Zack Williams | e940c7a | 2019-08-21 14:25:39 -0700 | [diff] [blame] | 1 | package jsoniter |
| 2 | |
| 3 | import ( |
| 4 | "encoding/json" |
| 5 | "fmt" |
| 6 | "io" |
| 7 | ) |
| 8 | |
| 9 | // ValueType the type for JSON element |
| 10 | type ValueType int |
| 11 | |
| 12 | const ( |
| 13 | // InvalidValue invalid JSON element |
| 14 | InvalidValue ValueType = iota |
| 15 | // StringValue JSON element "string" |
| 16 | StringValue |
| 17 | // NumberValue JSON element 100 or 0.10 |
| 18 | NumberValue |
| 19 | // NilValue JSON element null |
| 20 | NilValue |
| 21 | // BoolValue JSON element true or false |
| 22 | BoolValue |
| 23 | // ArrayValue JSON element [] |
| 24 | ArrayValue |
| 25 | // ObjectValue JSON element {} |
| 26 | ObjectValue |
| 27 | ) |
| 28 | |
| 29 | var hexDigits []byte |
| 30 | var valueTypes []ValueType |
| 31 | |
| 32 | func init() { |
| 33 | hexDigits = make([]byte, 256) |
| 34 | for i := 0; i < len(hexDigits); i++ { |
| 35 | hexDigits[i] = 255 |
| 36 | } |
| 37 | for i := '0'; i <= '9'; i++ { |
| 38 | hexDigits[i] = byte(i - '0') |
| 39 | } |
| 40 | for i := 'a'; i <= 'f'; i++ { |
| 41 | hexDigits[i] = byte((i - 'a') + 10) |
| 42 | } |
| 43 | for i := 'A'; i <= 'F'; i++ { |
| 44 | hexDigits[i] = byte((i - 'A') + 10) |
| 45 | } |
| 46 | valueTypes = make([]ValueType, 256) |
| 47 | for i := 0; i < len(valueTypes); i++ { |
| 48 | valueTypes[i] = InvalidValue |
| 49 | } |
| 50 | valueTypes['"'] = StringValue |
| 51 | valueTypes['-'] = NumberValue |
| 52 | valueTypes['0'] = NumberValue |
| 53 | valueTypes['1'] = NumberValue |
| 54 | valueTypes['2'] = NumberValue |
| 55 | valueTypes['3'] = NumberValue |
| 56 | valueTypes['4'] = NumberValue |
| 57 | valueTypes['5'] = NumberValue |
| 58 | valueTypes['6'] = NumberValue |
| 59 | valueTypes['7'] = NumberValue |
| 60 | valueTypes['8'] = NumberValue |
| 61 | valueTypes['9'] = NumberValue |
| 62 | valueTypes['t'] = BoolValue |
| 63 | valueTypes['f'] = BoolValue |
| 64 | valueTypes['n'] = NilValue |
| 65 | valueTypes['['] = ArrayValue |
| 66 | valueTypes['{'] = ObjectValue |
| 67 | } |
| 68 | |
| 69 | // Iterator is a io.Reader like object, with JSON specific read functions. |
| 70 | // Error is not returned as return value, but stored as Error member on this iterator instance. |
| 71 | type Iterator struct { |
| 72 | cfg *frozenConfig |
| 73 | reader io.Reader |
| 74 | buf []byte |
| 75 | head int |
| 76 | tail int |
David K. Bainbridge | bd6b288 | 2021-08-26 13:31:02 +0000 | [diff] [blame^] | 77 | depth int |
Zack Williams | e940c7a | 2019-08-21 14:25:39 -0700 | [diff] [blame] | 78 | captureStartedAt int |
| 79 | captured []byte |
| 80 | Error error |
| 81 | Attachment interface{} // open for customized decoder |
| 82 | } |
| 83 | |
| 84 | // NewIterator creates an empty Iterator instance |
| 85 | func NewIterator(cfg API) *Iterator { |
| 86 | return &Iterator{ |
| 87 | cfg: cfg.(*frozenConfig), |
| 88 | reader: nil, |
| 89 | buf: nil, |
| 90 | head: 0, |
| 91 | tail: 0, |
David K. Bainbridge | bd6b288 | 2021-08-26 13:31:02 +0000 | [diff] [blame^] | 92 | depth: 0, |
Zack Williams | e940c7a | 2019-08-21 14:25:39 -0700 | [diff] [blame] | 93 | } |
| 94 | } |
| 95 | |
| 96 | // Parse creates an Iterator instance from io.Reader |
| 97 | func Parse(cfg API, reader io.Reader, bufSize int) *Iterator { |
| 98 | return &Iterator{ |
| 99 | cfg: cfg.(*frozenConfig), |
| 100 | reader: reader, |
| 101 | buf: make([]byte, bufSize), |
| 102 | head: 0, |
| 103 | tail: 0, |
David K. Bainbridge | bd6b288 | 2021-08-26 13:31:02 +0000 | [diff] [blame^] | 104 | depth: 0, |
Zack Williams | e940c7a | 2019-08-21 14:25:39 -0700 | [diff] [blame] | 105 | } |
| 106 | } |
| 107 | |
| 108 | // ParseBytes creates an Iterator instance from byte array |
| 109 | func ParseBytes(cfg API, input []byte) *Iterator { |
| 110 | return &Iterator{ |
| 111 | cfg: cfg.(*frozenConfig), |
| 112 | reader: nil, |
| 113 | buf: input, |
| 114 | head: 0, |
| 115 | tail: len(input), |
David K. Bainbridge | bd6b288 | 2021-08-26 13:31:02 +0000 | [diff] [blame^] | 116 | depth: 0, |
Zack Williams | e940c7a | 2019-08-21 14:25:39 -0700 | [diff] [blame] | 117 | } |
| 118 | } |
| 119 | |
| 120 | // ParseString creates an Iterator instance from string |
| 121 | func ParseString(cfg API, input string) *Iterator { |
| 122 | return ParseBytes(cfg, []byte(input)) |
| 123 | } |
| 124 | |
| 125 | // Pool returns a pool can provide more iterator with same configuration |
| 126 | func (iter *Iterator) Pool() IteratorPool { |
| 127 | return iter.cfg |
| 128 | } |
| 129 | |
| 130 | // Reset reuse iterator instance by specifying another reader |
| 131 | func (iter *Iterator) Reset(reader io.Reader) *Iterator { |
| 132 | iter.reader = reader |
| 133 | iter.head = 0 |
| 134 | iter.tail = 0 |
David K. Bainbridge | bd6b288 | 2021-08-26 13:31:02 +0000 | [diff] [blame^] | 135 | iter.depth = 0 |
Zack Williams | e940c7a | 2019-08-21 14:25:39 -0700 | [diff] [blame] | 136 | return iter |
| 137 | } |
| 138 | |
| 139 | // ResetBytes reuse iterator instance by specifying another byte array as input |
| 140 | func (iter *Iterator) ResetBytes(input []byte) *Iterator { |
| 141 | iter.reader = nil |
| 142 | iter.buf = input |
| 143 | iter.head = 0 |
| 144 | iter.tail = len(input) |
David K. Bainbridge | bd6b288 | 2021-08-26 13:31:02 +0000 | [diff] [blame^] | 145 | iter.depth = 0 |
Zack Williams | e940c7a | 2019-08-21 14:25:39 -0700 | [diff] [blame] | 146 | return iter |
| 147 | } |
| 148 | |
| 149 | // WhatIsNext gets ValueType of relatively next json element |
| 150 | func (iter *Iterator) WhatIsNext() ValueType { |
| 151 | valueType := valueTypes[iter.nextToken()] |
| 152 | iter.unreadByte() |
| 153 | return valueType |
| 154 | } |
| 155 | |
| 156 | func (iter *Iterator) skipWhitespacesWithoutLoadMore() bool { |
| 157 | for i := iter.head; i < iter.tail; i++ { |
| 158 | c := iter.buf[i] |
| 159 | switch c { |
| 160 | case ' ', '\n', '\t', '\r': |
| 161 | continue |
| 162 | } |
| 163 | iter.head = i |
| 164 | return false |
| 165 | } |
| 166 | return true |
| 167 | } |
| 168 | |
| 169 | func (iter *Iterator) isObjectEnd() bool { |
| 170 | c := iter.nextToken() |
| 171 | if c == ',' { |
| 172 | return false |
| 173 | } |
| 174 | if c == '}' { |
| 175 | return true |
| 176 | } |
| 177 | iter.ReportError("isObjectEnd", "object ended prematurely, unexpected char "+string([]byte{c})) |
| 178 | return true |
| 179 | } |
| 180 | |
| 181 | func (iter *Iterator) nextToken() byte { |
| 182 | // a variation of skip whitespaces, returning the next non-whitespace token |
| 183 | for { |
| 184 | for i := iter.head; i < iter.tail; i++ { |
| 185 | c := iter.buf[i] |
| 186 | switch c { |
| 187 | case ' ', '\n', '\t', '\r': |
| 188 | continue |
| 189 | } |
| 190 | iter.head = i + 1 |
| 191 | return c |
| 192 | } |
| 193 | if !iter.loadMore() { |
| 194 | return 0 |
| 195 | } |
| 196 | } |
| 197 | } |
| 198 | |
| 199 | // ReportError record a error in iterator instance with current position. |
| 200 | func (iter *Iterator) ReportError(operation string, msg string) { |
| 201 | if iter.Error != nil { |
| 202 | if iter.Error != io.EOF { |
| 203 | return |
| 204 | } |
| 205 | } |
| 206 | peekStart := iter.head - 10 |
| 207 | if peekStart < 0 { |
| 208 | peekStart = 0 |
| 209 | } |
| 210 | peekEnd := iter.head + 10 |
| 211 | if peekEnd > iter.tail { |
| 212 | peekEnd = iter.tail |
| 213 | } |
| 214 | parsing := string(iter.buf[peekStart:peekEnd]) |
| 215 | contextStart := iter.head - 50 |
| 216 | if contextStart < 0 { |
| 217 | contextStart = 0 |
| 218 | } |
| 219 | contextEnd := iter.head + 50 |
| 220 | if contextEnd > iter.tail { |
| 221 | contextEnd = iter.tail |
| 222 | } |
| 223 | context := string(iter.buf[contextStart:contextEnd]) |
| 224 | iter.Error = fmt.Errorf("%s: %s, error found in #%v byte of ...|%s|..., bigger context ...|%s|...", |
| 225 | operation, msg, iter.head-peekStart, parsing, context) |
| 226 | } |
| 227 | |
| 228 | // CurrentBuffer gets current buffer as string for debugging purpose |
| 229 | func (iter *Iterator) CurrentBuffer() string { |
| 230 | peekStart := iter.head - 10 |
| 231 | if peekStart < 0 { |
| 232 | peekStart = 0 |
| 233 | } |
| 234 | return fmt.Sprintf("parsing #%v byte, around ...|%s|..., whole buffer ...|%s|...", iter.head, |
| 235 | string(iter.buf[peekStart:iter.head]), string(iter.buf[0:iter.tail])) |
| 236 | } |
| 237 | |
| 238 | func (iter *Iterator) readByte() (ret byte) { |
| 239 | if iter.head == iter.tail { |
| 240 | if iter.loadMore() { |
| 241 | ret = iter.buf[iter.head] |
| 242 | iter.head++ |
| 243 | return ret |
| 244 | } |
| 245 | return 0 |
| 246 | } |
| 247 | ret = iter.buf[iter.head] |
| 248 | iter.head++ |
| 249 | return ret |
| 250 | } |
| 251 | |
| 252 | func (iter *Iterator) loadMore() bool { |
| 253 | if iter.reader == nil { |
| 254 | if iter.Error == nil { |
| 255 | iter.head = iter.tail |
| 256 | iter.Error = io.EOF |
| 257 | } |
| 258 | return false |
| 259 | } |
| 260 | if iter.captured != nil { |
| 261 | iter.captured = append(iter.captured, |
| 262 | iter.buf[iter.captureStartedAt:iter.tail]...) |
| 263 | iter.captureStartedAt = 0 |
| 264 | } |
| 265 | for { |
| 266 | n, err := iter.reader.Read(iter.buf) |
| 267 | if n == 0 { |
| 268 | if err != nil { |
| 269 | if iter.Error == nil { |
| 270 | iter.Error = err |
| 271 | } |
| 272 | return false |
| 273 | } |
| 274 | } else { |
| 275 | iter.head = 0 |
| 276 | iter.tail = n |
| 277 | return true |
| 278 | } |
| 279 | } |
| 280 | } |
| 281 | |
| 282 | func (iter *Iterator) unreadByte() { |
| 283 | if iter.Error != nil { |
| 284 | return |
| 285 | } |
| 286 | iter.head-- |
| 287 | return |
| 288 | } |
| 289 | |
| 290 | // Read read the next JSON element as generic interface{}. |
| 291 | func (iter *Iterator) Read() interface{} { |
| 292 | valueType := iter.WhatIsNext() |
| 293 | switch valueType { |
| 294 | case StringValue: |
| 295 | return iter.ReadString() |
| 296 | case NumberValue: |
| 297 | if iter.cfg.configBeforeFrozen.UseNumber { |
| 298 | return json.Number(iter.readNumberAsString()) |
| 299 | } |
| 300 | return iter.ReadFloat64() |
| 301 | case NilValue: |
| 302 | iter.skipFourBytes('n', 'u', 'l', 'l') |
| 303 | return nil |
| 304 | case BoolValue: |
| 305 | return iter.ReadBool() |
| 306 | case ArrayValue: |
| 307 | arr := []interface{}{} |
| 308 | iter.ReadArrayCB(func(iter *Iterator) bool { |
| 309 | var elem interface{} |
| 310 | iter.ReadVal(&elem) |
| 311 | arr = append(arr, elem) |
| 312 | return true |
| 313 | }) |
| 314 | return arr |
| 315 | case ObjectValue: |
| 316 | obj := map[string]interface{}{} |
| 317 | iter.ReadMapCB(func(Iter *Iterator, field string) bool { |
| 318 | var elem interface{} |
| 319 | iter.ReadVal(&elem) |
| 320 | obj[field] = elem |
| 321 | return true |
| 322 | }) |
| 323 | return obj |
| 324 | default: |
| 325 | iter.ReportError("Read", fmt.Sprintf("unexpected value type: %v", valueType)) |
| 326 | return nil |
| 327 | } |
| 328 | } |
David K. Bainbridge | bd6b288 | 2021-08-26 13:31:02 +0000 | [diff] [blame^] | 329 | |
| 330 | // limit maximum depth of nesting, as allowed by https://tools.ietf.org/html/rfc7159#section-9 |
| 331 | const maxDepth = 10000 |
| 332 | |
| 333 | func (iter *Iterator) incrementDepth() (success bool) { |
| 334 | iter.depth++ |
| 335 | if iter.depth <= maxDepth { |
| 336 | return true |
| 337 | } |
| 338 | iter.ReportError("incrementDepth", "exceeded max depth") |
| 339 | return false |
| 340 | } |
| 341 | |
| 342 | func (iter *Iterator) decrementDepth() (success bool) { |
| 343 | iter.depth-- |
| 344 | if iter.depth >= 0 { |
| 345 | return true |
| 346 | } |
| 347 | iter.ReportError("decrementDepth", "unexpected negative nesting") |
| 348 | return false |
| 349 | } |