Serge Bazanski | cc25bdf | 2018-10-25 14:02:58 +0200 | [diff] [blame] | 1 | // Package jlexer contains a JSON lexer implementation. |
| 2 | // |
| 3 | // It is expected that it is mostly used with generated parser code, so the interface is tuned |
| 4 | // for a parser that knows what kind of data is expected. |
| 5 | package jlexer |
| 6 | |
| 7 | import ( |
| 8 | "encoding/base64" |
| 9 | "encoding/json" |
| 10 | "errors" |
| 11 | "fmt" |
| 12 | "io" |
| 13 | "strconv" |
| 14 | "unicode" |
| 15 | "unicode/utf16" |
| 16 | "unicode/utf8" |
| 17 | ) |
| 18 | |
| 19 | // tokenKind determines type of a token. |
| 20 | type tokenKind byte |
| 21 | |
| 22 | const ( |
| 23 | tokenUndef tokenKind = iota // No token. |
| 24 | tokenDelim // Delimiter: one of '{', '}', '[' or ']'. |
| 25 | tokenString // A string literal, e.g. "abc\u1234" |
| 26 | tokenNumber // Number literal, e.g. 1.5e5 |
| 27 | tokenBool // Boolean literal: true or false. |
| 28 | tokenNull // null keyword. |
| 29 | ) |
| 30 | |
| 31 | // token describes a single token: type, position in the input and value. |
| 32 | type token struct { |
| 33 | kind tokenKind // Type of a token. |
| 34 | |
| 35 | boolValue bool // Value if a boolean literal token. |
| 36 | byteValue []byte // Raw value of a token. |
| 37 | delimValue byte |
| 38 | } |
| 39 | |
| 40 | // Lexer is a JSON lexer: it iterates over JSON tokens in a byte slice. |
| 41 | type Lexer struct { |
| 42 | Data []byte // Input data given to the lexer. |
| 43 | |
| 44 | start int // Start of the current token. |
| 45 | pos int // Current unscanned position in the input stream. |
| 46 | token token // Last scanned token, if token.kind != tokenUndef. |
| 47 | |
| 48 | firstElement bool // Whether current element is the first in array or an object. |
| 49 | wantSep byte // A comma or a colon character, which need to occur before a token. |
| 50 | |
| 51 | UseMultipleErrors bool // If we want to use multiple errors. |
| 52 | fatalError error // Fatal error occurred during lexing. It is usually a syntax error. |
| 53 | multipleErrors []*LexerError // Semantic errors occurred during lexing. Marshalling will be continued after finding this errors. |
| 54 | } |
| 55 | |
| 56 | // FetchToken scans the input for the next token. |
| 57 | func (r *Lexer) FetchToken() { |
| 58 | r.token.kind = tokenUndef |
| 59 | r.start = r.pos |
| 60 | |
| 61 | // Check if r.Data has r.pos element |
| 62 | // If it doesn't, it mean corrupted input data |
| 63 | if len(r.Data) < r.pos { |
| 64 | r.errParse("Unexpected end of data") |
| 65 | return |
| 66 | } |
| 67 | // Determine the type of a token by skipping whitespace and reading the |
| 68 | // first character. |
| 69 | for _, c := range r.Data[r.pos:] { |
| 70 | switch c { |
| 71 | case ':', ',': |
| 72 | if r.wantSep == c { |
| 73 | r.pos++ |
| 74 | r.start++ |
| 75 | r.wantSep = 0 |
| 76 | } else { |
| 77 | r.errSyntax() |
| 78 | } |
| 79 | |
| 80 | case ' ', '\t', '\r', '\n': |
| 81 | r.pos++ |
| 82 | r.start++ |
| 83 | |
| 84 | case '"': |
| 85 | if r.wantSep != 0 { |
| 86 | r.errSyntax() |
| 87 | } |
| 88 | |
| 89 | r.token.kind = tokenString |
| 90 | r.fetchString() |
| 91 | return |
| 92 | |
| 93 | case '{', '[': |
| 94 | if r.wantSep != 0 { |
| 95 | r.errSyntax() |
| 96 | } |
| 97 | r.firstElement = true |
| 98 | r.token.kind = tokenDelim |
| 99 | r.token.delimValue = r.Data[r.pos] |
| 100 | r.pos++ |
| 101 | return |
| 102 | |
| 103 | case '}', ']': |
| 104 | if !r.firstElement && (r.wantSep != ',') { |
| 105 | r.errSyntax() |
| 106 | } |
| 107 | r.wantSep = 0 |
| 108 | r.token.kind = tokenDelim |
| 109 | r.token.delimValue = r.Data[r.pos] |
| 110 | r.pos++ |
| 111 | return |
| 112 | |
| 113 | case '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '-': |
| 114 | if r.wantSep != 0 { |
| 115 | r.errSyntax() |
| 116 | } |
| 117 | r.token.kind = tokenNumber |
| 118 | r.fetchNumber() |
| 119 | return |
| 120 | |
| 121 | case 'n': |
| 122 | if r.wantSep != 0 { |
| 123 | r.errSyntax() |
| 124 | } |
| 125 | |
| 126 | r.token.kind = tokenNull |
| 127 | r.fetchNull() |
| 128 | return |
| 129 | |
| 130 | case 't': |
| 131 | if r.wantSep != 0 { |
| 132 | r.errSyntax() |
| 133 | } |
| 134 | |
| 135 | r.token.kind = tokenBool |
| 136 | r.token.boolValue = true |
| 137 | r.fetchTrue() |
| 138 | return |
| 139 | |
| 140 | case 'f': |
| 141 | if r.wantSep != 0 { |
| 142 | r.errSyntax() |
| 143 | } |
| 144 | |
| 145 | r.token.kind = tokenBool |
| 146 | r.token.boolValue = false |
| 147 | r.fetchFalse() |
| 148 | return |
| 149 | |
| 150 | default: |
| 151 | r.errSyntax() |
| 152 | return |
| 153 | } |
| 154 | } |
| 155 | r.fatalError = io.EOF |
| 156 | return |
| 157 | } |
| 158 | |
| 159 | // isTokenEnd returns true if the char can follow a non-delimiter token |
| 160 | func isTokenEnd(c byte) bool { |
| 161 | return c == ' ' || c == '\t' || c == '\r' || c == '\n' || c == '[' || c == ']' || c == '{' || c == '}' || c == ',' || c == ':' |
| 162 | } |
| 163 | |
| 164 | // fetchNull fetches and checks remaining bytes of null keyword. |
| 165 | func (r *Lexer) fetchNull() { |
| 166 | r.pos += 4 |
| 167 | if r.pos > len(r.Data) || |
| 168 | r.Data[r.pos-3] != 'u' || |
| 169 | r.Data[r.pos-2] != 'l' || |
| 170 | r.Data[r.pos-1] != 'l' || |
| 171 | (r.pos != len(r.Data) && !isTokenEnd(r.Data[r.pos])) { |
| 172 | |
| 173 | r.pos -= 4 |
| 174 | r.errSyntax() |
| 175 | } |
| 176 | } |
| 177 | |
| 178 | // fetchTrue fetches and checks remaining bytes of true keyword. |
| 179 | func (r *Lexer) fetchTrue() { |
| 180 | r.pos += 4 |
| 181 | if r.pos > len(r.Data) || |
| 182 | r.Data[r.pos-3] != 'r' || |
| 183 | r.Data[r.pos-2] != 'u' || |
| 184 | r.Data[r.pos-1] != 'e' || |
| 185 | (r.pos != len(r.Data) && !isTokenEnd(r.Data[r.pos])) { |
| 186 | |
| 187 | r.pos -= 4 |
| 188 | r.errSyntax() |
| 189 | } |
| 190 | } |
| 191 | |
| 192 | // fetchFalse fetches and checks remaining bytes of false keyword. |
| 193 | func (r *Lexer) fetchFalse() { |
| 194 | r.pos += 5 |
| 195 | if r.pos > len(r.Data) || |
| 196 | r.Data[r.pos-4] != 'a' || |
| 197 | r.Data[r.pos-3] != 'l' || |
| 198 | r.Data[r.pos-2] != 's' || |
| 199 | r.Data[r.pos-1] != 'e' || |
| 200 | (r.pos != len(r.Data) && !isTokenEnd(r.Data[r.pos])) { |
| 201 | |
| 202 | r.pos -= 5 |
| 203 | r.errSyntax() |
| 204 | } |
| 205 | } |
| 206 | |
| 207 | // fetchNumber scans a number literal token. |
| 208 | func (r *Lexer) fetchNumber() { |
| 209 | hasE := false |
| 210 | afterE := false |
| 211 | hasDot := false |
| 212 | |
| 213 | r.pos++ |
| 214 | for i, c := range r.Data[r.pos:] { |
| 215 | switch { |
| 216 | case c >= '0' && c <= '9': |
| 217 | afterE = false |
| 218 | case c == '.' && !hasDot: |
| 219 | hasDot = true |
| 220 | case (c == 'e' || c == 'E') && !hasE: |
| 221 | hasE = true |
| 222 | hasDot = true |
| 223 | afterE = true |
| 224 | case (c == '+' || c == '-') && afterE: |
| 225 | afterE = false |
| 226 | default: |
| 227 | r.pos += i |
| 228 | if !isTokenEnd(c) { |
| 229 | r.errSyntax() |
| 230 | } else { |
| 231 | r.token.byteValue = r.Data[r.start:r.pos] |
| 232 | } |
| 233 | return |
| 234 | } |
| 235 | } |
| 236 | |
| 237 | r.pos = len(r.Data) |
| 238 | r.token.byteValue = r.Data[r.start:] |
| 239 | } |
| 240 | |
| 241 | // findStringLen tries to scan into the string literal for ending quote char to determine required size. |
| 242 | // The size will be exact if no escapes are present and may be inexact if there are escaped chars. |
| 243 | func findStringLen(data []byte) (isValid, hasEscapes bool, length int) { |
| 244 | delta := 0 |
| 245 | |
| 246 | for i := 0; i < len(data); i++ { |
| 247 | switch data[i] { |
| 248 | case '\\': |
| 249 | i++ |
| 250 | delta++ |
| 251 | if i < len(data) && data[i] == 'u' { |
| 252 | delta++ |
| 253 | } |
| 254 | case '"': |
| 255 | return true, (delta > 0), (i - delta) |
| 256 | } |
| 257 | } |
| 258 | |
| 259 | return false, false, len(data) |
| 260 | } |
| 261 | |
| 262 | // getu4 decodes \uXXXX from the beginning of s, returning the hex value, |
| 263 | // or it returns -1. |
| 264 | func getu4(s []byte) rune { |
| 265 | if len(s) < 6 || s[0] != '\\' || s[1] != 'u' { |
| 266 | return -1 |
| 267 | } |
| 268 | var val rune |
| 269 | for i := 2; i < len(s) && i < 6; i++ { |
| 270 | var v byte |
| 271 | c := s[i] |
| 272 | switch c { |
| 273 | case '0', '1', '2', '3', '4', '5', '6', '7', '8', '9': |
| 274 | v = c - '0' |
| 275 | case 'a', 'b', 'c', 'd', 'e', 'f': |
| 276 | v = c - 'a' + 10 |
| 277 | case 'A', 'B', 'C', 'D', 'E', 'F': |
| 278 | v = c - 'A' + 10 |
| 279 | default: |
| 280 | return -1 |
| 281 | } |
| 282 | |
| 283 | val <<= 4 |
| 284 | val |= rune(v) |
| 285 | } |
| 286 | return val |
| 287 | } |
| 288 | |
| 289 | // processEscape processes a single escape sequence and returns number of bytes processed. |
| 290 | func (r *Lexer) processEscape(data []byte) (int, error) { |
| 291 | if len(data) < 2 { |
| 292 | return 0, fmt.Errorf("syntax error at %v", string(data)) |
| 293 | } |
| 294 | |
| 295 | c := data[1] |
| 296 | switch c { |
| 297 | case '"', '/', '\\': |
| 298 | r.token.byteValue = append(r.token.byteValue, c) |
| 299 | return 2, nil |
| 300 | case 'b': |
| 301 | r.token.byteValue = append(r.token.byteValue, '\b') |
| 302 | return 2, nil |
| 303 | case 'f': |
| 304 | r.token.byteValue = append(r.token.byteValue, '\f') |
| 305 | return 2, nil |
| 306 | case 'n': |
| 307 | r.token.byteValue = append(r.token.byteValue, '\n') |
| 308 | return 2, nil |
| 309 | case 'r': |
| 310 | r.token.byteValue = append(r.token.byteValue, '\r') |
| 311 | return 2, nil |
| 312 | case 't': |
| 313 | r.token.byteValue = append(r.token.byteValue, '\t') |
| 314 | return 2, nil |
| 315 | case 'u': |
| 316 | rr := getu4(data) |
| 317 | if rr < 0 { |
| 318 | return 0, errors.New("syntax error") |
| 319 | } |
| 320 | |
| 321 | read := 6 |
| 322 | if utf16.IsSurrogate(rr) { |
| 323 | rr1 := getu4(data[read:]) |
| 324 | if dec := utf16.DecodeRune(rr, rr1); dec != unicode.ReplacementChar { |
| 325 | read += 6 |
| 326 | rr = dec |
| 327 | } else { |
| 328 | rr = unicode.ReplacementChar |
| 329 | } |
| 330 | } |
| 331 | var d [4]byte |
| 332 | s := utf8.EncodeRune(d[:], rr) |
| 333 | r.token.byteValue = append(r.token.byteValue, d[:s]...) |
| 334 | return read, nil |
| 335 | } |
| 336 | |
| 337 | return 0, errors.New("syntax error") |
| 338 | } |
| 339 | |
| 340 | // fetchString scans a string literal token. |
| 341 | func (r *Lexer) fetchString() { |
| 342 | r.pos++ |
| 343 | data := r.Data[r.pos:] |
| 344 | |
| 345 | isValid, hasEscapes, length := findStringLen(data) |
| 346 | if !isValid { |
| 347 | r.pos += length |
| 348 | r.errParse("unterminated string literal") |
| 349 | return |
| 350 | } |
| 351 | if !hasEscapes { |
| 352 | r.token.byteValue = data[:length] |
| 353 | r.pos += length + 1 |
| 354 | return |
| 355 | } |
| 356 | |
| 357 | r.token.byteValue = make([]byte, 0, length) |
| 358 | p := 0 |
| 359 | for i := 0; i < len(data); { |
| 360 | switch data[i] { |
| 361 | case '"': |
| 362 | r.pos += i + 1 |
| 363 | r.token.byteValue = append(r.token.byteValue, data[p:i]...) |
| 364 | i++ |
| 365 | return |
| 366 | |
| 367 | case '\\': |
| 368 | r.token.byteValue = append(r.token.byteValue, data[p:i]...) |
| 369 | off, err := r.processEscape(data[i:]) |
| 370 | if err != nil { |
| 371 | r.errParse(err.Error()) |
| 372 | return |
| 373 | } |
| 374 | i += off |
| 375 | p = i |
| 376 | |
| 377 | default: |
| 378 | i++ |
| 379 | } |
| 380 | } |
| 381 | r.errParse("unterminated string literal") |
| 382 | } |
| 383 | |
| 384 | // scanToken scans the next token if no token is currently available in the lexer. |
| 385 | func (r *Lexer) scanToken() { |
| 386 | if r.token.kind != tokenUndef || r.fatalError != nil { |
| 387 | return |
| 388 | } |
| 389 | |
| 390 | r.FetchToken() |
| 391 | } |
| 392 | |
| 393 | // consume resets the current token to allow scanning the next one. |
| 394 | func (r *Lexer) consume() { |
| 395 | r.token.kind = tokenUndef |
| 396 | r.token.delimValue = 0 |
| 397 | } |
| 398 | |
| 399 | // Ok returns true if no error (including io.EOF) was encountered during scanning. |
| 400 | func (r *Lexer) Ok() bool { |
| 401 | return r.fatalError == nil |
| 402 | } |
| 403 | |
| 404 | const maxErrorContextLen = 13 |
| 405 | |
| 406 | func (r *Lexer) errParse(what string) { |
| 407 | if r.fatalError == nil { |
| 408 | var str string |
| 409 | if len(r.Data)-r.pos <= maxErrorContextLen { |
| 410 | str = string(r.Data) |
| 411 | } else { |
| 412 | str = string(r.Data[r.pos:r.pos+maxErrorContextLen-3]) + "..." |
| 413 | } |
| 414 | r.fatalError = &LexerError{ |
| 415 | Reason: what, |
| 416 | Offset: r.pos, |
| 417 | Data: str, |
| 418 | } |
| 419 | } |
| 420 | } |
| 421 | |
| 422 | func (r *Lexer) errSyntax() { |
| 423 | r.errParse("syntax error") |
| 424 | } |
| 425 | |
| 426 | func (r *Lexer) errInvalidToken(expected string) { |
| 427 | if r.fatalError != nil { |
| 428 | return |
| 429 | } |
| 430 | if r.UseMultipleErrors { |
| 431 | r.pos = r.start |
| 432 | r.consume() |
| 433 | r.SkipRecursive() |
| 434 | switch expected { |
| 435 | case "[": |
| 436 | r.token.delimValue = ']' |
| 437 | r.token.kind = tokenDelim |
| 438 | case "{": |
| 439 | r.token.delimValue = '}' |
| 440 | r.token.kind = tokenDelim |
| 441 | } |
| 442 | r.addNonfatalError(&LexerError{ |
| 443 | Reason: fmt.Sprintf("expected %s", expected), |
| 444 | Offset: r.start, |
| 445 | Data: string(r.Data[r.start:r.pos]), |
| 446 | }) |
| 447 | return |
| 448 | } |
| 449 | |
| 450 | var str string |
| 451 | if len(r.token.byteValue) <= maxErrorContextLen { |
| 452 | str = string(r.token.byteValue) |
| 453 | } else { |
| 454 | str = string(r.token.byteValue[:maxErrorContextLen-3]) + "..." |
| 455 | } |
| 456 | r.fatalError = &LexerError{ |
| 457 | Reason: fmt.Sprintf("expected %s", expected), |
| 458 | Offset: r.pos, |
| 459 | Data: str, |
| 460 | } |
| 461 | } |
| 462 | |
| 463 | func (r *Lexer) GetPos() int { |
| 464 | return r.pos |
| 465 | } |
| 466 | |
| 467 | // Delim consumes a token and verifies that it is the given delimiter. |
| 468 | func (r *Lexer) Delim(c byte) { |
| 469 | if r.token.kind == tokenUndef && r.Ok() { |
| 470 | r.FetchToken() |
| 471 | } |
| 472 | |
| 473 | if !r.Ok() || r.token.delimValue != c { |
| 474 | r.consume() // errInvalidToken can change token if UseMultipleErrors is enabled. |
| 475 | r.errInvalidToken(string([]byte{c})) |
| 476 | } else { |
| 477 | r.consume() |
| 478 | } |
| 479 | } |
| 480 | |
| 481 | // IsDelim returns true if there was no scanning error and next token is the given delimiter. |
| 482 | func (r *Lexer) IsDelim(c byte) bool { |
| 483 | if r.token.kind == tokenUndef && r.Ok() { |
| 484 | r.FetchToken() |
| 485 | } |
| 486 | return !r.Ok() || r.token.delimValue == c |
| 487 | } |
| 488 | |
| 489 | // Null verifies that the next token is null and consumes it. |
| 490 | func (r *Lexer) Null() { |
| 491 | if r.token.kind == tokenUndef && r.Ok() { |
| 492 | r.FetchToken() |
| 493 | } |
| 494 | if !r.Ok() || r.token.kind != tokenNull { |
| 495 | r.errInvalidToken("null") |
| 496 | } |
| 497 | r.consume() |
| 498 | } |
| 499 | |
| 500 | // IsNull returns true if the next token is a null keyword. |
| 501 | func (r *Lexer) IsNull() bool { |
| 502 | if r.token.kind == tokenUndef && r.Ok() { |
| 503 | r.FetchToken() |
| 504 | } |
| 505 | return r.Ok() && r.token.kind == tokenNull |
| 506 | } |
| 507 | |
| 508 | // Skip skips a single token. |
| 509 | func (r *Lexer) Skip() { |
| 510 | if r.token.kind == tokenUndef && r.Ok() { |
| 511 | r.FetchToken() |
| 512 | } |
| 513 | r.consume() |
| 514 | } |
| 515 | |
| 516 | // SkipRecursive skips next array or object completely, or just skips a single token if not |
| 517 | // an array/object. |
| 518 | // |
| 519 | // Note: no syntax validation is performed on the skipped data. |
| 520 | func (r *Lexer) SkipRecursive() { |
| 521 | r.scanToken() |
| 522 | var start, end byte |
| 523 | |
| 524 | if r.token.delimValue == '{' { |
| 525 | start, end = '{', '}' |
| 526 | } else if r.token.delimValue == '[' { |
| 527 | start, end = '[', ']' |
| 528 | } else { |
| 529 | r.consume() |
| 530 | return |
| 531 | } |
| 532 | |
| 533 | r.consume() |
| 534 | |
| 535 | level := 1 |
| 536 | inQuotes := false |
| 537 | wasEscape := false |
| 538 | |
| 539 | for i, c := range r.Data[r.pos:] { |
| 540 | switch { |
| 541 | case c == start && !inQuotes: |
| 542 | level++ |
| 543 | case c == end && !inQuotes: |
| 544 | level-- |
| 545 | if level == 0 { |
| 546 | r.pos += i + 1 |
| 547 | return |
| 548 | } |
| 549 | case c == '\\' && inQuotes: |
| 550 | wasEscape = !wasEscape |
| 551 | continue |
| 552 | case c == '"' && inQuotes: |
| 553 | inQuotes = wasEscape |
| 554 | case c == '"': |
| 555 | inQuotes = true |
| 556 | } |
| 557 | wasEscape = false |
| 558 | } |
| 559 | r.pos = len(r.Data) |
| 560 | r.fatalError = &LexerError{ |
| 561 | Reason: "EOF reached while skipping array/object or token", |
| 562 | Offset: r.pos, |
| 563 | Data: string(r.Data[r.pos:]), |
| 564 | } |
| 565 | } |
| 566 | |
| 567 | // Raw fetches the next item recursively as a data slice |
| 568 | func (r *Lexer) Raw() []byte { |
| 569 | r.SkipRecursive() |
| 570 | if !r.Ok() { |
| 571 | return nil |
| 572 | } |
| 573 | return r.Data[r.start:r.pos] |
| 574 | } |
| 575 | |
| 576 | // IsStart returns whether the lexer is positioned at the start |
| 577 | // of an input string. |
| 578 | func (r *Lexer) IsStart() bool { |
| 579 | return r.pos == 0 |
| 580 | } |
| 581 | |
| 582 | // Consumed reads all remaining bytes from the input, publishing an error if |
| 583 | // there is anything but whitespace remaining. |
| 584 | func (r *Lexer) Consumed() { |
| 585 | if r.pos > len(r.Data) || !r.Ok() { |
| 586 | return |
| 587 | } |
| 588 | |
| 589 | for _, c := range r.Data[r.pos:] { |
| 590 | if c != ' ' && c != '\t' && c != '\r' && c != '\n' { |
| 591 | r.AddError(&LexerError{ |
| 592 | Reason: "invalid character '" + string(c) + "' after top-level value", |
| 593 | Offset: r.pos, |
| 594 | Data: string(r.Data[r.pos:]), |
| 595 | }) |
| 596 | return |
| 597 | } |
| 598 | |
| 599 | r.pos++ |
| 600 | r.start++ |
| 601 | } |
| 602 | } |
| 603 | |
| 604 | func (r *Lexer) unsafeString() (string, []byte) { |
| 605 | if r.token.kind == tokenUndef && r.Ok() { |
| 606 | r.FetchToken() |
| 607 | } |
| 608 | if !r.Ok() || r.token.kind != tokenString { |
| 609 | r.errInvalidToken("string") |
| 610 | return "", nil |
| 611 | } |
| 612 | bytes := r.token.byteValue |
| 613 | ret := bytesToStr(r.token.byteValue) |
| 614 | r.consume() |
| 615 | return ret, bytes |
| 616 | } |
| 617 | |
| 618 | // UnsafeString returns the string value if the token is a string literal. |
| 619 | // |
| 620 | // Warning: returned string may point to the input buffer, so the string should not outlive |
| 621 | // the input buffer. Intended pattern of usage is as an argument to a switch statement. |
| 622 | func (r *Lexer) UnsafeString() string { |
| 623 | ret, _ := r.unsafeString() |
| 624 | return ret |
| 625 | } |
| 626 | |
| 627 | // UnsafeBytes returns the byte slice if the token is a string literal. |
| 628 | func (r *Lexer) UnsafeBytes() []byte { |
| 629 | _, ret := r.unsafeString() |
| 630 | return ret |
| 631 | } |
| 632 | |
| 633 | // String reads a string literal. |
| 634 | func (r *Lexer) String() string { |
| 635 | if r.token.kind == tokenUndef && r.Ok() { |
| 636 | r.FetchToken() |
| 637 | } |
| 638 | if !r.Ok() || r.token.kind != tokenString { |
| 639 | r.errInvalidToken("string") |
| 640 | return "" |
| 641 | } |
| 642 | ret := string(r.token.byteValue) |
| 643 | r.consume() |
| 644 | return ret |
| 645 | } |
| 646 | |
| 647 | // Bytes reads a string literal and base64 decodes it into a byte slice. |
| 648 | func (r *Lexer) Bytes() []byte { |
| 649 | if r.token.kind == tokenUndef && r.Ok() { |
| 650 | r.FetchToken() |
| 651 | } |
| 652 | if !r.Ok() || r.token.kind != tokenString { |
| 653 | r.errInvalidToken("string") |
| 654 | return nil |
| 655 | } |
| 656 | ret := make([]byte, base64.StdEncoding.DecodedLen(len(r.token.byteValue))) |
| 657 | n, err := base64.StdEncoding.Decode(ret, r.token.byteValue) |
| 658 | if err != nil { |
| 659 | r.fatalError = &LexerError{ |
| 660 | Reason: err.Error(), |
| 661 | } |
| 662 | return nil |
| 663 | } |
| 664 | |
| 665 | r.consume() |
| 666 | return ret[:n] |
| 667 | } |
| 668 | |
| 669 | // Bool reads a true or false boolean keyword. |
| 670 | func (r *Lexer) Bool() bool { |
| 671 | if r.token.kind == tokenUndef && r.Ok() { |
| 672 | r.FetchToken() |
| 673 | } |
| 674 | if !r.Ok() || r.token.kind != tokenBool { |
| 675 | r.errInvalidToken("bool") |
| 676 | return false |
| 677 | } |
| 678 | ret := r.token.boolValue |
| 679 | r.consume() |
| 680 | return ret |
| 681 | } |
| 682 | |
| 683 | func (r *Lexer) number() string { |
| 684 | if r.token.kind == tokenUndef && r.Ok() { |
| 685 | r.FetchToken() |
| 686 | } |
| 687 | if !r.Ok() || r.token.kind != tokenNumber { |
| 688 | r.errInvalidToken("number") |
| 689 | return "" |
| 690 | } |
| 691 | ret := bytesToStr(r.token.byteValue) |
| 692 | r.consume() |
| 693 | return ret |
| 694 | } |
| 695 | |
| 696 | func (r *Lexer) Uint8() uint8 { |
| 697 | s := r.number() |
| 698 | if !r.Ok() { |
| 699 | return 0 |
| 700 | } |
| 701 | |
| 702 | n, err := strconv.ParseUint(s, 10, 8) |
| 703 | if err != nil { |
| 704 | r.addNonfatalError(&LexerError{ |
| 705 | Offset: r.start, |
| 706 | Reason: err.Error(), |
| 707 | Data: s, |
| 708 | }) |
| 709 | } |
| 710 | return uint8(n) |
| 711 | } |
| 712 | |
| 713 | func (r *Lexer) Uint16() uint16 { |
| 714 | s := r.number() |
| 715 | if !r.Ok() { |
| 716 | return 0 |
| 717 | } |
| 718 | |
| 719 | n, err := strconv.ParseUint(s, 10, 16) |
| 720 | if err != nil { |
| 721 | r.addNonfatalError(&LexerError{ |
| 722 | Offset: r.start, |
| 723 | Reason: err.Error(), |
| 724 | Data: s, |
| 725 | }) |
| 726 | } |
| 727 | return uint16(n) |
| 728 | } |
| 729 | |
| 730 | func (r *Lexer) Uint32() uint32 { |
| 731 | s := r.number() |
| 732 | if !r.Ok() { |
| 733 | return 0 |
| 734 | } |
| 735 | |
| 736 | n, err := strconv.ParseUint(s, 10, 32) |
| 737 | if err != nil { |
| 738 | r.addNonfatalError(&LexerError{ |
| 739 | Offset: r.start, |
| 740 | Reason: err.Error(), |
| 741 | Data: s, |
| 742 | }) |
| 743 | } |
| 744 | return uint32(n) |
| 745 | } |
| 746 | |
| 747 | func (r *Lexer) Uint64() uint64 { |
| 748 | s := r.number() |
| 749 | if !r.Ok() { |
| 750 | return 0 |
| 751 | } |
| 752 | |
| 753 | n, err := strconv.ParseUint(s, 10, 64) |
| 754 | if err != nil { |
| 755 | r.addNonfatalError(&LexerError{ |
| 756 | Offset: r.start, |
| 757 | Reason: err.Error(), |
| 758 | Data: s, |
| 759 | }) |
| 760 | } |
| 761 | return n |
| 762 | } |
| 763 | |
| 764 | func (r *Lexer) Uint() uint { |
| 765 | return uint(r.Uint64()) |
| 766 | } |
| 767 | |
| 768 | func (r *Lexer) Int8() int8 { |
| 769 | s := r.number() |
| 770 | if !r.Ok() { |
| 771 | return 0 |
| 772 | } |
| 773 | |
| 774 | n, err := strconv.ParseInt(s, 10, 8) |
| 775 | if err != nil { |
| 776 | r.addNonfatalError(&LexerError{ |
| 777 | Offset: r.start, |
| 778 | Reason: err.Error(), |
| 779 | Data: s, |
| 780 | }) |
| 781 | } |
| 782 | return int8(n) |
| 783 | } |
| 784 | |
| 785 | func (r *Lexer) Int16() int16 { |
| 786 | s := r.number() |
| 787 | if !r.Ok() { |
| 788 | return 0 |
| 789 | } |
| 790 | |
| 791 | n, err := strconv.ParseInt(s, 10, 16) |
| 792 | if err != nil { |
| 793 | r.addNonfatalError(&LexerError{ |
| 794 | Offset: r.start, |
| 795 | Reason: err.Error(), |
| 796 | Data: s, |
| 797 | }) |
| 798 | } |
| 799 | return int16(n) |
| 800 | } |
| 801 | |
| 802 | func (r *Lexer) Int32() int32 { |
| 803 | s := r.number() |
| 804 | if !r.Ok() { |
| 805 | return 0 |
| 806 | } |
| 807 | |
| 808 | n, err := strconv.ParseInt(s, 10, 32) |
| 809 | if err != nil { |
| 810 | r.addNonfatalError(&LexerError{ |
| 811 | Offset: r.start, |
| 812 | Reason: err.Error(), |
| 813 | Data: s, |
| 814 | }) |
| 815 | } |
| 816 | return int32(n) |
| 817 | } |
| 818 | |
| 819 | func (r *Lexer) Int64() int64 { |
| 820 | s := r.number() |
| 821 | if !r.Ok() { |
| 822 | return 0 |
| 823 | } |
| 824 | |
| 825 | n, err := strconv.ParseInt(s, 10, 64) |
| 826 | if err != nil { |
| 827 | r.addNonfatalError(&LexerError{ |
| 828 | Offset: r.start, |
| 829 | Reason: err.Error(), |
| 830 | Data: s, |
| 831 | }) |
| 832 | } |
| 833 | return n |
| 834 | } |
| 835 | |
| 836 | func (r *Lexer) Int() int { |
| 837 | return int(r.Int64()) |
| 838 | } |
| 839 | |
| 840 | func (r *Lexer) Uint8Str() uint8 { |
| 841 | s, b := r.unsafeString() |
| 842 | if !r.Ok() { |
| 843 | return 0 |
| 844 | } |
| 845 | |
| 846 | n, err := strconv.ParseUint(s, 10, 8) |
| 847 | if err != nil { |
| 848 | r.addNonfatalError(&LexerError{ |
| 849 | Offset: r.start, |
| 850 | Reason: err.Error(), |
| 851 | Data: string(b), |
| 852 | }) |
| 853 | } |
| 854 | return uint8(n) |
| 855 | } |
| 856 | |
| 857 | func (r *Lexer) Uint16Str() uint16 { |
| 858 | s, b := r.unsafeString() |
| 859 | if !r.Ok() { |
| 860 | return 0 |
| 861 | } |
| 862 | |
| 863 | n, err := strconv.ParseUint(s, 10, 16) |
| 864 | if err != nil { |
| 865 | r.addNonfatalError(&LexerError{ |
| 866 | Offset: r.start, |
| 867 | Reason: err.Error(), |
| 868 | Data: string(b), |
| 869 | }) |
| 870 | } |
| 871 | return uint16(n) |
| 872 | } |
| 873 | |
| 874 | func (r *Lexer) Uint32Str() uint32 { |
| 875 | s, b := r.unsafeString() |
| 876 | if !r.Ok() { |
| 877 | return 0 |
| 878 | } |
| 879 | |
| 880 | n, err := strconv.ParseUint(s, 10, 32) |
| 881 | if err != nil { |
| 882 | r.addNonfatalError(&LexerError{ |
| 883 | Offset: r.start, |
| 884 | Reason: err.Error(), |
| 885 | Data: string(b), |
| 886 | }) |
| 887 | } |
| 888 | return uint32(n) |
| 889 | } |
| 890 | |
| 891 | func (r *Lexer) Uint64Str() uint64 { |
| 892 | s, b := r.unsafeString() |
| 893 | if !r.Ok() { |
| 894 | return 0 |
| 895 | } |
| 896 | |
| 897 | n, err := strconv.ParseUint(s, 10, 64) |
| 898 | if err != nil { |
| 899 | r.addNonfatalError(&LexerError{ |
| 900 | Offset: r.start, |
| 901 | Reason: err.Error(), |
| 902 | Data: string(b), |
| 903 | }) |
| 904 | } |
| 905 | return n |
| 906 | } |
| 907 | |
| 908 | func (r *Lexer) UintStr() uint { |
| 909 | return uint(r.Uint64Str()) |
| 910 | } |
| 911 | |
| 912 | func (r *Lexer) UintptrStr() uintptr { |
| 913 | return uintptr(r.Uint64Str()) |
| 914 | } |
| 915 | |
| 916 | func (r *Lexer) Int8Str() int8 { |
| 917 | s, b := r.unsafeString() |
| 918 | if !r.Ok() { |
| 919 | return 0 |
| 920 | } |
| 921 | |
| 922 | n, err := strconv.ParseInt(s, 10, 8) |
| 923 | if err != nil { |
| 924 | r.addNonfatalError(&LexerError{ |
| 925 | Offset: r.start, |
| 926 | Reason: err.Error(), |
| 927 | Data: string(b), |
| 928 | }) |
| 929 | } |
| 930 | return int8(n) |
| 931 | } |
| 932 | |
| 933 | func (r *Lexer) Int16Str() int16 { |
| 934 | s, b := r.unsafeString() |
| 935 | if !r.Ok() { |
| 936 | return 0 |
| 937 | } |
| 938 | |
| 939 | n, err := strconv.ParseInt(s, 10, 16) |
| 940 | if err != nil { |
| 941 | r.addNonfatalError(&LexerError{ |
| 942 | Offset: r.start, |
| 943 | Reason: err.Error(), |
| 944 | Data: string(b), |
| 945 | }) |
| 946 | } |
| 947 | return int16(n) |
| 948 | } |
| 949 | |
| 950 | func (r *Lexer) Int32Str() int32 { |
| 951 | s, b := r.unsafeString() |
| 952 | if !r.Ok() { |
| 953 | return 0 |
| 954 | } |
| 955 | |
| 956 | n, err := strconv.ParseInt(s, 10, 32) |
| 957 | if err != nil { |
| 958 | r.addNonfatalError(&LexerError{ |
| 959 | Offset: r.start, |
| 960 | Reason: err.Error(), |
| 961 | Data: string(b), |
| 962 | }) |
| 963 | } |
| 964 | return int32(n) |
| 965 | } |
| 966 | |
| 967 | func (r *Lexer) Int64Str() int64 { |
| 968 | s, b := r.unsafeString() |
| 969 | if !r.Ok() { |
| 970 | return 0 |
| 971 | } |
| 972 | |
| 973 | n, err := strconv.ParseInt(s, 10, 64) |
| 974 | if err != nil { |
| 975 | r.addNonfatalError(&LexerError{ |
| 976 | Offset: r.start, |
| 977 | Reason: err.Error(), |
| 978 | Data: string(b), |
| 979 | }) |
| 980 | } |
| 981 | return n |
| 982 | } |
| 983 | |
| 984 | func (r *Lexer) IntStr() int { |
| 985 | return int(r.Int64Str()) |
| 986 | } |
| 987 | |
| 988 | func (r *Lexer) Float32() float32 { |
| 989 | s := r.number() |
| 990 | if !r.Ok() { |
| 991 | return 0 |
| 992 | } |
| 993 | |
| 994 | n, err := strconv.ParseFloat(s, 32) |
| 995 | if err != nil { |
| 996 | r.addNonfatalError(&LexerError{ |
| 997 | Offset: r.start, |
| 998 | Reason: err.Error(), |
| 999 | Data: s, |
| 1000 | }) |
| 1001 | } |
| 1002 | return float32(n) |
| 1003 | } |
| 1004 | |
| 1005 | func (r *Lexer) Float32Str() float32 { |
| 1006 | s, b := r.unsafeString() |
| 1007 | if !r.Ok() { |
| 1008 | return 0 |
| 1009 | } |
| 1010 | n, err := strconv.ParseFloat(s, 32) |
| 1011 | if err != nil { |
| 1012 | r.addNonfatalError(&LexerError{ |
| 1013 | Offset: r.start, |
| 1014 | Reason: err.Error(), |
| 1015 | Data: string(b), |
| 1016 | }) |
| 1017 | } |
| 1018 | return float32(n) |
| 1019 | } |
| 1020 | |
| 1021 | func (r *Lexer) Float64() float64 { |
| 1022 | s := r.number() |
| 1023 | if !r.Ok() { |
| 1024 | return 0 |
| 1025 | } |
| 1026 | |
| 1027 | n, err := strconv.ParseFloat(s, 64) |
| 1028 | if err != nil { |
| 1029 | r.addNonfatalError(&LexerError{ |
| 1030 | Offset: r.start, |
| 1031 | Reason: err.Error(), |
| 1032 | Data: s, |
| 1033 | }) |
| 1034 | } |
| 1035 | return n |
| 1036 | } |
| 1037 | |
| 1038 | func (r *Lexer) Float64Str() float64 { |
| 1039 | s, b := r.unsafeString() |
| 1040 | if !r.Ok() { |
| 1041 | return 0 |
| 1042 | } |
| 1043 | n, err := strconv.ParseFloat(s, 64) |
| 1044 | if err != nil { |
| 1045 | r.addNonfatalError(&LexerError{ |
| 1046 | Offset: r.start, |
| 1047 | Reason: err.Error(), |
| 1048 | Data: string(b), |
| 1049 | }) |
| 1050 | } |
| 1051 | return n |
| 1052 | } |
| 1053 | |
| 1054 | func (r *Lexer) Error() error { |
| 1055 | return r.fatalError |
| 1056 | } |
| 1057 | |
| 1058 | func (r *Lexer) AddError(e error) { |
| 1059 | if r.fatalError == nil { |
| 1060 | r.fatalError = e |
| 1061 | } |
| 1062 | } |
| 1063 | |
| 1064 | func (r *Lexer) AddNonFatalError(e error) { |
| 1065 | r.addNonfatalError(&LexerError{ |
| 1066 | Offset: r.start, |
| 1067 | Data: string(r.Data[r.start:r.pos]), |
| 1068 | Reason: e.Error(), |
| 1069 | }) |
| 1070 | } |
| 1071 | |
| 1072 | func (r *Lexer) addNonfatalError(err *LexerError) { |
| 1073 | if r.UseMultipleErrors { |
| 1074 | // We don't want to add errors with the same offset. |
| 1075 | if len(r.multipleErrors) != 0 && r.multipleErrors[len(r.multipleErrors)-1].Offset == err.Offset { |
| 1076 | return |
| 1077 | } |
| 1078 | r.multipleErrors = append(r.multipleErrors, err) |
| 1079 | return |
| 1080 | } |
| 1081 | r.fatalError = err |
| 1082 | } |
| 1083 | |
| 1084 | func (r *Lexer) GetNonFatalErrors() []*LexerError { |
| 1085 | return r.multipleErrors |
| 1086 | } |
| 1087 | |
| 1088 | // JsonNumber fetches and json.Number from 'encoding/json' package. |
| 1089 | // Both int, float or string, contains them are valid values |
| 1090 | func (r *Lexer) JsonNumber() json.Number { |
| 1091 | if r.token.kind == tokenUndef && r.Ok() { |
| 1092 | r.FetchToken() |
| 1093 | } |
| 1094 | if !r.Ok() { |
| 1095 | r.errInvalidToken("json.Number") |
| 1096 | return json.Number("") |
| 1097 | } |
| 1098 | |
| 1099 | switch r.token.kind { |
| 1100 | case tokenString: |
| 1101 | return json.Number(r.String()) |
| 1102 | case tokenNumber: |
| 1103 | return json.Number(r.Raw()) |
| 1104 | case tokenNull: |
| 1105 | r.Null() |
| 1106 | return json.Number("") |
| 1107 | default: |
| 1108 | r.errSyntax() |
| 1109 | return json.Number("") |
| 1110 | } |
| 1111 | } |
| 1112 | |
| 1113 | // Interface fetches an interface{} analogous to the 'encoding/json' package. |
| 1114 | func (r *Lexer) Interface() interface{} { |
| 1115 | if r.token.kind == tokenUndef && r.Ok() { |
| 1116 | r.FetchToken() |
| 1117 | } |
| 1118 | |
| 1119 | if !r.Ok() { |
| 1120 | return nil |
| 1121 | } |
| 1122 | switch r.token.kind { |
| 1123 | case tokenString: |
| 1124 | return r.String() |
| 1125 | case tokenNumber: |
| 1126 | return r.Float64() |
| 1127 | case tokenBool: |
| 1128 | return r.Bool() |
| 1129 | case tokenNull: |
| 1130 | r.Null() |
| 1131 | return nil |
| 1132 | } |
| 1133 | |
| 1134 | if r.token.delimValue == '{' { |
| 1135 | r.consume() |
| 1136 | |
| 1137 | ret := map[string]interface{}{} |
| 1138 | for !r.IsDelim('}') { |
| 1139 | key := r.String() |
| 1140 | r.WantColon() |
| 1141 | ret[key] = r.Interface() |
| 1142 | r.WantComma() |
| 1143 | } |
| 1144 | r.Delim('}') |
| 1145 | |
| 1146 | if r.Ok() { |
| 1147 | return ret |
| 1148 | } else { |
| 1149 | return nil |
| 1150 | } |
| 1151 | } else if r.token.delimValue == '[' { |
| 1152 | r.consume() |
| 1153 | |
| 1154 | var ret []interface{} |
| 1155 | for !r.IsDelim(']') { |
| 1156 | ret = append(ret, r.Interface()) |
| 1157 | r.WantComma() |
| 1158 | } |
| 1159 | r.Delim(']') |
| 1160 | |
| 1161 | if r.Ok() { |
| 1162 | return ret |
| 1163 | } else { |
| 1164 | return nil |
| 1165 | } |
| 1166 | } |
| 1167 | r.errSyntax() |
| 1168 | return nil |
| 1169 | } |
| 1170 | |
| 1171 | // WantComma requires a comma to be present before fetching next token. |
| 1172 | func (r *Lexer) WantComma() { |
| 1173 | r.wantSep = ',' |
| 1174 | r.firstElement = false |
| 1175 | } |
| 1176 | |
| 1177 | // WantColon requires a colon to be present before fetching next token. |
| 1178 | func (r *Lexer) WantColon() { |
| 1179 | r.wantSep = ':' |
| 1180 | r.firstElement = false |
| 1181 | } |