clerk/journal/lexer/lexer.go (view raw)
Oleksandr Smirnov
Oleksandr Smirnov
olexsmir@gmail.com journal/parser: fix include directive, 1 month ago
olexsmir@gmail.com journal/parser: fix include directive, 1 month ago
| 1 | package lexer |
| 2 | |
| 3 | import ( |
| 4 | "strings" |
| 5 | "unicode" |
| 6 | "unicode/utf8" |
| 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 | |
| 59 | func New(file string, input []byte) *Lexer { |
| 60 | l := &Lexer{ |
| 61 | file: file, |
| 62 | input: input, |
| 63 | line: 1, |
| 64 | } |
| 65 | l.advance() |
| 66 | if l.ch == '\uFEFF' { // start of the input |
| 67 | l.advance() |
| 68 | } |
| 69 | return l |
| 70 | } |
| 71 | |
| 72 | // Next returns next token in the input |
| 73 | func (l *Lexer) Next() token.Token { |
| 74 | switch l.mode { |
| 75 | case ModeDefault: |
| 76 | return l.lexDefault() |
| 77 | case ModeComment: |
| 78 | return l.lexComment() |
| 79 | case ModeTransaction: |
| 80 | return l.lexTransaction() |
| 81 | case ModePosting: |
| 82 | return l.lexPosting() |
| 83 | case ModePeriodic: |
| 84 | return l.lexPeriodic() |
| 85 | case ModeAutomated: |
| 86 | return l.lexAutomated() |
| 87 | case ModeDirective: |
| 88 | return l.lexDirective() |
| 89 | } |
| 90 | panic("unreachable") |
| 91 | } |
| 92 | |
| 93 | func (l *Lexer) lexDefault() token.Token { |
| 94 | switch { |
| 95 | case l.ch == 0: |
| 96 | return l.token(token.EOF, "") |
| 97 | case l.ch == '\n': |
| 98 | return l.lexNewline() |
| 99 | case l.ch == '\r': |
| 100 | l.col = 0 |
| 101 | l.advance() |
| 102 | return l.lexNewline() |
| 103 | case l.ch == ' ' || l.ch == '\t': |
| 104 | tok := l.lexIndent() |
| 105 | l.mode = ModePosting |
| 106 | l.postingExpectAccount = true |
| 107 | return tok |
| 108 | case l.ch == ';' || l.ch == '#' || l.ch == '%': |
| 109 | l.mode = ModeComment |
| 110 | return l.lexSingle(token.SEMICOLON) // todo: ?? |
| 111 | case l.ch == '*': // * at col 0 == comment |
| 112 | l.mode = ModeComment |
| 113 | return l.lexSingle(token.STAR) |
| 114 | case l.ch == '~': |
| 115 | l.mode = ModePeriodic |
| 116 | return l.lexSingle(token.TILDE) |
| 117 | case l.ch == '=': |
| 118 | l.mode = ModeAutomated |
| 119 | return l.lexSingle(token.EQ) |
| 120 | case l.ch == '+': |
| 121 | return l.lexSingle(token.PLUS) |
| 122 | case l.ch == '-': |
| 123 | return l.lexSingle(token.MINUS) |
| 124 | case l.ch == '.': |
| 125 | return l.lexSingle(token.TEXT) |
| 126 | case l.ch == '!': |
| 127 | return l.lexSingle(token.BANG) |
| 128 | case l.ch == '@': |
| 129 | return l.lexSingle(token.AT) |
| 130 | case l.isAlpha(): |
| 131 | return l.lexKeyword() |
| 132 | case l.isDigit(): |
| 133 | if !l.isDate() { |
| 134 | s := l.save() |
| 135 | for l.isDigit() || l.ch == '-' || l.ch == '/' || l.ch == '.' { |
| 136 | l.advance() |
| 137 | } |
| 138 | return token.Token{Type: token.ILLEGAL, Literal: string(l.input[s.offset:l.pos]), Span: l.span(s)} |
| 139 | } |
| 140 | tok := l.lexDate() |
| 141 | l.mode = ModeTransaction |
| 142 | l.transactionPastStatus = false |
| 143 | return tok |
| 144 | default: |
| 145 | s := l.save() |
| 146 | l.advance() |
| 147 | return token.Token{Type: token.ILLEGAL, Literal: string(l.input[s.offset:l.pos]), Span: l.span(s)} |
| 148 | } |
| 149 | } |
| 150 | |
| 151 | func (l *Lexer) lexComment() token.Token { |
| 152 | if l.ch == '\n' || l.ch == 0 { |
| 153 | l.mode = ModeDefault |
| 154 | return l.lexNewline() |
| 155 | } |
| 156 | |
| 157 | for l.ch == ' ' || l.ch == '\t' { |
| 158 | l.lexWhitespace() |
| 159 | } |
| 160 | |
| 161 | if l.ch == '\n' || l.ch == 0 { |
| 162 | l.mode = ModeDefault |
| 163 | return l.lexNewline() |
| 164 | } |
| 165 | |
| 166 | s := l.save() |
| 167 | for l.ch != '\n' && l.ch != 0 { |
| 168 | l.advance() |
| 169 | } |
| 170 | return token.Token{Type: token.TEXT, Literal: string(l.input[s.offset:l.pos]), Span: l.span(s)} |
| 171 | } |
| 172 | |
| 173 | func (l *Lexer) lexTransaction() token.Token { |
| 174 | if l.readingNoteAfterPipe { |
| 175 | l.readingNoteAfterPipe = false |
| 176 | return l.lexNote() |
| 177 | } |
| 178 | |
| 179 | switch l.ch { |
| 180 | case 0: |
| 181 | return l.token(token.EOF, "") |
| 182 | case '\n': |
| 183 | l.mode = ModeDefault |
| 184 | return l.lexNewline() |
| 185 | case '\r': |
| 186 | l.col = 0 |
| 187 | l.advance() |
| 188 | return l.lexNewline() |
| 189 | case ' ', '\t': |
| 190 | return l.lexWhitespace() |
| 191 | case ';': |
| 192 | l.mode = ModeComment |
| 193 | return l.lexSingle(token.SEMICOLON) |
| 194 | case '*': |
| 195 | if !l.transactionPastStatus { |
| 196 | l.transactionPastStatus = true |
| 197 | return l.lexSingle(token.STAR) |
| 198 | } |
| 199 | return l.lexText() |
| 200 | case '!': |
| 201 | if !l.transactionPastStatus { |
| 202 | l.transactionPastStatus = true |
| 203 | return l.lexSingle(token.BANG) |
| 204 | } |
| 205 | return l.lexText() |
| 206 | case '|': |
| 207 | l.transactionPastStatus = true |
| 208 | l.readingNoteAfterPipe = true |
| 209 | return l.lexSingle(token.PIPE) |
| 210 | case '+': |
| 211 | return l.lexSingle(token.PLUS) |
| 212 | case '-': |
| 213 | return l.lexSingle(token.MINUS) |
| 214 | case '=': |
| 215 | return l.lexEquals() |
| 216 | case '"', '\'': |
| 217 | return l.lexString() |
| 218 | default: // description / payee |
| 219 | if l.isDate() { // secondsry date after = |
| 220 | return l.lexDate() |
| 221 | } |
| 222 | return l.lexText() |
| 223 | } |
| 224 | } |
| 225 | |
| 226 | func (l *Lexer) lexNote() token.Token { |
| 227 | for l.ch == ' ' || l.ch == '\t' { |
| 228 | l.advance() |
| 229 | } |
| 230 | if l.ch == 0 || l.ch == '\n' || l.ch == ';' { |
| 231 | return l.Next() |
| 232 | } |
| 233 | s := l.save() |
| 234 | for l.ch != 0 && l.ch != '\n' && l.ch != ';' { |
| 235 | l.advance() |
| 236 | } |
| 237 | lit := string(l.input[s.offset:l.pos]) |
| 238 | for len(lit) > 0 && (lit[len(lit)-1] == ' ' || lit[len(lit)-1] == '\t') { |
| 239 | lit = lit[:len(lit)-1] |
| 240 | } |
| 241 | return token.Token{Type: token.TEXT, Literal: lit, Span: l.span(s)} |
| 242 | } |
| 243 | |
| 244 | func (l *Lexer) lexPeriodic() token.Token { |
| 245 | switch l.ch { |
| 246 | case 0: |
| 247 | return l.token(token.EOF, "") |
| 248 | case '\n': |
| 249 | l.mode = ModeDefault |
| 250 | return l.lexNewline() |
| 251 | case '\r': |
| 252 | l.col = 0 |
| 253 | l.advance() |
| 254 | return l.lexNewline() |
| 255 | case ';': |
| 256 | l.mode = ModeComment |
| 257 | return l.lexSingle(token.SEMICOLON) |
| 258 | case ' ', '\t': |
| 259 | return l.lexWhitespace() |
| 260 | default: |
| 261 | return l.lexText() |
| 262 | } |
| 263 | } |
| 264 | |
| 265 | func (l *Lexer) lexAutomated() token.Token { |
| 266 | switch l.ch { |
| 267 | case 0: |
| 268 | return l.token(token.EOF, "") |
| 269 | case '\n': |
| 270 | l.mode = ModeDefault |
| 271 | return l.lexNewline() |
| 272 | case '\r': |
| 273 | l.col = 0 |
| 274 | l.advance() |
| 275 | return l.lexNewline() |
| 276 | case ' ', '\t': |
| 277 | return l.lexWhitespace() |
| 278 | case ';': |
| 279 | l.mode = ModeComment |
| 280 | return l.lexSingle(token.SEMICOLON) |
| 281 | default: |
| 282 | return l.lexText() |
| 283 | } |
| 284 | } |
| 285 | |
| 286 | func (l *Lexer) lexPosting() token.Token { |
| 287 | switch { |
| 288 | case l.ch == 0: |
| 289 | l.postingExpectAccount = false |
| 290 | return l.token(token.EOF, "") |
| 291 | case l.ch == '\n': |
| 292 | l.postingExpectAccount = false |
| 293 | l.mode = ModeDefault |
| 294 | return l.lexNewline() |
| 295 | case l.ch == ';': |
| 296 | l.postingExpectAccount = false |
| 297 | l.mode = ModeComment |
| 298 | return l.lexSingle(token.SEMICOLON) |
| 299 | case l.ch == ' ' || l.ch == '\t': |
| 300 | return l.lexWhitespace() |
| 301 | case l.postingExpectAccount && l.ch == '*': |
| 302 | return l.lexSingle(token.STAR) |
| 303 | case l.postingExpectAccount && l.ch == '!': |
| 304 | return l.lexSingle(token.BANG) |
| 305 | case l.ch == '=': |
| 306 | return l.lexEquals() |
| 307 | case l.ch == '@': |
| 308 | return l.lexAt() |
| 309 | case l.ch == '{': |
| 310 | return l.lexLBrace() |
| 311 | case l.ch == '}': |
| 312 | return l.lexRBrace() |
| 313 | case l.ch == '(': |
| 314 | if !l.postingExpectAccount { |
| 315 | return l.lexParenExpr() |
| 316 | } |
| 317 | return l.lexSingle(token.LPAREN) |
| 318 | case l.ch == ')': |
| 319 | return l.lexSingle(token.RPAREN) |
| 320 | case l.ch == '[': |
| 321 | return l.lexSingle(token.LBRACKET) |
| 322 | case l.ch == ']': |
| 323 | return l.lexSingle(token.RBRACKET) |
| 324 | case l.postingExpectAccount && l.ch != '*' && l.ch != '!' && l.ch != '(' && l.ch != '[': |
| 325 | l.postingExpectAccount = false |
| 326 | return l.lexAccountText() |
| 327 | case l.ch == '*': // after account name |
| 328 | return l.lexSingle(token.STAR) |
| 329 | case l.isDigit(), l.ch == '.': |
| 330 | return l.lexNumber() |
| 331 | case l.ch == '-': |
| 332 | return l.lexSingle(token.MINUS) |
| 333 | case l.ch == '+': |
| 334 | return l.lexSingle(token.PLUS) |
| 335 | case l.isCommodityStart(): |
| 336 | return l.lexCommodityMark() |
| 337 | case l.ch >= 'a' && l.ch <= 'z': |
| 338 | return l.lexCommodityMark() |
| 339 | default: |
| 340 | return l.lexAccountText() |
| 341 | } |
| 342 | } |
| 343 | |
| 344 | func (l *Lexer) lexDirective() token.Token { |
| 345 | switch l.ch { |
| 346 | case '\n', 0: |
| 347 | l.mode = ModeDefault |
| 348 | return l.lexNewline() |
| 349 | case ';': |
| 350 | l.mode = ModeComment |
| 351 | return l.lexSingle(token.SEMICOLON) |
| 352 | case ' ', '\t': |
| 353 | return l.lexWhitespace() |
| 354 | case '=': |
| 355 | return l.lexSingle(token.EQ) |
| 356 | case '+': |
| 357 | return l.lexSingle(token.PLUS) |
| 358 | case '-': |
| 359 | return l.lexSingle(token.MINUS) |
| 360 | case '"', '\'': |
| 361 | return l.lexString() |
| 362 | default: |
| 363 | if l.isCommodityStart() { |
| 364 | return l.lexCommodityMark() |
| 365 | } |
| 366 | if l.isTime() { |
| 367 | return l.lexTime() |
| 368 | } |
| 369 | if l.isDate() { |
| 370 | return l.lexDate() |
| 371 | } |
| 372 | if l.isDigit() { |
| 373 | return l.lexNumber() |
| 374 | } |
| 375 | return l.lexText() |
| 376 | } |
| 377 | // case l.ch == '/': // regex in 'alias' |
| 378 | // return l.lexSingle(token.SLASH) |
| 379 | } |
| 380 | |
| 381 | func (l *Lexer) lexSingle(kind token.Type) token.Token { |
| 382 | s := l.save() |
| 383 | l.advance() |
| 384 | return token.Token{ |
| 385 | Type: kind, |
| 386 | Literal: string(l.input[s.offset:l.pos]), |
| 387 | Span: l.span(s), |
| 388 | } |
| 389 | } |
| 390 | |
| 391 | func (l *Lexer) lexNewline() token.Token { |
| 392 | s := l.save() |
| 393 | l.advance() |
| 394 | l.mode = ModeDefault |
| 395 | return token.Token{Type: token.NEWLINE, Literal: "\n", Span: l.span(s)} |
| 396 | } |
| 397 | |
| 398 | func (l *Lexer) lexWhitespace() token.Token { |
| 399 | s := l.save() |
| 400 | for l.ch == ' ' || l.ch == '\t' { |
| 401 | l.advance() |
| 402 | } |
| 403 | lit := string(l.input[s.offset:l.pos]) |
| 404 | return token.Token{Type: token.WHITESPACE, Literal: lit, Span: l.span(s)} |
| 405 | } |
| 406 | |
| 407 | func (l *Lexer) lexIndent() token.Token { |
| 408 | s := l.save() |
| 409 | for l.ch == ' ' || l.ch == '\t' { |
| 410 | l.advance() |
| 411 | } |
| 412 | lit := string(l.input[s.offset:l.pos]) |
| 413 | return token.Token{Type: token.INDENT, Literal: lit, Span: l.span(s)} |
| 414 | } |
| 415 | |
| 416 | func (l *Lexer) lexEquals() token.Token { |
| 417 | s := l.save() |
| 418 | l.advance() |
| 419 | if l.ch == '=' { |
| 420 | l.advance() |
| 421 | switch l.ch { |
| 422 | case '=': |
| 423 | l.advance() |
| 424 | return token.Token{Type: token.EQEQEQ, Literal: "===", Span: l.span(s)} |
| 425 | case '*': |
| 426 | l.advance() |
| 427 | return token.Token{Type: token.EQEQEQ, Literal: "==*", Span: l.span(s)} |
| 428 | default: |
| 429 | return token.Token{Type: token.EQEQ, Literal: "==", Span: l.span(s)} |
| 430 | } |
| 431 | } |
| 432 | return token.Token{Type: token.EQ, Literal: "=", Span: l.span(s)} |
| 433 | } |
| 434 | |
| 435 | func (l *Lexer) lexAt() token.Token { |
| 436 | s := l.save() |
| 437 | l.advance() |
| 438 | if l.ch == '@' { |
| 439 | l.advance() |
| 440 | return token.Token{Type: token.ATAT, Literal: "@@", Span: l.span(s)} |
| 441 | } |
| 442 | return token.Token{Type: token.AT, Literal: "@", Span: l.span(s)} |
| 443 | } |
| 444 | |
| 445 | func (l *Lexer) lexText() token.Token { |
| 446 | s := l.save() |
| 447 | l.advance() |
| 448 | for l.ch != '\n' && l.ch != ';' && l.ch != 0 && l.ch != ' ' && l.ch != '\t' { |
| 449 | l.advance() |
| 450 | } |
| 451 | lit := string(l.input[s.offset:l.pos]) |
| 452 | return token.Token{Type: token.TEXT, Literal: lit, Span: l.span(s)} |
| 453 | } |
| 454 | |
| 455 | func (l *Lexer) lexAccountText() token.Token { |
| 456 | s := l.save() |
| 457 | for l.ch != '\n' && l.ch != ';' && l.ch != 0 && l.ch != ')' && l.ch != ']' { |
| 458 | // two spaces = end of account name |
| 459 | if l.isTwoSpaces() { |
| 460 | break |
| 461 | } |
| 462 | l.advance() |
| 463 | } |
| 464 | lit := string(l.input[s.offset:l.pos]) |
| 465 | return token.Token{Type: token.TEXT, Literal: lit, Span: l.span(s)} |
| 466 | } |
| 467 | |
| 468 | func (l *Lexer) lexParenExpr() token.Token { |
| 469 | s := l.save() |
| 470 | depth := 0 |
| 471 | for l.ch != '\n' && l.ch != 0 { |
| 472 | if l.ch == '(' { |
| 473 | depth++ |
| 474 | } else if l.ch == ')' { |
| 475 | depth-- |
| 476 | if depth == 0 { |
| 477 | l.advance() |
| 478 | break |
| 479 | } |
| 480 | } |
| 481 | l.advance() |
| 482 | } |
| 483 | lit := string(l.input[s.offset:l.pos]) |
| 484 | return token.Token{Type: token.PARENEXPR, Literal: lit, Span: l.span(s)} |
| 485 | } |
| 486 | |
| 487 | func (l *Lexer) lexNumber() token.Token { |
| 488 | s := l.save() |
| 489 | for { |
| 490 | if l.isDigit() || l.ch == '.' || l.ch == ',' || l.ch == '_' || l.ch == '\'' { |
| 491 | l.advance() |
| 492 | } else if l.ch == ' ' && (l.peek() >= '0' && l.peek() <= '9') { |
| 493 | l.advance() |
| 494 | } else { |
| 495 | break |
| 496 | } |
| 497 | } |
| 498 | lit := string(l.input[s.offset:l.pos]) |
| 499 | kind := token.INT |
| 500 | if strings.ContainsAny(lit, "., ") { |
| 501 | kind = token.DECIMAL |
| 502 | } |
| 503 | return token.Token{Type: kind, Literal: lit, Span: l.span(s)} |
| 504 | } |
| 505 | |
| 506 | func (l *Lexer) lexKeyword() token.Token { |
| 507 | s := l.save() |
| 508 | for l.ch != 0 && l.ch != '\n' && l.ch != '\r' && l.ch != ' ' && l.ch != '\t' && l.ch != ';' { |
| 509 | l.advance() |
| 510 | } |
| 511 | lit := string(l.input[s.offset:l.pos]) |
| 512 | kind := l.keyword(lit) |
| 513 | if kind == token.ILLEGAL { // todo: report an error ?? |
| 514 | kind = token.TEXT |
| 515 | } else { |
| 516 | l.mode = ModeDirective |
| 517 | } |
| 518 | return token.Token{Type: kind, Literal: lit, Span: l.span(s)} |
| 519 | } |
| 520 | |
| 521 | func (l *Lexer) lexDate() token.Token { |
| 522 | s := l.save() |
| 523 | for l.isDigit() || (l.isDateSep() && l.peekIsDigit()) { |
| 524 | l.advance() |
| 525 | } |
| 526 | return token.Token{Type: token.DATE, Literal: string(l.input[s.offset:l.pos]), Span: l.span(s)} |
| 527 | } |
| 528 | |
| 529 | func isSymbolChar(r rune) bool { |
| 530 | return r == '$' || unicode.In(r, unicode.Sc) |
| 531 | } |
| 532 | |
| 533 | func (l *Lexer) lexString() token.Token { |
| 534 | s := l.save() |
| 535 | quote := l.ch |
| 536 | l.advance() // consume the quote character |
| 537 | for l.ch != quote && l.ch != '\n' && l.ch != 0 { |
| 538 | l.advance() |
| 539 | } |
| 540 | if l.ch == quote { |
| 541 | l.advance() |
| 542 | } |
| 543 | return token.Token{Type: token.STRING, Literal: string(l.input[s.offset:l.pos]), Span: l.span(s)} |
| 544 | } |
| 545 | |
| 546 | func (l *Lexer) lexCommodityMark() token.Token { |
| 547 | s := l.save() |
| 548 | |
| 549 | if l.ch == '"' { |
| 550 | l.advance() |
| 551 | for l.ch != '"' && l.ch != '\n' && l.ch != 0 { |
| 552 | l.advance() |
| 553 | } |
| 554 | if l.ch == '"' { |
| 555 | l.advance() |
| 556 | } |
| 557 | return token.Token{Type: token.COMMODITYMARK, Literal: string(l.input[s.offset:l.pos]), Span: l.span(s)} |
| 558 | } |
| 559 | |
| 560 | if unicode.IsLetter(l.ch) { |
| 561 | for unicode.IsLetter(l.ch) || unicode.IsDigit(l.ch) { |
| 562 | l.advance() |
| 563 | } |
| 564 | return token.Token{Type: token.COMMODITYMARK, Literal: string(l.input[s.offset:l.pos]), Span: l.span(s)} |
| 565 | } |
| 566 | |
| 567 | if isSymbolChar(l.ch) { |
| 568 | for isSymbolChar(l.ch) { |
| 569 | l.advance() |
| 570 | } |
| 571 | return token.Token{Type: token.COMMODITYMARK, Literal: string(l.input[s.offset:l.pos]), Span: l.span(s)} |
| 572 | } |
| 573 | |
| 574 | l.advance() |
| 575 | return token.Token{Type: token.COMMODITYMARK, Literal: string(l.input[s.offset:l.pos]), Span: l.span(s)} |
| 576 | } |
| 577 | |
| 578 | func (l *Lexer) lexLBrace() token.Token { |
| 579 | s := l.save() |
| 580 | l.advance() |
| 581 | if l.ch == '{' { |
| 582 | l.advance() |
| 583 | return token.Token{Type: token.LBRACELBRACE, Literal: "{{", Span: l.span(s)} |
| 584 | } |
| 585 | return token.Token{Type: token.LBRACE, Literal: "{", Span: l.span(s)} |
| 586 | } |
| 587 | |
| 588 | func (l *Lexer) lexRBrace() token.Token { |
| 589 | s := l.save() |
| 590 | l.advance() |
| 591 | if l.ch == '}' { |
| 592 | l.advance() |
| 593 | return token.Token{Type: token.RBRACERBRACE, Literal: "}}", Span: l.span(s)} |
| 594 | } |
| 595 | return token.Token{Type: token.RBRACE, Literal: "}", Span: l.span(s)} |
| 596 | } |
| 597 | |
| 598 | func (l *Lexer) advance() { |
| 599 | if l.rpos >= len(l.input) { |
| 600 | l.ch = 0 |
| 601 | l.chSize = 0 |
| 602 | } else { |
| 603 | r, size := utf8.DecodeRune(l.input[l.rpos:]) |
| 604 | l.ch = r |
| 605 | l.chSize = size |
| 606 | } |
| 607 | l.pos = l.rpos |
| 608 | l.rpos += l.chSize |
| 609 | if l.ch == '\n' || l.ch == '\r' { |
| 610 | l.line++ |
| 611 | l.col = 0 |
| 612 | } else { |
| 613 | l.col++ |
| 614 | } |
| 615 | } |
| 616 | |
| 617 | func (l *Lexer) peek() rune { |
| 618 | r, _ := utf8.DecodeRune(l.input[l.rpos:]) |
| 619 | return r |
| 620 | } |
| 621 | |
| 622 | func (l *Lexer) peekN(n int) byte { |
| 623 | if l.pos+n >= len(l.input) { |
| 624 | return 0 |
| 625 | } |
| 626 | return l.input[l.pos+n] |
| 627 | } |
| 628 | |
| 629 | func (l *Lexer) isDigit() bool { return l.ch >= '0' && l.ch <= '9' } |
| 630 | func (l *Lexer) isAlpha() bool { |
| 631 | return (l.ch >= 'a' && l.ch <= 'z') || |
| 632 | (l.ch >= 'A' && l.ch <= 'Z') |
| 633 | } |
| 634 | |
| 635 | func (l *Lexer) isTwoSpaces() bool { return l.ch == ' ' && l.peek() == ' ' } |
| 636 | |
| 637 | func (l *Lexer) isDateSep() bool { return l.ch == '-' || l.ch == '/' || l.ch == '.' } |
| 638 | |
| 639 | func (l *Lexer) peekIsDigit() bool { |
| 640 | r := l.peek() |
| 641 | return r >= '0' && r <= '9' |
| 642 | } |
| 643 | |
| 644 | func (l *Lexer) isCommodityStart() bool { |
| 645 | if l.ch == '$' || (l.ch >= 'A' && l.ch <= 'Z') { |
| 646 | return true |
| 647 | } |
| 648 | if l.ch < utf8.RuneSelf { |
| 649 | return false |
| 650 | } |
| 651 | return unicode.In(l.ch, unicode.Sc) || unicode.IsLetter(l.ch) |
| 652 | } |
| 653 | |
| 654 | func (l *Lexer) isDate() bool { |
| 655 | if !l.isDigit() { |
| 656 | return false |
| 657 | } |
| 658 | // YYYY/M/D or YYYY/MM/DD |
| 659 | if l.peekN(1) >= '0' && l.peekN(1) <= '9' && |
| 660 | l.peekN(2) >= '0' && l.peekN(2) <= '9' && |
| 661 | l.peekN(3) >= '0' && l.peekN(3) <= '9' { |
| 662 | sep := l.peekN(4) |
| 663 | if sep == '/' || sep == '-' || sep == '.' { |
| 664 | if l.peekN(5) >= '0' && l.peekN(5) <= '9' { |
| 665 | if l.peekN(6) == sep { |
| 666 | return l.peekN(7) >= '0' && l.peekN(7) <= '9' |
| 667 | } |
| 668 | if l.peekN(7) == sep { |
| 669 | return l.peekN(8) >= '0' && l.peekN(8) <= '9' |
| 670 | } |
| 671 | } |
| 672 | } |
| 673 | return false |
| 674 | } |
| 675 | // M/D or MM/DD(year inferred, only / and - separators; . is ambiguous with decimal numbers like 1.01) |
| 676 | if (l.peekN(1) == '/' || l.peekN(1) == '-') && |
| 677 | l.peekN(2) >= '0' && l.peekN(2) <= '9' && |
| 678 | l.ch >= '1' && l.ch <= '9' { |
| 679 | return validDay(l.peekN(2), l.peekN(3)) |
| 680 | } |
| 681 | if (l.peekN(2) == '/' || l.peekN(2) == '-') && |
| 682 | l.peekN(3) >= '0' && l.peekN(3) <= '9' { |
| 683 | m := int(l.ch-'0')*10 + int(l.peekN(1)-'0') |
| 684 | return m >= 1 && m <= 12 && validDay(l.peekN(3), l.peekN(4)) |
| 685 | } |
| 686 | return false |
| 687 | } |
| 688 | |
| 689 | func validDay(first, second byte) bool { |
| 690 | d := int(first - '0') |
| 691 | if second >= '0' && second <= '9' { |
| 692 | d = d*10 + int(second-'0') |
| 693 | } |
| 694 | return d >= 1 && d <= 31 |
| 695 | } |
| 696 | |
| 697 | func (l *Lexer) isTime() bool { |
| 698 | if !l.isDigit() { |
| 699 | return false |
| 700 | } |
| 701 | return l.peekN(2) == ':' |
| 702 | } |
| 703 | |
| 704 | func (l *Lexer) lexTime() token.Token { |
| 705 | s := l.save() |
| 706 | for l.isDigit() || l.ch == ':' { |
| 707 | l.advance() |
| 708 | } |
| 709 | return token.Token{Type: token.TIME, Literal: string(l.input[s.offset:l.pos]), Span: l.span(s)} |
| 710 | } |
| 711 | |
| 712 | type savedPos struct{ offset, line, col int } |
| 713 | |
| 714 | func (l *Lexer) save() savedPos { |
| 715 | return savedPos{l.pos, l.line, l.col} |
| 716 | } |
| 717 | |
| 718 | func (l *Lexer) span(s savedPos) token.Span { |
| 719 | return token.Span{ |
| 720 | Start: token.Pos{File: l.file, Offset: s.offset, Line: s.line, Col: s.col}, |
| 721 | End: token.Pos{File: l.file, Offset: l.pos, Line: l.line, Col: l.col}, |
| 722 | } |
| 723 | } |
| 724 | |
| 725 | func (l *Lexer) token(kind token.Type, literal string) token.Token { |
| 726 | s := savedPos{l.pos, l.line, l.col} |
| 727 | return token.Token{Type: kind, Literal: literal, Span: l.span(s)} |
| 728 | } |
| 729 | |
| 730 | func (l *Lexer) keyword(s string) token.Type { |
| 731 | switch s { |
| 732 | case "comment": |
| 733 | return token.COMMENTKW |
| 734 | case "account": |
| 735 | return token.ACCOUNT |
| 736 | case "commodity": |
| 737 | return token.COMMODITY |
| 738 | case "include": |
| 739 | return token.INCLUDE |
| 740 | case "alias": |
| 741 | return token.ALIAS |
| 742 | case "payee": |
| 743 | return token.PAYEE |
| 744 | case "tag": |
| 745 | return token.TAG |
| 746 | case "apply": |
| 747 | return token.APPLY |
| 748 | case "end": |
| 749 | return token.END |
| 750 | case "Y", "year": |
| 751 | return token.YEAR |
| 752 | case "decimal-mark": |
| 753 | return token.DECIMALMARK |
| 754 | case "D": |
| 755 | return token.D |
| 756 | case "P": |
| 757 | return token.P |
| 758 | case "N": |
| 759 | return token.N |
| 760 | case "C": |
| 761 | return token.C |
| 762 | default: |
| 763 | return token.ILLEGAL |
| 764 | } |
| 765 | } |