clerk/journal/parser/parser.go (view raw)
Oleksandr Smirnov
Oleksandr Smirnov
olexsmir@gmail.com parser: increase preallocated size of account segments, 1 month ago
olexsmir@gmail.com parser: increase preallocated size of account segments, 1 month ago
| 1 | package parser |
| 2 | |
| 3 | import ( |
| 4 | "fmt" |
| 5 | "strconv" |
| 6 | "strings" |
| 7 | "unicode" |
| 8 | "unicode/utf8" |
| 9 | |
| 10 | "olexsmir.xyz/clerk/internal/decimal" |
| 11 | "olexsmir.xyz/clerk/journal/ast" |
| 12 | "olexsmir.xyz/clerk/journal/lexer" |
| 13 | "olexsmir.xyz/clerk/journal/token" |
| 14 | ) |
| 15 | |
| 16 | type Parser struct { |
| 17 | lexer *lexer.Lexer |
| 18 | errors []*ast.ParseError |
| 19 | cur token.Token |
| 20 | peek token.Token |
| 21 | |
| 22 | defaultYear int // set by year directive, used for short date inference |
| 23 | } |
| 24 | |
| 25 | func New(lex *lexer.Lexer) *Parser { |
| 26 | p := &Parser{lexer: lex} |
| 27 | p.advance() // populate .peek |
| 28 | p.advance() // populate .cur |
| 29 | return p |
| 30 | } |
| 31 | |
| 32 | func NewWithYear(lex *lexer.Lexer, year int) *Parser { |
| 33 | p := &Parser{lexer: lex, defaultYear: year} |
| 34 | p.advance() // populate .peek |
| 35 | p.advance() // populate .cur |
| 36 | return p |
| 37 | } |
| 38 | |
| 39 | func (p *Parser) ParseJournal() *ast.Journal { |
| 40 | f := &ast.Journal{} |
| 41 | for p.cur.Type != token.EOF { |
| 42 | if e := p.parseEntry(); e != nil { |
| 43 | f.Entries = append(f.Entries, e) |
| 44 | } |
| 45 | } |
| 46 | f.Errors = p.errors |
| 47 | return f |
| 48 | } |
| 49 | |
| 50 | func (p *Parser) parseEntry() ast.Entry { |
| 51 | if p.got(token.BANG) || p.got(token.AT) { |
| 52 | if isDirectiveKeyword(p.peek.Type) { |
| 53 | p.advance() // consume prefix |
| 54 | } |
| 55 | } |
| 56 | |
| 57 | switch p.cur.Type { |
| 58 | case token.ILLEGAL: |
| 59 | p.errorf("illegal character %q", p.cur.Literal) |
| 60 | p.advance() |
| 61 | return nil |
| 62 | case token.INDENT: |
| 63 | p.errorf("unexpected indent") |
| 64 | p.syncToNextline() |
| 65 | return nil |
| 66 | case token.DATE: |
| 67 | return p.parseTransaction() |
| 68 | case token.TILDE: |
| 69 | return p.parsePeriodicTransaction() |
| 70 | case token.EQ: |
| 71 | return p.parseAutomatedTransaction() |
| 72 | case token.NEWLINE: |
| 73 | return p.parseBlankLine() |
| 74 | case token.SEMICOLON, token.HASH, token.PERCENT, token.STAR: |
| 75 | return p.parseComment() |
| 76 | case token.ACCOUNT: |
| 77 | return p.parseAccountDirective() |
| 78 | case token.COMMODITY: |
| 79 | return p.parseCommodityDirective() |
| 80 | case token.INCLUDE: |
| 81 | return p.parseIncludeDirective() |
| 82 | case token.ALIAS: |
| 83 | return p.parseAliasDirective() |
| 84 | case token.PAYEE: |
| 85 | return p.parsePayeeDirective() |
| 86 | case token.TAG: |
| 87 | return p.parseTagDirective() |
| 88 | case token.YEAR: |
| 89 | return p.parseYearDirective() |
| 90 | case token.DECIMALMARK: |
| 91 | return p.parseDecimalMarkDirective() |
| 92 | case token.D: |
| 93 | return p.parseDefaultCommodityDirective() |
| 94 | case token.P: |
| 95 | return p.parseMarketPriceDirective() |
| 96 | case token.N: |
| 97 | return p.parseIgnoredDirective() |
| 98 | case token.C: |
| 99 | return p.parseConversionDirective() |
| 100 | case token.APPLY: |
| 101 | return p.parseApplyDirective() |
| 102 | case token.END: |
| 103 | return p.parseEndDirective() |
| 104 | case token.COMMENTKW: |
| 105 | return p.parseCommentBlockDirective() |
| 106 | default: |
| 107 | p.errorf("unexpected token %s", p.cur.Type) |
| 108 | p.sync() |
| 109 | return nil |
| 110 | } |
| 111 | } |
| 112 | |
| 113 | func (p *Parser) parseTransaction() *ast.Transaction { |
| 114 | s := p.cur.Span |
| 115 | tx := &ast.Transaction{} |
| 116 | |
| 117 | tx.Date = p.parseDate() |
| 118 | |
| 119 | p.skipWhitespace() |
| 120 | |
| 121 | // optional secondary date |
| 122 | if p.got(token.EQ) { |
| 123 | p.advance() |
| 124 | p.skipWhitespace() |
| 125 | d := p.parseDate() |
| 126 | tx.SecondDate = &d |
| 127 | } |
| 128 | |
| 129 | p.skipWhitespace() |
| 130 | |
| 131 | // optional status |
| 132 | tx.Status = p.parseStatus() |
| 133 | |
| 134 | // optional code - the lexer emits "(CODE)" as a single TEXT token; split it here |
| 135 | if p.got(token.TEXT) { |
| 136 | if lit := p.cur.Literal; len(lit) >= 2 && lit[0] == '(' && lit[len(lit)-1] == ')' { |
| 137 | tx.Code = &ast.Code{Value: lit[1 : len(lit)-1], Span: p.cur.Span} |
| 138 | p.advance() |
| 139 | p.skipWhitespace() |
| 140 | } |
| 141 | } |
| 142 | |
| 143 | // optional payee | note |
| 144 | if p.got(token.TEXT) || p.got(token.STRING) { |
| 145 | tx.Payee = p.parsePayee() |
| 146 | |
| 147 | // check for | separator |
| 148 | p.skipWhitespace() |
| 149 | |
| 150 | if p.got(token.PIPE) { |
| 151 | p.advance() |
| 152 | if p.got(token.TEXT) { |
| 153 | sn := p.cur.Span |
| 154 | n := p.cur.Literal |
| 155 | p.advance() |
| 156 | tx.Note = &ast.Note{Value: n, Span: p.span(sn)} |
| 157 | } |
| 158 | } |
| 159 | } |
| 160 | |
| 161 | tx.Comment = p.parseOptInlineComment() |
| 162 | p.expectNewline() |
| 163 | |
| 164 | tx.HeaderComments, tx.Postings = p.parseHeaderCommentsAndPostings() |
| 165 | |
| 166 | tx.Span = p.span(s) |
| 167 | return tx |
| 168 | } |
| 169 | |
| 170 | func unquote(s string) string { |
| 171 | if len(s) >= 2 && ((s[0] == '"' && s[len(s)-1] == '"') || (s[0] == '\'' && s[len(s)-1] == '\'')) { |
| 172 | return s[1 : len(s)-1] |
| 173 | } |
| 174 | return s |
| 175 | } |
| 176 | |
| 177 | func (p *Parser) parsePayee() *ast.Payee { |
| 178 | s := p.cur.Span |
| 179 | |
| 180 | if p.got(token.STRING) { |
| 181 | name := unquote(p.cur.Literal) |
| 182 | p.advance() |
| 183 | return &ast.Payee{Name: name, Span: p.span(s)} |
| 184 | } |
| 185 | |
| 186 | // keep spaces/tags between text tokens; stop before trailing whitespace |
| 187 | var name strings.Builder |
| 188 | for isPayeeWord(p.cur.Type) || (isPayeeWord(p.peek.Type) && p.got(token.WHITESPACE)) { |
| 189 | _, _ = name.WriteString(p.cur.Literal) |
| 190 | p.advance() |
| 191 | } |
| 192 | return &ast.Payee{Name: unquote(name.String()), Span: p.span(s)} |
| 193 | } |
| 194 | |
| 195 | func isPayeeWord(t token.Type) bool { |
| 196 | switch t { |
| 197 | case token.TEXT, token.INT, token.DECIMAL, token.COMMODITYMARK: |
| 198 | return true |
| 199 | } |
| 200 | return false |
| 201 | } |
| 202 | |
| 203 | func (p *Parser) parsePeriodicTransaction() *ast.PeriodicTransaction { |
| 204 | s := p.cur.Span |
| 205 | p.expect(token.TILDE) |
| 206 | p.skipWhitespace() |
| 207 | |
| 208 | pt := &ast.PeriodicTransaction{} |
| 209 | |
| 210 | pt.Period = p.parsePeriod() |
| 211 | |
| 212 | if desc, dspan := p.parseOptPeriodicDescription(); desc != "" { |
| 213 | pt.Description = &ast.Description{Value: desc, Span: dspan} |
| 214 | } |
| 215 | |
| 216 | comment := p.parseOptInlineComment() |
| 217 | p.expectNewline() |
| 218 | |
| 219 | pt.HeaderComments, pt.Postings = p.parseHeaderCommentsAndPostings() |
| 220 | |
| 221 | pt.Span = p.span(s) |
| 222 | pt.Comment = comment |
| 223 | return pt |
| 224 | } |
| 225 | |
| 226 | func (p *Parser) parseAutomatedTransaction() *ast.AutomatedTransaction { |
| 227 | s := p.cur.Span |
| 228 | p.expect(token.EQ) |
| 229 | p.skipWhitespace() |
| 230 | |
| 231 | at := &ast.AutomatedTransaction{} |
| 232 | |
| 233 | // expression |
| 234 | sd := p.cur.Span |
| 235 | expr := p.parseDirectiveExpr() |
| 236 | at.Expr = ast.Expr{Value: expr, Span: p.span(sd)} |
| 237 | at.Comment = p.parseOptInlineComment() |
| 238 | p.expectNewline() |
| 239 | |
| 240 | at.HeaderComments, at.Postings = p.parseHeaderCommentsAndPostings() |
| 241 | |
| 242 | at.Span = p.span(s) |
| 243 | return at |
| 244 | } |
| 245 | |
| 246 | func (p *Parser) parseHeaderCommentsAndPostings() (comments []*ast.Comment, postings []ast.Posting) { |
| 247 | for p.got(token.INDENT) && p.willGet(token.SEMICOLON) { |
| 248 | p.advance() // consume indent |
| 249 | comments = append(comments, p.parseComment()) |
| 250 | } |
| 251 | |
| 252 | postings = make([]ast.Posting, 0, 2) // most transactions have 2 postings, small optimization |
| 253 | for p.got(token.INDENT) { |
| 254 | if posting, ok := p.parsePosting(); ok { |
| 255 | postings = append(postings, posting) |
| 256 | } |
| 257 | } |
| 258 | |
| 259 | return comments, postings |
| 260 | } |
| 261 | |
| 262 | func (p *Parser) parsePeriod() ast.Period { |
| 263 | s := p.cur.Span |
| 264 | |
| 265 | var periodBuf strings.Builder |
| 266 | |
| 267 | for !p.got(token.NEWLINE) && !p.got(token.EOF) && |
| 268 | !p.got(token.SEMICOLON) && !p.got(token.HASH) && !p.got(token.PERCENT) && !p.got(token.STAR) { |
| 269 | |
| 270 | if p.got(token.WHITESPACE) { |
| 271 | if len(p.cur.Literal) >= 2 { |
| 272 | break |
| 273 | } |
| 274 | if p.willGet(token.NEWLINE) || p.willGet(token.EOF) || |
| 275 | p.willGet(token.SEMICOLON) || p.willGet(token.HASH) || |
| 276 | p.willGet(token.PERCENT) || p.willGet(token.STAR) { |
| 277 | p.advance() |
| 278 | continue |
| 279 | } |
| 280 | } |
| 281 | |
| 282 | periodBuf.WriteString(p.cur.Literal) |
| 283 | p.advance() |
| 284 | } |
| 285 | |
| 286 | str := periodBuf.String() |
| 287 | period := ast.Period{Raw: str, Span: p.span(s)} |
| 288 | |
| 289 | if _, after, ok := strings.Cut(str, " from "); ok { |
| 290 | end := strings.Index(after, " ") |
| 291 | dateStr := after |
| 292 | if end >= 0 { |
| 293 | dateStr = after[:end] |
| 294 | } |
| 295 | if d := parseSimpleDate(dateStr); d.Year > 0 { |
| 296 | fromOff := strings.Index(str, dateStr) |
| 297 | d.Span = periodDateSpan(period, str, dateStr, fromOff) |
| 298 | period.From = &d |
| 299 | rest := after |
| 300 | if end >= 0 { |
| 301 | rest = after[end:] |
| 302 | } |
| 303 | if _, toAfter, ok := strings.Cut(rest, " to "); ok { |
| 304 | if toEnd := strings.Index(toAfter, " "); toEnd >= 0 { |
| 305 | toAfter = toAfter[:toEnd] |
| 306 | } |
| 307 | if d := parseSimpleDate(toAfter); d.Year > 0 { |
| 308 | d.Span = periodDateSpan(period, str, toAfter, fromOff+len(dateStr)) |
| 309 | period.To = &d |
| 310 | } |
| 311 | } |
| 312 | } |
| 313 | } |
| 314 | return period |
| 315 | } |
| 316 | |
| 317 | // periodDateSpan returns the source span of dateStr, which occurs in the |
| 318 | // period text at or after searchFrom. The period span and text cover the same |
| 319 | // bytes, so offsets line up 1:1. |
| 320 | func periodDateSpan(period ast.Period, text, dateStr string, searchFrom int) token.Span { |
| 321 | off := strings.Index(text[searchFrom:], dateStr) |
| 322 | abs := period.Span.Start.Offset + searchFrom + off |
| 323 | return token.Span{ |
| 324 | File: period.Span.File, |
| 325 | Start: token.Pos{Offset: abs}, |
| 326 | End: token.Pos{Offset: abs + len(dateStr)}, |
| 327 | } |
| 328 | } |
| 329 | |
| 330 | func (p *Parser) parseComment() *ast.Comment { |
| 331 | s := p.cur.Span |
| 332 | c := p.parseCommentRest(s) |
| 333 | p.expectNewline() |
| 334 | c.Span = p.span(s) // comment spans its line through the newline |
| 335 | return c |
| 336 | } |
| 337 | |
| 338 | func (p *Parser) parseAccountDirective() *ast.AccountDirective { |
| 339 | s := p.cur.Span |
| 340 | p.expect(token.ACCOUNT) |
| 341 | p.skipWhitespace() |
| 342 | |
| 343 | account := p.parseAccount() |
| 344 | comment := p.parseOptInlineComment() |
| 345 | p.expectNewline() |
| 346 | |
| 347 | var subs []ast.AccountSubdirective |
| 348 | for p.got(token.INDENT) { |
| 349 | p.advance() |
| 350 | p.skipWhitespace() |
| 351 | if p.got(token.NEWLINE) || p.got(token.EOF) { |
| 352 | // whitespace-only line: block continues |
| 353 | p.expectNewline() |
| 354 | continue |
| 355 | } |
| 356 | switch { |
| 357 | case p.got(token.SEMICOLON): |
| 358 | // comment line: directive mode lexes only ';' as a comment marker |
| 359 | ns := p.cur.Span |
| 360 | c := p.parseCommentRest(ns) |
| 361 | p.expectNewline() |
| 362 | subs = append(subs, ast.AccountSubdirective{Kind: ast.SubdirectiveComment, NameSpan: ns, Comment: c}) |
| 363 | case p.got(token.TEXT) && isCommentMarker(p.cur.Literal): |
| 364 | // '#', '%' and '*' lex as TEXT in directive mode; treat them as comment lines |
| 365 | c := p.parseTextComment() |
| 366 | subs = append(subs, ast.AccountSubdirective{Kind: ast.SubdirectiveComment, NameSpan: c.Span, Comment: c}) |
| 367 | case p.got(token.TEXT): |
| 368 | name := p.cur.Literal |
| 369 | kind, ok := accountSubdirectiveKind(name) |
| 370 | if !ok { |
| 371 | p.errorf("unknown subdirective %q", name) |
| 372 | p.skipToNewline() |
| 373 | continue |
| 374 | } |
| 375 | kw := p.cur.Span |
| 376 | p.advance() |
| 377 | value, vspan := p.parseSubdirectiveValue() |
| 378 | if value == "" { |
| 379 | p.errorf("expected value for subdirective %q", name) |
| 380 | } |
| 381 | c := p.parseOptInlineComment() |
| 382 | p.expectNewline() |
| 383 | subs = append(subs, ast.AccountSubdirective{ |
| 384 | Kind: kind, |
| 385 | NameSpan: kw, |
| 386 | Value: value, |
| 387 | ValueSpan: vspan, |
| 388 | Comment: c, |
| 389 | }) |
| 390 | default: |
| 391 | p.errorf("expected subdirective name, got %s", p.cur.Type) |
| 392 | p.skipToNewline() |
| 393 | } |
| 394 | } |
| 395 | |
| 396 | return &ast.AccountDirective{ |
| 397 | Account: account, |
| 398 | Subdirectives: subs, |
| 399 | Comment: comment, |
| 400 | Span: p.span(s), |
| 401 | } |
| 402 | } |
| 403 | |
| 404 | func (p *Parser) parseCommodityDirective() *ast.CommodityDirective { |
| 405 | s := p.cur.Span |
| 406 | p.expect(token.COMMODITY) |
| 407 | p.skipWhitespace() |
| 408 | |
| 409 | var commodity string |
| 410 | var commoditySpan token.Span |
| 411 | var format *ast.FormatSubDirective |
| 412 | |
| 413 | switch p.cur.Type { |
| 414 | case token.COMMODITYMARK, token.TEXT, token.STRING: |
| 415 | cs := p.cur.Span |
| 416 | commodity = unquote(p.cur.Literal) |
| 417 | p.advance() |
| 418 | commoditySpan = token.Span{File: cs.File, Start: cs.Start, End: p.cur.Span.Start} |
| 419 | hadSpace := p.got(token.WHITESPACE) |
| 420 | p.skipWhitespace() |
| 421 | if p.got(token.INT) || p.got(token.DECIMAL) || p.got(token.TEXT) { |
| 422 | amt := p.parseAmount() |
| 423 | amt.Commodity = commodity |
| 424 | amt.CommoditySpan = commoditySpan |
| 425 | amt.CommodityPos = ast.CommodityBefore |
| 426 | amt.HasSpace = hadSpace |
| 427 | format = &ast.FormatSubDirective{Amount: amt} |
| 428 | } |
| 429 | case token.INT, token.DECIMAL: |
| 430 | amt := p.parseAmount() |
| 431 | commodity = amt.Commodity |
| 432 | commoditySpan = amt.CommoditySpan |
| 433 | format = &ast.FormatSubDirective{Amount: amt} |
| 434 | default: |
| 435 | p.errorf("expected commodity name or amount, got %s", p.cur.Type) |
| 436 | } |
| 437 | |
| 438 | if commodity == "" { |
| 439 | p.errorf("expected commodity name, got %s", p.cur.Type) |
| 440 | } |
| 441 | |
| 442 | // hledger parity: an inline format amount must include a decimal mark |
| 443 | if format != nil && format.Amount.QuantityFmt.Decimal == 0 { |
| 444 | p.errorfAt(format.Amount.Span, "Please include a decimal point or decimal comma in commodity directives, to help us parse correctly. It may be followed by zero or more decimal digits.") |
| 445 | } |
| 446 | |
| 447 | comment := p.parseOptInlineComment() |
| 448 | p.expectNewline() |
| 449 | |
| 450 | var blockComments []*ast.Comment |
| 451 | for p.got(token.INDENT) { |
| 452 | p.advance() |
| 453 | p.skipWhitespace() |
| 454 | if p.got(token.NEWLINE) || p.got(token.EOF) { |
| 455 | // whitespace-only line: block continues |
| 456 | p.expectNewline() |
| 457 | continue |
| 458 | } |
| 459 | switch { |
| 460 | case p.got(token.TEXT) && p.cur.Literal == "format": |
| 461 | kw := p.cur.Span |
| 462 | p.advance() |
| 463 | p.skipWhitespace() |
| 464 | amt := p.parseAmount() |
| 465 | // hledger parity: the format symbol must match the declared commodity, |
| 466 | // and the amount must include a decimal mark; the node is kept either |
| 467 | // way so the printer can round-trip the input. |
| 468 | if amt.Commodity != commodity { |
| 469 | p.errorfAt(amt.Span, "commodity directive symbol %q and format directive symbol %q should be the same", commodity, amt.Commodity) |
| 470 | } else if amt.QuantityFmt.Decimal == 0 { |
| 471 | p.errorfAt(amt.Span, "Please include a decimal point or decimal comma in commodity directives, to help us parse correctly. It may be followed by zero or more decimal digits.") |
| 472 | } |
| 473 | c := p.parseOptInlineComment() |
| 474 | p.expectNewline() |
| 475 | format = &ast.FormatSubDirective{KeywordSpan: kw, Amount: amt, Comment: c} |
| 476 | case p.got(token.SEMICOLON): // comment line |
| 477 | c := p.parseCommentRest(p.cur.Span) |
| 478 | p.expectNewline() |
| 479 | blockComments = append(blockComments, c) |
| 480 | case p.got(token.TEXT) && isCommentMarker(p.cur.Literal): |
| 481 | // '#', '%' and '*' lex as TEXT in directive mode; treat them as comment lines |
| 482 | blockComments = append(blockComments, p.parseTextComment()) |
| 483 | case p.got(token.TEXT): |
| 484 | p.errorf("unknown subdirective %q", p.cur.Literal) |
| 485 | p.skipToNewline() |
| 486 | default: |
| 487 | p.errorf("expected subdirective name, got %s", p.cur.Type) |
| 488 | p.skipToNewline() |
| 489 | } |
| 490 | } |
| 491 | |
| 492 | cd := &ast.CommodityDirective{ |
| 493 | Commodity: commodity, |
| 494 | CommoditySpan: commoditySpan, |
| 495 | FormatSub: format, |
| 496 | BlockComments: blockComments, |
| 497 | Comment: comment, |
| 498 | Span: p.span(s), |
| 499 | } |
| 500 | return cd |
| 501 | } |
| 502 | |
| 503 | func (p *Parser) parseIncludeDirective() *ast.IncludeDirective { |
| 504 | s := p.cur.Span |
| 505 | p.expect(token.INCLUDE) |
| 506 | p.skipWhitespace() |
| 507 | |
| 508 | id := &ast.IncludeDirective{} |
| 509 | |
| 510 | if p.got(token.TEXT) { |
| 511 | id.Path = p.cur.Literal |
| 512 | p.advance() |
| 513 | } else { |
| 514 | p.errorf("expected file path, got %s", p.cur.Type) |
| 515 | } |
| 516 | |
| 517 | id.Comment = p.parseOptInlineComment() |
| 518 | p.expectNewline() |
| 519 | id.Span = p.span(s) |
| 520 | return id |
| 521 | } |
| 522 | |
| 523 | func (p *Parser) parseAliasDirective() *ast.AliasDirective { |
| 524 | s := p.cur.Span |
| 525 | alias := &ast.AliasDirective{} |
| 526 | p.expect(token.ALIAS) |
| 527 | p.skipWhitespace() |
| 528 | alias.From = p.parseAccount() |
| 529 | p.skipWhitespace() |
| 530 | p.expect(token.EQ) |
| 531 | p.skipWhitespace() |
| 532 | alias.To = p.parseAccount() |
| 533 | alias.Comment = p.parseOptInlineComment() |
| 534 | p.expectNewline() |
| 535 | alias.Span = p.span(s) |
| 536 | return alias |
| 537 | } |
| 538 | |
| 539 | func (p *Parser) parsePayeeDirective() *ast.PayeeDirective { |
| 540 | s := p.cur.Span |
| 541 | p.expect(token.PAYEE) |
| 542 | p.skipWhitespace() |
| 543 | |
| 544 | var name *ast.Payee |
| 545 | if p.got(token.TEXT) || p.got(token.STRING) || p.got(token.COMMODITYMARK) { |
| 546 | name = p.parsePayee() |
| 547 | } |
| 548 | |
| 549 | comment := p.parseOptInlineComment() |
| 550 | p.expectNewline() |
| 551 | |
| 552 | return &ast.PayeeDirective{ |
| 553 | Name: name, |
| 554 | Comment: comment, |
| 555 | Span: p.span(s), |
| 556 | } |
| 557 | } |
| 558 | |
| 559 | func (p *Parser) parseTagDirective() *ast.TagDirective { |
| 560 | s := p.cur.Span |
| 561 | p.expect(token.TAG) |
| 562 | p.skipWhitespace() |
| 563 | |
| 564 | name := "" |
| 565 | if p.got(token.TEXT) || p.got(token.COMMODITYMARK) || p.got(token.STRING) { |
| 566 | name = unquote(p.cur.Literal) |
| 567 | p.advance() |
| 568 | } |
| 569 | |
| 570 | comment := p.parseOptInlineComment() |
| 571 | p.expectNewline() |
| 572 | |
| 573 | return &ast.TagDirective{ |
| 574 | Name: name, |
| 575 | Comment: comment, |
| 576 | Span: p.span(s), |
| 577 | } |
| 578 | } |
| 579 | |
| 580 | func (p *Parser) parseYearDirective() *ast.YearDirective { |
| 581 | s := p.cur.Span |
| 582 | year := &ast.YearDirective{} |
| 583 | p.expect(token.YEAR) |
| 584 | p.skipWhitespace() |
| 585 | |
| 586 | if p.got(token.INT) { |
| 587 | year.Year, _ = strconv.Atoi(p.cur.Literal) |
| 588 | p.defaultYear = year.Year |
| 589 | p.advance() |
| 590 | } else { |
| 591 | p.errorf("expected year, got %s", p.cur.Type) |
| 592 | } |
| 593 | |
| 594 | year.Comment = p.parseOptInlineComment() |
| 595 | p.expectNewline() |
| 596 | year.Span = p.span(s) |
| 597 | |
| 598 | return year |
| 599 | } |
| 600 | |
| 601 | func (p *Parser) parseDecimalMarkDirective() *ast.DecimalMarkDirective { |
| 602 | s := p.cur.Span |
| 603 | mark := &ast.DecimalMarkDirective{} |
| 604 | p.expect(token.DECIMALMARK) |
| 605 | p.skipWhitespace() |
| 606 | |
| 607 | mark.Mark = byte('.') |
| 608 | if p.got(token.TEXT) { |
| 609 | if len(p.cur.Literal) > 0 { |
| 610 | mark.Mark = p.cur.Literal[0] |
| 611 | } |
| 612 | p.advance() |
| 613 | } |
| 614 | |
| 615 | mark.Comment = p.parseOptInlineComment() |
| 616 | p.expectNewline() |
| 617 | mark.Span = p.span(s) |
| 618 | return mark |
| 619 | } |
| 620 | |
| 621 | func (p *Parser) parseDefaultCommodityDirective() *ast.DefaultCommodityDirective { |
| 622 | s := p.cur.Span |
| 623 | com := &ast.DefaultCommodityDirective{} |
| 624 | p.expect(token.D) |
| 625 | p.skipWhitespace() |
| 626 | com.Amount = p.parseAmount() |
| 627 | com.Comment = p.parseOptInlineComment() |
| 628 | p.expectNewline() |
| 629 | com.Span = p.span(s) |
| 630 | return com |
| 631 | } |
| 632 | |
| 633 | func (p *Parser) parseConversionDirective() *ast.ConversionDirective { |
| 634 | s := p.cur.Span |
| 635 | cd := &ast.ConversionDirective{} |
| 636 | p.expect(token.C) |
| 637 | p.skipWhitespace() |
| 638 | |
| 639 | if p.isAmountStart() { |
| 640 | cd.From = p.parseAmount() |
| 641 | } else { |
| 642 | p.errorf("expected amount, got %s", p.cur.Type) |
| 643 | } |
| 644 | |
| 645 | p.skipWhitespace() |
| 646 | if p.got(token.EQ) { |
| 647 | p.advance() |
| 648 | p.skipWhitespace() |
| 649 | if p.isAmountStart() { |
| 650 | cd.To = p.parseAmount() |
| 651 | } else { |
| 652 | p.errorf("expected amount, got %s", p.cur.Type) |
| 653 | } |
| 654 | } |
| 655 | |
| 656 | cd.Comment = p.parseOptInlineComment() |
| 657 | p.expectNewline() |
| 658 | cd.Span = p.span(s) |
| 659 | return cd |
| 660 | } |
| 661 | |
| 662 | func (p *Parser) parseIgnoredDirective() *ast.IgnoredDirective { |
| 663 | s := p.cur.Span |
| 664 | p.expect(token.N) |
| 665 | p.skipWhitespace() |
| 666 | |
| 667 | id := &ast.IgnoredDirective{} |
| 668 | if p.got(token.TEXT) || p.got(token.COMMODITYMARK) || p.got(token.STRING) { |
| 669 | id.Text = unquote(p.cur.Literal) |
| 670 | p.advance() |
| 671 | } |
| 672 | id.Comment = p.parseOptInlineComment() |
| 673 | |
| 674 | p.expectNewline() |
| 675 | id.Span = p.span(s) |
| 676 | return id |
| 677 | } |
| 678 | |
| 679 | func (p *Parser) parseMarketPriceDirective() *ast.MarketPriceDirective { |
| 680 | s := p.cur.Span |
| 681 | p.expect(token.P) |
| 682 | p.skipWhitespace() |
| 683 | |
| 684 | mp := &ast.MarketPriceDirective{} |
| 685 | mp.DateTime.Date = p.parseDate() |
| 686 | p.skipWhitespace() |
| 687 | |
| 688 | if p.got(token.TIME) { |
| 689 | mp.DateTime.Time = new(p.parseTime()) |
| 690 | p.skipWhitespace() |
| 691 | } |
| 692 | |
| 693 | if p.got(token.COMMODITYMARK) || p.got(token.STRING) { |
| 694 | mp.Commodity = unquote(p.cur.Literal) |
| 695 | p.advance() |
| 696 | } else { |
| 697 | p.errorf("expected commodity symbol, got %s", p.cur.Type) |
| 698 | } |
| 699 | p.skipWhitespace() |
| 700 | |
| 701 | mp.Amount = p.parseAmount() |
| 702 | |
| 703 | mp.Comment = p.parseOptInlineComment() |
| 704 | |
| 705 | p.expectNewline() |
| 706 | mp.Span = p.span(s) |
| 707 | return mp |
| 708 | } |
| 709 | |
| 710 | func (p *Parser) parseTime() ast.Time { |
| 711 | s := p.cur.Span |
| 712 | tok, _ := p.expect(token.TIME) |
| 713 | lit := tok.Literal |
| 714 | |
| 715 | parts := strings.Split(lit, ":") |
| 716 | if len(parts) < 2 { |
| 717 | p.errorf("invalid time format: %q", lit) |
| 718 | return ast.Time{Span: p.span(s)} |
| 719 | } |
| 720 | |
| 721 | hour, _ := strconv.Atoi(parts[0]) |
| 722 | minute, _ := strconv.Atoi(parts[1]) |
| 723 | second := 0 |
| 724 | if len(parts) > 2 { |
| 725 | second, _ = strconv.Atoi(parts[2]) |
| 726 | } |
| 727 | |
| 728 | if hour < 0 || hour > 23 { |
| 729 | p.errorf("invalid hour %d in time %q", hour, lit) |
| 730 | } |
| 731 | if minute < 0 || minute > 59 { |
| 732 | p.errorf("invalid minute %d in time %q", minute, lit) |
| 733 | } |
| 734 | if second < 0 || second > 59 { |
| 735 | p.errorf("invalid second %d in time %q", second, lit) |
| 736 | } |
| 737 | |
| 738 | return ast.Time{ |
| 739 | Hour: hour, |
| 740 | Minute: minute, |
| 741 | Second: second, |
| 742 | Span: p.span(s), |
| 743 | } |
| 744 | } |
| 745 | |
| 746 | func (p *Parser) parseApplyDirective() *ast.ApplyDirective { |
| 747 | s := p.cur.Span |
| 748 | p.expect(token.APPLY) |
| 749 | p.skipWhitespace() |
| 750 | |
| 751 | expr := p.parseDirectiveExpr() |
| 752 | comment := p.parseOptInlineComment() |
| 753 | p.expectNewline() |
| 754 | |
| 755 | return &ast.ApplyDirective{ |
| 756 | Expr: expr, |
| 757 | Comment: comment, |
| 758 | Span: p.span(s), |
| 759 | } |
| 760 | } |
| 761 | |
| 762 | func (p *Parser) parseEndDirective() *ast.EndDirective { |
| 763 | s := p.cur.Span |
| 764 | p.expect(token.END) |
| 765 | p.skipWhitespace() |
| 766 | |
| 767 | expr := p.parseDirectiveExpr() |
| 768 | comment := p.parseOptInlineComment() |
| 769 | p.expectNewline() |
| 770 | |
| 771 | return &ast.EndDirective{ |
| 772 | Expr: expr, |
| 773 | Comment: comment, |
| 774 | Span: p.span(s), |
| 775 | } |
| 776 | } |
| 777 | |
| 778 | func (p *Parser) parseCommentBlockDirective() *ast.CommentBlockDirective { |
| 779 | start := p.cur.Span |
| 780 | p.expect(token.COMMENTKW) |
| 781 | p.skipWhitespace() |
| 782 | |
| 783 | header := p.parseDirectiveExpr() |
| 784 | comment := p.parseOptInlineComment() |
| 785 | p.expectNewline() |
| 786 | |
| 787 | var content strings.Builder |
| 788 | for p.cur.Type != token.EOF { |
| 789 | if p.got(token.END) { |
| 790 | if p.willGet(token.NEWLINE) || p.willGet(token.EOF) { |
| 791 | p.advance() |
| 792 | p.expectNewline() |
| 793 | break |
| 794 | } |
| 795 | if p.willGet(token.WHITESPACE) { |
| 796 | endTok := p.cur |
| 797 | p.advance() |
| 798 | wsTok := p.cur |
| 799 | p.advance() |
| 800 | if p.got(token.TEXT) && p.cur.Literal == "comment" { // todo: this should check if it's an actual COMMENTKW token |
| 801 | p.advance() |
| 802 | p.parseDirectiveExpr() |
| 803 | p.parseOptInlineComment() |
| 804 | p.expectNewline() |
| 805 | break |
| 806 | } |
| 807 | content.WriteString(endTok.Literal) |
| 808 | content.WriteString(wsTok.Literal) |
| 809 | continue |
| 810 | } |
| 811 | } |
| 812 | content.WriteString(p.cur.Literal) |
| 813 | p.advance() |
| 814 | } |
| 815 | |
| 816 | return &ast.CommentBlockDirective{ |
| 817 | Header: header, |
| 818 | Content: content.String(), |
| 819 | Comment: comment, |
| 820 | Span: p.span(start), |
| 821 | } |
| 822 | } |
| 823 | |
| 824 | func (p *Parser) parseStatus() ast.Status { |
| 825 | s := p.cur.Span |
| 826 | st := ast.Status{} |
| 827 | switch p.cur.Type { |
| 828 | case token.STAR: |
| 829 | st.Value = ast.StatusCleared |
| 830 | case token.BANG: |
| 831 | st.Value = ast.StatusPending |
| 832 | } |
| 833 | if st.Value != ast.StatusNone { |
| 834 | p.advance() |
| 835 | p.skipWhitespace() |
| 836 | } |
| 837 | st.Span = p.span(s) |
| 838 | return st |
| 839 | } |
| 840 | |
| 841 | func (p *Parser) isAmountStart() bool { |
| 842 | switch p.cur.Type { |
| 843 | default: |
| 844 | return false |
| 845 | case token.COMMODITYMARK, token.STRING, token.INT, token.DECIMAL, token.MINUS, token.PLUS, token.PARENEXPR, token.STAR: |
| 846 | return true |
| 847 | } |
| 848 | } |
| 849 | |
| 850 | func (p *Parser) parseAmount() ast.Amount { |
| 851 | s := p.cur.Span |
| 852 | amt := ast.Amount{QuantityFmt: ast.QuantityFormat{}} |
| 853 | |
| 854 | p.parseAmountSign(&amt) |
| 855 | p.skipWhitespace() |
| 856 | |
| 857 | // commodity before quantity: $10.00, eur 10.00 |
| 858 | if p.got(token.COMMODITYMARK) || p.got(token.TEXT) || p.got(token.STRING) { |
| 859 | cs := p.cur.Span |
| 860 | amt.Commodity = unquote(p.cur.Literal) |
| 861 | amt.CommodityPos = ast.CommodityBefore |
| 862 | p.advance() |
| 863 | amt.CommoditySpan = token.Span{File: cs.File, Start: cs.Start, End: p.cur.Span.Start} |
| 864 | if p.got(token.WHITESPACE) { |
| 865 | amt.HasSpace = true |
| 866 | p.skipWhitespace() |
| 867 | } |
| 868 | } |
| 869 | |
| 870 | // optional sign after commodity: $ -10 |
| 871 | p.parseAmountSign(&amt) |
| 872 | p.skipWhitespace() |
| 873 | |
| 874 | p.parseQuantityInto(&amt) |
| 875 | |
| 876 | // commodity after quantity: 10.00 UAH, 10.00 "EUR" (only if not set) |
| 877 | if amt.Commodity == "" { |
| 878 | switch p.cur.Type { |
| 879 | case token.WHITESPACE: |
| 880 | p.skipWhitespace() |
| 881 | if p.got(token.COMMODITYMARK) || p.got(token.TEXT) || p.got(token.STRING) { |
| 882 | cs := p.cur.Span |
| 883 | amt.HasSpace = true |
| 884 | amt.Commodity = unquote(p.cur.Literal) |
| 885 | amt.CommodityPos = ast.CommodityAfter |
| 886 | p.advance() |
| 887 | amt.CommoditySpan = token.Span{File: cs.File, Start: cs.Start, End: p.cur.Span.Start} |
| 888 | } |
| 889 | case token.COMMODITYMARK, token.TEXT, token.STRING: |
| 890 | cs := p.cur.Span |
| 891 | amt.Commodity = unquote(p.cur.Literal) |
| 892 | amt.CommodityPos = ast.CommodityAfter |
| 893 | p.advance() |
| 894 | amt.CommoditySpan = token.Span{File: cs.File, Start: cs.Start, End: p.cur.Span.Start} |
| 895 | } |
| 896 | } |
| 897 | |
| 898 | amt.Span = p.span(s) |
| 899 | return amt |
| 900 | } |
| 901 | |
| 902 | // parseAmountSign consumes an optional leading +/- into IsNegative. |
| 903 | func (p *Parser) parseAmountSign(amt *ast.Amount) { |
| 904 | switch p.cur.Type { |
| 905 | case token.MINUS: |
| 906 | amt.IsNegative = true |
| 907 | p.advance() |
| 908 | case token.PLUS: |
| 909 | p.advance() |
| 910 | } |
| 911 | } |
| 912 | |
| 913 | func (p *Parser) parseAmountWithOptExpr() ast.Amount { |
| 914 | if p.got(token.STAR) { |
| 915 | p.advance() |
| 916 | p.skipWhitespace() |
| 917 | amt := p.parseAmount() |
| 918 | amt.IsExpr = true |
| 919 | return amt |
| 920 | } |
| 921 | if p.got(token.PARENEXPR) { |
| 922 | lit := p.cur.Literal |
| 923 | amt := ast.Amount{ |
| 924 | IsExpr: true, |
| 925 | QuantityFmt: ast.QuantityFormat{}, |
| 926 | } |
| 927 | if len(lit) >= 2 && lit[0] == '(' && lit[len(lit)-1] == ')' { |
| 928 | amt.Expr = strings.Trim(lit[1:len(lit)-1], " \t") |
| 929 | } |
| 930 | amt.Span = p.cur.Span |
| 931 | p.advance() |
| 932 | return amt |
| 933 | } |
| 934 | return p.parseAmount() |
| 935 | } |
| 936 | |
| 937 | func (p *Parser) parsePosting() (ast.Posting, bool) { |
| 938 | s := p.cur.Span |
| 939 | posting := ast.Posting{} |
| 940 | p.expect(token.INDENT) |
| 941 | |
| 942 | // exit if it's empty line |
| 943 | if p.got(token.NEWLINE) || p.got(token.EOF) { |
| 944 | p.syncToNextline() |
| 945 | return ast.Posting{}, false |
| 946 | } |
| 947 | |
| 948 | // optional status, outside of brackets, '! (account)' |
| 949 | posting.Status = p.parseStatus() |
| 950 | |
| 951 | // detect virtual posting brackets |
| 952 | switch p.cur.Type { |
| 953 | case token.LPAREN: |
| 954 | posting.Type = ast.PostingVirtualUnbalanced |
| 955 | p.advance() |
| 956 | case token.LBRACKET: |
| 957 | posting.Type = ast.PostingVirtualBalanced |
| 958 | p.advance() |
| 959 | } |
| 960 | |
| 961 | // optional status, inside of brackets, '(* account)' |
| 962 | if p.got(token.STAR) || p.got(token.BANG) { |
| 963 | posting.Status = p.parseStatus() |
| 964 | } |
| 965 | |
| 966 | // validate, must be account text |
| 967 | if p.cur.Type != token.TEXT { |
| 968 | p.errorf("expected account name, got %s", p.cur.Type) |
| 969 | p.syncToNextline() |
| 970 | return ast.Posting{}, false |
| 971 | } |
| 972 | |
| 973 | posting.Account = p.parseAccount() |
| 974 | |
| 975 | // consume closing bracket |
| 976 | switch p.cur.Type { |
| 977 | case token.RPAREN: |
| 978 | p.advance() |
| 979 | case token.RBRACKET: |
| 980 | p.advance() |
| 981 | } |
| 982 | |
| 983 | // optional amount - after two spaces |
| 984 | if p.got(token.WHITESPACE) { |
| 985 | p.skipWhitespace() |
| 986 | if p.isAmountStart() { |
| 987 | amt := p.parseAmountWithOptExpr() |
| 988 | posting.Amount = &amt |
| 989 | } |
| 990 | } |
| 991 | |
| 992 | // optional cost '@' or '@@' |
| 993 | p.skipWhitespace() |
| 994 | if p.got(token.AT) || p.got(token.ATAT) { |
| 995 | posting.Cost = p.parseCost() |
| 996 | } |
| 997 | |
| 998 | // optional balance assertion or assignment |
| 999 | p.skipWhitespace() |
| 1000 | if p.got(token.COLON) && p.willGet(token.EQ) { |
| 1001 | p.advance() // consume ':' of ':=' |
| 1002 | posting.Balance = p.parseBalanceAssertion() |
| 1003 | posting.Balance.IsAssignment = true |
| 1004 | } else if p.got(token.EQ) || p.got(token.EQEQ) || p.got(token.EQEQEQ) || p.got(token.EQSTAR) { |
| 1005 | posting.Balance = p.parseBalanceAssertion() |
| 1006 | } |
| 1007 | |
| 1008 | posting.Comment = p.parseOptInlineComment() |
| 1009 | p.expectNewline() |
| 1010 | |
| 1011 | // continuation comments |
| 1012 | for p.got(token.INDENT) && p.willGet(token.SEMICOLON) { |
| 1013 | p.advance() |
| 1014 | c := p.parseComment() |
| 1015 | posting.Comments = append(posting.Comments, *c) |
| 1016 | } |
| 1017 | |
| 1018 | posting.Span = p.span(s) |
| 1019 | return posting, true |
| 1020 | } |
| 1021 | |
| 1022 | func (p *Parser) parseCost() *ast.Cost { |
| 1023 | s := p.cur.Span |
| 1024 | isTotal := p.got(token.ATAT) |
| 1025 | p.advance() // consume '@' '@@' |
| 1026 | p.skipWhitespace() |
| 1027 | return &ast.Cost{ |
| 1028 | IsTotal: isTotal, |
| 1029 | Amount: p.parseAmount(), |
| 1030 | Span: p.span(s), |
| 1031 | } |
| 1032 | } |
| 1033 | |
| 1034 | func (p *Parser) parseBalanceAssertion() *ast.BalanceAssertion { |
| 1035 | s := p.cur.Span |
| 1036 | |
| 1037 | ba := &ast.BalanceAssertion{} |
| 1038 | switch p.cur.Type { |
| 1039 | case token.EQ: // basic assertion |
| 1040 | case token.EQSTAR: // inclusive assertion |
| 1041 | ba.IsInclusive = true |
| 1042 | case token.EQEQ: // strict assertion |
| 1043 | ba.IsStrict = true |
| 1044 | case token.EQEQEQ: // strict inclusive assertion |
| 1045 | ba.IsStrict = true |
| 1046 | ba.IsInclusive = true |
| 1047 | } |
| 1048 | p.advance() |
| 1049 | p.skipWhitespace() |
| 1050 | |
| 1051 | ba.Amount = p.parseAmount() |
| 1052 | p.skipWhitespace() |
| 1053 | if p.got(token.AT) || p.got(token.ATAT) { |
| 1054 | c := p.parseCost() |
| 1055 | ba.Cost = c |
| 1056 | } |
| 1057 | ba.Span = p.span(s) |
| 1058 | return ba |
| 1059 | } |
| 1060 | |
| 1061 | func (p *Parser) readAccountSegment() (ast.SubAccount, bool) { |
| 1062 | switch p.cur.Type { |
| 1063 | case token.TEXT: |
| 1064 | sub := ast.SubAccount{Name: p.cur.Literal, Span: p.cur.Span} |
| 1065 | p.advance() |
| 1066 | |
| 1067 | // handle multi work segment, e.g: "credit card" |
| 1068 | if p.got(token.WHITESPACE) && p.willGet(token.TEXT) && len(p.peek.Literal) > 0 && p.peek.Literal[0] != '(' { |
| 1069 | sub.Name += " " |
| 1070 | p.advance() |
| 1071 | sub.Name += p.cur.Literal |
| 1072 | p.advance() |
| 1073 | } |
| 1074 | return sub, true |
| 1075 | |
| 1076 | case token.COMMODITYMARK: |
| 1077 | sub := ast.SubAccount{Name: p.cur.Literal, Span: p.cur.Span} |
| 1078 | p.advance() |
| 1079 | // merge "EUR" + "-HRK" to "EUR-HRK" |
| 1080 | for p.got(token.TEXT) { |
| 1081 | sub.Name += p.cur.Literal |
| 1082 | p.advance() |
| 1083 | } |
| 1084 | return sub, true |
| 1085 | |
| 1086 | default: |
| 1087 | return ast.SubAccount{}, false |
| 1088 | } |
| 1089 | } |
| 1090 | |
| 1091 | func (p *Parser) parseAccount() ast.Account { |
| 1092 | s := p.cur.Span |
| 1093 | acc := ast.Account{Name: make([]ast.SubAccount, 0, 6)} |
| 1094 | |
| 1095 | sub, ok := p.readAccountSegment() |
| 1096 | if !ok { |
| 1097 | p.errorf("expected account, got %s", p.cur.Type) |
| 1098 | return ast.Account{} |
| 1099 | } |
| 1100 | acc.Name = append(acc.Name, sub) |
| 1101 | |
| 1102 | for p.got(token.COLON) { |
| 1103 | p.advance() |
| 1104 | sub, ok := p.readAccountSegment() |
| 1105 | if !ok { |
| 1106 | break |
| 1107 | } |
| 1108 | acc.Name = append(acc.Name, sub) |
| 1109 | } |
| 1110 | |
| 1111 | acc.Span = p.span(s) |
| 1112 | return acc |
| 1113 | } |
| 1114 | |
| 1115 | func (p *Parser) parseDate() ast.Date { |
| 1116 | s := p.cur.Span |
| 1117 | tok, ok := p.expect(token.DATE) |
| 1118 | if !ok { |
| 1119 | return ast.Date{Span: p.span(s)} |
| 1120 | } |
| 1121 | |
| 1122 | year, month, day, sep, err := ParseDateLiteral(tok.Literal) |
| 1123 | if err != nil { |
| 1124 | p.errorf("%v", err) |
| 1125 | return ast.Date{Span: p.span(s)} |
| 1126 | } |
| 1127 | if year == 0 { |
| 1128 | year = p.defaultYear |
| 1129 | } |
| 1130 | |
| 1131 | return ast.Date{Year: year, Month: month, Day: day, Sep: sep, Span: p.span(s)} |
| 1132 | } |
| 1133 | |
| 1134 | func (p *Parser) parseOptInlineComment() *ast.Comment { |
| 1135 | p.skipWhitespace() |
| 1136 | if !p.got(token.SEMICOLON) { |
| 1137 | return nil |
| 1138 | } |
| 1139 | return p.parseCommentRest(p.cur.Span) |
| 1140 | } |
| 1141 | |
| 1142 | // parseCommentRest consumes a comment marker at p.cur, then optional text; |
| 1143 | // s anchors the span at the marker's start. |
| 1144 | func (p *Parser) parseCommentRest(s token.Span) *ast.Comment { |
| 1145 | marker := p.cur.Literal[0] |
| 1146 | p.advance() |
| 1147 | p.skipWhitespace() |
| 1148 | |
| 1149 | var tags []ast.Tag |
| 1150 | text := "" |
| 1151 | if p.got(token.TEXT) { |
| 1152 | text = p.cur.Literal |
| 1153 | tags = parseCommentTags(text, p.cur.Span) |
| 1154 | p.advance() |
| 1155 | } |
| 1156 | |
| 1157 | return &ast.Comment{ |
| 1158 | Marker: marker, |
| 1159 | Tags: tags, |
| 1160 | Text: text, |
| 1161 | Span: p.span(s), |
| 1162 | } |
| 1163 | } |
| 1164 | |
| 1165 | func (p *Parser) parseOptPeriodicDescription() (string, token.Span) { |
| 1166 | if p.cur.Type != token.WHITESPACE || len(p.cur.Literal) < 2 { |
| 1167 | return "", token.Span{} |
| 1168 | } |
| 1169 | |
| 1170 | p.skipWhitespace() |
| 1171 | |
| 1172 | if p.cur.Type != token.TEXT { |
| 1173 | return "", token.Span{} |
| 1174 | } |
| 1175 | |
| 1176 | s := p.cur.Span |
| 1177 | desc := p.parseDescription() |
| 1178 | return desc, p.span(s) |
| 1179 | } |
| 1180 | |
| 1181 | func (p *Parser) parseDescription() string { |
| 1182 | var desc strings.Builder |
| 1183 | for p.got(token.TEXT) || (p.got(token.WHITESPACE) && p.willGet(token.TEXT)) { |
| 1184 | _, _ = desc.WriteString(p.cur.Literal) |
| 1185 | p.advance() |
| 1186 | } |
| 1187 | return desc.String() |
| 1188 | } |
| 1189 | |
| 1190 | func (p *Parser) parseDirectiveExpr() string { |
| 1191 | var b strings.Builder |
| 1192 | for p.cur.Type != token.NEWLINE && p.cur.Type != token.EOF && p.cur.Type != token.SEMICOLON { |
| 1193 | _, _ = b.WriteString(p.cur.Literal) |
| 1194 | p.advance() |
| 1195 | } |
| 1196 | return b.String() |
| 1197 | } |
| 1198 | |
| 1199 | func (p *Parser) parseQuantityInto(amt *ast.Amount) { |
| 1200 | if p.cur.Type != token.INT && p.cur.Type != token.DECIMAL && p.cur.Type != token.TEXT { |
| 1201 | p.errorf("expected quantity, got %s", p.cur.Type) |
| 1202 | return |
| 1203 | } |
| 1204 | |
| 1205 | lit := p.cur.Literal |
| 1206 | p.advance() |
| 1207 | |
| 1208 | // detect format metadata before normalizing |
| 1209 | amt.QuantityFmt = detectFormat(lit) |
| 1210 | |
| 1211 | // normalize for decimal.NewFromString |
| 1212 | // remove thousands separators, replace decimal mark with '.' |
| 1213 | normalized := normalizeLiteral(lit, amt.QuantityFmt.Thousands, amt.QuantityFmt.Decimal) |
| 1214 | |
| 1215 | q, err := decimal.FromString(normalized) |
| 1216 | if err != nil { |
| 1217 | p.errorf("invalid quantity %q: %v", lit, err) |
| 1218 | return |
| 1219 | } |
| 1220 | |
| 1221 | if amt.IsNegative { |
| 1222 | q = q.Neg() |
| 1223 | } |
| 1224 | amt.Quantity = q |
| 1225 | } |
| 1226 | |
| 1227 | func (p *Parser) parseBlankLine() *ast.BlankLine { |
| 1228 | s := p.cur.Span |
| 1229 | p.expectNewline() |
| 1230 | return &ast.BlankLine{Span: s} |
| 1231 | } |
| 1232 | |
| 1233 | func (p *Parser) expectNewline() { |
| 1234 | if p.got(token.NEWLINE) || p.got(token.EOF) { |
| 1235 | if p.got(token.NEWLINE) { |
| 1236 | p.advance() |
| 1237 | } |
| 1238 | return |
| 1239 | } |
| 1240 | p.errorf("expected %s, got %s", token.NEWLINE, p.cur.Type) |
| 1241 | } |
| 1242 | |
| 1243 | func (p *Parser) advance() token.Token { |
| 1244 | prev := p.cur |
| 1245 | p.cur = p.peek |
| 1246 | p.peek = p.lexer.Next() |
| 1247 | return prev |
| 1248 | } |
| 1249 | |
| 1250 | func (p *Parser) got(kind token.Type) bool { return p.cur.Type == kind } |
| 1251 | func (p *Parser) willGet(kind token.Type) bool { return p.peek.Type == kind } |
| 1252 | |
| 1253 | func (p *Parser) expect(kind token.Type) (token.Token, bool) { |
| 1254 | if p.got(kind) { |
| 1255 | return p.advance(), true |
| 1256 | } |
| 1257 | p.errorf("expected %s, got %s", kind, p.cur.Type) |
| 1258 | return p.cur, false |
| 1259 | } |
| 1260 | |
| 1261 | func (p *Parser) errorf(format string, args ...any) { |
| 1262 | p.errors = append(p.errors, &ast.ParseError{ |
| 1263 | Span: p.cur.Span, |
| 1264 | Message: fmt.Sprintf(format, args...), |
| 1265 | }) |
| 1266 | } |
| 1267 | |
| 1268 | // errorfAt records a parse error pointing at the start of span. |
| 1269 | func (p *Parser) errorfAt(span token.Span, format string, args ...any) { |
| 1270 | p.errors = append(p.errors, &ast.ParseError{ |
| 1271 | Span: token.Span{File: span.File, Start: span.Start, End: span.Start}, |
| 1272 | Message: fmt.Sprintf(format, args...), |
| 1273 | }) |
| 1274 | } |
| 1275 | |
| 1276 | func accountSubdirectiveKind(name string) (ast.AccountSubdirectiveKind, bool) { |
| 1277 | switch name { |
| 1278 | case "alias": |
| 1279 | return ast.SubdirectiveAlias, true |
| 1280 | case "type": |
| 1281 | return ast.SubdirectiveType, true |
| 1282 | case "note": |
| 1283 | return ast.SubdirectiveNote, true |
| 1284 | } |
| 1285 | return 0, false |
| 1286 | } |
| 1287 | |
| 1288 | func isCommentMarker(s string) bool { |
| 1289 | return len(s) > 0 && (s[0] == '#' || s[0] == '%' || s[0] == '*') |
| 1290 | } |
| 1291 | |
| 1292 | // skipToNewline consumes the rest of the current line. |
| 1293 | func (p *Parser) skipToNewline() { |
| 1294 | for !p.got(token.NEWLINE) && !p.got(token.EOF) { |
| 1295 | p.advance() |
| 1296 | } |
| 1297 | p.expectNewline() |
| 1298 | } |
| 1299 | |
| 1300 | func (p *Parser) parseSubdirectiveValue() (string, token.Span) { |
| 1301 | var b strings.Builder |
| 1302 | var first, last token.Span |
| 1303 | var single, pendingWS string |
| 1304 | for !p.got(token.SEMICOLON) && !p.got(token.NEWLINE) && !p.got(token.EOF) { |
| 1305 | t := p.cur |
| 1306 | if t.Type == token.WHITESPACE || t.Type == token.INDENT { |
| 1307 | if first.Start.Offset > 0 { |
| 1308 | pendingWS = t.Literal // only the last whitespace run matters |
| 1309 | } |
| 1310 | p.advance() |
| 1311 | continue |
| 1312 | } |
| 1313 | if first.Start.Offset == 0 { |
| 1314 | first, last = t.Span, t.Span |
| 1315 | single = t.Literal |
| 1316 | } else { |
| 1317 | if single != "" { |
| 1318 | b.WriteString(single) |
| 1319 | single = "" |
| 1320 | } |
| 1321 | if pendingWS != "" { |
| 1322 | b.WriteString(pendingWS) |
| 1323 | pendingWS = "" |
| 1324 | } |
| 1325 | b.WriteString(t.Literal) |
| 1326 | last = t.Span |
| 1327 | } |
| 1328 | p.advance() |
| 1329 | } |
| 1330 | if first.Start.Offset == 0 { |
| 1331 | return "", token.Span{} |
| 1332 | } |
| 1333 | if single != "" { |
| 1334 | return single, token.Span{File: first.File, Start: first.Start, End: last.End} |
| 1335 | } |
| 1336 | return strings.TrimSpace(b.String()), token.Span{File: first.File, Start: first.Start, End: last.End} |
| 1337 | } |
| 1338 | |
| 1339 | func (p *Parser) parseTextComment() *ast.Comment { |
| 1340 | s := p.cur.Span |
| 1341 | marker := p.cur.Literal[0] |
| 1342 | var b strings.Builder |
| 1343 | b.WriteString(p.cur.Literal) |
| 1344 | p.advance() |
| 1345 | for !p.got(token.NEWLINE) && !p.got(token.EOF) { |
| 1346 | b.WriteString(p.cur.Literal) |
| 1347 | p.advance() |
| 1348 | } |
| 1349 | text := strings.TrimSpace(strings.TrimPrefix(strings.TrimSpace(b.String()), string(marker))) |
| 1350 | span := p.span(s) // marker to line end, without the newline |
| 1351 | p.expectNewline() |
| 1352 | return &ast.Comment{Marker: marker, Text: text, Span: span} |
| 1353 | } |
| 1354 | |
| 1355 | func isDirectiveKeyword(t token.Type) bool { |
| 1356 | switch t { |
| 1357 | case token.COMMENTKW, token.ACCOUNT, token.COMMODITY, token.INCLUDE, |
| 1358 | token.ALIAS, token.PAYEE, token.TAG, token.APPLY, token.END, |
| 1359 | token.YEAR, token.DECIMALMARK, token.D, token.P, token.N, token.C: |
| 1360 | return true |
| 1361 | } |
| 1362 | return false |
| 1363 | } |
| 1364 | |
| 1365 | func (p *Parser) sync() { |
| 1366 | for { |
| 1367 | switch p.cur.Type { |
| 1368 | case token.EOF: |
| 1369 | return |
| 1370 | case token.NEWLINE: |
| 1371 | p.advance() |
| 1372 | t := p.cur.Type |
| 1373 | if isDirectiveKeyword(t) || t == token.DATE || t == token.TILDE || t == token.EQ { |
| 1374 | return |
| 1375 | } |
| 1376 | default: |
| 1377 | p.advance() |
| 1378 | } |
| 1379 | } |
| 1380 | } |
| 1381 | |
| 1382 | func (p *Parser) syncToNextline() { |
| 1383 | for p.cur.Type != token.NEWLINE && p.cur.Type != token.EOF { |
| 1384 | p.advance() |
| 1385 | } |
| 1386 | if p.got(token.NEWLINE) { |
| 1387 | p.advance() |
| 1388 | } |
| 1389 | } |
| 1390 | |
| 1391 | func (p *Parser) skipWhitespace() { |
| 1392 | for p.got(token.WHITESPACE) { |
| 1393 | p.advance() |
| 1394 | } |
| 1395 | } |
| 1396 | |
| 1397 | func (p *Parser) span(s token.Span) token.Span { |
| 1398 | return token.Span{File: s.File, Start: s.Start, End: p.cur.Span.Start} |
| 1399 | } |
| 1400 | |
| 1401 | func normalizeLiteral(lit string, thousands, decimal byte) string { |
| 1402 | // fast path: no separators to strip and the decimal mark is already '.' |
| 1403 | if thousands == 0 && (decimal == 0 || decimal == '.') { |
| 1404 | return lit |
| 1405 | } |
| 1406 | var b strings.Builder |
| 1407 | for _, ch := range []byte(lit) { |
| 1408 | if thousands != 0 && ch == thousands { |
| 1409 | continue // skip thousands separator |
| 1410 | } |
| 1411 | if ch == decimal { |
| 1412 | b.WriteByte('.') |
| 1413 | } else { |
| 1414 | b.WriteByte(ch) |
| 1415 | } |
| 1416 | } |
| 1417 | return b.String() |
| 1418 | } |
| 1419 | |
| 1420 | func detectFormat(lit string) ast.QuantityFormat { |
| 1421 | var seps []int |
| 1422 | for i, ch := range []byte(lit) { |
| 1423 | if ch == '.' || ch == ',' || ch == ' ' || ch == '_' || ch == '\'' { |
| 1424 | seps = append(seps, i) |
| 1425 | } |
| 1426 | } |
| 1427 | |
| 1428 | if len(seps) == 0 { |
| 1429 | return ast.QuantityFormat{} |
| 1430 | } |
| 1431 | |
| 1432 | last := seps[len(seps)-1] |
| 1433 | dec := lit[last] |
| 1434 | if dec != '.' && dec != ',' { |
| 1435 | // the last separator is a thousands mark; the literal has no decimal mark |
| 1436 | dec = 0 |
| 1437 | } |
| 1438 | var thou byte |
| 1439 | if len(seps) > 1 { |
| 1440 | thou = lit[seps[0]] |
| 1441 | } else if dec == 0 { |
| 1442 | // single space/underscore/apostrophe is always thousands |
| 1443 | thou = lit[last] |
| 1444 | } |
| 1445 | |
| 1446 | // calculate precision when the last separator is a real decimal |
| 1447 | prec := 0 |
| 1448 | if thou == 0 || len(seps) > 1 { |
| 1449 | prec = len(lit) - last - 1 |
| 1450 | } |
| 1451 | |
| 1452 | return ast.QuantityFormat{Decimal: dec, Thousands: thou, Precision: prec} |
| 1453 | } |
| 1454 | |
| 1455 | // parseSimpleDate parses full YYYY/MM/DD date literal embedded in free text. |
| 1456 | func parseSimpleDate(s string) ast.Date { |
| 1457 | year, month, day, sep, err := ParseDateLiteral(s) |
| 1458 | if err != nil { |
| 1459 | return ast.Date{} |
| 1460 | } |
| 1461 | return ast.Date{Year: year, Month: month, Day: day, Sep: sep} |
| 1462 | } |
| 1463 | |
| 1464 | // ParseDateLiteral parses and validates a date literal. |
| 1465 | // It accepts full YYYY/MM/DD and partial MM/DD forms, with '-', '/' or '.' as separators. |
| 1466 | func ParseDateLiteral(lit string) (year, month, day int, sep byte, err error) { |
| 1467 | sep = dateSeparator(lit) |
| 1468 | if sep == 0 { |
| 1469 | return 0, 0, 0, 0, fmt.Errorf("invalid date format: %q", lit) |
| 1470 | } |
| 1471 | |
| 1472 | parts := strings.Split(lit, string(sep)) |
| 1473 | if len(parts) != 2 && len(parts) != 3 { |
| 1474 | return 0, 0, 0, 0, fmt.Errorf("invalid date format: %q", lit) |
| 1475 | } |
| 1476 | |
| 1477 | nums := make([]int, len(parts)) |
| 1478 | for i, part := range parts { |
| 1479 | if nums[i], err = strconv.Atoi(part); err != nil { |
| 1480 | return 0, 0, 0, 0, fmt.Errorf("invalid date literal: %q", lit) |
| 1481 | } |
| 1482 | } |
| 1483 | |
| 1484 | month = nums[len(parts)-2] |
| 1485 | if month < 1 || month > 12 { |
| 1486 | return 0, 0, 0, 0, fmt.Errorf("invalid month %d in %q", month, lit) |
| 1487 | } |
| 1488 | |
| 1489 | day = nums[len(parts)-1] |
| 1490 | if day < 1 || day > 31 { |
| 1491 | return 0, 0, 0, 0, fmt.Errorf("invalid day %d in %q", day, lit) |
| 1492 | } |
| 1493 | |
| 1494 | if len(parts) == 2 { |
| 1495 | return 0, month, day, sep, nil |
| 1496 | } |
| 1497 | return nums[0], month, day, sep, nil |
| 1498 | } |
| 1499 | |
| 1500 | func dateSeparator(lit string) byte { |
| 1501 | for i := 0; i < len(lit); i++ { |
| 1502 | if lit[i] == '/' || lit[i] == '-' || lit[i] == '.' { |
| 1503 | return lit[i] |
| 1504 | } |
| 1505 | } |
| 1506 | return 0 |
| 1507 | } |
| 1508 | |
| 1509 | // parseCommentTags extacts tags from comment text. |
| 1510 | // A tag is a word immediately followed by a ':', with an optional value that ends at a comma or the end of a line. |
| 1511 | // https://hledger.org/1.52/hledger.html?highlight=tags#tags |
| 1512 | func parseCommentTags(text string, base token.Span) []ast.Tag { |
| 1513 | var tags []ast.Tag |
| 1514 | for i := 0; i < len(text); { |
| 1515 | colon := strings.IndexByte(text[i:], ':') |
| 1516 | if colon < 0 { |
| 1517 | break |
| 1518 | } |
| 1519 | colon += i |
| 1520 | |
| 1521 | keyStart := colon |
| 1522 | for keyStart > i { |
| 1523 | r, size := utf8.DecodeLastRuneInString(text[:keyStart]) |
| 1524 | if unicode.IsSpace(r) { |
| 1525 | break |
| 1526 | } |
| 1527 | keyStart -= size |
| 1528 | } |
| 1529 | if keyStart == colon { // nothing before the colon = not a tag |
| 1530 | i = colon + 1 |
| 1531 | continue |
| 1532 | } |
| 1533 | key := text[keyStart:colon] |
| 1534 | |
| 1535 | valueEnd := colon + 1 |
| 1536 | for valueEnd < len(text) && text[valueEnd] != ',' { |
| 1537 | valueEnd++ |
| 1538 | } |
| 1539 | value := strings.TrimSpace(text[colon+1 : valueEnd]) |
| 1540 | |
| 1541 | tags = append(tags, ast.Tag{ |
| 1542 | Key: key, |
| 1543 | Value: value, |
| 1544 | Span: token.Span{ |
| 1545 | File: base.File, |
| 1546 | Start: tagPos(base.Start, text, keyStart), |
| 1547 | End: tagPos(base.Start, text, valueEnd), |
| 1548 | }, |
| 1549 | }) |
| 1550 | i = valueEnd |
| 1551 | if i < len(text) && text[i] == ',' { |
| 1552 | i++ |
| 1553 | } |
| 1554 | } |
| 1555 | |
| 1556 | return tags |
| 1557 | } |
| 1558 | |
| 1559 | func tagPos(base token.Pos, text string, off int) token.Pos { |
| 1560 | return token.Pos{ |
| 1561 | Offset: base.Offset + off, |
| 1562 | Line: base.Line, |
| 1563 | Col: base.Col + utf8.RuneCountInString(text[:off]), |
| 1564 | } |
| 1565 | } |