clerk/journal/lexer/lexer.go (view raw)
| 1 | package lexer |
| 2 | |
| 3 | import ( |
| 4 | "unicode" |
| 5 | "unicode/utf8" |
| 6 | "unsafe" |
| 7 | |
| 8 | "olexsmir.xyz/clerk/journal/token" |
| 9 | ) |
| 10 | |
| 11 | type mode uint |
| 12 | |
| 13 | const ( |
| 14 | // start of a line, nothing consumed |
| 15 | modeDefault mode = iota |
| 16 | |
| 17 | // after ; # * % ; at start of line, or anywhere inline |
| 18 | // everything until \n is comment text |
| 19 | modeComment |
| 20 | |
| 21 | // after lexing a date at column 0 |
| 22 | // expects: optional status, optional code, description, comment |
| 23 | modeTransaction |
| 24 | |
| 25 | // after lexing an indent at start of line |
| 26 | // expects: account name, then two spaces, then amount |
| 27 | modePosting |
| 28 | |
| 29 | // after ~, period expression |
| 30 | // expects: period, optional description (after 2+ spaces), optional comment |
| 31 | modePeriodic |
| 32 | |
| 33 | // after =, automates transaction |
| 34 | // expects: expression |
| 35 | modeAutomated |
| 36 | |
| 37 | // after a directive keyword like account, commodity, include |
| 38 | // expects: rest of directive content |
| 39 | modeDirective |
| 40 | ) |
| 41 | |
| 42 | type Lexer struct { |
| 43 | file string |
| 44 | input []byte |
| 45 | mode mode |
| 46 | |
| 47 | ch rune // current rune (0 = EOF/sentinel) |
| 48 | chSize int // byte size of current rune |
| 49 | pos int // current byte offset (points at ch) |
| 50 | rpos int // next byte offset to read (one ahead of pos) |
| 51 | col int // current column (1-based) |
| 52 | line int // current line (1-based) |
| 53 | |
| 54 | transactionPastStatus bool |
| 55 | postingExpectAccount bool |
| 56 | readingNoteAfterPipe bool |
| 57 | |
| 58 | includePath bool // true after [token.INCLUDE] is omitted |
| 59 | |
| 60 | // subdirective is set while inside an account/commodity directive block; |
| 61 | // every indented line then lexes as directive content until a |
| 62 | // non-indented line start clears it in lexDefault |
| 63 | subdirective bool |
| 64 | } |
| 65 | |
| 66 | func New(file string, input []byte) *Lexer { |
| 67 | l := &Lexer{ |
| 68 | file: file, |
| 69 | input: input, |
| 70 | line: 1, |
| 71 | } |
| 72 | l.advance() |
| 73 | if l.ch == '\uFEFF' { // start of the input |
| 74 | l.advance() |
| 75 | } |
| 76 | return l |
| 77 | } |
| 78 | |
| 79 | // Next returns next token in the input |
| 80 | func (l *Lexer) Next() token.Token { |
| 81 | switch l.mode { |
| 82 | case modeDefault: |
| 83 | return l.lexDefault() |
| 84 | case modeComment: |
| 85 | return l.lexComment() |
| 86 | case modeTransaction: |
| 87 | return l.lexTransaction() |
| 88 | case modePosting: |
| 89 | return l.lexPosting() |
| 90 | case modePeriodic: |
| 91 | return l.lexPeriodic() |
| 92 | case modeAutomated: |
| 93 | return l.lexAutomated() |
| 94 | case modeDirective: |
| 95 | return l.lexDirective() |
| 96 | } |
| 97 | panic("unreachable") |
| 98 | } |
| 99 | |
| 100 | func (l *Lexer) lexDefault() token.Token { |
| 101 | if l.includePath { |
| 102 | l.includePath = false |
| 103 | } |
| 104 | if l.subdirective && l.ch != ' ' && l.ch != '\t' { |
| 105 | l.subdirective = false |
| 106 | } |
| 107 | switch { |
| 108 | case l.ch == 0: |
| 109 | return l.token(token.EOF, "") |
| 110 | case l.ch == '\n': |
| 111 | return l.lexNewline() |
| 112 | case l.ch == '\r': |
| 113 | l.col = 0 |
| 114 | l.advance() |
| 115 | return l.lexNewline() |
| 116 | case l.ch == ' ' || l.ch == '\t': |
| 117 | tok := l.lexIndent() |
| 118 | if l.subdirective { |
| 119 | // keep directive mode for every line of a directive block; the |
| 120 | // flag is cleared at the next non-indented line start in lexDefault |
| 121 | l.mode = modeDirective |
| 122 | } else { |
| 123 | l.mode = modePosting |
| 124 | } |
| 125 | l.postingExpectAccount = true |
| 126 | return tok |
| 127 | case l.ch == ';' || l.ch == '#' || l.ch == '%': |
| 128 | l.mode = modeComment |
| 129 | return l.lexSingle(token.SEMICOLON) // todo: ?? |
| 130 | case l.ch == '*': // * at col 0 == comment |
| 131 | l.mode = modeComment |
| 132 | return l.lexSingle(token.STAR) |
| 133 | case l.ch == '~': |
| 134 | l.mode = modePeriodic |
| 135 | return l.lexSingle(token.TILDE) |
| 136 | case l.ch == '=': |
| 137 | l.mode = modeAutomated |
| 138 | return l.lexSingle(token.EQ) |
| 139 | case l.ch == '+': |
| 140 | return l.lexSingle(token.PLUS) |
| 141 | case l.ch == '-': |
| 142 | return l.lexSingle(token.MINUS) |
| 143 | case l.ch == '.': |
| 144 | return l.lexSingle(token.TEXT) |
| 145 | case l.ch == '!': |
| 146 | return l.lexSingle(token.BANG) |
| 147 | case l.ch == '@': |
| 148 | return l.lexSingle(token.AT) |
| 149 | case l.isAlpha(): |
| 150 | return l.lexKeyword() |
| 151 | case l.isDigit(): |
| 152 | if !l.isDate() { |
| 153 | s := l.save() |
| 154 | for l.isDigit() || l.ch == '-' || l.ch == '/' || l.ch == '.' { |
| 155 | l.advance() |
| 156 | } |
| 157 | return token.Token{Type: token.ILLEGAL, Literal: string(l.input[s.offset:l.pos]), Span: l.span(s)} |
| 158 | } |
| 159 | tok := l.lexDate() |
| 160 | l.mode = modeTransaction |
| 161 | l.transactionPastStatus = false |
| 162 | return tok |
| 163 | default: |
| 164 | s := l.save() |
| 165 | l.advance() |
| 166 | return token.Token{Type: token.ILLEGAL, Literal: l.lit(s), Span: l.span(s)} |
| 167 | } |
| 168 | } |
| 169 | |
| 170 | func (l *Lexer) lexComment() token.Token { |
| 171 | if l.ch == '\n' || l.ch == '\r' || l.ch == 0 { |
| 172 | l.mode = modeDefault |
| 173 | if l.ch == '\r' { |
| 174 | l.col = 0 |
| 175 | l.advance() |
| 176 | } |
| 177 | return l.lexNewline() |
| 178 | } |
| 179 | |
| 180 | for l.ch == ' ' || l.ch == '\t' { |
| 181 | l.lexWhitespace() |
| 182 | } |
| 183 | |
| 184 | if l.ch == '\n' || l.ch == '\r' || l.ch == 0 { |
| 185 | l.mode = modeDefault |
| 186 | if l.ch == '\r' { |
| 187 | l.col = 0 |
| 188 | l.advance() |
| 189 | } |
| 190 | return l.lexNewline() |
| 191 | } |
| 192 | |
| 193 | s := l.save() |
| 194 | for l.ch != '\n' && l.ch != '\r' && l.ch != 0 { |
| 195 | l.advance() |
| 196 | } |
| 197 | return token.Token{Type: token.TEXT, Literal: l.lit(s), Span: l.span(s)} |
| 198 | } |
| 199 | |
| 200 | func (l *Lexer) lexTransaction() token.Token { |
| 201 | if l.readingNoteAfterPipe { |
| 202 | l.readingNoteAfterPipe = false |
| 203 | return l.lexNote() |
| 204 | } |
| 205 | |
| 206 | switch l.ch { |
| 207 | case 0: |
| 208 | return l.token(token.EOF, "") |
| 209 | case '\n': |
| 210 | l.mode = modeDefault |
| 211 | return l.lexNewline() |
| 212 | case '\r': |
| 213 | l.col = 0 |
| 214 | l.advance() |
| 215 | return l.lexNewline() |
| 216 | case ' ', '\t': |
| 217 | return l.lexWhitespace() |
| 218 | case ';': |
| 219 | l.mode = modeComment |
| 220 | return l.lexSingle(token.SEMICOLON) |
| 221 | case '*': |
| 222 | if !l.transactionPastStatus { |
| 223 | l.transactionPastStatus = true |
| 224 | return l.lexSingle(token.STAR) |
| 225 | } |
| 226 | return l.lexText() |
| 227 | case '!': |
| 228 | if !l.transactionPastStatus { |
| 229 | l.transactionPastStatus = true |
| 230 | return l.lexSingle(token.BANG) |
| 231 | } |
| 232 | return l.lexText() |
| 233 | case '|': |
| 234 | l.transactionPastStatus = true |
| 235 | l.readingNoteAfterPipe = true |
| 236 | return l.lexSingle(token.PIPE) |
| 237 | case '+': |
| 238 | return l.lexSingle(token.PLUS) |
| 239 | case '-': |
| 240 | return l.lexSingle(token.MINUS) |
| 241 | case '=': |
| 242 | return l.lexEquals() |
| 243 | case '"', '\'': |
| 244 | return l.lexString() |
| 245 | default: // description / payee |
| 246 | if l.isDate() { // secondsry date after = |
| 247 | return l.lexDate() |
| 248 | } |
| 249 | return l.lexText() |
| 250 | } |
| 251 | } |
| 252 | |
| 253 | func (l *Lexer) lexNote() token.Token { |
| 254 | for l.ch == ' ' || l.ch == '\t' { |
| 255 | l.advance() |
| 256 | } |
| 257 | if l.ch == 0 || l.ch == '\n' || l.ch == '\r' || l.ch == ';' { |
| 258 | return l.Next() |
| 259 | } |
| 260 | s := l.save() |
| 261 | for l.ch != 0 && l.ch != '\n' && l.ch != '\r' && l.ch != ';' { |
| 262 | l.advance() |
| 263 | } |
| 264 | lit := l.lit(s) |
| 265 | for len(lit) > 0 && (lit[len(lit)-1] == ' ' || lit[len(lit)-1] == '\t') { |
| 266 | lit = lit[:len(lit)-1] |
| 267 | } |
| 268 | return token.Token{Type: token.TEXT, Literal: lit, Span: l.span(s)} |
| 269 | } |
| 270 | |
| 271 | func (l *Lexer) lexPeriodic() token.Token { |
| 272 | switch l.ch { |
| 273 | case 0: |
| 274 | return l.token(token.EOF, "") |
| 275 | case '\n': |
| 276 | l.mode = modeDefault |
| 277 | return l.lexNewline() |
| 278 | case '\r': |
| 279 | l.col = 0 |
| 280 | l.advance() |
| 281 | return l.lexNewline() |
| 282 | case ';': |
| 283 | l.mode = modeComment |
| 284 | return l.lexSingle(token.SEMICOLON) |
| 285 | case ' ', '\t': |
| 286 | return l.lexWhitespace() |
| 287 | default: |
| 288 | return l.lexText() |
| 289 | } |
| 290 | } |
| 291 | |
| 292 | func (l *Lexer) lexAutomated() token.Token { |
| 293 | switch l.ch { |
| 294 | case 0: |
| 295 | return l.token(token.EOF, "") |
| 296 | case '\n': |
| 297 | l.mode = modeDefault |
| 298 | return l.lexNewline() |
| 299 | case '\r': |
| 300 | l.col = 0 |
| 301 | l.advance() |
| 302 | return l.lexNewline() |
| 303 | case ' ', '\t': |
| 304 | return l.lexWhitespace() |
| 305 | case ';': |
| 306 | l.mode = modeComment |
| 307 | return l.lexSingle(token.SEMICOLON) |
| 308 | default: |
| 309 | return l.lexText() |
| 310 | } |
| 311 | } |
| 312 | |
| 313 | func (l *Lexer) lexPosting() token.Token { |
| 314 | switch { |
| 315 | case l.ch == 0: |
| 316 | l.postingExpectAccount = false |
| 317 | return l.token(token.EOF, "") |
| 318 | case l.ch == '\n': |
| 319 | l.postingExpectAccount = false |
| 320 | l.mode = modeDefault |
| 321 | return l.lexNewline() |
| 322 | case l.ch == '\r': |
| 323 | l.postingExpectAccount = false |
| 324 | l.col = 0 |
| 325 | l.advance() |
| 326 | l.mode = modeDefault |
| 327 | return l.lexNewline() |
| 328 | case l.ch == ';': |
| 329 | l.postingExpectAccount = false |
| 330 | l.mode = modeComment |
| 331 | return l.lexSingle(token.SEMICOLON) |
| 332 | case l.ch == ' ' || l.ch == '\t': |
| 333 | return l.lexWhitespace() |
| 334 | case l.postingExpectAccount && l.ch == '*': |
| 335 | return l.lexSingle(token.STAR) |
| 336 | case l.postingExpectAccount && l.ch == '!': |
| 337 | return l.lexSingle(token.BANG) |
| 338 | case l.ch == '=': |
| 339 | return l.lexEquals() |
| 340 | case l.ch == '@': |
| 341 | return l.lexAt() |
| 342 | case l.ch == '{': |
| 343 | return l.lexLBrace() |
| 344 | case l.ch == '}': |
| 345 | return l.lexRBrace() |
| 346 | case l.ch == '(': |
| 347 | if !l.postingExpectAccount { |
| 348 | return l.lexParenExpr() |
| 349 | } |
| 350 | return l.lexSingle(token.LPAREN) |
| 351 | case l.ch == ')': |
| 352 | return l.lexSingle(token.RPAREN) |
| 353 | case l.ch == '[': |
| 354 | return l.lexSingle(token.LBRACKET) |
| 355 | case l.ch == ']': |
| 356 | return l.lexSingle(token.RBRACKET) |
| 357 | case l.ch == ':': |
| 358 | return l.lexSingle(token.COLON) |
| 359 | case l.postingExpectAccount && l.ch != '*' && l.ch != '!' && l.ch != '(' && l.ch != '[': |
| 360 | return l.lexAccountNamePosting() |
| 361 | case l.ch == '*': // after account name |
| 362 | return l.lexSingle(token.STAR) |
| 363 | case l.isDigit(), l.ch == '.': |
| 364 | return l.lexNumber() |
| 365 | case l.ch == '-': |
| 366 | return l.lexSingle(token.MINUS) |
| 367 | case l.ch == '+': |
| 368 | return l.lexSingle(token.PLUS) |
| 369 | case l.isCommodityStart(): |
| 370 | return l.lexCommodityMark() |
| 371 | case l.ch == '"' || l.ch == '\'': |
| 372 | return l.lexString() |
| 373 | case l.ch >= 'a' && l.ch <= 'z': |
| 374 | return l.lexCommodityMark() |
| 375 | default: |
| 376 | return l.lexAccountNamePosting() |
| 377 | } |
| 378 | } |
| 379 | |
| 380 | func (l *Lexer) lexDirective() token.Token { |
| 381 | switch { |
| 382 | case l.ch == '\n', l.ch == 0: |
| 383 | l.mode = modeDefault |
| 384 | return l.lexNewline() |
| 385 | case l.ch == '\r': |
| 386 | l.mode = modeDefault |
| 387 | l.col = 0 |
| 388 | l.advance() |
| 389 | return l.lexNewline() |
| 390 | case l.ch == ';': |
| 391 | l.mode = modeComment |
| 392 | return l.lexSingle(token.SEMICOLON) |
| 393 | case l.ch == ' ', l.ch == '\t': |
| 394 | return l.lexWhitespace() |
| 395 | case l.ch == '=': |
| 396 | return l.lexSingle(token.EQ) |
| 397 | case l.ch == '+': |
| 398 | return l.lexSingle(token.PLUS) |
| 399 | case l.ch == '-': |
| 400 | return l.lexSingle(token.MINUS) |
| 401 | case l.ch == '"', l.ch == '\'': |
| 402 | return l.lexString() |
| 403 | case l.ch == ':': |
| 404 | return l.lexSingle(token.COLON) |
| 405 | case l.ch == ')': |
| 406 | return l.lexSingle(token.RPAREN) |
| 407 | case l.ch == ']': |
| 408 | return l.lexSingle(token.RBRACKET) |
| 409 | case l.includePath: |
| 410 | return l.lexPath() |
| 411 | case l.isCommodityStart(): |
| 412 | return l.lexCommodityMark() |
| 413 | case l.isTime(): |
| 414 | return l.lexTime() |
| 415 | case l.isDate(): |
| 416 | return l.lexDate() |
| 417 | case l.isDigit(): |
| 418 | return l.lexNumber() |
| 419 | default: |
| 420 | return l.lexAccountNameDirective() |
| 421 | } |
| 422 | } |
| 423 | |
| 424 | func (l *Lexer) lexPath() token.Token { |
| 425 | s := l.save() |
| 426 | for l.ch != '\n' && l.ch != '\r' && l.ch != ';' && l.ch != 0 && l.ch != ' ' && l.ch != '\t' { |
| 427 | l.advance() |
| 428 | } |
| 429 | return token.Token{ |
| 430 | Type: token.TEXT, |
| 431 | Literal: l.lit(s), |
| 432 | Span: l.span(s), |
| 433 | } |
| 434 | } |
| 435 | |
| 436 | func (l *Lexer) lexSingle(kind token.Type) token.Token { |
| 437 | s := l.save() |
| 438 | l.advance() |
| 439 | return token.Token{ |
| 440 | Type: kind, |
| 441 | Literal: l.lit(s), |
| 442 | Span: l.span(s), |
| 443 | } |
| 444 | } |
| 445 | |
| 446 | func (l *Lexer) lexNewline() token.Token { |
| 447 | s := l.save() |
| 448 | l.advance() |
| 449 | l.mode = modeDefault |
| 450 | return token.Token{Type: token.NEWLINE, Literal: "\n", Span: l.span(s)} |
| 451 | } |
| 452 | |
| 453 | func (l *Lexer) lexWhitespace() token.Token { |
| 454 | s := l.save() |
| 455 | for l.ch == ' ' || l.ch == '\t' { |
| 456 | l.advance() |
| 457 | } |
| 458 | return token.Token{Type: token.WHITESPACE, Literal: l.lit(s), Span: l.span(s)} |
| 459 | } |
| 460 | |
| 461 | func (l *Lexer) lexIndent() token.Token { |
| 462 | s := l.save() |
| 463 | for l.ch == ' ' || l.ch == '\t' { |
| 464 | l.advance() |
| 465 | } |
| 466 | return token.Token{Type: token.INDENT, Literal: l.lit(s), Span: l.span(s)} |
| 467 | } |
| 468 | |
| 469 | func (l *Lexer) lexEquals() token.Token { |
| 470 | s := l.save() |
| 471 | l.advance() |
| 472 | if l.ch == '=' { |
| 473 | l.advance() |
| 474 | switch l.ch { |
| 475 | case '=': |
| 476 | l.advance() |
| 477 | return token.Token{Type: token.EQEQEQ, Literal: "===", Span: l.span(s)} |
| 478 | case '*': |
| 479 | l.advance() |
| 480 | return token.Token{Type: token.EQEQEQ, Literal: "==*", Span: l.span(s)} |
| 481 | default: |
| 482 | return token.Token{Type: token.EQEQ, Literal: "==", Span: l.span(s)} |
| 483 | } |
| 484 | } |
| 485 | if l.ch == '*' { |
| 486 | l.advance() |
| 487 | return token.Token{Type: token.EQSTAR, Literal: "=*", Span: l.span(s)} |
| 488 | } |
| 489 | return token.Token{Type: token.EQ, Literal: "=", Span: l.span(s)} |
| 490 | } |
| 491 | |
| 492 | func (l *Lexer) lexAt() token.Token { |
| 493 | s := l.save() |
| 494 | l.advance() |
| 495 | if l.ch == '@' { |
| 496 | l.advance() |
| 497 | return token.Token{Type: token.ATAT, Literal: "@@", Span: l.span(s)} |
| 498 | } |
| 499 | return token.Token{Type: token.AT, Literal: "@", Span: l.span(s)} |
| 500 | } |
| 501 | |
| 502 | func (l *Lexer) lexText() token.Token { |
| 503 | s := l.save() |
| 504 | l.advance() |
| 505 | for l.ch != '\n' && l.ch != '\r' && l.ch != ';' && l.ch != 0 && l.ch != ' ' && l.ch != '\t' { |
| 506 | l.advance() |
| 507 | } |
| 508 | lit := string(l.input[s.offset:l.pos]) |
| 509 | return token.Token{Type: token.TEXT, Literal: lit, Span: l.span(s)} |
| 510 | } |
| 511 | |
| 512 | // lexAccountNameDirective reads accout name in directive context. |
| 513 | // stops at any whitespace, supports multi-word names("Taxi Fare"). |
| 514 | func (l *Lexer) lexAccountNameDirective() token.Token { |
| 515 | s := l.save() |
| 516 | for l.ch != '\n' && l.ch != '\r' && l.ch != ';' && l.ch != 0 && l.ch != ')' && l.ch != ']' && l.ch != ':' && l.ch != ' ' && l.ch != '\t' { |
| 517 | l.advance() |
| 518 | } |
| 519 | return token.Token{Type: token.TEXT, Literal: l.lit(s), Span: l.span(s)} |
| 520 | } |
| 521 | |
| 522 | // lexAccountNamePosting reads an account name in posting context. |
| 523 | // stops at two consecutive spaces. |
| 524 | func (l *Lexer) lexAccountNamePosting() token.Token { |
| 525 | s := l.save() |
| 526 | for l.ch != '\n' && l.ch != '\r' && l.ch != ';' && l.ch != 0 && l.ch != ')' && l.ch != ']' && l.ch != ':' { |
| 527 | if l.isTwoSpaces() { |
| 528 | break |
| 529 | } |
| 530 | l.advance() |
| 531 | } |
| 532 | if l.ch != ':' { |
| 533 | l.postingExpectAccount = false |
| 534 | } |
| 535 | return token.Token{Type: token.TEXT, Literal: l.lit(s), Span: l.span(s)} |
| 536 | } |
| 537 | |
| 538 | func (l *Lexer) lexParenExpr() token.Token { |
| 539 | s := l.save() |
| 540 | depth := 0 |
| 541 | for l.ch != '\n' && l.ch != '\r' && l.ch != 0 { |
| 542 | if l.ch == '(' { |
| 543 | depth++ |
| 544 | } else if l.ch == ')' { |
| 545 | depth-- |
| 546 | if depth == 0 { |
| 547 | l.advance() |
| 548 | break |
| 549 | } |
| 550 | } |
| 551 | l.advance() |
| 552 | } |
| 553 | return token.Token{Type: token.PARENEXPR, Literal: l.lit(s), Span: l.span(s)} |
| 554 | } |
| 555 | |
| 556 | func (l *Lexer) lexNumber() token.Token { |
| 557 | s := l.save() |
| 558 | isDecimal := false |
| 559 | for { |
| 560 | if l.isDigit() || l.ch == '.' || l.ch == ',' || l.ch == '_' || l.ch == '\'' { |
| 561 | if l.ch == '.' || l.ch == ',' { |
| 562 | isDecimal = true |
| 563 | } |
| 564 | l.advance() |
| 565 | } else if l.ch == ' ' && (l.peek() >= '0' && l.peek() <= '9') { |
| 566 | isDecimal = true |
| 567 | l.advance() |
| 568 | } else if l.ch == 'e' || l.ch == 'E' { |
| 569 | // exponent: consume only when E[sign]digit, so `10E ` stays an integer |
| 570 | p := l.pos + 1 |
| 571 | if p < len(l.input) && (l.input[p] == '+' || l.input[p] == '-') { |
| 572 | p++ |
| 573 | } |
| 574 | if p >= len(l.input) || l.input[p] < '0' || l.input[p] > '9' { |
| 575 | break |
| 576 | } |
| 577 | isDecimal = true |
| 578 | l.advance() // e/E |
| 579 | if l.ch == '+' || l.ch == '-' { |
| 580 | l.advance() |
| 581 | } |
| 582 | for l.isDigit() { |
| 583 | l.advance() |
| 584 | } |
| 585 | } else { |
| 586 | break |
| 587 | } |
| 588 | } |
| 589 | lit := l.lit(s) |
| 590 | kind := token.INT |
| 591 | if isDecimal { |
| 592 | kind = token.DECIMAL |
| 593 | } |
| 594 | return token.Token{Type: kind, Literal: lit, Span: l.span(s)} |
| 595 | } |
| 596 | |
| 597 | func (l *Lexer) lexKeyword() token.Token { |
| 598 | s := l.save() |
| 599 | for l.ch != 0 && l.ch != '\n' && l.ch != '\r' && l.ch != ' ' && l.ch != '\t' && l.ch != ';' { |
| 600 | l.advance() |
| 601 | } |
| 602 | lit := l.lit(s) |
| 603 | kind := l.keyword(lit) |
| 604 | if kind == token.ILLEGAL { // todo: report an error ?? |
| 605 | kind = token.TEXT |
| 606 | } else { |
| 607 | l.mode = modeDirective |
| 608 | l.subdirective = kind == token.ACCOUNT || kind == token.COMMODITY |
| 609 | l.includePath = kind == token.INCLUDE |
| 610 | } |
| 611 | return token.Token{Type: kind, Literal: lit, Span: l.span(s)} |
| 612 | } |
| 613 | |
| 614 | func (l *Lexer) lexDate() token.Token { |
| 615 | s := l.save() |
| 616 | for l.isDigit() || (l.isDateSep() && l.peekIsDigit()) { |
| 617 | l.advance() |
| 618 | } |
| 619 | return token.Token{Type: token.DATE, Literal: l.lit(s), Span: l.span(s)} |
| 620 | } |
| 621 | |
| 622 | func isSymbolChar(r rune) bool { |
| 623 | return r == '$' || unicode.In(r, unicode.Sc) |
| 624 | } |
| 625 | |
| 626 | func (l *Lexer) lexString() token.Token { |
| 627 | s := l.save() |
| 628 | quote := l.ch |
| 629 | l.advance() // consume the quote character |
| 630 | for l.ch != quote && l.ch != '\n' && l.ch != 0 { |
| 631 | l.advance() |
| 632 | } |
| 633 | if l.ch == quote { |
| 634 | l.advance() |
| 635 | } |
| 636 | return token.Token{Type: token.STRING, Literal: l.lit(s), Span: l.span(s)} |
| 637 | } |
| 638 | |
| 639 | func (l *Lexer) lexCommodityMark() token.Token { |
| 640 | s := l.save() |
| 641 | |
| 642 | if l.ch == '"' { |
| 643 | l.advance() |
| 644 | for l.ch != '"' && l.ch != '\n' && l.ch != 0 { |
| 645 | l.advance() |
| 646 | } |
| 647 | if l.ch == '"' { |
| 648 | l.advance() |
| 649 | } |
| 650 | return token.Token{Type: token.COMMODITYMARK, Literal: l.lit(s), Span: l.span(s)} |
| 651 | } |
| 652 | |
| 653 | if unicode.IsLetter(l.ch) { |
| 654 | for unicode.IsLetter(l.ch) { |
| 655 | l.advance() |
| 656 | } |
| 657 | return token.Token{Type: token.COMMODITYMARK, Literal: l.lit(s), Span: l.span(s)} |
| 658 | } |
| 659 | |
| 660 | if isSymbolChar(l.ch) { |
| 661 | for isSymbolChar(l.ch) { |
| 662 | l.advance() |
| 663 | } |
| 664 | return token.Token{Type: token.COMMODITYMARK, Literal: l.lit(s), Span: l.span(s)} |
| 665 | } |
| 666 | |
| 667 | l.advance() |
| 668 | return token.Token{Type: token.COMMODITYMARK, Literal: l.lit(s), Span: l.span(s)} |
| 669 | } |
| 670 | |
| 671 | func (l *Lexer) lexLBrace() token.Token { |
| 672 | s := l.save() |
| 673 | l.advance() |
| 674 | if l.ch == '{' { |
| 675 | l.advance() |
| 676 | return token.Token{Type: token.LBRACELBRACE, Literal: "{{", Span: l.span(s)} |
| 677 | } |
| 678 | return token.Token{Type: token.LBRACE, Literal: "{", Span: l.span(s)} |
| 679 | } |
| 680 | |
| 681 | func (l *Lexer) lexRBrace() token.Token { |
| 682 | s := l.save() |
| 683 | l.advance() |
| 684 | if l.ch == '}' { |
| 685 | l.advance() |
| 686 | return token.Token{Type: token.RBRACERBRACE, Literal: "}}", Span: l.span(s)} |
| 687 | } |
| 688 | return token.Token{Type: token.RBRACE, Literal: "}", Span: l.span(s)} |
| 689 | } |
| 690 | |
| 691 | func (l *Lexer) advance() { |
| 692 | if l.rpos >= len(l.input) { |
| 693 | l.ch = 0 |
| 694 | l.chSize = 0 |
| 695 | } else if b := l.input[l.rpos]; b < utf8.RuneSelf { // ASCII fast path |
| 696 | l.ch = rune(b) |
| 697 | l.chSize = 1 |
| 698 | } else { |
| 699 | l.ch, l.chSize = utf8.DecodeRune(l.input[l.rpos:]) |
| 700 | } |
| 701 | l.pos = l.rpos |
| 702 | l.rpos += l.chSize |
| 703 | if l.ch == '\n' || l.ch == '\r' { |
| 704 | l.line++ |
| 705 | l.col = 0 |
| 706 | } else { |
| 707 | l.col++ |
| 708 | } |
| 709 | } |
| 710 | |
| 711 | func (l *Lexer) peek() rune { |
| 712 | if l.rpos < len(l.input) && l.input[l.rpos] < utf8.RuneSelf { |
| 713 | return rune(l.input[l.rpos]) |
| 714 | } |
| 715 | r, _ := utf8.DecodeRune(l.input[l.rpos:]) |
| 716 | return r |
| 717 | } |
| 718 | |
| 719 | func (l *Lexer) peekN(n int) byte { |
| 720 | if l.pos+n >= len(l.input) { |
| 721 | return 0 |
| 722 | } |
| 723 | return l.input[l.pos+n] |
| 724 | } |
| 725 | |
| 726 | func (l *Lexer) isDigit() bool { return l.ch >= '0' && l.ch <= '9' } |
| 727 | func (l *Lexer) isAlpha() bool { |
| 728 | return (l.ch >= 'a' && l.ch <= 'z') || |
| 729 | (l.ch >= 'A' && l.ch <= 'Z') |
| 730 | } |
| 731 | |
| 732 | func (l *Lexer) isTwoSpaces() bool { |
| 733 | return l.ch == ' ' && l.rpos < len(l.input) && l.input[l.rpos] == ' ' |
| 734 | } |
| 735 | |
| 736 | func (l *Lexer) isDateSep() bool { return l.ch == '-' || l.ch == '/' || l.ch == '.' } |
| 737 | |
| 738 | func (l *Lexer) peekIsDigit() bool { |
| 739 | r := l.peek() |
| 740 | return r >= '0' && r <= '9' |
| 741 | } |
| 742 | |
| 743 | func (l *Lexer) isCommodityStart() bool { |
| 744 | if l.ch == '$' || (l.ch >= 'A' && l.ch <= 'Z') { |
| 745 | return true |
| 746 | } |
| 747 | if l.ch < utf8.RuneSelf { |
| 748 | return false |
| 749 | } |
| 750 | return unicode.In(l.ch, unicode.Sc) || unicode.IsLetter(l.ch) |
| 751 | } |
| 752 | |
| 753 | func (l *Lexer) isDate() bool { |
| 754 | if !l.isDigit() { |
| 755 | return false |
| 756 | } |
| 757 | // YYYY/M/D or YYYY/MM/DD |
| 758 | if l.peekN(1) >= '0' && l.peekN(1) <= '9' && |
| 759 | l.peekN(2) >= '0' && l.peekN(2) <= '9' && |
| 760 | l.peekN(3) >= '0' && l.peekN(3) <= '9' { |
| 761 | sep := l.peekN(4) |
| 762 | if sep == '/' || sep == '-' || sep == '.' { |
| 763 | if l.peekN(5) >= '0' && l.peekN(5) <= '9' { |
| 764 | if l.peekN(6) == sep { |
| 765 | return l.peekN(7) >= '0' && l.peekN(7) <= '9' |
| 766 | } |
| 767 | if l.peekN(7) == sep { |
| 768 | return l.peekN(8) >= '0' && l.peekN(8) <= '9' |
| 769 | } |
| 770 | } |
| 771 | } |
| 772 | return false |
| 773 | } |
| 774 | // M/D or MM/DD(year inferred, only / and - separators; . is ambiguous with decimal numbers like 1.01) |
| 775 | if (l.peekN(1) == '/' || l.peekN(1) == '-') && |
| 776 | l.peekN(2) >= '0' && l.peekN(2) <= '9' && |
| 777 | l.ch >= '1' && l.ch <= '9' { |
| 778 | return validDay(l.peekN(2), l.peekN(3)) |
| 779 | } |
| 780 | if (l.peekN(2) == '/' || l.peekN(2) == '-') && |
| 781 | l.peekN(3) >= '0' && l.peekN(3) <= '9' { |
| 782 | m := int(l.ch-'0')*10 + int(l.peekN(1)-'0') |
| 783 | return m >= 1 && m <= 12 && validDay(l.peekN(3), l.peekN(4)) |
| 784 | } |
| 785 | return false |
| 786 | } |
| 787 | |
| 788 | func validDay(first, second byte) bool { |
| 789 | d := int(first - '0') |
| 790 | if second >= '0' && second <= '9' { |
| 791 | d = d*10 + int(second-'0') |
| 792 | } |
| 793 | return d >= 1 && d <= 31 |
| 794 | } |
| 795 | |
| 796 | func (l *Lexer) isTime() bool { |
| 797 | if !l.isDigit() { |
| 798 | return false |
| 799 | } |
| 800 | return l.peekN(2) == ':' |
| 801 | } |
| 802 | |
| 803 | func (l *Lexer) lexTime() token.Token { |
| 804 | s := l.save() |
| 805 | for l.isDigit() || l.ch == ':' { |
| 806 | l.advance() |
| 807 | } |
| 808 | return token.Token{Type: token.TIME, Literal: l.lit(s), Span: l.span(s)} |
| 809 | } |
| 810 | |
| 811 | type savedPos struct{ offset, line, col int } |
| 812 | |
| 813 | func (l *Lexer) save() savedPos { |
| 814 | return savedPos{l.pos, l.line, l.col} |
| 815 | } |
| 816 | |
| 817 | func (l *Lexer) span(s savedPos) token.Span { |
| 818 | return token.Span{ |
| 819 | File: l.file, |
| 820 | Start: token.Pos{Offset: s.offset, Line: s.line, Col: s.col}, |
| 821 | End: token.Pos{Offset: l.pos, Line: l.line, Col: l.col}, |
| 822 | } |
| 823 | } |
| 824 | |
| 825 | func (l *Lexer) token(kind token.Type, literal string) token.Token { |
| 826 | s := savedPos{l.pos, l.line, l.col} |
| 827 | return token.Token{Type: kind, Literal: literal, Span: l.span(s)} |
| 828 | } |
| 829 | |
| 830 | func (l *Lexer) lit(s savedPos) string { |
| 831 | if s.offset == l.pos { |
| 832 | return "" |
| 833 | } |
| 834 | return unsafe.String(&l.input[s.offset], l.pos-s.offset) |
| 835 | } |
| 836 | |
| 837 | func (l *Lexer) keyword(s string) token.Type { |
| 838 | switch s { |
| 839 | case "comment": |
| 840 | return token.COMMENTKW |
| 841 | case "account": |
| 842 | return token.ACCOUNT |
| 843 | case "commodity": |
| 844 | return token.COMMODITY |
| 845 | case "include": |
| 846 | return token.INCLUDE |
| 847 | case "alias": |
| 848 | return token.ALIAS |
| 849 | case "payee": |
| 850 | return token.PAYEE |
| 851 | case "tag": |
| 852 | return token.TAG |
| 853 | case "apply": |
| 854 | return token.APPLY |
| 855 | case "end": |
| 856 | return token.END |
| 857 | case "Y", "year": |
| 858 | return token.YEAR |
| 859 | case "decimal-mark": |
| 860 | return token.DECIMALMARK |
| 861 | case "D": |
| 862 | return token.D |
| 863 | case "P": |
| 864 | return token.P |
| 865 | case "N": |
| 866 | return token.N |
| 867 | case "C": |
| 868 | return token.C |
| 869 | default: |
| 870 | return token.ILLEGAL |
| 871 | } |
| 872 | } |