Mercurial > repos > tabletprog

comparison modules/parser.tp @ 220:a1a80af71b05

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Implement onePlus macro. Fix some bugs in the other matching macros. Implement integer literal parsing rules.
author Michael Pavone <pavone@retrodev.com>
date Mon, 23 Dec 2013 14:44:31 -0800
parents b799192e404b
children c6e321a538d4
comparison
equal deleted inserted replaced
219:b70be565d54c 220:a1a80af71b05
64 #{ 64 #{
65 valid? <- { true } 65 valid? <- { true }
66 matchcall <- quote: (_applyMatch: :tomatch { 66 matchcall <- quote: (_applyMatch: :tomatch {
67 lm <- left 67 lm <- left
68 if: (lm matched?) { 68 if: (lm matched?) {
69 orig <- tomatch
69 tomatch <- tomatch from: (lm matchlen) 70 tomatch <- tomatch from: (lm matchlen)
70 rm <- right 71 rm <- right
71 if: (rm matched?) { 72 if: (rm matched?) {
72 total <- (rm matchlen) + (lm matchlen) 73 total <- (rm matchlen) + (lm matchlen)
73 #{ 74 #{
74 matched? <- { true } 75 matched? <- { true }
75 matchlen <- { total } 76 matchlen <- { total }
76 basicYield? <- { true } 77 basicYield? <- { true }
77 yield <- { tomatch from: 0 withLen: total } 78 yield <- { orig from: 0 withLength: total }
78 } 79 }
79 } else: { 80 } else: {
80 rm 81 rm
81 } 82 }
82 } else: { 83 } else: {
206 } else: { 207 } else: {
207 print: "#error Argument to charClass macro must be a compile-time constant\n" 208 print: "#error Argument to charClass macro must be a compile-time constant\n"
208 } 209 }
209 } 210 }
210 211
211 zeroPlus <- macro: :matchexpr { 212 _nPlus <- :matchexpr min {
212 funexpr <- false 213 funexpr <- false
213 valid <- false 214 valid <- false
214 mc <- _makeMatchCall: matchexpr 215 mc <- _makeMatchCall: matchexpr
215 if: (mc valid?) { 216 if: (mc valid?) {
216 mcall <- mc matchcall 217 mcall <- mc matchcall
217 quote: :tomatch { 218 quote: :tomatch {
218 cur <- 0 219 cur <- 0
220 count <- 0
219 n <- tomatch byte_length 221 n <- tomatch byte_length
220 orig <- tomatch 222 orig <- tomatch
221 match <- true 223 match <- true
222 allBasic? <- true 224 allBasic? <- true
223 yieldvals <- [] 225 yieldvals <- []
224 while: { match && cur < n } do: { 226 while: { match && cur < n } do: {
225 res <- mcall 227 res <- mcall
226 match <- res matched? 228 match <- res matched?
227 if: match { 229 if: match {
230 count <- count + 1
228 //TODO: Use some kind of lightweight substring wrapper here 231 //TODO: Use some kind of lightweight substring wrapper here
229 tomatch <- tomatch from: (res matchlen) 232 tomatch <- tomatch from: (res matchlen)
230 if: allBasic? { 233 if: allBasic? {
231 ifnot: (res basicYield?) { 234 ifnot: (res basicYield?) {
232 allBasic? <- false 235 allBasic? <- false
240 } 243 }
241 allBasic? <- allBasic? && (res basicYield?) 244 allBasic? <- allBasic? && (res basicYield?)
242 cur <- cur + (res matchlen) 245 cur <- cur + (res matchlen)
243 } 246 }
244 } 247 }
245 if: cur > 0 { 248 if: count >= min {
246 if: allBasic? { 249 if: allBasic? {
247 #{ 250 #{
248 matched? <- { true } 251 matched? <- { true }
249 matchlen <- { cur } 252 matchlen <- { cur }
250 basicYield? <- { true } 253 basicYield? <- { true }
264 matched? <- { false } 267 matched? <- { false }
265 } 268 }
266 } 269 }
267 } 270 }
268 } else: { 271 } else: {
269 print: "#error Invalid zeroPlus macro call: " . (mc message) . "\n" 272 print: "#error Invalid nPlus macro call: " . (mc message) . "\n"
270 } 273 }
274 }
275
276 zeroPlus <- macro: :matchexpr {
277 _nPlus: matchexpr 0
278 }
279
280 onePlus <- macro: :matchexpr {
281 _nPlus: matchexpr 1
271 } 282 }
272 283
273 matchOne <- macro: :options { 284 matchOne <- macro: :options {
274 options <- (options value) map: :option { 285 options <- (options value) map: :option {
275 _makeMatchCall: option 286 _makeMatchCall: option
402 413
403 414
404 _alpha <- charClass: "a-zA-Z" 415 _alpha <- charClass: "a-zA-Z"
405 alpha <- zeroPlus: _alpha 416 alpha <- zeroPlus: _alpha
406 alphaNum <- zeroPlus: (charClass: "a-zA-Z0-9") 417 alphaNum <- zeroPlus: (charClass: "a-zA-Z0-9")
418
419 blockComment <- match: "/*" . (zeroPlus: (matchOne: [(charClass: "^*") "*" . (charClass: "^/")])) . "*/" yield: { false }
420
407 hws <- zeroPlus: (matchOne: [ 421 hws <- zeroPlus: (matchOne: [
408 (charClass: " \t") 422 (charClass: " \t")
409 "/*" . (zeroPlus: (matchOne: [(charClass: "^*") "*" . (charClass: "^/")])) . "*/" 423 blockComment
410 ]) 424 ])
411 425
426 ws <- zeroPlus: (matchOne: [
427 (charClass: " \n\t\r")
428 "//" . (zeroPlus: (charClass: "^\n")) . "\n"
429 blockComment
430 ])
431
432 escape <- matchOne: [
433 (match: "\\n" yield: {"\n"})
434 (match: "\\r" yield: {"\n"})
435 (match: "\\t" yield: {"\n"})
436 (match: "\\\\" yield: {"\\"})
437 (match: "\\\"" yield: {"\""})
438 ]
439
440 string <- match: "\"" . Chars . "\"" where: {
441 Chars <- zeroPlus: (matchOne: [
442 (charClass: "^\"\\")
443 escape
444 ])
445 } yield: {
446 Chars join: ""
447 }
448
449 bdigit <- matchOne: [
450 (match: "0" yield: {0i64})
451 (match: "1" yield: {1i64})
452 ]
453
412 digit <- matchOne: [ 454 digit <- matchOne: [
413 (match: "0" yield: {0}) 455 bdigit
414 (match: "1" yield: {1}) 456 (match: "2" yield: {2i64})
415 (match: "2" yield: {2}) 457 (match: "3" yield: {3i64})
416 (match: "3" yield: {3}) 458 (match: "4" yield: {4i64})
417 (match: "4" yield: {4}) 459 (match: "5" yield: {5i64})
418 (match: "5" yield: {5}) 460 (match: "6" yield: {6i64})
419 (match: "6" yield: {6}) 461 (match: "7" yield: {7i64})
420 (match: "7" yield: {7}) 462 (match: "8" yield: {8i64})
421 (match: "8" yield: {8}) 463 (match: "9" yield: {9i64})
422 (match: "9" yield: {9})
423 ] 464 ]
424 465
425 posint <- match: Digits where: { 466 hdigit <- matchOne: [
426 Digits <- zeroPlus: digit 467 digit
468 (match: (charClass: "aA") yield: {10i64})
469 (match: (charClass: "bB") yield: {11i64})
470 (match: (charClass: "cC") yield: {12i64})
471 (match: (charClass: "dD") yield: {13i64})
472 (match: (charClass: "eE") yield: {14i64})
473 (match: (charClass: "fF") yield: {15i64})
474 ]
475
476 binary <- match: "0b" . Digits . Suffix where: {
477 Digits <- onePlus: bdigit
478 Suffix <- matchOne: [
479 (charClass: "ui") . (matchOne: ["8" "16" "32" "64"])
480 ""
481 ]
427 } yield: { 482 } yield: {
428 num <- Digits fold: 0 with: :acc el { 483 num <- Digits fold: 0 with: :acc el {
429 print: "Element " . el . "\n" 484 acc * 2i64 + el
430 acc * 10 + el 485 }
486 signed <- true
487 litbits <- 32
488 if: (Suffix length) > 0 {
489 if: (Suffix from: 0 withLength: 1) = "u" {
490 signed <- false
491 }
492 litbits <- (Suffix from: 1) int32
431 } 493 }
432 #{ 494 #{
433 litval <- num 495 litval <- num
496 signed? <- signed
497 bits <- litbits
498 }
499 }
500
501 decimal <- match: Sign . Digits . Suffix where: {
502 Sign <- matchOne: ["-" ""]
503 Digits <- onePlus: digit
504 Suffix <- matchOne: [
505 (charClass: "ui") . (matchOne: ["8" "16" "32" "64"])
506 ""
507 ]
508 } yield: {
509 num <- Digits fold: 0 with: :acc el {
510 acc * 10i64 + el
511 }
512 if: Sign = "-" {
513 num <- 0i64 - num
514 }
515 signed <- true
516 litbits <- 32
517 if: (Suffix length) > 0 {
518 if: (Suffix from: 0 withLength: 1) = "u" {
519 signed <- false
520 }
521 print: (Suffix from: 1) . "\n"
522 litbits <- (Suffix from: 1) int32
523 }
524 #{
525 litval <- num
526 signed? <- signed
527 bits <- litbits
528 }
529 }
530
531 hex <- match: "0x" . Digits . Suffix where: {
532 Digits <- onePlus: hdigit
533 Suffix <- matchOne: [
534 (charClass: "ui") . (matchOne: ["8" "16" "32" "64"])
535 ""
536 ]
537 } yield: {
538 num <- Digits fold: 0 with: :acc el {
539 acc * 16i64 + el
540 }
541 signed <- true
542 litbits <- 32
543 if: (Suffix length) > 0 {
544 if: (Suffix from: 0 withLength: 1) = "u" {
545 signed <- false
546 }
547 litbits <- (Suffix from: 1) int32
548 }
549 #{
550 litval <- num
551 signed? <- signed
552 bits <- litbits
553 }
554 }
555
556 testmatchintlit <- :val matchfun {
557 res <- matchfun: val
558 if: (res matched?) {
559 y <- res yield
560 print: val . " matched with litval " . (y litval) . ", bits " . (y bits) . " and singned? " . (y signed?) . "\n"
561 } else: {
562 print: val . " did not match\n"
434 } 563 }
435 } 564 }
436 565
437 main <- { 566 main <- {
438 cmatch <- alpha: "czx0123" 567 cmatch <- alpha: "czx0123"
465 print: "3 matched with yield " . (tmatch yield) . ", yield + 1 = " . ((tmatch yield) + 1) . "\n" 594 print: "3 matched with yield " . (tmatch yield) . ", yield + 1 = " . ((tmatch yield) + 1) . "\n"
466 } else: { 595 } else: {
467 print: "3 did not match\n" 596 print: "3 did not match\n"
468 } 597 }
469 598
470 posintm <- posint: "345" 599 testmatchintlit: "345" :s {decimal: s}
471 if: (posintm matched?) { 600 testmatchintlit: "-567" :s {decimal: s}
472 print: "345 matched with intlit value " . ((posintm yield) litval) . "\n" 601 testmatchintlit: "123u16" :s {decimal: s}
473 } else: { 602 testmatchintlit: "0x20" :s {hex: s}
474 print: "345 did not match\n" 603 testmatchintlit: "0x42u64" :s {hex: s}
475 } 604 testmatchintlit: "0b10101" :s {binary: s}
476 } 605 }
477 } 606 }