |
| 48 | + hexDigits[i] = byte((i - 'A') + 10) |
| 49 | + } |
| 50 | + valueTypes = make([]ValueType, 256) |
| 51 | + for i := 0; i < len(valueTypes); i++ { |
| 52 | + valueTypes[i] = InvalidValue |
| 53 | + } |
| 54 | + valueTypes['"'] = StringValue |
| 55 | + valueTypes['-'] = NumberValue |
| 56 | + valueTypes['0'] = NumberValue |
| 57 | + valueTypes['1'] = NumberValue |
| 58 | + valueTypes['2'] = NumberValue |
| 59 | + valueTypes['3'] = NumberValue |
| 60 | + valueTypes['4'] = NumberValue |
| 61 | + valueTypes['5'] = NumberValue |
| 62 | + valueTypes['6'] = NumberValue |
| 63 | + valueTypes['7'] = NumberValue |
| 64 | + valueTypes['8'] = NumberValue |
| 65 | + valueTypes['9'] = NumberValue |
| 66 | + valueTypes['t'] = BoolValue |
| 67 | + valueTypes['f'] = BoolValue |
| 68 | + valueTypes['n'] = NilValue |
| 69 | + valueTypes['['] = ArrayValue |
| 70 | + valueTypes['{'] = ObjectValue |
| 71 | +} |
| 72 | + |
| 73 | +// Iterator is a io.Reader like object, with JSON specific read functions. |
| 74 | +// Error is not returned as return value, but stored as Error member on this iterator instance. |
| 75 | +type Iterator struct { |
| 76 | + buf []byte |
| 77 | + head int |
| 78 | + Error error |
| 79 | +} |
| 80 | + |
| 81 | +// ParseBytes creates an Iterator instance from byte array |
| 82 | +func ParseBytes(input []byte) *Iterator { |
| 83 | + return &Iterator{ |
| 84 | + buf: input, |
| 85 | + head: 0, |
| 86 | + } |
| 87 | +} |
| 88 | + |
| 89 | +// WhatIsNext gets ValueType of relatively next json element |
| 90 | +func (iter *Iterator) WhatIsNext() ValueType { |
| 91 | + valueType := valueTypes[iter.nextToken()] |
| 92 | + iter.unreadByte() |
| 93 | + return valueType |
| 94 | +} |
| 95 | + |
| 96 | +func (iter *Iterator) skipWhitespacesWithoutLoadMore() bool { |
| 97 | + for i := iter.head; i < len(iter.buf); i++ { |
| 98 | + c := iter.buf[i] |
| 99 | + switch c { |
| 100 | + case ' ', '\n', '\t', '\r': |
| 101 | + continue |
| 102 | + } |
| 103 | + iter.head = i |
| 104 | + return false |
| 105 | + } |
| 106 | + return true |
| 107 | +} |
| 108 | + |
| 109 | +func (iter *Iterator) isObjectEnd() bool { |
| 110 | + c := iter.nextToken() |
| 111 | + if c == ',' { |
| 112 | + return false |
| 113 | + } |
| 114 | + if c == '}' { |
| 115 | + return true |
| 116 | + } |
| 117 | + iter.ReportError("isObjectEnd", "object ended prematurely, unexpected char "+string([]byte{c})) |
| 118 | + return true |
| 119 | +} |
| 120 | + |
| 121 | +// ReportError record a error in iterator instance with current position. |
| 122 | +func (iter *Iterator) ReportError(operation string, msg string) { |
| 123 | + if iter.Error != nil { |
| 124 | + if iter.Error != io.EOF { |
| 125 | + return |
| 126 | + } |
| 127 | + } |
| 128 | + peekStart := iter.head - 10 |
| 129 | + if peekStart < 0 { |
| 130 | + peekStart = 0 |
| 131 | + } |
| 132 | + peekEnd := iter.head + 10 |
| 133 | + if peekEnd > len(iter.buf) { |
| 134 | + peekEnd = len(iter.buf) |
| 135 | + } |
| 136 | + parsing := string(iter.buf[peekStart:peekEnd]) |
| 137 | + contextStart := iter.head - 50 |
| 138 | + if contextStart < 0 { |
| 139 | + contextStart = 0 |
| 140 | + } |
| 141 | + contextEnd := iter.head + 50 |
| 142 | + if contextEnd > len(iter.buf) { |
| 143 | + contextEnd = len(iter.buf) |
| 144 | + } |
| 145 | + context := string(iter.buf[contextStart:contextEnd]) |
| 146 | + iter.Error = fmt.Errorf("%s: %s, error found in #%v byte of ...|%s|..., bigger context ...|%s|...", |
| 147 | + operation, msg, iter.head-peekStart, parsing, context) |
| 148 | +} |
| 149 | + |
| 150 | +// CurrentBuffer gets current buffer as string for debugging purpose |
| 151 | +func (iter *Iterator) CurrentBuffer() string { |
| 152 | + peekStart := iter.head - 10 |
| 153 | + if peekStart < 0 { |
| 154 | + peekStart = 0 |
| 155 | + } |
| 156 | + return fmt.Sprintf("parsing #%v byte, around ...|%s|..., whole buffer ...|%s|...", iter.head, |
| 157 | + string(iter.buf[peekStart:iter.head]), string(iter.buf[0:len(iter.buf)])) |
| 158 | +} |
| 159 | + |
| 160 | +func (iter *Iterator) readByte() (ret byte) { |
| 161 | + ret = iter.buf[iter.head] |
| 162 | + iter.head++ |
| 163 | + return ret |
| 164 | +} |
| 165 | + |
| 166 | +func (iter *Iterator) unreadByte() { |
| 167 | + if iter.Error != nil { |
| 168 | + return |
| 169 | + } |
| 170 | + iter.head-- |
| 171 | + return |
| 172 | +} |
| 173 | + |
| 174 | +func (iter *Iterator) nextToken() byte { |
| 175 | + // a variation of skip whitespaces, returning the next non-whitespace token |
| 176 | + for { |
| 177 | + for i := iter.head; i < len(iter.buf); i++ { |
| 178 | + c := iter.buf[i] |
| 179 | + switch c { |
| 180 | + case ' ', '\n', '\t', '\r': |
| 181 | + continue |
| 182 | + } |
| 183 | + iter.head = i + 1 |
| 184 | + return c |
| 185 | + } |
| 186 | + return 0 |
| 187 | + } |
| 188 | +} |
0 commit comments