regex.bend
bend-kit-regex@0.5.0.0 · 1,486 lines · 0 definitions
1# Linear-time regular expressions: RE2 syntax, Pike VM, capture groups. Source: https://github.com/paymog/bend-kit/tree/main/regex
2import Base
3import bend-kit-unicode@0.1.0.0/unicode.bend as U
4import bend-kit-bytes@0.3.0.0/bytes.bend as Bytes
5
6# Positions count code points of the String, not octets.
7# Leftmost-first semantics, as RE2 and Perl: the first alternative that matches wins.
8
9# Syntax
10# ------
11
12# A set member: a code point range, or a Unicode general category ("L", "Lu", ...).
13type Item is Data:
14 Rng{lo: U32, hi: U32}
15 Prop{neg: Bool, name: String}
16
17# Assertion kinds: 0 start of text, 1 end of text, 2 word boundary, 3 not a word boundary.
18type Node is Data:
19 NEmpty{}
20 NSet{neg: Bool, items: List<&2, Item>}
21 NAssert{k: U32}
22 NCat{a: Node, b: Node}
23 NAlt{a: Node, b: Node}
24 NPlus{greedy: Bool, a: Node}
25 NQuest{greedy: Bool, a: Node}
26 NGroup{i: U32, a: Node}
27
28type Tok is Data:
29 TAtom{n: Node}
30 TOpen{cap: Bool}
31 TClose{}
32 TBar{}
33 TRep{min: U32, max: Maybe<&2, U32>, greedy: Bool}
34
35def lit(+c: U32) -> Node:
36 NSet{False{}, [Rng{c, c}]}
37
38def is_digit(+c: U32) -> Bool:
39 U32.is_le(48, c) && U32.is_le(c, 57)
40
41def is_word(+c: U32) -> Bool:
42 is_digit(c) || (U32.is_le(65, c) && U32.is_le(c, 90)) || (U32.is_le(97, c) && U32.is_le(c, 122)) || U32.is_eq(c, 95)
43
44def is_alnum(+c: U32) -> Bool:
45 Bool.and(is_word(c), U32.is_ne(c, 95))
46
47# \d \w \s are ASCII, as in RE2.
48def perl.d() -> List<&2, Item>:
49 [Rng{48, 57}]
50
51def perl.D() -> List<&2, Item>:
52 [Rng{0, 47}, Rng{58, 1114111}]
53
54def perl.w() -> List<&2, Item>:
55 [Rng{48, 57}, Rng{65, 90}, Rng{95, 95}, Rng{97, 122}]
56
57def perl.W() -> List<&2, Item>:
58 [Rng{0, 47}, Rng{58, 64}, Rng{91, 94}, Rng{96, 96}, Rng{123, 1114111}]
59
60def perl.s() -> List<&2, Item>:
61 [Rng{9, 10}, Rng{12, 13}, Rng{32, 32}]
62
63def perl.S() -> List<&2, Item>:
64 [Rng{0, 8}, Rng{11, 11}, Rng{14, 31}, Rng{33, 1114111}]
65
66def cats() -> List<&2, String>:
67 ["L", "Lu", "Ll", "Lt", "Lm", "Lo", "M", "Mn", "Mc", "Me", "N", "Nd", "Nl", "No",
68 "P", "Pc", "Pd", "Ps", "Pe", "Pi", "Pf", "Po", "S", "Sm", "Sc", "Sk", "So",
69 "Z", "Zs", "Zl", "Zp", "C", "Cc", "Cf", "Cs", "Co", "Cn"]
70
71# Escapes
72# -------
73
74type Esc is Data:
75 EPoint{c: U32, rest: String}
76 EItems{items: List<&2, Item>, rest: String}
77 EAssert{k: U32, rest: String}
78 EBad{}
79
80def prop.if(ok: Bool, neg: Bool, name: String, t: String) -> Esc:
81 match ok:
82 case True{}:
83 EItems{[Prop{neg, name}], t}
84 case False{}:
85 EBad{}
86
87def prop.ok(neg: Bool, +name: String, t: String) -> Esc:
88 prop.if(List.contains(~String, ~String.eq, cats(), name), neg, name, t)
89
90def prop.name(s: String, neg: Bool, acc: String) -> Esc:
91 match s:
92 case SNil{}:
93 EBad{}
94 case SCon{'}', t}:
95 prop.ok(neg, String.reverse(acc), t)
96 case SCon{c, t}:
97 prop.name(t, neg, SCon{c, acc})
98
99# \pL or \p{Lu}; \P negates.
100def prop(neg: Bool, s: String) -> Esc:
101 match s:
102 case SNil{}:
103 EBad{}
104 case SCon{'{', t}:
105 prop.name(t, neg, SNil{})
106 case SCon{c, t}:
107 prop.ok(neg, SCon{c, SNil{}}, t)
108
109# An escaped letter or digit with no meaning is an error, as in RE2.
110def esc.other(alnum: Bool, +c: U32, t: String) -> Esc:
111 match alnum:
112 case True{}:
113 EBad{}
114 case False{}:
115 EPoint{c, t}
116
117# The text after a backslash.
118def esc(s: String) -> Esc:
119 match s:
120 case SNil{}:
121 EBad{}
122 case SCon{'d', t}:
123 EItems{perl.d(), t}
124 case SCon{'D', t}:
125 EItems{perl.D(), t}
126 case SCon{'w', t}:
127 EItems{perl.w(), t}
128 case SCon{'W', t}:
129 EItems{perl.W(), t}
130 case SCon{'s', t}:
131 EItems{perl.s(), t}
132 case SCon{'S', t}:
133 EItems{perl.S(), t}
134 case SCon{'p', t}:
135 prop(False{}, t)
136 case SCon{'P', t}:
137 prop(True{}, t)
138 case SCon{'A', t}:
139 EAssert{0, t}
140 case SCon{'z', t}:
141 EAssert{1, t}
142 case SCon{'b', t}:
143 EAssert{2, t}
144 case SCon{'B', t}:
145 EAssert{3, t}
146 case SCon{'n', t}:
147 EPoint{10, t}
148 case SCon{'t', t}:
149 EPoint{9, t}
150 case SCon{'r', t}:
151 EPoint{13, t}
152 case SCon{'f', t}:
153 EPoint{12, t}
154 case SCon{'v', t}:
155 EPoint{11, t}
156 case SCon{Chr{+c}, t}:
157 esc.other(is_alnum(c), c, t)
158
159# Classes
160# -------
161
162type Cs is Data:
163 CsGo{s: String, first: Bool, items: List<&2, Item>}
164 CsDone{items: List<&2, Item>, rest: String}
165 CsFail{}
166
167def class.range(ok: Bool, +c: U32, +d: U32, u: String, items: List<&2, Item>) -> Cs:
168 match ok:
169 case True{}:
170 CsGo{u, False{}, Rng{c, d} <> items}
171 case False{}:
172 CsFail{}
173
174def class.hi(e: Esc, +c: U32, items: List<&2, Item>) -> Cs:
175 match e:
176 case EPoint{+d, u}:
177 class.range(U32.is_le(c, d), c, d, u, items)
178 case EItems{xs, u}:
179 CsFail{}
180 case EAssert{k, u}:
181 CsFail{}
182 case EBad{}:
183 CsFail{}
184
185# c is one member; a "-" and a code point after it make a range.
186def class.lo(+c: U32, t: String, items: List<&2, Item>) -> Cs:
187 match t:
188 case SCon{'-', SCon{']', u}}:
189 CsDone{Rng{45, 45} <> Rng{c, c} <> items, u}
190 case SCon{'-', SCon{'\\', u}}:
191 class.hi(esc(u), c, items)
192 case SCon{'-', SCon{Chr{+d}, u}}:
193 class.range(U32.is_le(c, d), c, d, u, items)
194 case _:
195 CsGo{t, False{}, Rng{c, c} <> items}
196
197def class.esc(e: Esc, items: List<&2, Item>) -> Cs:
198 match e:
199 case EPoint{+c, t}:
200 class.lo(c, t, items)
201 case EItems{xs, t}:
202 CsGo{t, False{}, List.append(&2, Item, xs, items)}
203 case EAssert{k, t}:
204 CsFail{}
205 case EBad{}:
206 CsFail{}
207
208# A "]" right after "[" or "[^" is a member.
209def class.step(s: String, first: Bool, items: List<&2, Item>) -> Cs:
210 match s:
211 case SNil{}:
212 CsFail{}
213 case SCon{']', t}:
214 match first:
215 case True{}:
216 class.lo(93, t, items)
217 case False{}:
218 CsDone{items, t}
219 case SCon{'\\', t}:
220 class.esc(esc(t), items)
221 case SCon{Chr{+c}, t}:
222 class.lo(c, t, items)
223
224# fuel: every step eats at least one char.
225def class.run(fuel: Nat, st: Cs) -> Cs:
226 match fuel:
227 case 0n:
228 CsFail{}
229 case 1n+f:
230 match st:
231 case CsGo{s, first, items}:
232 class.run(f, class.step(s, first, items))
233 case CsDone{items, rest}:
234 CsDone{items, rest}
235 case CsFail{}:
236 CsFail{}
237
238def class(+s: String, first: Bool) -> Cs:
239 class.run(Nat.add(String.length(s), 1n), CsGo{s, first, Nil{}})
240
241# Lexer
242# -----
243
244type Lx is Data:
245 LGo{s: String, toks: List<&2, Tok>}
246 LDone{toks: List<&2, Tok>}
247 LFail{}
248
249# A run of decimal digits: its value (capped at 100000), its length, and the rest.
250type Num is Data:
251 Num{v: U32, n: U32, rest: String}
252
253def num.add(+acc: U32, +c: U32) -> U32:
254 Bool.pick(U32, U32.is_lt(acc, 100000), (acc * 10 + (c - 48) : U32), 100000)
255
256def num.go(t: String, +c: U32, +acc: U32, +n: U32, dig: Bool) -> Num:
257 match t:
258 case SNil{}:
259 match dig:
260 case True{}:
261 Num{num.add(acc, c), (n + 1 : U32), SNil{}}
262 case False{}:
263 Num{acc, n, SCon{Chr{c}, SNil{}}}
264 case SCon{Chr{+d}, u}:
265 match dig:
266 case True{}:
267 num.go(u, d, num.add(acc, c), (n + 1 : U32), is_digit(d))
268 case False{}:
269 Num{acc, n, SCon{Chr{c}, SCon{Chr{d}, u}}}
270
271def num(s: String) -> Num:
272 match s:
273 case SNil{}:
274 Num{0, 0, SNil{}}
275 case SCon{Chr{+c}, t}:
276 num.go(t, c, 0, 0, is_digit(c))
277
278# A trailing "?" makes a repetition lazy.
279def lex.rep(+min: U32, max: Maybe<&2, U32>, t: String, toks: List<&2, Tok>) -> Lx:
280 match t:
281 case SCon{'?', u}:
282 LGo{u, TRep{min, max, False{}} <> toks}
283 case _:
284 LGo{t, TRep{min, max, True{}} <> toks}
285
286def Num.n(m: Num) -> U32:
287 Num{v, n, r} = m
288 n
289
290# A "{" that does not start {n}, {n,} or {n,m} is a literal, as in RE2.
291def lex.brace.max(none: Bool, +min: U32, m: Num, orig: String, toks: List<&2, Tok>) -> Lx:
292 match none:
293 case True{}:
294 LGo{orig, TAtom{lit(123)} <> toks}
295 case False{}:
296 Num{+v, n, r} = m
297 match r:
298 case SCon{'}', u}:
299 lex.rep(min, Some{v}, u, toks)
300 case _:
301 LGo{orig, TAtom{lit(123)} <> toks}
302
303def lex.brace.min(none: Bool, m: Num, +orig: String, toks: List<&2, Tok>) -> Lx:
304 match none:
305 case True{}:
306 LGo{orig, TAtom{lit(123)} <> toks}
307 case False{}:
308 Num{+v, n, r} = m
309 match r:
310 case SCon{'}', u}:
311 lex.rep(v, Some{v}, u, toks)
312 case SCon{',', SCon{'}', u}}:
313 lex.rep(v, None{}, u, toks)
314 case SCon{',', u}:
315 +m2 = num(u)
316 lex.brace.max(U32.is_eq(Num.n(m2), 0), v, m2, orig, toks)
317 case _:
318 LGo{orig, TAtom{lit(123)} <> toks}
319
320def lex.class(neg: Bool, r: Cs, toks: List<&2, Tok>) -> Lx:
321 match r:
322 case CsDone{items, rest}:
323 LGo{rest, TAtom{NSet{neg, items}} <> toks}
324 case CsGo{s, first, items}:
325 LFail{}
326 case CsFail{}:
327 LFail{}
328
329def lex.esc(e: Esc, toks: List<&2, Tok>) -> Lx:
330 match e:
331 case EPoint{c, t}:
332 LGo{t, TAtom{lit(c)} <> toks}
333 case EItems{xs, t}:
334 LGo{t, TAtom{NSet{False{}, xs}} <> toks}
335 case EAssert{k, t}:
336 LGo{t, TAtom{NAssert{k}} <> toks}
337 case EBad{}:
338 LFail{}
339
340def lex.step(s: String, toks: List<&2, Tok>) -> Lx:
341 match s:
342 case SNil{}:
343 LDone{toks}
344 case SCon{'\\', t}:
345 lex.esc(esc(t), toks)
346 case SCon{'[', SCon{'^', t}}:
347 lex.class(True{}, class(t, True{}), toks)
348 case SCon{'[', t}:
349 lex.class(False{}, class(t, True{}), toks)
350 case SCon{'(', SCon{'?', SCon{':', t}}}:
351 LGo{t, TOpen{False{}} <> toks}
352 case SCon{'(', SCon{'?', t}}:
353 LFail{}
354 case SCon{'(', t}:
355 LGo{t, TOpen{True{}} <> toks}
356 case SCon{')', t}:
357 LGo{t, TClose{} <> toks}
358 case SCon{'|', t}:
359 LGo{t, TBar{} <> toks}
360 case SCon{'*', t}:
361 lex.rep(0, None{}, t, toks)
362 case SCon{'+', t}:
363 lex.rep(1, None{}, t, toks)
364 case SCon{'?', t}:
365 lex.rep(0, Some{1}, t, toks)
366 case SCon{'{', +t}:
367 +m = num(t)
368 lex.brace.min(U32.is_eq(Num.n(m), 0), m, t, toks)
369 case SCon{'.', t}:
370 LGo{t, TAtom{NSet{True{}, [Rng{10, 10}]}} <> toks}
371 case SCon{'^', t}:
372 LGo{t, TAtom{NAssert{0}} <> toks}
373 case SCon{'$', t}:
374 LGo{t, TAtom{NAssert{1}} <> toks}
375 case SCon{Chr{+c}, t}:
376 LGo{t, TAtom{lit(c)} <> toks}
377
378# fuel: every step but the last eats at least one char.
379def lex(fuel: Nat, st: Lx) -> Maybe<&2, List<&2, Tok>>:
380 match fuel:
381 case 0n:
382 None{}
383 case 1n+f:
384 match st:
385 case LGo{s, toks}:
386 lex(f, lex.step(s, toks))
387 case LDone{toks}:
388 Some{List.reverse(&2, Tok, toks)}
389 case LFail{}:
390 None{}
391
392# Parser
393# ------
394
395# An open group: its capture index (None for (?:...)), and the enclosing alternatives and sequence.
396type Frame is Data:
397 Frame{cap: Maybe<&2, U32>, alts: List<&2, Node>, cur: List<&2, Node>}
398
399# n: the next capture index. rep: the last token was a repetition. alts and cur are reversed.
400type Ps is Data:
401 Ps{n: U32, rep: Bool, stack: List<&2, Frame>, alts: List<&2, Node>, cur: List<&2, Node>}
402 PsFail{}
403def cat(a: Node, b: Node) -> Node:
404 match a:
405 case NEmpty{}:
406 b
407 case _:
408 match b:
409 case NEmpty{}:
410 a
411 case _:
412 NCat{a, b}
413
414def seq.go(xs: List<&2, Node>, acc: Node) -> Node:
415 match xs:
416 case Nil{}:
417 acc
418 case Con{h, t}:
419 seq.go(t, cat(h, acc))
420
421def seq(xs: List<&2, Node>) -> Node:
422 seq.go(xs, NEmpty{})
423
424def alt.go(xs: List<&2, Node>, acc: Node) -> Node:
425 match xs:
426 case Nil{}:
427 acc
428 case Con{h, t}:
429 alt.go(t, NAlt{h, acc})
430
431def group(cap: Maybe<&2, U32>, a: Node) -> Node:
432 match cap:
433 case None{}:
434 a
435 case Some{i}:
436 NGroup{i, a}
437
438def rep.copies(k: Nat, +h: Node, acc: Node) -> Node:
439 match k:
440 case 0n:
441 acc
442 case 1n+p:
443 rep.copies(p, h, cat(h, acc))
444
445def rep.opt(k: Nat, +g: Bool, +h: Node) -> Node:
446 match k:
447 case 0n:
448 NEmpty{}
449 case 1n+p:
450 NQuest{g, cat(h, rep.opt(p, g, h))}
451
452# x* is (x+)?, so an empty x cannot loop; x{n,} is n-1 copies then x+; x{n,m} is n copies then (x(x...)?)?, as RE2 builds them.
453def rep.inf(k: Nat, +g: Bool, +h: Node) -> Node:
454 match k:
455 case 0n:
456 NQuest{g, NPlus{g, h}}
457 case 1n+p:
458 rep.copies(p, h, NPlus{g, h})
459
460def rep.fin(max: Maybe<&2, U32>, +min: U32, +g: Bool, +h: Node) -> Node:
461 match max:
462 case None{}:
463 rep.inf(U32.to_nat(min), g, h)
464 case Some{m}:
465 rep.copies(U32.to_nat(min), h, rep.opt(U32.to_nat((m - min : U32)), g, h))
466
467def rep(+h: Node, +min: U32, max: Maybe<&2, U32>, +g: Bool) -> Node:
468 rep.fin(max, min, g, h)
469
470# Counts go up to 1000, and a repetition cannot repeat, as in RE2.
471def rep.ok(+min: U32, max: Maybe<&2, U32>) -> Bool:
472 match max:
473 case None{}:
474 U32.is_le(min, 1000)
475 case Some{+m}:
476 U32.is_le(m, 1000) && U32.is_le(min, m)
477
478def parse.rep(ok: Bool, cur: List<&2, Node>, +min: U32, max: Maybe<&2, U32>, +g: Bool, +n: U32, stack: List<&2, Frame>, alts: List<&2, Node>) -> Ps:
479 match ok:
480 case False{}:
481 PsFail{}
482 case True{}:
483 match cur:
484 case Nil{}:
485 PsFail{}
486 case Con{h, t}:
487 Ps{n, True{}, stack, alts, rep(h, min, max, g) <> t}
488
489def parse.open(cap: Bool, +n: U32, stack: List<&2, Frame>, alts: List<&2, Node>, cur: List<&2, Node>) -> Ps:
490 match cap:
491 case True{}:
492 Ps{(n + 1 : U32), False{}, Frame{Some{n}, alts, cur} <> stack, Nil{}, Nil{}}
493 case False{}:
494 Ps{n, False{}, Frame{None{}, alts, cur} <> stack, Nil{}, Nil{}}
495
496def parse.close(stack: List<&2, Frame>, +n: U32, alts: List<&2, Node>, cur: List<&2, Node>) -> Ps:
497 match stack:
498 case Nil{}:
499 PsFail{}
500 case Con{Frame{cap, fa, fc}, up}:
501 Ps{n, False{}, up, fa, group(cap, alt.go(alts, seq(cur))) <> fc}
502
503def parse.tok(tok: Tok, +n: U32, rep: Bool, stack: List<&2, Frame>, alts: List<&2, Node>, cur: List<&2, Node>) -> Ps:
504 match tok:
505 case TAtom{a}:
506 Ps{n, False{}, stack, alts, a <> cur}
507 case TRep{+min, +max, g}:
508 parse.rep(Bool.and(Bool.not(rep), rep.ok(min, max)), cur, min, max, g, n, stack, alts)
509 case TOpen{cap}:
510 parse.open(cap, n, stack, alts, cur)
511 case TClose{}:
512 parse.close(stack, n, alts, cur)
513 case TBar{}:
514 Ps{n, False{}, stack, seq(cur) <> alts, Nil{}}
515
516def parse.step(st: Ps, tok: Tok) -> Ps:
517 match st:
518 case PsFail{}:
519 PsFail{}
520 case Ps{n, rep, stack, alts, cur}:
521 parse.tok(tok, n, rep, stack, alts, cur)
522
523def parse(toks: List<&2, Tok>, st: Ps) -> Ps:
524 match toks:
525 case Nil{}:
526 st
527 case Con{tok, t}:
528 parse(t, parse.step(st, tok))
529
530# Compiler
531# --------
532
533type Inst is Data:
534 ISet{neg: Bool, items: List<&2, Item>}
535 IAssert{k: U32}
536 ISplit{x: U32, y: U32}
537 IJmp{x: U32}
538 ISave{k: U32}
539 IMatch{}
540
541def size(n: Node) -> U32:
542 match n:
543 case NEmpty{}:
544 0
545 case NSet{neg, items}:
546 1
547 case NAssert{k}:
548 1
549 case NCat{a, b}:
550 (size(a) + size(b) : U32)
551 case NAlt{a, b}:
552 (2 + size(a) + size(b) : U32)
553 case NPlus{g, a}:
554 (1 + size(a) : U32)
555 case NQuest{g, a}:
556 (1 + size(a) : U32)
557 case NGroup{i, a}:
558 (2 + size(a) : U32)
559
560# The instructions of n, placed at pc, in front of rest. ISplit tries x first.
561def emit(n: Node, +pc: U32, rest: List<&2, Inst>) -> List<&2, Inst>:
562 match n:
563 case NEmpty{}:
564 rest
565 case NSet{neg, items}:
566 ISet{neg, items} <> rest
567 case NAssert{k}:
568 IAssert{k} <> rest
569 case NCat{+a, b}:
570 emit(a, pc, emit(b, (pc + size(a) : U32), rest))
571 case NAlt{+a, +b}:
572 +j = (pc + 1 + size(a) : U32)
573 ISplit{(pc + 1 : U32), (j + 1 : U32)} <> emit(a, (pc + 1 : U32), IJmp{(j + 1 + size(b) : U32)} <> emit(b, (j + 1 : U32), rest))
574 case NPlus{+g, +a}:
575 +out = (pc + 1 + size(a) : U32)
576 emit(a, pc, ISplit{Bool.pick(U32, g, pc, out), Bool.pick(U32, g, out, pc)} <> rest)
577 case NQuest{+g, +a}:
578 +out = (pc + 1 + size(a) : U32)
579 ISplit{Bool.pick(U32, g, (pc + 1 : U32), out), Bool.pick(U32, g, out, (pc + 1 : U32))} <> emit(a, (pc + 1 : U32), rest)
580 case NGroup{+i, a}:
581 ISave{(2 * i : U32)} <> emit(a, (pc + 1 : U32), ISave{(2 * i + 1 : U32)} <> rest)
582
583# A binary trie keyed by n >= 1: the path is the bits of n below its top bit, low bit
584# first, so key n costs log2(n) steps and small keys stay near the root.
585type Trie<-V: Data> is Data:
586 TTip{}
587 TNode{val: Maybe<&2, V>, lo: Trie<V>, hi: Trie<V>}
588
589# here: n is 1. left: n is even.
590def trie.get(-V: Data, t: Trie<V>, here: Bool, left: Bool, +n: U32) -> Maybe<&2, V>:
591 match t:
592 case TTip{}:
593 None{}
594 case TNode{v, lo, hi}:
595 match here:
596 case True{}:
597 v
598 case False{}:
599 match left:
600 case True{}:
601 trie.get(V, lo, U32.is_eq(U32.shr(n), 1), U32.is_even(U32.shr(n)), U32.shr(n))
602 case False{}:
603 trie.get(V, hi, U32.is_eq(U32.shr(n), 1), U32.is_even(U32.shr(n)), U32.shr(n))
604
605# fuel: a U32 key has at most 32 bits.
606def trie.put(-V: Data, fuel: Nat, t: Trie<V>, here: Bool, left: Bool, +n: U32, v: V) -> Trie<V>:
607 match fuel:
608 case 0n:
609 t
610 case 1n+f:
611 match t:
612 case TNode{x, lo, hi}:
613 match here:
614 case True{}:
615 TNode{Some{v}, lo, hi}
616 case False{}:
617 match left:
618 case True{}:
619 TNode{x, trie.put(V, f, lo, U32.is_eq(U32.shr(n), 1), U32.is_even(U32.shr(n)), U32.shr(n), v), hi}
620 case False{}:
621 TNode{x, lo, trie.put(V, f, hi, U32.is_eq(U32.shr(n), 1), U32.is_even(U32.shr(n)), U32.shr(n), v)}
622 case TTip{}:
623 match here:
624 case True{}:
625 TNode{Some{v}, TTip{}, TTip{}}
626 case False{}:
627 match left:
628 case True{}:
629 TNode{None{}, trie.put(V, f, TTip{}, U32.is_eq(U32.shr(n), 1), U32.is_even(U32.shr(n)), U32.shr(n), v), TTip{}}
630 case False{}:
631 TNode{None{}, TTip{}, trie.put(V, f, TTip{}, U32.is_eq(U32.shr(n), 1), U32.is_even(U32.shr(n)), U32.shr(n), v)}
632
633# The value at key k, stored as n = k + 1.
634def trie.at(-V: Data, t: Trie<V>, +k: U32) -> Maybe<&2, V>:
635 +n = (k + 1 : U32)
636 trie.get(V, t, U32.is_eq(n, 1), U32.is_even(n), n)
637
638def trie.set(-V: Data, t: Trie<V>, +k: U32, v: V) -> Trie<V>:
639 +n = (k + 1 : U32)
640 trie.put(V, 32n, t, U32.is_eq(n, 1), U32.is_even(n), n, v)
641
642def trie.from(-V: Data, xs: List<&2, V>, +k: U32, t: Trie<V>) -> Trie<V>:
643 match xs:
644 case Nil{}:
645 t
646 case Con{h, r}:
647 trie.from(V, r, (k + 1 : U32), trie.set(V, t, k, h))
648
649# The walk from pc 0 over the instructions that consume no char: the sets it reaches, or
650# FsAny when it reaches an assertion or a match, so that any char may start a match.
651type Fs is Data:
652 Fs{stack: List<&2, U32>, seen: List<&2, U32>, sets: List<&2, Inst>}
653 FsAny{}
654
655def first.inst(i: Maybe<&2, Inst>, +pc: U32, stack: List<&2, U32>, seen: List<&2, U32>, sets: List<&2, Inst>) -> Fs:
656 match i:
657 case Some{IJmp{x}}:
658 Fs{x <> stack, seen, sets}
659 case Some{ISplit{x, y}}:
660 Fs{x <> y <> stack, seen, sets}
661 case Some{ISave{k}}:
662 Fs{(pc + 1 : U32) <> stack, seen, sets}
663 case Some{ISet{neg, items}}:
664 Fs{stack, seen, ISet{neg, items} <> sets}
665 case _:
666 FsAny{}
667
668def first.seen(hit: Bool, +prog: List<&2, Inst>, +pc: U32, stack: List<&2, U32>, seen: List<&2, U32>, sets: List<&2, Inst>) -> Fs:
669 match hit:
670 case True{}:
671 Fs{stack, seen, sets}
672 case False{}:
673 first.inst(List.get(&2, Inst, prog, U32.to_nat(pc)), pc, stack, pc <> seen, sets)
674
675# fuel: each pc expands once and pushes at most two pcs.
676def first(fuel: Nat, +prog: List<&2, Inst>, st: Fs) -> Maybe<&2, List<&2, Inst>>:
677 match fuel:
678 case 0n:
679 None{}
680 case 1n+f:
681 match st:
682 case FsAny{}:
683 None{}
684 case Fs{Nil{}, seen, sets}:
685 Some{sets}
686 case Fs{Con{+pc, t}, +seen, sets}:
687 first(f, prog, first.seen(List.contains(~U32, ~U32.is_eq, seen, pc), prog, pc, t, seen, sets))
688
689# Bit-parallel NFA for is_match: bit pc stands for the set at pc, waiting on the next char.
690# The epsilon walk from a pc: the sets it reaches and whether it reaches the match, where at0
691# says the position is the start of the text and at1 the end.
692type Ew is Data:
693 Ew{stack: List<&2, U32>, seen: List<&2, U32>, mask: U32, hit: Bool}
694
695def eps.ctx(k: U32, +at0: Bool, +at1: Bool) -> Bool:
696 match k:
697 case 0:
698 at0
699 case 1:
700 at1
701 case _:
702 False{}
703
704def eps.assert(ok: Bool, +pc: U32, stack: List<&2, U32>, seen: List<&2, U32>, mask: U32, hit: Bool) -> Ew:
705 match ok:
706 case True{}:
707 Ew{(pc + 1 : U32) <> stack, seen, mask, hit}
708 case False{}:
709 Ew{stack, seen, mask, hit}
710
711def eps.inst(i: Maybe<&2, Inst>, +pc: U32, +at0: Bool, +at1: Bool, stack: List<&2, U32>, seen: List<&2, U32>, +mask: U32, hit: Bool) -> Ew:
712 match i:
713 case Some{IJmp{x}}:
714 Ew{x <> stack, seen, mask, hit}
715 case Some{ISplit{x, y}}:
716 Ew{x <> y <> stack, seen, mask, hit}
717 case Some{ISave{k}}:
718 Ew{(pc + 1 : U32) <> stack, seen, mask, hit}
719 case Some{IAssert{k}}:
720 eps.assert(eps.ctx(k, at0, at1), pc, stack, seen, mask, hit)
721 case Some{ISet{neg, items}}:
722 Ew{stack, seen, (mask .|. (1 << U32.to_nat(pc)) : U32), hit}
723 case Some{IMatch{}}:
724 Ew{stack, seen, mask, True{}}
725 case None{}:
726 Ew{stack, seen, mask, hit}
727
728def eps.seen(hit: Bool, +prog: List<&2, Inst>, +pc: U32, +at0: Bool, +at1: Bool, w: Ew) -> Ew:
729 match hit:
730 case True{}:
731 w
732 case False{}:
733 Ew{stack, seen, mask, h} = w
734 eps.inst(List.get(&2, Inst, prog, U32.to_nat(pc)), pc, at0, at1, stack, pc <> seen, mask, h)
735
736# fuel: each pc expands once and pushes at most two pcs.
737def eps.go(fuel: Nat, +prog: List<&2, Inst>, +at0: Bool, +at1: Bool, w: Ew) -> Ew:
738 match fuel:
739 case 0n:
740 w
741 case 1n+f:
742 match w:
743 case Ew{Nil{}, seen, mask, hit}:
744 Ew{Nil{}, seen, mask, hit}
745 case Ew{Con{+pc, t}, +seen, mask, hit}:
746 eps.go(f, prog, at0, at1, eps.seen(List.contains(~U32, ~U32.is_eq, seen, pc), prog, pc, at0, at1, Ew{t, seen, mask, hit}))
747
748def eps(+fuel: Nat, +prog: List<&2, Inst>, +pc: U32, +at0: Bool, +at1: Bool) -> Ew:
749 eps.go(fuel, prog, at0, at1, Ew{[pc], Nil{}, 0, False{}})
750
751def Ew.mask(w: Ew) -> U32:
752 Ew{s, v, mask, hit} = w
753 mask
754
755def Ew.hit(w: Ew) -> Bool:
756 Ew{s, v, mask, hit} = w
757 hit
758
759# The set at bit mask; follow: the sets live after it takes a char; fin, end: whether the match
760# is then reached inside the text, or at its end.
761type Bit is Data:
762 Bit{mask: U32, neg: Bool, items: List<&2, Item>, follow: U32, fin: Bool, end: Bool}
763
764# at0: the sets live at the start of the text, and hit0 whether the empty match is there;
765# hit01 the same for the empty text; atn and hitn the same inside the text, hit1 at its end.
766type Bits is Data:
767 Bits{sets: List<&2, Bit>, at0: U32, hit0: Bool, hit01: Bool, atn: U32, hit1: Bool}
768
769def bits.set(+fuel: Nat, +prog: List<&2, Inst>, +pc: U32, neg: Bool, items: List<&2, Item>) -> Bit:
770 +w = eps(fuel, prog, (pc + 1 : U32), False{}, False{})
771 Bit{(1 << U32.to_nat(pc) : U32), neg, items, Ew.mask(w), Ew.hit(w), Ew.hit(eps(fuel, prog, (pc + 1 : U32), False{}, True{}))}
772
773def bits.sets(xs: List<&2, Inst>, +fuel: Nat, +prog: List<&2, Inst>, +pc: U32) -> List<&2, Bit>:
774 match xs:
775 case Nil{}:
776 Nil{}
777 case Con{ISet{neg, items}, t}:
778 bits.set(fuel, prog, pc, neg, items) <> bits.sets(t, fuel, prog, (pc + 1 : U32))
779 case Con{i, t}:
780 bits.sets(t, fuel, prog, (pc + 1 : U32))
781
782# Word boundaries depend on the chars on both sides, which one mask per set cannot track.
783def bits.plain(xs: List<&2, Inst>) -> Bool:
784 match xs:
785 case Nil{}:
786 True{}
787 case Con{IAssert{+k}, t}:
788 U32.is_lt(k, 2) && bits.plain(t)
789 case Con{i, t}:
790 bits.plain(t)
791
792def bits.if(ok: Bool, +fuel: Nat, +prog: List<&2, Inst>) -> Maybe<&2, Bits>:
793 match ok:
794 case False{}:
795 None{}
796 case True{}:
797 +w0 = eps(fuel, prog, 0, True{}, False{})
798 +wn = eps(fuel, prog, 0, False{}, False{})
799 Some{Bits{bits.sets(prog, fuel, prog, 0), Ew.mask(w0), Ew.hit(w0), Ew.hit(eps(fuel, prog, 0, True{}, True{})), Ew.mask(wn), Ew.hit(eps(fuel, prog, 0, False{}, True{}))}}
800
801def bits(+fuel: Nat, +prog: List<&2, Inst>) -> Maybe<&2, Bits>:
802 bits.if(Nat.is_le(List.length(&2, Inst, prog), 32n) && bits.plain(prog), fuel, prog)
803
804def item.has(i: Item, +c: U32) -> Bool:
805 match i:
806 case Rng{+lo, +hi}:
807 U32.is_le(lo, c) && U32.is_le(c, hi)
808 case Prop{neg, name}:
809 Bool.xor(neg, String.starts_with(U.category(Chr{c}), name))
810
811def items.has(xs: List<&2, Item>, +c: U32) -> Bool:
812 match xs:
813 case Nil{}:
814 False{}
815 case Con{h, t}:
816 item.has(h, c) || items.has(t, c)
817
818def starts(xs: List<&2, Inst>, +c: U32) -> Bool:
819 match xs:
820 case Nil{}:
821 False{}
822 case Con{ISet{neg, set}, t}:
823 Bool.xor(neg, items.has(set, c)) || starts(t, c)
824 case Con{i, t}:
825 starts(t, c)
826
827# The ASCII bytes that may start a match, one bit per byte, 32 to a word.
828type Asc is Data:
829 Asc{m0: U32, m1: U32, m2: U32, m3: U32}
830
831def asc.bit(hit: Bool, +c: U32, a: Asc) -> Asc:
832 match hit:
833 case False{}:
834 a
835 case True{}:
836 Asc{+m0, +m1, +m2, +m3} = a
837 +b = (1 << U32.to_nat((c .&. 31 : U32)) : U32)
838 Asc{Bool.pick(U32, U32.is_eq((c >> 5n : U32), 0), (m0 .|. b : U32), m0), Bool.pick(U32, U32.is_eq((c >> 5n : U32), 1), (m1 .|. b : U32), m1), Bool.pick(U32, U32.is_eq((c >> 5n : U32), 2), (m2 .|. b : U32), m2), Bool.pick(U32, U32.is_eq((c >> 5n : U32), 3), (m3 .|. b : U32), m3)}
839
840# c: the byte for step n, counting down from 127 to 0.
841def asc.go(n: Nat, +c: U32, +sets: List<&2, Inst>, a: Asc) -> Asc:
842 match n:
843 case 0n:
844 a
845 case 1n+p:
846 asc.go(p, (c - 1 : U32), sets, asc.bit(starts(sets, c), c, a))
847
848def asc(start: Maybe<&2, List<&2, Inst>>) -> Maybe<&2, Asc>:
849 match start:
850 case None{}:
851 None{}
852 case Some{sets}:
853 Some{asc.go(128n, 127, sets, Asc{0, 0, 0, 0})}
854
855# prog: the instructions by pc; fuel: enough steps for one closure; slots: two per group, group 0 included;
856# start: the sets one of which the first char of a match is in, or None when a match may start anywhere;
857# bits: the bit-parallel form, for programs of at most 32 instructions without word boundaries;
858# asc: start as a table of ASCII bytes, for skipping through Bytes.
859type Regex is Data:
860 Regex{prog: Trie<Inst>, fuel: Nat, slots: Nat, start: Maybe<&2, List<&2, Inst>>, bits: Maybe<&2, Bits>, asc: Maybe<&2, Asc>}
861
862def build(+node: Node, +n: U32) -> Regex:
863 +insts = {ISave{0} <> emit(node, 1, [ISave{1}, IMatch{}]) : List<&2, Inst>}
864 +fuel = Nat.add(Nat.mul(3n, List.length(&2, Inst, insts)), 2n)
865 +start = first(fuel, insts, Fs{[0], Nil{}, Nil{}})
866 Regex{trie.from(Inst, insts, 0, TTip{}), fuel, U32.to_nat((2 * n : U32)), start, bits(fuel, insts), asc(start)}
867
868def compile.fin(st: Ps) -> Maybe<&2, Regex>:
869 match st:
870 case PsFail{}:
871 None{}
872 case Ps{n, rep, stack, alts, cur}:
873 match stack:
874 case Nil{}:
875 Some{build(alt.go(alts, seq(cur)), n)}
876 case Con{f, up}:
877 None{}
878
879def compile.parse(toks: Maybe<&2, List<&2, Tok>>) -> Maybe<&2, Regex>:
880 match toks:
881 case None{}:
882 None{}
883 case Some{ts}:
884 compile.fin(parse(ts, Ps{1, False{}, Nil{}, Nil{}, Nil{}}))
885
886# The pattern, or None for a syntax error.
887def compile(+pat: String) -> Maybe<&2, Regex>:
888 compile.parse(lex(Nat.add(String.length(pat), 2n), LGo{pat, Nil{}}))
889
890# Matcher
891# -------
892
893def word(m: Maybe<&2, Char>) -> Bool:
894 match m:
895 case None{}:
896 False{}
897 case Some{Chr{+c}}:
898 is_word(c)
899
900def assert.ok(k: U32, +prev: Maybe<&2, Char>, +next: Maybe<&2, Char>) -> Bool:
901 match k:
902 case 0:
903 Maybe.is_none(&2, Char, prev)
904 case 1:
905 Maybe.is_none(&2, Char, next)
906 case 2:
907 Bool.xor(word(prev), word(next))
908 case _:
909 Bool.not(Bool.xor(word(prev), word(next)))
910
911# A thread: its pc and its capture slots.
912type Th is Data:
913 Th{pc: U32, caps: List<&2, Maybe<&2, U32>>}
914
915# The epsilon closure as a depth-first walk: stack is the work left, seen the pcs
916# visited at this position, out the threads that wait on a char or match (reversed).
917type Cl is Data:
918 Cl{stack: List<&2, Th>, seen: Trie<Unit>, out: List<&2, Th>}
919
920def close.assert(ok: Bool, +pc: U32, caps: List<&2, Maybe<&2, U32>>, stack: List<&2, Th>, seen: Trie<Unit>, out: List<&2, Th>) -> Cl:
921 match ok:
922 case True{}:
923 Cl{Th{(pc + 1 : U32), caps} <> stack, seen, out}
924 case False{}:
925 Cl{stack, seen, out}
926
927def close.inst(i: Maybe<&2, Inst>, +pc: U32, +caps: List<&2, Maybe<&2, U32>>, stack: List<&2, Th>, seen: Trie<Unit>, out: List<&2, Th>, +pos: U32, +prev: Maybe<&2, Char>, +next: Maybe<&2, Char>) -> Cl:
928 match i:
929 case None{}:
930 Cl{stack, seen, out}
931 case Some{IJmp{x}}:
932 Cl{Th{x, caps} <> stack, seen, out}
933 case Some{ISplit{x, y}}:
934 Cl{Th{x, caps} <> Th{y, caps} <> stack, seen, out}
935 case Some{ISave{k}}:
936 Cl{Th{(pc + 1 : U32), List.set(&2, Maybe<&2, U32>, caps, U32.to_nat(k), Some{pos})} <> stack, seen, out}
937 case Some{IAssert{k}}:
938 close.assert(assert.ok(k, prev, next), pc, caps, stack, seen, out)
939 case Some{ISet{neg, items}}:
940 Cl{stack, seen, Th{pc, caps} <> out}
941 case Some{IMatch{}}:
942 Cl{stack, seen, Th{pc, caps} <> out}
943
944# ponytail: trie lookup and membership cost O(log m) per thread step, so a char costs
945# O(m log m) for m instructions; a sparse set over an Array would make a step O(1).
946def close.seen(hit: Bool, +prog: Trie<Inst>, +pc: U32, caps: List<&2, Maybe<&2, U32>>, stack: List<&2, Th>, seen: Trie<Unit>, out: List<&2, Th>, +pos: U32, +prev: Maybe<&2, Char>, +next: Maybe<&2, Char>) -> Cl:
947 match hit:
948 case True{}:
949 Cl{stack, seen, out}
950 case False{}:
951 close.inst(trie.at(Inst, prog, pc), pc, caps, stack, trie.set(Unit, seen, pc, Unit{}), out, pos, prev, next)
952
953def close.step(th: Th, stack: List<&2, Th>, +seen: Trie<Unit>, out: List<&2, Th>, +prog: Trie<Inst>, +pos: U32, +prev: Maybe<&2, Char>, +next: Maybe<&2, Char>) -> Cl:
954 Th{+pc, caps} = th
955 close.seen(Maybe.is_some(&2, Unit, trie.at(Unit, seen, pc)), prog, pc, caps, stack, seen, out, pos, prev, next)
956
957# fuel: each pc expands once and pushes at most two threads.
958def close(fuel: Nat, +prog: Trie<Inst>, +pos: U32, +prev: Maybe<&2, Char>, +next: Maybe<&2, Char>, st: Cl) -> List<&2, Th>:
959 match fuel:
960 case 0n:
961 Cl{stack, seen, out} = st
962 List.reverse(&2, Th, out)
963 case 1n+f:
964 match st:
965 case Cl{Nil{}, seen, out}:
966 List.reverse(&2, Th, out)
967 case Cl{Con{th, t}, seen, out}:
968 close(f, prog, pos, prev, next, close.step(th, t, seen, out, prog, pos, prev, next))
969
970# A scan of the threads at one char, in priority order: items are the threads that
971# step past it (reversed), best the latest match. A match drops every later thread.
972type Sc is Data:
973 Sc{items: List<&2, Th>, best: Maybe<&2, List<&2, Maybe<&2, U32>>>, stop: Bool}
974
975def scan.set(hit: Bool, +pc: U32, caps: List<&2, Maybe<&2, U32>>, items: List<&2, Th>, best: Maybe<&2, List<&2, Maybe<&2, U32>>>) -> Sc:
976 match hit:
977 case True{}:
978 Sc{Th{(pc + 1 : U32), caps} <> items, best, False{}}
979 case False{}:
980 Sc{items, best, False{}}
981
982def scan.char(c: Maybe<&2, Char>, neg: Bool, set: List<&2, Item>, +pc: U32, caps: List<&2, Maybe<&2, U32>>, items: List<&2, Th>, best: Maybe<&2, List<&2, Maybe<&2, U32>>>) -> Sc:
983 match c:
984 case None{}:
985 Sc{items, best, False{}}
986 case Some{Chr{+x}}:
987 scan.set(Bool.xor(neg, items.has(set, x)), pc, caps, items, best)
988
989# any: only whether a match exists counts, so a match also drops the earlier threads.
990def scan.hit(any: Bool, caps: List<&2, Maybe<&2, U32>>, items: List<&2, Th>) -> Sc:
991 match any:
992 case True{}:
993 Sc{Nil{}, Some{caps}, True{}}
994 case False{}:
995 Sc{items, Some{caps}, True{}}
996
997def scan.inst(i: Maybe<&2, Inst>, c: Maybe<&2, Char>, +any: Bool, +pc: U32, caps: List<&2, Maybe<&2, U32>>, items: List<&2, Th>, best: Maybe<&2, List<&2, Maybe<&2, U32>>>) -> Sc:
998 match i:
999 case Some{IMatch{}}:
1000 scan.hit(any, caps, items)
1001 case Some{ISet{neg, set}}:
1002 scan.char(c, neg, set, pc, caps, items, best)
1003 case _:
1004 Sc{items, best, False{}}
1005
1006def scan.go(stop: Bool, th: Th, +prog: Trie<Inst>, c: Maybe<&2, Char>, +any: Bool, items: List<&2, Th>, best: Maybe<&2, List<&2, Maybe<&2, U32>>>) -> Sc:
1007 match stop:
1008 case True{}:
1009 Sc{items, best, True{}}
1010 case False{}:
1011 Th{+pc, caps} = th
1012 scan.inst(trie.at(Inst, prog, pc), c, any, pc, caps, items, best)
1013
1014def scan.step(st: Sc, th: Th, +prog: Trie<Inst>, c: Maybe<&2, Char>, +any: Bool) -> Sc:
1015 Sc{items, best, stop} = st
1016 scan.go(stop, th, prog, c, any, items, best)
1017
1018def scan(xs: List<&2, Th>, +prog: Trie<Inst>, +c: Maybe<&2, Char>, +any: Bool, st: Sc) -> Sc:
1019 match xs:
1020 case Nil{}:
1021 st
1022 case Con{th, t}:
1023 scan(t, prog, c, any, scan.step(st, th, prog, c, any))
1024
1025# The live threads and the best match so far.
1026type Vm is Data:
1027 Vm{ths: List<&2, Th>, best: Maybe<&2, List<&2, Maybe<&2, U32>>>}
1028
1029# Until a match is found, a new thread starts at each position, below every other.
1030def seed(best: Maybe<&2, List<&2, Maybe<&2, U32>>>, xs: List<&2, Th>, init: List<&2, Maybe<&2, U32>>) -> List<&2, Th>:
1031 match best:
1032 case None{}:
1033 List.append(&2, Th, xs, [Th{0, init}])
1034 case Some{b}:
1035 xs
1036
1037# No live thread and no match yet: the seed is the only thread, and it dies unless next is in start.
1038def seed.skip(items: List<&2, Th>, best: Maybe<&2, List<&2, Maybe<&2, U32>>>, start: Maybe<&2, List<&2, Inst>>, next: Maybe<&2, Char>) -> Bool:
1039 match items best start next:
1040 case Nil{} None{} Some{sets} None{}:
1041 True{}
1042 case Nil{} None{} Some{sets} Some{Chr{+c}}:
1043 Bool.not(starts(sets, c))
1044 case _ _ _ _:
1045 False{}
1046
1047def run.seed(skip: Bool, items: List<&2, Th>, +best: Maybe<&2, List<&2, Maybe<&2, U32>>>, +prog: Trie<Inst>, +fuel: Nat, +init: List<&2, Maybe<&2, U32>>, +pos: U32, +prev: Maybe<&2, Char>, +next: Maybe<&2, Char>) -> Vm:
1048 match skip:
1049 case True{}:
1050 Vm{Nil{}, None{}}
1051 case False{}:
1052 Vm{close(fuel, prog, pos, prev, next, Cl{seed(best, List.reverse(&2, Th, items), init), TTip{}, Nil{}}), best}
1053
1054def run.close(sc: Sc, +prog: Trie<Inst>, +fuel: Nat, +init: List<&2, Maybe<&2, U32>>, +start: Maybe<&2, List<&2, Inst>>, +pos: U32, +prev: Maybe<&2, Char>, +next: Maybe<&2, Char>) -> Vm:
1055 Sc{+items, +best, stop} = sc
1056 run.seed(seed.skip(items, best, start, next), items, best, prog, fuel, init, pos, prev, next)
1057
1058def run.step(st: Vm, +prog: Trie<Inst>, +fuel: Nat, +init: List<&2, Maybe<&2, U32>>, +start: Maybe<&2, List<&2, Inst>>, +any: Bool, +pos: U32, +c: Char, +next: Maybe<&2, Char>) -> Vm:
1059 Vm{ths, best} = st
1060 run.close(scan(ths, prog, Some{c}, any, Sc{Nil{}, best, False{}}), prog, fuel, init, start, pos, Some{c}, next)
1061
1062def run.end(sc: Sc) -> Maybe<&2, List<&2, Maybe<&2, U32>>>:
1063 Sc{items, best, stop} = sc
1064 best
1065
1066# pos: the position of s's head. Once a match exists and no thread is live, the rest of s cannot change it.
1067def run(s: String, +prog: Trie<Inst>, +fuel: Nat, +init: List<&2, Maybe<&2, U32>>, +start: Maybe<&2, List<&2, Inst>>, +any: Bool, +pos: U32, st: Vm) -> Maybe<&2, List<&2, Maybe<&2, U32>>>:
1068 match s:
1069 case SNil{}:
1070 Vm{ths, best} = st
1071 run.end(scan(ths, prog, None{}, any, Sc{Nil{}, best, False{}}))
1072 case SCon{+c, +t}:
1073 match st:
1074 case Vm{Nil{}, Some{b}}:
1075 Some{b}
1076 case Vm{ths, best}:
1077 +p = (pos + 1 : U32)
1078 run(t, prog, fuel, init, start, any, p, run.step(Vm{ths, best}, prog, fuel, init, start, any, p, c, String.get(t, 0n)))
1079
1080type Span is Data:
1081 Span{start: U32, end: U32}
1082
1083def spans(caps: List<&2, Maybe<&2, U32>>) -> List<&2, Maybe<&2, Span>>:
1084 match caps:
1085 case Con{Some{a}, Con{Some{b}, t}}:
1086 Some{Span{a, b}} <> spans(t)
1087 case Con{x, Con{y, t}}:
1088 None{} <> spans(t)
1089 case _:
1090 Nil{}
1091
1092def find.spans(m: Maybe<&2, List<&2, Maybe<&2, U32>>>) -> Maybe<&2, List<&2, Maybe<&2, Span>>>:
1093 match m:
1094 case None{}:
1095 None{}
1096 case Some{caps}:
1097 Some{spans(caps)}
1098
1099# slots: capture slots to keep; is_match keeps none, so each ISave is free.
1100def exec(re: Regex, +s: String, +any: Bool) -> Maybe<&2, List<&2, Maybe<&2, U32>>>:
1101 Regex{+prog, +fuel, +slots, +start, bits, asc} = re
1102 +init = List.replicate(Maybe<&2, U32>, Bool.pick(Nat, any, 0n, slots), None{})
1103 run(s, prog, fuel, init, start, any, 0, run.close(Sc{Nil{}, None{}, False{}}, prog, fuel, init, start, 0, None{}, String.get(s, 0n)))
1104
1105# The leftmost match in s: the span of group 0, then of each group; None for a group that did not take part.
1106def find(re: Regex, +s: String) -> Maybe<&2, List<&2, Maybe<&2, Span>>>:
1107 find.spans(exec(re, s, False{}))
1108
1109# The Pike VM form of is_match: stops at the first match of any priority and records no captures.
1110def is_match.vm(re: Regex, +s: String) -> Bool:
1111 Maybe.is_some(&2, List<&2, Maybe<&2, U32>>, exec(re, s, True{}))
1112
1113# One char through the bit NFA: next the sets live after it, fin and end as in Bit.
1114type Bs is Data:
1115 Bs{next: U32, fin: Bool, end: Bool}
1116
1117def bits.take(hit: Bool, +follow: U32, +fin: Bool, +end: Bool, acc: Bs) -> Bs:
1118 match hit:
1119 case False{}:
1120 acc
1121 case True{}:
1122 Bs{+nx, f, e} = acc
1123 Bs{(nx .|. follow : U32), f || fin, e || end}
1124
1125def bits.char(live: Bool, +c: U32, neg: Bool, items: List<&2, Item>, +follow: U32, +fin: Bool, +end: Bool, acc: Bs) -> Bs:
1126 match live:
1127 case False{}:
1128 acc
1129 case True{}:
1130 bits.take(Bool.xor(neg, items.has(items, c)), follow, fin, end, acc)
1131
1132# ponytail: one test per set in the program, not a table lookup; use per-byte tables if sets grow many.
1133def bits.step(xs: List<&2, Bit>, +cur: U32, +c: U32, acc: Bs) -> Bs:
1134 match xs:
1135 case Nil{}:
1136 acc
1137 case Con{Bit{+mask, neg, items, follow, fin, end}, t}:
1138 bits.step(t, cur, c, bits.char(U32.is_ne((cur .&. mask : U32), 0), c, neg, items, follow, fin, end, acc))
1139
1140# Only the sets live at a fresh start wait on c, and c starts none of them: nothing moves.
1141def bits.idle(start: Maybe<&2, List<&2, Inst>>, +cur: U32, +atn: U32, +c: U32) -> Bool:
1142 match start:
1143 case None{}:
1144 False{}
1145 case Some{sets}:
1146 U32.is_eq(cur, atn) && Bool.not(starts(sets, c))
1147
1148# hit1: a match starts and ends at the end of the text, so end starts from it.
1149def bits.go(idle: Bool, +sets: List<&2, Bit>, +atn: U32, +hit1: Bool, +cur: U32, +c: U32) -> Bs:
1150 match idle:
1151 case True{}:
1152 Bs{atn, False{}, hit1}
1153 case False{}:
1154 bits.step(sets, cur, c, Bs{atn, False{}, hit1})
1155
1156# r: the sets live at s's head, whether a match was found, and whether one ends at the end of
1157# the text, if the text ends here. Inside the text an assertion can only fail, so fin implies end,
1158# and a match at fin is final.
1159def bits.run(s: String, +sets: List<&2, Bit>, +start: Maybe<&2, List<&2, Inst>>, +atn: U32, +hit1: Bool, r: Bs) -> Bool:
1160 match s:
1161 case SNil{}:
1162 Bs{cur, done, ok} = r
1163 done || ok
1164 case SCon{Chr{+c}, t}:
1165 match r:
1166 case Bs{cur, True{}, ok}:
1167 True{}
1168 case Bs{+cur, False{}, ok}:
1169 bits.run(t, sets, start, atn, hit1, bits.go(bits.idle(start, cur, atn, c), sets, atn, hit1, cur, c))
1170
1171def bits.is_match(b: Bits, start: Maybe<&2, List<&2, Inst>>, s: String) -> Bool:
1172 Bits{sets, at0, hit0, hit01, atn, hit1} = b
1173 bits.run(s, sets, start, atn, hit1, Bs{at0, hit0, hit01})
1174
1175# The bit NFA form of is_match, or None when the program is too large or has a word boundary.
1176def is_match.bits(re: Regex, s: String) -> Maybe<&2, Bool>:
1177 Regex{prog, fuel, slots, start, bits, asc} = re
1178 match bits:
1179 case None{}:
1180 None{}
1181 case Some{b}:
1182 Some{bits.is_match(b, start, s)}
1183
1184def is_match.pick(m: Maybe<&2, Bool>, re: Regex, +s: String) -> Bool:
1185 match m:
1186 case Some{x}:
1187 x
1188 case None{}:
1189 is_match.vm(re, s)
1190
1191# Whether s has a match; the bit NFA runs when the program allows it, else the Pike VM.
1192def is_match(+re: Regex, +s: String) -> Bool:
1193 is_match.pick(is_match.bits(re, s), re, s)
1194
1195# Bytes
1196# -----
1197# The same matchers over a UTF-8 Bytes buffer; positions are byte offsets, as in RE2 and Go.
1198
1199# A code point decoded from UTF-8 and its width in bytes, or DEnd at the end of the buffer.
1200# An invalid byte is U+FFFD of width 1, as in RE2.
1201type Dc is Data:
1202 Dc{c: U32, w: U32}
1203 DEnd{}
1204
1205def Dc.char(d: Dc) -> Maybe<&2, Char>:
1206 match d:
1207 case DEnd{}:
1208 None{}
1209 case Dc{c, w}:
1210 Some{Chr{c}}
1211
1212def pk.if(ok: Bool, a: Array<U32>, +i: U32) -> Array<U32> & U32:
1213 match ok:
1214 case True{}:
1215 Bytes.peek(a, i)
1216 case False{}:
1217 (a, 0)
1218
1219# Byte i, or 0 past len, which is no continuation byte.
1220def pk(a: Array<U32>, +len: U32, +i: U32) -> Array<U32> & U32:
1221 pk.if(U32.is_lt(i, len), a, i)
1222
1223def cont(+b: U32) -> Bool:
1224 U32.is_eq((b .&. 192 : U32), 128)
1225
1226def dec.pick(ok2: Bool, ok3: Bool, ok4: Bool, +c2: U32, +c3: U32, +c4: U32) -> Dc:
1227 match ok2:
1228 case True{}:
1229 Dc{c2, 2}
1230 case False{}:
1231 match ok3:
1232 case True{}:
1233 Dc{c3, 3}
1234 case False{}:
1235 match ok4:
1236 case True{}:
1237 Dc{c4, 4}
1238 case False{}:
1239 Dc{65533, 1}
1240
1241# Overlong forms, surrogates, and code points past U+10FFFF are invalid.
1242def dec.n(+b0: U32, +b1: U32, +b2: U32, +b3: U32) -> Dc:
1243 +x1 = (b1 .&. 63 : U32)
1244 +x2 = (b2 .&. 63 : U32)
1245 +x3 = (b3 .&. 63 : U32)
1246 +c2 = (((b0 .&. 31) << 6n) .|. x1 : U32)
1247 +c3 = ((((b0 .&. 15) << 12n) .|. (x1 << 6n)) .|. x2 : U32)
1248 +c4 = (((((b0 .&. 7) << 18n) .|. (x1 << 12n)) .|. (x2 << 6n)) .|. x3 : U32)
1249 +ok2 = U32.is_le(192, b0) && U32.is_lt(b0, 224) && cont(b1) && U32.is_le(128, c2)
1250 +ok3 = U32.is_le(224, b0) && U32.is_lt(b0, 240) && cont(b1) && cont(b2) && U32.is_le(2048, c3) && (U32.is_lt(c3, 55296) || U32.is_lt(57343, c3))
1251 +ok4 = U32.is_le(240, b0) && U32.is_lt(b0, 248) && cont(b1) && cont(b2) && cont(b3) && U32.is_le(65536, c4) && U32.is_le(c4, 1114111)
1252 dec.pick(ok2, ok3, ok4, c2, c3, c4)
1253
1254def dec.b3(+b0: U32, +b1: U32, +b2: U32, r: Array<U32> & U32) -> Array<U32> & Dc:
1255 (a, +b3) = r
1256 (a, dec.n(b0, b1, b2, b3))
1257
1258def dec.b2(+len: U32, +i: U32, +b0: U32, +b1: U32, r: Array<U32> & U32) -> Array<U32> & Dc:
1259 (a, +b2) = r
1260 dec.b3(b0, b1, b2, pk(a, len, (i + 3 : U32)))
1261
1262def dec.b1(+len: U32, +i: U32, +b0: U32, r: Array<U32> & U32) -> Array<U32> & Dc:
1263 (a, +b1) = r
1264 dec.b2(len, i, b0, b1, pk(a, len, (i + 2 : U32)))
1265
1266def dec.lead(ascii: Bool, +len: U32, +i: U32, +b0: U32, a: Array<U32>) -> Array<U32> & Dc:
1267 match ascii:
1268 case True{}:
1269 (a, Dc{b0, 1})
1270 case False{}:
1271 dec.b1(len, i, b0, pk(a, len, (i + 1 : U32)))
1272
1273def dec.at(+len: U32, +i: U32, r: Array<U32> & U32) -> Array<U32> & Dc:
1274 (a, +b0) = r
1275 dec.lead(U32.is_lt(b0, 128), len, i, b0, a)
1276
1277def dec.if(ok: Bool, a: Array<U32>, +len: U32, +i: U32) -> Array<U32> & Dc:
1278 match ok:
1279 case True{}:
1280 dec.at(len, i, Bytes.peek(a, i))
1281 case False{}:
1282 (a, DEnd{})
1283
1284# The char at byte i, or DEnd at len.
1285def dec(a: Array<U32>, +len: U32, +i: U32) -> Array<U32> & Dc:
1286 dec.if(U32.is_lt(i, len), a, len, i)
1287
1288def asc.word(k: U32, a: Asc) -> U32:
1289 match k:
1290 case 0:
1291 Asc{m0, m1, m2, m3} = a
1292 m0
1293 case 1:
1294 Asc{m0, m1, m2, m3} = a
1295 m1
1296 case 2:
1297 Asc{m0, m1, m2, m3} = a
1298 m2
1299 case _:
1300 Asc{m0, m1, m2, m3} = a
1301 m3
1302
1303# Any non-ASCII byte may start a match: it may lead a char in start.
1304def asc.has(+b: U32, a: Asc) -> Bool:
1305 U32.is_le(128, b) || U32.is_ne((U32.shrn(asc.word((b >> 5n : U32), a), U32.to_nat((b .&. 31 : U32))) .&. 1 : U32), 0)
1306
1307def probe.of(+m: Asc, r: Array<U32> & U32) -> Array<U32> & Bool:
1308 (a, +v) = r
1309 (a, asc.has(v, m))
1310
1311# Past len counts as a hit, so a skip stops there.
1312def probe.if(ok: Bool, a: Array<U32>, +k: U32, +m: Asc) -> Array<U32> & Bool:
1313 match ok:
1314 case True{}:
1315 probe.of(m, Bytes.peek(a, k))
1316 case False{}:
1317 (a, True{})
1318
1319def probe(a: Array<U32>, +len: U32, +k: U32, +m: Asc) -> Array<U32> & Bool:
1320 probe.if(U32.is_lt(k, len), a, k, m)
1321
1322# The first offset at or after k whose byte may start a match, or len; r holds the probe of k.
1323def skip(fuel: Nat, r: Array<U32> & Bool, +len: U32, +k: U32, +m: Asc) -> Array<U32> & U32:
1324 match fuel:
1325 case 0n:
1326 (a, h) = r
1327 (a, k)
1328 case 1n+f:
1329 (a, hit) = r
1330 match hit:
1331 case True{}:
1332 (a, k)
1333 case False{}:
1334 skip(f, probe(a, len, (k + 1 : U32), m), len, (k + 1 : U32), m)
1335
1336def skip.from(a: Array<U32>, +len: U32, +k: U32, +m: Asc) -> Array<U32> & U32:
1337 skip(U32.to_nat((len - k : U32)), probe(a, len, k, m), len, k, m)
1338
1339# The Pike VM over Bytes: the buffer, the offset i of the next char, that char, and the threads waiting on it.
1340type Rb is Type:
1341 Rb{a: Array<U32>, i: U32, d: Dc, st: Vm}
1342
1343def rb.step(r: Array<U32> & Dc, +j: U32, st: Vm, +c: U32, +prog: Trie<Inst>, +fuel: Nat, +init: List<&2, Maybe<&2, U32>>, +start: Maybe<&2, List<&2, Inst>>, +any: Bool) -> Rb:
1344 (a, +d) = r
1345 Rb{a, j, d, run.step(st, prog, fuel, init, start, any, j, Chr{c}, Dc.char(d))}
1346
1347# A fresh seed at k. start is Some here, so no assertion is reachable before a char and prev does not count.
1348def rb.seed(r: Array<U32> & Dc, +k: U32, +prog: Trie<Inst>, +fuel: Nat, +init: List<&2, Maybe<&2, U32>>, +start: Maybe<&2, List<&2, Inst>>) -> Rb:
1349 (a, +d) = r
1350 Rb{a, k, d, run.close(Sc{Nil{}, None{}, False{}}, prog, fuel, init, start, k, None{}, Dc.char(d))}
1351
1352def rb.skip(r: Array<U32> & U32, +len: U32, +prog: Trie<Inst>, +fuel: Nat, +init: List<&2, Maybe<&2, U32>>, +start: Maybe<&2, List<&2, Inst>>) -> Rb:
1353 (a, +k) = r
1354 rb.seed(dec(a, len, k), k, prog, fuel, init, start)
1355
1356# No thread is live and no match exists. With asc, the seed at the char before j was skipped
1357# because that char cannot start a match: jump to the next byte that may. Without asc, the
1358# seed died on an assertion, so step on as usual; prev matters then.
1359def rb.idle(asc: Maybe<&2, Asc>, a: Array<U32>, +len: U32, +j: U32, +c: U32, +prog: Trie<Inst>, +fuel: Nat, +init: List<&2, Maybe<&2, U32>>, +start: Maybe<&2, List<&2, Inst>>, +any: Bool) -> Rb:
1360 match asc:
1361 case None{}:
1362 rb.step(dec(a, len, j), j, Vm{Nil{}, None{}}, c, prog, fuel, init, start, any)
1363 case Some{+m}:
1364 rb.skip(skip.from(a, len, j, m), len, prog, fuel, init, start)
1365
1366# fuel: each step eats at least one byte, and the last one sees DEnd.
1367def runb(fuel: Nat, r: Rb, +len: U32, +prog: Trie<Inst>, +cfuel: Nat, +init: List<&2, Maybe<&2, U32>>, +start: Maybe<&2, List<&2, Inst>>, +asc: Maybe<&2, Asc>, +any: Bool) -> Array<U32> & Maybe<&2, List<&2, Maybe<&2, U32>>>:
1368 match fuel:
1369 case 0n:
1370 Rb{a, i, d, st} = r
1371 (a, None{})
1372 case 1n+f:
1373 Rb{a, +i, d, st} = r
1374 match d:
1375 case DEnd{}:
1376 Vm{ths, best} = st
1377 (a, run.end(scan(ths, prog, None{}, any, Sc{Nil{}, best, False{}})))
1378 case Dc{+c, +w}:
1379 match st:
1380 case Vm{Nil{}, Some{b}}:
1381 (a, Some{b})
1382 case Vm{Nil{}, None{}}:
1383 runb(f, rb.idle(asc, a, len, (i + w : U32), c, prog, cfuel, init, start, any), len, prog, cfuel, init, start, asc, any)
1384 case Vm{ths, best}:
1385 +j = (i + w : U32)
1386 runb(f, rb.step(dec(a, len, j), j, Vm{ths, best}, c, prog, cfuel, init, start, any), len, prog, cfuel, init, start, asc, any)
1387
1388def exec.bytes.fin(+len: U32, r: Array<U32> & Maybe<&2, List<&2, Maybe<&2, U32>>>) -> Bytes.Bytes & Maybe<&2, List<&2, Maybe<&2, U32>>>:
1389 (a, m) = r
1390 (Bytes.Bytes{len, a}, m)
1391
1392def exec.bytes(re: Regex, b: Bytes.Bytes, +any: Bool) -> Bytes.Bytes & Maybe<&2, List<&2, Maybe<&2, U32>>>:
1393 Regex{+prog, +fuel, +slots, +start, bits, +asc} = re
1394 Bytes.Bytes{+len, buf} = b
1395 +init = List.replicate(Maybe<&2, U32>, Bool.pick(Nat, any, 0n, slots), None{})
1396 exec.bytes.fin(len, runb(Nat.add(U32.to_nat(len), 2n), rb.seed(dec(buf, len, 0), 0, prog, fuel, init, start), len, prog, fuel, init, start, asc, any))
1397
1398def find.bytes.fin(r: Bytes.Bytes & Maybe<&2, List<&2, Maybe<&2, U32>>>) -> Bytes.Bytes & Maybe<&2, List<&2, Maybe<&2, Span>>>:
1399 (b, m) = r
1400 (b, find.spans(m))
1401
1402# find over UTF-8 bytes: the buffer back, and spans as byte offsets.
1403def find.bytes(re: Regex, b: Bytes.Bytes) -> Bytes.Bytes & Maybe<&2, List<&2, Maybe<&2, Span>>>:
1404 find.bytes.fin(exec.bytes(re, b, False{}))
1405
1406def is_match.bytes.fin(r: Bytes.Bytes & Maybe<&2, List<&2, Maybe<&2, U32>>>) -> Bytes.Bytes & Bool:
1407 (b, m) = r
1408 (b, Maybe.is_some(&2, List<&2, Maybe<&2, U32>>, m))
1409
1410def is_match.bytes.vm(re: Regex, b: Bytes.Bytes) -> Bytes.Bytes & Bool:
1411 is_match.bytes.fin(exec.bytes(re, b, True{}))
1412
1413# The bit NFA over Bytes: the buffer, the offset i of the next char, that char, and the NFA state.
1414type Rt is Type:
1415 Rt{a: Array<U32>, i: U32, d: Dc, st: Bs}
1416
1417def rt.at(r: Array<U32> & Dc, +j: U32, st: Bs) -> Rt:
1418 (a, +d) = r
1419 Rt{a, j, d, st}
1420
1421def rt.skip(r: Array<U32> & U32, +len: U32, +atn: U32, +hit1: Bool) -> Rt:
1422 (a, +k) = r
1423 rt.at(dec(a, len, k), k, Bs{atn, False{}, hit1})
1424
1425# idle: only the sets live at a fresh start wait on c, and c starts none of them.
1426def rt.next(idle: Bool, asc: Maybe<&2, Asc>, a: Array<U32>, +len: U32, +j: U32, +sets: List<&2, Bit>, +atn: U32, +hit1: Bool, +cur: U32, +c: U32) -> Rt:
1427 match idle:
1428 case True{}:
1429 match asc:
1430 case None{}:
1431 rt.at(dec(a, len, j), j, Bs{atn, False{}, hit1})
1432 case Some{+m}:
1433 rt.skip(skip.from(a, len, j, m), len, atn, hit1)
1434 case False{}:
1435 rt.at(dec(a, len, j), j, bits.step(sets, cur, c, Bs{atn, False{}, hit1}))
1436
1437def bitsb(fuel: Nat, r: Rt, +len: U32, +sets: List<&2, Bit>, +start: Maybe<&2, List<&2, Inst>>, +asc: Maybe<&2, Asc>, +atn: U32, +hit1: Bool) -> Array<U32> & Bool:
1438 match fuel:
1439 case 0n:
1440 Rt{a, i, d, st} = r
1441 (a, False{})
1442 case 1n+f:
1443 Rt{a, +i, d, st} = r
1444 match d:
1445 case DEnd{}:
1446 Bs{cur, done, ok} = st
1447 (a, done || ok)
1448 case Dc{+c, +w}:
1449 match st:
1450 case Bs{cur, True{}, ok}:
1451 (a, True{})
1452 case Bs{+cur, False{}, ok}:
1453 bitsb(f, rt.next(bits.idle(start, cur, atn, c), asc, a, len, (i + w : U32), sets, atn, hit1, cur, c), len, sets, start, asc, atn, hit1)
1454
1455def bits.bytes.fin(+len: U32, r: Array<U32> & Bool) -> Bytes.Bytes & Maybe<&2, Bool>:
1456 (a, x) = r
1457 (Bytes.Bytes{len, a}, Some{x})
1458
1459def bits.bytes(b: Bits, start: Maybe<&2, List<&2, Inst>>, asc: Maybe<&2, Asc>, +len: U32, buf: Array<U32>) -> Bytes.Bytes & Maybe<&2, Bool>:
1460 Bits{sets, at0, hit0, hit01, atn, hit1} = b
1461 bits.bytes.fin(len, bitsb(Nat.add(U32.to_nat(len), 2n), rt.at(dec(buf, len, 0), 0, Bs{at0, hit0, hit01}), len, sets, start, asc, atn, hit1))
1462
1463def bits.bytes.if(m: Maybe<&2, Bits>, start: Maybe<&2, List<&2, Inst>>, asc: Maybe<&2, Asc>, b: Bytes.Bytes) -> Bytes.Bytes & Maybe<&2, Bool>:
1464 match m:
1465 case None{}:
1466 (b, None{})
1467 case Some{x}:
1468 Bytes.Bytes{+len, buf} = b
1469 bits.bytes(x, start, asc, len, buf)
1470
1471# The bit NFA form of is_match over Bytes, or None when the program does not allow it.
1472def is_match.bytes.bits(re: Regex, b: Bytes.Bytes) -> Bytes.Bytes & Maybe<&2, Bool>:
1473 Regex{prog, fuel, slots, start, bt, asc} = re
1474 bits.bytes.if(bt, start, asc, b)
1475
1476def is_match.bytes.pick(r: Bytes.Bytes & Maybe<&2, Bool>, re: Regex) -> Bytes.Bytes & Bool:
1477 (b, m) = r
1478 match m:
1479 case Some{x}:
1480 (b, x)
1481 case None{}:
1482 is_match.bytes.vm(re, b)
1483
1484# is_match over UTF-8 bytes: the buffer back, and whether it has a match.
1485def is_match.bytes(+re: Regex, b: Bytes.Bytes) -> Bytes.Bytes & Bool:
1486 is_match.bytes.pick(is_match.bytes.bits(re, b), re)