1 /*-
2  * This code is derived from OpenBSD's libc/regex, original license follows:
3  *
4  * Copyright (c) 1992, 1993, 1994 Henry Spencer.
5  * Copyright (c) 1992, 1993, 1994
6  *	The Regents of the University of California.  All rights reserved.
7  *
8  * This code is derived from software contributed to Berkeley by
9  * Henry Spencer.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
14  * 1. Redistributions of source code must retain the above copyright
15  *    notice, this list of conditions and the following disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  *    notice, this list of conditions and the following disclaimer in the
18  *    documentation and/or other materials provided with the distribution.
19  * 3. Neither the name of the University nor the names of its contributors
20  *    may be used to endorse or promote products derived from this software
21  *    without specific prior written permission.
22  *
23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33  * SUCH DAMAGE.
34  *
35  *	@(#)regcomp.c	8.5 (Berkeley) 3/20/94
36  */
37 
38 #include <sys/types.h>
39 #include <stdio.h>
40 #include <string.h>
41 #include <ctype.h>
42 #include <limits.h>
43 #include <stdlib.h>
44 #include "regex_impl.h"
45 
46 #include "regutils.h"
47 #include "regex2.h"
48 
49 #include "regcclass.h"
50 #include "regcname.h"
51 
52 #include "llvm/Config/config.h"
53 #if HAVE_STDINT_H
54 #include <stdint.h>
55 #else
56 /* Pessimistically bound memory use */
57 #define SIZE_MAX UINT_MAX
58 #endif
59 
60 /*
61  * parse structure, passed up and down to avoid global variables and
62  * other clumsinesses
63  */
64 struct parse {
65 	char *next;		/* next character in RE */
66 	char *end;		/* end of string (-> NUL normally) */
67 	int error;		/* has an error been seen? */
68 	sop *strip;		/* malloced strip */
69 	sopno ssize;		/* malloced strip size (allocated) */
70 	sopno slen;		/* malloced strip length (used) */
71 	int ncsalloc;		/* number of csets allocated */
72 	struct re_guts *g;
73 #	define	NPAREN	10	/* we need to remember () 1-9 for back refs */
74 	sopno pbegin[NPAREN];	/* -> ( ([0] unused) */
75 	sopno pend[NPAREN];	/* -> ) ([0] unused) */
76 };
77 
78 static void p_ere(struct parse *, int);
79 static void p_ere_exp(struct parse *);
80 static void p_str(struct parse *);
81 static void p_bre(struct parse *, int, int);
82 static int p_simp_re(struct parse *, int);
83 static int p_count(struct parse *);
84 static void p_bracket(struct parse *);
85 static void p_b_term(struct parse *, cset *);
86 static void p_b_cclass(struct parse *, cset *);
87 static void p_b_eclass(struct parse *, cset *);
88 static char p_b_symbol(struct parse *);
89 static char p_b_coll_elem(struct parse *, int);
90 static char othercase(int);
91 static void bothcases(struct parse *, int);
92 static void ordinary(struct parse *, int);
93 static void nonnewline(struct parse *);
94 static void repeat(struct parse *, sopno, int, int);
95 static int seterr(struct parse *, int);
96 static cset *allocset(struct parse *);
97 static void freeset(struct parse *, cset *);
98 static int freezeset(struct parse *, cset *);
99 static int firstch(struct parse *, cset *);
100 static int nch(struct parse *, cset *);
101 static void mcadd(struct parse *, cset *, const char *);
102 static void mcinvert(struct parse *, cset *);
103 static void mccase(struct parse *, cset *);
104 static int isinsets(struct re_guts *, int);
105 static int samesets(struct re_guts *, int, int);
106 static void categorize(struct parse *, struct re_guts *);
107 static sopno dupl(struct parse *, sopno, sopno);
108 static void doemit(struct parse *, sop, size_t);
109 static void doinsert(struct parse *, sop, size_t, sopno);
110 static void dofwd(struct parse *, sopno, sop);
111 static void enlarge(struct parse *, sopno);
112 static void stripsnug(struct parse *, struct re_guts *);
113 static void findmust(struct parse *, struct re_guts *);
114 static sopno pluscount(struct parse *, struct re_guts *);
115 
116 static char nuls[10];		/* place to point scanner in event of error */
117 
118 /*
119  * macros for use with parse structure
120  * BEWARE:  these know that the parse structure is named `p' !!!
121  */
122 #define	PEEK()	(*p->next)
123 #define	PEEK2()	(*(p->next+1))
124 #define	MORE()	(p->next < p->end)
125 #define	MORE2()	(p->next+1 < p->end)
126 #define	SEE(c)	(MORE() && PEEK() == (c))
127 #define	SEETWO(a, b)	(MORE() && MORE2() && PEEK() == (a) && PEEK2() == (b))
128 #define	EAT(c)	((SEE(c)) ? (NEXT(), 1) : 0)
129 #define	EATTWO(a, b)	((SEETWO(a, b)) ? (NEXT2(), 1) : 0)
130 #define	NEXT()	(p->next++)
131 #define	NEXT2()	(p->next += 2)
132 #define	NEXTn(n)	(p->next += (n))
133 #define	GETNEXT()	(*p->next++)
134 #define	SETERROR(e)	seterr(p, (e))
135 #define	REQUIRE(co, e)	(void)((co) || SETERROR(e))
136 #define	MUSTSEE(c, e)	(REQUIRE(MORE() && PEEK() == (c), e))
137 #define	MUSTEAT(c, e)	(REQUIRE(MORE() && GETNEXT() == (c), e))
138 #define	MUSTNOTSEE(c, e)	(REQUIRE(!MORE() || PEEK() != (c), e))
139 #define	EMIT(op, sopnd)	doemit(p, (sop)(op), (size_t)(sopnd))
140 #define	INSERT(op, pos)	doinsert(p, (sop)(op), HERE()-(pos)+1, pos)
141 #define	AHEAD(pos)		dofwd(p, pos, HERE()-(pos))
142 #define	ASTERN(sop, pos)	EMIT(sop, HERE()-pos)
143 #define	HERE()		(p->slen)
144 #define	THERE()		(p->slen - 1)
145 #define	THERETHERE()	(p->slen - 2)
146 #define	DROP(n)	(p->slen -= (n))
147 
148 #ifdef	_POSIX2_RE_DUP_MAX
149 #define	DUPMAX	_POSIX2_RE_DUP_MAX
150 #else
151 #define	DUPMAX	255
152 #endif
153 #define	INFINITY	(DUPMAX + 1)
154 
155 #ifndef NDEBUG
156 static int never = 0;		/* for use in asserts; shuts lint up */
157 #else
158 #define	never	0		/* some <assert.h>s have bugs too */
159 #endif
160 
161 /*
162  - llvm_regcomp - interface for parser and compilation
163  */
164 int				/* 0 success, otherwise REG_something */
llvm_regcomp(llvm_regex_t * preg,const char * pattern,int cflags)165 llvm_regcomp(llvm_regex_t *preg, const char *pattern, int cflags)
166 {
167 	struct parse pa;
168 	struct re_guts *g;
169 	struct parse *p = &pa;
170 	int i;
171 	size_t len;
172 #ifdef REDEBUG
173 #	define	GOODFLAGS(f)	(f)
174 #else
175 #	define	GOODFLAGS(f)	((f)&~REG_DUMP)
176 #endif
177 
178 	cflags = GOODFLAGS(cflags);
179 	if ((cflags&REG_EXTENDED) && (cflags&REG_NOSPEC))
180 		return(REG_INVARG);
181 
182 	if (cflags&REG_PEND) {
183 		if (preg->re_endp < pattern)
184 			return(REG_INVARG);
185 		len = preg->re_endp - pattern;
186 	} else
187 		len = strlen((const char *)pattern);
188 
189 	/* do the mallocs early so failure handling is easy */
190 	g = (struct re_guts *)malloc(sizeof(struct re_guts) +
191 							(NC-1)*sizeof(cat_t));
192 	if (g == NULL)
193 		return(REG_ESPACE);
194 	p->ssize = len/(size_t)2*(size_t)3 + (size_t)1;	/* ugh */
195 	p->strip = (sop *)calloc(p->ssize, sizeof(sop));
196 	p->slen = 0;
197 	if (p->strip == NULL) {
198 		free((char *)g);
199 		return(REG_ESPACE);
200 	}
201 
202 	/* set things up */
203 	p->g = g;
204 	p->next = (char *)pattern;	/* convenience; we do not modify it */
205 	p->end = p->next + len;
206 	p->error = 0;
207 	p->ncsalloc = 0;
208 	for (i = 0; i < NPAREN; i++) {
209 		p->pbegin[i] = 0;
210 		p->pend[i] = 0;
211 	}
212 	g->csetsize = NC;
213 	g->sets = NULL;
214 	g->setbits = NULL;
215 	g->ncsets = 0;
216 	g->cflags = cflags;
217 	g->iflags = 0;
218 	g->nbol = 0;
219 	g->neol = 0;
220 	g->must = NULL;
221 	g->mlen = 0;
222 	g->nsub = 0;
223 	g->ncategories = 1;	/* category 0 is "everything else" */
224 	g->categories = &g->catspace[-(CHAR_MIN)];
225 	(void) memset((char *)g->catspace, 0, NC*sizeof(cat_t));
226 	g->backrefs = 0;
227 
228 	/* do it */
229 	EMIT(OEND, 0);
230 	g->firststate = THERE();
231 	if (cflags&REG_EXTENDED)
232 		p_ere(p, OUT);
233 	else if (cflags&REG_NOSPEC)
234 		p_str(p);
235 	else
236 		p_bre(p, OUT, OUT);
237 	EMIT(OEND, 0);
238 	g->laststate = THERE();
239 
240 	/* tidy up loose ends and fill things in */
241 	categorize(p, g);
242 	stripsnug(p, g);
243 	findmust(p, g);
244 	g->nplus = pluscount(p, g);
245 	g->magic = MAGIC2;
246 	preg->re_nsub = g->nsub;
247 	preg->re_g = g;
248 	preg->re_magic = MAGIC1;
249 #ifndef REDEBUG
250 	/* not debugging, so can't rely on the assert() in llvm_regexec() */
251 	if (g->iflags&REGEX_BAD)
252 		SETERROR(REG_ASSERT);
253 #endif
254 
255 	/* win or lose, we're done */
256 	if (p->error != 0)	/* lose */
257 		llvm_regfree(preg);
258 	return(p->error);
259 }
260 
261 /*
262  - p_ere - ERE parser top level, concatenation and alternation
263  */
264 static void
p_ere(struct parse * p,int stop)265 p_ere(struct parse *p, int stop)	/* character this ERE should end at */
266 {
267 	char c;
268 	sopno prevback = 0;
269 	sopno prevfwd = 0;
270 	sopno conc;
271 	int first = 1;		/* is this the first alternative? */
272 
273 	for (;;) {
274 		/* do a bunch of concatenated expressions */
275 		conc = HERE();
276 		while (MORE() && (c = PEEK()) != '|' && c != stop)
277 			p_ere_exp(p);
278 		REQUIRE(HERE() != conc, REG_EMPTY);	/* require nonempty */
279 
280 		if (!EAT('|'))
281 			break;		/* NOTE BREAK OUT */
282 
283 		if (first) {
284 			INSERT(OCH_, conc);	/* offset is wrong */
285 			prevfwd = conc;
286 			prevback = conc;
287 			first = 0;
288 		}
289 		ASTERN(OOR1, prevback);
290 		prevback = THERE();
291 		AHEAD(prevfwd);			/* fix previous offset */
292 		prevfwd = HERE();
293 		EMIT(OOR2, 0);			/* offset is very wrong */
294 	}
295 
296 	if (!first) {		/* tail-end fixups */
297 		AHEAD(prevfwd);
298 		ASTERN(O_CH, prevback);
299 	}
300 
301 	assert(!MORE() || SEE(stop));
302 }
303 
304 /*
305  - p_ere_exp - parse one subERE, an atom possibly followed by a repetition op
306  */
307 static void
p_ere_exp(struct parse * p)308 p_ere_exp(struct parse *p)
309 {
310 	char c;
311 	sopno pos;
312 	int count;
313 	int count2;
314 	int backrefnum;
315 	sopno subno;
316 	int wascaret = 0;
317 
318 	assert(MORE());		/* caller should have ensured this */
319 	c = GETNEXT();
320 
321 	pos = HERE();
322 	switch (c) {
323 	case '(':
324 		REQUIRE(MORE(), REG_EPAREN);
325 		p->g->nsub++;
326 		subno = p->g->nsub;
327 		if (subno < NPAREN)
328 			p->pbegin[subno] = HERE();
329 		EMIT(OLPAREN, subno);
330 		if (!SEE(')'))
331 			p_ere(p, ')');
332 		if (subno < NPAREN) {
333 			p->pend[subno] = HERE();
334 			assert(p->pend[subno] != 0);
335 		}
336 		EMIT(ORPAREN, subno);
337 		MUSTEAT(')', REG_EPAREN);
338 		break;
339 #ifndef POSIX_MISTAKE
340 	case ')':		/* happens only if no current unmatched ( */
341 		/*
342 		 * You may ask, why the ifndef?  Because I didn't notice
343 		 * this until slightly too late for 1003.2, and none of the
344 		 * other 1003.2 regular-expression reviewers noticed it at
345 		 * all.  So an unmatched ) is legal POSIX, at least until
346 		 * we can get it fixed.
347 		 */
348 		SETERROR(REG_EPAREN);
349 		break;
350 #endif
351 	case '^':
352 		EMIT(OBOL, 0);
353 		p->g->iflags |= USEBOL;
354 		p->g->nbol++;
355 		wascaret = 1;
356 		break;
357 	case '$':
358 		EMIT(OEOL, 0);
359 		p->g->iflags |= USEEOL;
360 		p->g->neol++;
361 		break;
362 	case '|':
363 		SETERROR(REG_EMPTY);
364 		break;
365 	case '*':
366 	case '+':
367 	case '?':
368 		SETERROR(REG_BADRPT);
369 		break;
370 	case '.':
371 		if (p->g->cflags&REG_NEWLINE)
372 			nonnewline(p);
373 		else
374 			EMIT(OANY, 0);
375 		break;
376 	case '[':
377 		p_bracket(p);
378 		break;
379 	case '\\':
380 		REQUIRE(MORE(), REG_EESCAPE);
381 		c = GETNEXT();
382 		if (c >= '1' && c <= '9') {
383 			/* \[0-9] is taken to be a back-reference to a previously specified
384 			 * matching group. backrefnum will hold the number. The matching
385 			 * group must exist (i.e. if \4 is found there must have been at
386 			 * least 4 matching groups specified in the pattern previously).
387 			 */
388 			backrefnum = c - '0';
389 			if (p->pend[backrefnum] == 0) {
390 				SETERROR(REG_ESUBREG);
391 				break;
392 			}
393 
394 			/* Make sure everything checks out and emit the sequence
395 			 * that marks a back-reference to the parse structure.
396 			 */
397 			assert(backrefnum <= p->g->nsub);
398 			EMIT(OBACK_, backrefnum);
399 			assert(p->pbegin[backrefnum] != 0);
400 			assert(OP(p->strip[p->pbegin[backrefnum]]) != OLPAREN);
401 			assert(OP(p->strip[p->pend[backrefnum]]) != ORPAREN);
402 			(void) dupl(p, p->pbegin[backrefnum]+1, p->pend[backrefnum]);
403 			EMIT(O_BACK, backrefnum);
404 			p->g->backrefs = 1;
405 		} else {
406 			/* Other chars are simply themselves when escaped with a backslash.
407 			 */
408 			ordinary(p, c);
409 		}
410 		break;
411 	case '{':		/* okay as ordinary except if digit follows */
412 		REQUIRE(!MORE() || !isdigit((uch)PEEK()), REG_BADRPT);
413 		/* FALLTHROUGH */
414 	default:
415 		ordinary(p, c);
416 		break;
417 	}
418 
419 	if (!MORE())
420 		return;
421 	c = PEEK();
422 	/* we call { a repetition if followed by a digit */
423 	if (!( c == '*' || c == '+' || c == '?' ||
424 				(c == '{' && MORE2() && isdigit((uch)PEEK2())) ))
425 		return;		/* no repetition, we're done */
426 	NEXT();
427 
428 	REQUIRE(!wascaret, REG_BADRPT);
429 	switch (c) {
430 	case '*':	/* implemented as +? */
431 		/* this case does not require the (y|) trick, noKLUDGE */
432 		INSERT(OPLUS_, pos);
433 		ASTERN(O_PLUS, pos);
434 		INSERT(OQUEST_, pos);
435 		ASTERN(O_QUEST, pos);
436 		break;
437 	case '+':
438 		INSERT(OPLUS_, pos);
439 		ASTERN(O_PLUS, pos);
440 		break;
441 	case '?':
442 		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
443 		INSERT(OCH_, pos);		/* offset slightly wrong */
444 		ASTERN(OOR1, pos);		/* this one's right */
445 		AHEAD(pos);			/* fix the OCH_ */
446 		EMIT(OOR2, 0);			/* offset very wrong... */
447 		AHEAD(THERE());			/* ...so fix it */
448 		ASTERN(O_CH, THERETHERE());
449 		break;
450 	case '{':
451 		count = p_count(p);
452 		if (EAT(',')) {
453 			if (isdigit((uch)PEEK())) {
454 				count2 = p_count(p);
455 				REQUIRE(count <= count2, REG_BADBR);
456 			} else		/* single number with comma */
457 				count2 = INFINITY;
458 		} else		/* just a single number */
459 			count2 = count;
460 		repeat(p, pos, count, count2);
461 		if (!EAT('}')) {	/* error heuristics */
462 			while (MORE() && PEEK() != '}')
463 				NEXT();
464 			REQUIRE(MORE(), REG_EBRACE);
465 			SETERROR(REG_BADBR);
466 		}
467 		break;
468 	}
469 
470 	if (!MORE())
471 		return;
472 	c = PEEK();
473 	if (!( c == '*' || c == '+' || c == '?' ||
474 				(c == '{' && MORE2() && isdigit((uch)PEEK2())) ) )
475 		return;
476 	SETERROR(REG_BADRPT);
477 }
478 
479 /*
480  - p_str - string (no metacharacters) "parser"
481  */
482 static void
p_str(struct parse * p)483 p_str(struct parse *p)
484 {
485 	REQUIRE(MORE(), REG_EMPTY);
486 	while (MORE())
487 		ordinary(p, GETNEXT());
488 }
489 
490 /*
491  - p_bre - BRE parser top level, anchoring and concatenation
492  * Giving end1 as OUT essentially eliminates the end1/end2 check.
493  *
494  * This implementation is a bit of a kludge, in that a trailing $ is first
495  * taken as an ordinary character and then revised to be an anchor.  The
496  * only undesirable side effect is that '$' gets included as a character
497  * category in such cases.  This is fairly harmless; not worth fixing.
498  * The amount of lookahead needed to avoid this kludge is excessive.
499  */
500 static void
p_bre(struct parse * p,int end1,int end2)501 p_bre(struct parse *p,
502     int end1,		/* first terminating character */
503     int end2)		/* second terminating character */
504 {
505 	sopno start = HERE();
506 	int first = 1;			/* first subexpression? */
507 	int wasdollar = 0;
508 
509 	if (EAT('^')) {
510 		EMIT(OBOL, 0);
511 		p->g->iflags |= USEBOL;
512 		p->g->nbol++;
513 	}
514 	while (MORE() && !SEETWO(end1, end2)) {
515 		wasdollar = p_simp_re(p, first);
516 		first = 0;
517 	}
518 	if (wasdollar) {	/* oops, that was a trailing anchor */
519 		DROP(1);
520 		EMIT(OEOL, 0);
521 		p->g->iflags |= USEEOL;
522 		p->g->neol++;
523 	}
524 
525 	REQUIRE(HERE() != start, REG_EMPTY);	/* require nonempty */
526 }
527 
528 /*
529  - p_simp_re - parse a simple RE, an atom possibly followed by a repetition
530  */
531 static int			/* was the simple RE an unbackslashed $? */
p_simp_re(struct parse * p,int starordinary)532 p_simp_re(struct parse *p,
533     int starordinary)		/* is a leading * an ordinary character? */
534 {
535 	int c;
536 	int count;
537 	int count2;
538 	sopno pos;
539 	int i;
540 	sopno subno;
541 #	define	BACKSL	(1<<CHAR_BIT)
542 
543         pos = HERE(); /* repetition op, if any, covers from here */
544 
545         assert(MORE()); /* caller should have ensured this */
546         c = GETNEXT();
547 	if (c == '\\') {
548 		REQUIRE(MORE(), REG_EESCAPE);
549 		c = BACKSL | GETNEXT();
550 	}
551 	switch (c) {
552 	case '.':
553 		if (p->g->cflags&REG_NEWLINE)
554 			nonnewline(p);
555 		else
556 			EMIT(OANY, 0);
557 		break;
558 	case '[':
559 		p_bracket(p);
560 		break;
561 	case BACKSL|'{':
562 		SETERROR(REG_BADRPT);
563 		break;
564 	case BACKSL|'(':
565 		p->g->nsub++;
566 		subno = p->g->nsub;
567 		if (subno < NPAREN)
568 			p->pbegin[subno] = HERE();
569 		EMIT(OLPAREN, subno);
570 		/* the MORE here is an error heuristic */
571 		if (MORE() && !SEETWO('\\', ')'))
572 			p_bre(p, '\\', ')');
573 		if (subno < NPAREN) {
574 			p->pend[subno] = HERE();
575 			assert(p->pend[subno] != 0);
576 		}
577 		EMIT(ORPAREN, subno);
578 		REQUIRE(EATTWO('\\', ')'), REG_EPAREN);
579 		break;
580 	case BACKSL|')':	/* should not get here -- must be user */
581 	case BACKSL|'}':
582 		SETERROR(REG_EPAREN);
583 		break;
584 	case BACKSL|'1':
585 	case BACKSL|'2':
586 	case BACKSL|'3':
587 	case BACKSL|'4':
588 	case BACKSL|'5':
589 	case BACKSL|'6':
590 	case BACKSL|'7':
591 	case BACKSL|'8':
592 	case BACKSL|'9':
593 		i = (c&~BACKSL) - '0';
594 		assert(i < NPAREN);
595 		if (p->pend[i] != 0) {
596 			assert(i <= p->g->nsub);
597 			EMIT(OBACK_, i);
598 			assert(p->pbegin[i] != 0);
599 			assert(OP(p->strip[p->pbegin[i]]) == OLPAREN);
600 			assert(OP(p->strip[p->pend[i]]) == ORPAREN);
601 			(void) dupl(p, p->pbegin[i]+1, p->pend[i]);
602 			EMIT(O_BACK, i);
603 		} else
604 			SETERROR(REG_ESUBREG);
605 		p->g->backrefs = 1;
606 		break;
607 	case '*':
608 		REQUIRE(starordinary, REG_BADRPT);
609 		/* FALLTHROUGH */
610 	default:
611 		ordinary(p, (char)c);
612 		break;
613 	}
614 
615 	if (EAT('*')) {		/* implemented as +? */
616 		/* this case does not require the (y|) trick, noKLUDGE */
617 		INSERT(OPLUS_, pos);
618 		ASTERN(O_PLUS, pos);
619 		INSERT(OQUEST_, pos);
620 		ASTERN(O_QUEST, pos);
621 	} else if (EATTWO('\\', '{')) {
622 		count = p_count(p);
623 		if (EAT(',')) {
624 			if (MORE() && isdigit((uch)PEEK())) {
625 				count2 = p_count(p);
626 				REQUIRE(count <= count2, REG_BADBR);
627 			} else		/* single number with comma */
628 				count2 = INFINITY;
629 		} else		/* just a single number */
630 			count2 = count;
631 		repeat(p, pos, count, count2);
632 		if (!EATTWO('\\', '}')) {	/* error heuristics */
633 			while (MORE() && !SEETWO('\\', '}'))
634 				NEXT();
635 			REQUIRE(MORE(), REG_EBRACE);
636 			SETERROR(REG_BADBR);
637 		}
638 	} else if (c == '$')	/* $ (but not \$) ends it */
639 		return(1);
640 
641 	return(0);
642 }
643 
644 /*
645  - p_count - parse a repetition count
646  */
647 static int			/* the value */
p_count(struct parse * p)648 p_count(struct parse *p)
649 {
650 	int count = 0;
651 	int ndigits = 0;
652 
653 	while (MORE() && isdigit((uch)PEEK()) && count <= DUPMAX) {
654 		count = count*10 + (GETNEXT() - '0');
655 		ndigits++;
656 	}
657 
658 	REQUIRE(ndigits > 0 && count <= DUPMAX, REG_BADBR);
659 	return(count);
660 }
661 
662 /*
663  - p_bracket - parse a bracketed character list
664  *
665  * Note a significant property of this code:  if the allocset() did SETERROR,
666  * no set operations are done.
667  */
668 static void
p_bracket(struct parse * p)669 p_bracket(struct parse *p)
670 {
671 	cset *cs;
672 	int invert = 0;
673 
674 	/* Dept of Truly Sickening Special-Case Kludges */
675 	if (p->next + 5 < p->end && strncmp(p->next, "[:<:]]", 6) == 0) {
676 		EMIT(OBOW, 0);
677 		NEXTn(6);
678 		return;
679 	}
680 	if (p->next + 5 < p->end && strncmp(p->next, "[:>:]]", 6) == 0) {
681 		EMIT(OEOW, 0);
682 		NEXTn(6);
683 		return;
684 	}
685 
686 	if ((cs = allocset(p)) == NULL) {
687 		/* allocset did set error status in p */
688 		return;
689 	}
690 
691 	if (EAT('^'))
692 		invert++;	/* make note to invert set at end */
693 	if (EAT(']'))
694 		CHadd(cs, ']');
695 	else if (EAT('-'))
696 		CHadd(cs, '-');
697 	while (MORE() && PEEK() != ']' && !SEETWO('-', ']'))
698 		p_b_term(p, cs);
699 	if (EAT('-'))
700 		CHadd(cs, '-');
701 	MUSTEAT(']', REG_EBRACK);
702 
703 	if (p->error != 0) {	/* don't mess things up further */
704 		freeset(p, cs);
705 		return;
706 	}
707 
708 	if (p->g->cflags&REG_ICASE) {
709 		int i;
710 		int ci;
711 
712 		for (i = p->g->csetsize - 1; i >= 0; i--)
713 			if (CHIN(cs, i) && isalpha(i)) {
714 				ci = othercase(i);
715 				if (ci != i)
716 					CHadd(cs, ci);
717 			}
718 		if (cs->multis != NULL)
719 			mccase(p, cs);
720 	}
721 	if (invert) {
722 		int i;
723 
724 		for (i = p->g->csetsize - 1; i >= 0; i--)
725 			if (CHIN(cs, i))
726 				CHsub(cs, i);
727 			else
728 				CHadd(cs, i);
729 		if (p->g->cflags&REG_NEWLINE)
730 			CHsub(cs, '\n');
731 		if (cs->multis != NULL)
732 			mcinvert(p, cs);
733 	}
734 
735 	assert(cs->multis == NULL);		/* xxx */
736 
737 	if (nch(p, cs) == 1) {		/* optimize singleton sets */
738 		ordinary(p, firstch(p, cs));
739 		freeset(p, cs);
740 	} else
741 		EMIT(OANYOF, freezeset(p, cs));
742 }
743 
744 /*
745  - p_b_term - parse one term of a bracketed character list
746  */
747 static void
p_b_term(struct parse * p,cset * cs)748 p_b_term(struct parse *p, cset *cs)
749 {
750 	char c;
751 	char start, finish;
752 	int i;
753 
754 	/* classify what we've got */
755 	switch ((MORE()) ? PEEK() : '\0') {
756 	case '[':
757 		c = (MORE2()) ? PEEK2() : '\0';
758 		break;
759 	case '-':
760 		SETERROR(REG_ERANGE);
761 		return;			/* NOTE RETURN */
762 		break;
763 	default:
764 		c = '\0';
765 		break;
766 	}
767 
768 	switch (c) {
769 	case ':':		/* character class */
770 		NEXT2();
771 		REQUIRE(MORE(), REG_EBRACK);
772 		c = PEEK();
773 		REQUIRE(c != '-' && c != ']', REG_ECTYPE);
774 		p_b_cclass(p, cs);
775 		REQUIRE(MORE(), REG_EBRACK);
776 		REQUIRE(EATTWO(':', ']'), REG_ECTYPE);
777 		break;
778 	case '=':		/* equivalence class */
779 		NEXT2();
780 		REQUIRE(MORE(), REG_EBRACK);
781 		c = PEEK();
782 		REQUIRE(c != '-' && c != ']', REG_ECOLLATE);
783 		p_b_eclass(p, cs);
784 		REQUIRE(MORE(), REG_EBRACK);
785 		REQUIRE(EATTWO('=', ']'), REG_ECOLLATE);
786 		break;
787 	default:		/* symbol, ordinary character, or range */
788 /* xxx revision needed for multichar stuff */
789 		start = p_b_symbol(p);
790 		if (SEE('-') && MORE2() && PEEK2() != ']') {
791 			/* range */
792 			NEXT();
793 			if (EAT('-'))
794 				finish = '-';
795 			else
796 				finish = p_b_symbol(p);
797 		} else
798 			finish = start;
799 /* xxx what about signed chars here... */
800 		REQUIRE(start <= finish, REG_ERANGE);
801 		for (i = start; i <= finish; i++)
802 			CHadd(cs, i);
803 		break;
804 	}
805 }
806 
807 /*
808  - p_b_cclass - parse a character-class name and deal with it
809  */
810 static void
p_b_cclass(struct parse * p,cset * cs)811 p_b_cclass(struct parse *p, cset *cs)
812 {
813 	char *sp = p->next;
814 	struct cclass *cp;
815 	size_t len;
816 	const char *u;
817 	char c;
818 
819 	while (MORE() && isalpha((uch)PEEK()))
820 		NEXT();
821 	len = p->next - sp;
822 	for (cp = cclasses; cp->name != NULL; cp++)
823 		if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0')
824 			break;
825 	if (cp->name == NULL) {
826 		/* oops, didn't find it */
827 		SETERROR(REG_ECTYPE);
828 		return;
829 	}
830 
831 	u = cp->chars;
832 	while ((c = *u++) != '\0')
833 		CHadd(cs, c);
834 	for (u = cp->multis; *u != '\0'; u += strlen(u) + 1)
835 		MCadd(p, cs, u);
836 }
837 
838 /*
839  - p_b_eclass - parse an equivalence-class name and deal with it
840  *
841  * This implementation is incomplete. xxx
842  */
843 static void
p_b_eclass(struct parse * p,cset * cs)844 p_b_eclass(struct parse *p, cset *cs)
845 {
846 	char c;
847 
848 	c = p_b_coll_elem(p, '=');
849 	CHadd(cs, c);
850 }
851 
852 /*
853  - p_b_symbol - parse a character or [..]ed multicharacter collating symbol
854  */
855 static char			/* value of symbol */
p_b_symbol(struct parse * p)856 p_b_symbol(struct parse *p)
857 {
858 	char value;
859 
860 	REQUIRE(MORE(), REG_EBRACK);
861 	if (!EATTWO('[', '.'))
862 		return(GETNEXT());
863 
864 	/* collating symbol */
865 	value = p_b_coll_elem(p, '.');
866 	REQUIRE(EATTWO('.', ']'), REG_ECOLLATE);
867 	return(value);
868 }
869 
870 /*
871  - p_b_coll_elem - parse a collating-element name and look it up
872  */
873 static char			/* value of collating element */
p_b_coll_elem(struct parse * p,int endc)874 p_b_coll_elem(struct parse *p,
875     int endc)			/* name ended by endc,']' */
876 {
877 	char *sp = p->next;
878 	struct cname *cp;
879 	int len;
880 
881 	while (MORE() && !SEETWO(endc, ']'))
882 		NEXT();
883 	if (!MORE()) {
884 		SETERROR(REG_EBRACK);
885 		return(0);
886 	}
887 	len = p->next - sp;
888 	for (cp = cnames; cp->name != NULL; cp++)
889 		if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0')
890 			return(cp->code);	/* known name */
891 	if (len == 1)
892 		return(*sp);	/* single character */
893 	SETERROR(REG_ECOLLATE);			/* neither */
894 	return(0);
895 }
896 
897 /*
898  - othercase - return the case counterpart of an alphabetic
899  */
900 static char			/* if no counterpart, return ch */
othercase(int ch)901 othercase(int ch)
902 {
903 	ch = (uch)ch;
904 	assert(isalpha(ch));
905 	if (isupper(ch))
906 		return ((uch)tolower(ch));
907 	else if (islower(ch))
908 		return ((uch)toupper(ch));
909 	else			/* peculiar, but could happen */
910 		return(ch);
911 }
912 
913 /*
914  - bothcases - emit a dualcase version of a two-case character
915  *
916  * Boy, is this implementation ever a kludge...
917  */
918 static void
bothcases(struct parse * p,int ch)919 bothcases(struct parse *p, int ch)
920 {
921 	char *oldnext = p->next;
922 	char *oldend = p->end;
923 	char bracket[3];
924 
925 	ch = (uch)ch;
926 	assert(othercase(ch) != ch);	/* p_bracket() would recurse */
927 	p->next = bracket;
928 	p->end = bracket+2;
929 	bracket[0] = ch;
930 	bracket[1] = ']';
931 	bracket[2] = '\0';
932 	p_bracket(p);
933 	assert(p->next == bracket+2);
934 	p->next = oldnext;
935 	p->end = oldend;
936 }
937 
938 /*
939  - ordinary - emit an ordinary character
940  */
941 static void
ordinary(struct parse * p,int ch)942 ordinary(struct parse *p, int ch)
943 {
944 	cat_t *cap = p->g->categories;
945 
946 	if ((p->g->cflags&REG_ICASE) && isalpha((uch)ch) && othercase(ch) != ch)
947 		bothcases(p, ch);
948 	else {
949 		EMIT(OCHAR, (uch)ch);
950 		if (cap[ch] == 0)
951 			cap[ch] = p->g->ncategories++;
952 	}
953 }
954 
955 /*
956  - nonnewline - emit REG_NEWLINE version of OANY
957  *
958  * Boy, is this implementation ever a kludge...
959  */
960 static void
nonnewline(struct parse * p)961 nonnewline(struct parse *p)
962 {
963 	char *oldnext = p->next;
964 	char *oldend = p->end;
965 	char bracket[4];
966 
967 	p->next = bracket;
968 	p->end = bracket+3;
969 	bracket[0] = '^';
970 	bracket[1] = '\n';
971 	bracket[2] = ']';
972 	bracket[3] = '\0';
973 	p_bracket(p);
974 	assert(p->next == bracket+3);
975 	p->next = oldnext;
976 	p->end = oldend;
977 }
978 
979 /*
980  - repeat - generate code for a bounded repetition, recursively if needed
981  */
982 static void
repeat(struct parse * p,sopno start,int from,int to)983 repeat(struct parse *p,
984     sopno start,		/* operand from here to end of strip */
985     int from,			/* repeated from this number */
986     int to)			/* to this number of times (maybe INFINITY) */
987 {
988 	sopno finish = HERE();
989 #	define	N	2
990 #	define	INF	3
991 #	define	REP(f, t)	((f)*8 + (t))
992 #	define	MAP(n)	(((n) <= 1) ? (n) : ((n) == INFINITY) ? INF : N)
993 	sopno copy;
994 
995 	if (p->error != 0)	/* head off possible runaway recursion */
996 		return;
997 
998 	assert(from <= to);
999 
1000 	switch (REP(MAP(from), MAP(to))) {
1001 	case REP(0, 0):			/* must be user doing this */
1002 		DROP(finish-start);	/* drop the operand */
1003 		break;
1004 	case REP(0, 1):			/* as x{1,1}? */
1005 	case REP(0, N):			/* as x{1,n}? */
1006 	case REP(0, INF):		/* as x{1,}? */
1007 		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1008 		INSERT(OCH_, start);		/* offset is wrong... */
1009 		repeat(p, start+1, 1, to);
1010 		ASTERN(OOR1, start);
1011 		AHEAD(start);			/* ... fix it */
1012 		EMIT(OOR2, 0);
1013 		AHEAD(THERE());
1014 		ASTERN(O_CH, THERETHERE());
1015 		break;
1016 	case REP(1, 1):			/* trivial case */
1017 		/* done */
1018 		break;
1019 	case REP(1, N):			/* as x?x{1,n-1} */
1020 		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1021 		INSERT(OCH_, start);
1022 		ASTERN(OOR1, start);
1023 		AHEAD(start);
1024 		EMIT(OOR2, 0);			/* offset very wrong... */
1025 		AHEAD(THERE());			/* ...so fix it */
1026 		ASTERN(O_CH, THERETHERE());
1027 		copy = dupl(p, start+1, finish+1);
1028 		assert(copy == finish+4);
1029 		repeat(p, copy, 1, to-1);
1030 		break;
1031 	case REP(1, INF):		/* as x+ */
1032 		INSERT(OPLUS_, start);
1033 		ASTERN(O_PLUS, start);
1034 		break;
1035 	case REP(N, N):			/* as xx{m-1,n-1} */
1036 		copy = dupl(p, start, finish);
1037 		repeat(p, copy, from-1, to-1);
1038 		break;
1039 	case REP(N, INF):		/* as xx{n-1,INF} */
1040 		copy = dupl(p, start, finish);
1041 		repeat(p, copy, from-1, to);
1042 		break;
1043 	default:			/* "can't happen" */
1044 		SETERROR(REG_ASSERT);	/* just in case */
1045 		break;
1046 	}
1047 }
1048 
1049 /*
1050  - seterr - set an error condition
1051  */
1052 static int			/* useless but makes type checking happy */
seterr(struct parse * p,int e)1053 seterr(struct parse *p, int e)
1054 {
1055 	if (p->error == 0)	/* keep earliest error condition */
1056 		p->error = e;
1057 	p->next = nuls;		/* try to bring things to a halt */
1058 	p->end = nuls;
1059 	return(0);		/* make the return value well-defined */
1060 }
1061 
1062 /*
1063  - allocset - allocate a set of characters for []
1064  */
1065 static cset *
allocset(struct parse * p)1066 allocset(struct parse *p)
1067 {
1068 	int no = p->g->ncsets++;
1069 	size_t nc;
1070 	size_t nbytes;
1071 	cset *cs;
1072 	size_t css = (size_t)p->g->csetsize;
1073 	int i;
1074 
1075 	if (no >= p->ncsalloc) {	/* need another column of space */
1076 		void *ptr;
1077 
1078 		p->ncsalloc += CHAR_BIT;
1079 		nc = p->ncsalloc;
1080 		if (nc > SIZE_MAX / sizeof(cset))
1081 			goto nomem;
1082 		assert(nc % CHAR_BIT == 0);
1083 		nbytes = nc / CHAR_BIT * css;
1084 
1085 		ptr = (cset *)realloc((char *)p->g->sets, nc * sizeof(cset));
1086 		if (ptr == NULL)
1087 			goto nomem;
1088 		p->g->sets = ptr;
1089 
1090 		ptr = (uch *)realloc((char *)p->g->setbits, nbytes);
1091 		if (ptr == NULL)
1092 			goto nomem;
1093 		p->g->setbits = ptr;
1094 
1095 		for (i = 0; i < no; i++)
1096 			p->g->sets[i].ptr = p->g->setbits + css*(i/CHAR_BIT);
1097 
1098 		(void) memset((char *)p->g->setbits + (nbytes - css), 0, css);
1099 	}
1100 	/* XXX should not happen */
1101 	if (p->g->sets == NULL || p->g->setbits == NULL)
1102 		goto nomem;
1103 
1104 	cs = &p->g->sets[no];
1105 	cs->ptr = p->g->setbits + css*((no)/CHAR_BIT);
1106 	cs->mask = 1 << ((no) % CHAR_BIT);
1107 	cs->hash = 0;
1108 	cs->smultis = 0;
1109 	cs->multis = NULL;
1110 
1111 	return(cs);
1112 nomem:
1113 	free(p->g->sets);
1114 	p->g->sets = NULL;
1115 	free(p->g->setbits);
1116 	p->g->setbits = NULL;
1117 
1118 	SETERROR(REG_ESPACE);
1119 	/* caller's responsibility not to do set ops */
1120 	return(NULL);
1121 }
1122 
1123 /*
1124  - freeset - free a now-unused set
1125  */
1126 static void
freeset(struct parse * p,cset * cs)1127 freeset(struct parse *p, cset *cs)
1128 {
1129 	size_t i;
1130 	cset *top = &p->g->sets[p->g->ncsets];
1131 	size_t css = (size_t)p->g->csetsize;
1132 
1133 	for (i = 0; i < css; i++)
1134 		CHsub(cs, i);
1135 	if (cs == top-1)	/* recover only the easy case */
1136 		p->g->ncsets--;
1137 }
1138 
1139 /*
1140  - freezeset - final processing on a set of characters
1141  *
1142  * The main task here is merging identical sets.  This is usually a waste
1143  * of time (although the hash code minimizes the overhead), but can win
1144  * big if REG_ICASE is being used.  REG_ICASE, by the way, is why the hash
1145  * is done using addition rather than xor -- all ASCII [aA] sets xor to
1146  * the same value!
1147  */
1148 static int			/* set number */
freezeset(struct parse * p,cset * cs)1149 freezeset(struct parse *p, cset *cs)
1150 {
1151 	uch h = cs->hash;
1152 	size_t i;
1153 	cset *top = &p->g->sets[p->g->ncsets];
1154 	cset *cs2;
1155 	size_t css = (size_t)p->g->csetsize;
1156 
1157 	/* look for an earlier one which is the same */
1158 	for (cs2 = &p->g->sets[0]; cs2 < top; cs2++)
1159 		if (cs2->hash == h && cs2 != cs) {
1160 			/* maybe */
1161 			for (i = 0; i < css; i++)
1162 				if (!!CHIN(cs2, i) != !!CHIN(cs, i))
1163 					break;		/* no */
1164 			if (i == css)
1165 				break;			/* yes */
1166 		}
1167 
1168 	if (cs2 < top) {	/* found one */
1169 		freeset(p, cs);
1170 		cs = cs2;
1171 	}
1172 
1173 	return((int)(cs - p->g->sets));
1174 }
1175 
1176 /*
1177  - firstch - return first character in a set (which must have at least one)
1178  */
1179 static int			/* character; there is no "none" value */
firstch(struct parse * p,cset * cs)1180 firstch(struct parse *p, cset *cs)
1181 {
1182 	size_t i;
1183 	size_t css = (size_t)p->g->csetsize;
1184 
1185 	for (i = 0; i < css; i++)
1186 		if (CHIN(cs, i))
1187 			return((char)i);
1188 	assert(never);
1189 	return(0);		/* arbitrary */
1190 }
1191 
1192 /*
1193  - nch - number of characters in a set
1194  */
1195 static int
nch(struct parse * p,cset * cs)1196 nch(struct parse *p, cset *cs)
1197 {
1198 	size_t i;
1199 	size_t css = (size_t)p->g->csetsize;
1200 	int n = 0;
1201 
1202 	for (i = 0; i < css; i++)
1203 		if (CHIN(cs, i))
1204 			n++;
1205 	return(n);
1206 }
1207 
1208 /*
1209  - mcadd - add a collating element to a cset
1210  */
1211 static void
mcadd(struct parse * p,cset * cs,const char * cp)1212 mcadd( struct parse *p, cset *cs, const char *cp)
1213 {
1214 	size_t oldend = cs->smultis;
1215 	void *np;
1216 
1217 	cs->smultis += strlen(cp) + 1;
1218 	np = realloc(cs->multis, cs->smultis);
1219 	if (np == NULL) {
1220 		if (cs->multis)
1221 			free(cs->multis);
1222 		cs->multis = NULL;
1223 		SETERROR(REG_ESPACE);
1224 		return;
1225 	}
1226 	cs->multis = np;
1227 
1228 	llvm_strlcpy(cs->multis + oldend - 1, cp, cs->smultis - oldend + 1);
1229 }
1230 
1231 /*
1232  - mcinvert - invert the list of collating elements in a cset
1233  *
1234  * This would have to know the set of possibilities.  Implementation
1235  * is deferred.
1236  */
1237 /* ARGSUSED */
1238 static void
mcinvert(struct parse * p,cset * cs)1239 mcinvert(struct parse *p, cset *cs)
1240 {
1241 	assert(cs->multis == NULL);	/* xxx */
1242 }
1243 
1244 /*
1245  - mccase - add case counterparts of the list of collating elements in a cset
1246  *
1247  * This would have to know the set of possibilities.  Implementation
1248  * is deferred.
1249  */
1250 /* ARGSUSED */
1251 static void
mccase(struct parse * p,cset * cs)1252 mccase(struct parse *p, cset *cs)
1253 {
1254 	assert(cs->multis == NULL);	/* xxx */
1255 }
1256 
1257 /*
1258  - isinsets - is this character in any sets?
1259  */
1260 static int			/* predicate */
isinsets(struct re_guts * g,int c)1261 isinsets(struct re_guts *g, int c)
1262 {
1263 	uch *col;
1264 	int i;
1265 	int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1266 	unsigned uc = (uch)c;
1267 
1268 	for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1269 		if (col[uc] != 0)
1270 			return(1);
1271 	return(0);
1272 }
1273 
1274 /*
1275  - samesets - are these two characters in exactly the same sets?
1276  */
1277 static int			/* predicate */
samesets(struct re_guts * g,int c1,int c2)1278 samesets(struct re_guts *g, int c1, int c2)
1279 {
1280 	uch *col;
1281 	int i;
1282 	int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1283 	unsigned uc1 = (uch)c1;
1284 	unsigned uc2 = (uch)c2;
1285 
1286 	for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1287 		if (col[uc1] != col[uc2])
1288 			return(0);
1289 	return(1);
1290 }
1291 
1292 /*
1293  - categorize - sort out character categories
1294  */
1295 static void
categorize(struct parse * p,struct re_guts * g)1296 categorize(struct parse *p, struct re_guts *g)
1297 {
1298 	cat_t *cats = g->categories;
1299 	int c;
1300 	int c2;
1301 	cat_t cat;
1302 
1303 	/* avoid making error situations worse */
1304 	if (p->error != 0)
1305 		return;
1306 
1307 	for (c = CHAR_MIN; c <= CHAR_MAX; c++)
1308 		if (cats[c] == 0 && isinsets(g, c)) {
1309 			cat = g->ncategories++;
1310 			cats[c] = cat;
1311 			for (c2 = c+1; c2 <= CHAR_MAX; c2++)
1312 				if (cats[c2] == 0 && samesets(g, c, c2))
1313 					cats[c2] = cat;
1314 		}
1315 }
1316 
1317 /*
1318  - dupl - emit a duplicate of a bunch of sops
1319  */
1320 static sopno			/* start of duplicate */
dupl(struct parse * p,sopno start,sopno finish)1321 dupl(struct parse *p,
1322     sopno start,		/* from here */
1323     sopno finish)		/* to this less one */
1324 {
1325 	sopno ret = HERE();
1326 	sopno len = finish - start;
1327 
1328 	assert(finish >= start);
1329 	if (len == 0)
1330 		return(ret);
1331 	enlarge(p, p->ssize + len);	/* this many unexpected additions */
1332 	assert(p->ssize >= p->slen + len);
1333 	(void) memmove((char *)(p->strip + p->slen),
1334 		(char *)(p->strip + start), (size_t)len*sizeof(sop));
1335 	p->slen += len;
1336 	return(ret);
1337 }
1338 
1339 /*
1340  - doemit - emit a strip operator
1341  *
1342  * It might seem better to implement this as a macro with a function as
1343  * hard-case backup, but it's just too big and messy unless there are
1344  * some changes to the data structures.  Maybe later.
1345  */
1346 static void
doemit(struct parse * p,sop op,size_t opnd)1347 doemit(struct parse *p, sop op, size_t opnd)
1348 {
1349 	/* avoid making error situations worse */
1350 	if (p->error != 0)
1351 		return;
1352 
1353 	/* deal with oversize operands ("can't happen", more or less) */
1354 	assert(opnd < 1<<OPSHIFT);
1355 
1356 	/* deal with undersized strip */
1357 	if (p->slen >= p->ssize)
1358 		enlarge(p, (p->ssize+1) / 2 * 3);	/* +50% */
1359 	assert(p->slen < p->ssize);
1360 
1361 	/* finally, it's all reduced to the easy case */
1362 	p->strip[p->slen++] = SOP(op, opnd);
1363 }
1364 
1365 /*
1366  - doinsert - insert a sop into the strip
1367  */
1368 static void
doinsert(struct parse * p,sop op,size_t opnd,sopno pos)1369 doinsert(struct parse *p, sop op, size_t opnd, sopno pos)
1370 {
1371 	sopno sn;
1372 	sop s;
1373 	int i;
1374 
1375 	/* avoid making error situations worse */
1376 	if (p->error != 0)
1377 		return;
1378 
1379 	sn = HERE();
1380 	EMIT(op, opnd);		/* do checks, ensure space */
1381 	assert(HERE() == sn+1);
1382 	s = p->strip[sn];
1383 
1384 	/* adjust paren pointers */
1385 	assert(pos > 0);
1386 	for (i = 1; i < NPAREN; i++) {
1387 		if (p->pbegin[i] >= pos) {
1388 			p->pbegin[i]++;
1389 		}
1390 		if (p->pend[i] >= pos) {
1391 			p->pend[i]++;
1392 		}
1393 	}
1394 
1395 	memmove((char *)&p->strip[pos+1], (char *)&p->strip[pos],
1396 						(HERE()-pos-1)*sizeof(sop));
1397 	p->strip[pos] = s;
1398 }
1399 
1400 /*
1401  - dofwd - complete a forward reference
1402  */
1403 static void
dofwd(struct parse * p,sopno pos,sop value)1404 dofwd(struct parse *p, sopno pos, sop value)
1405 {
1406 	/* avoid making error situations worse */
1407 	if (p->error != 0)
1408 		return;
1409 
1410 	assert(value < 1<<OPSHIFT);
1411 	p->strip[pos] = OP(p->strip[pos]) | value;
1412 }
1413 
1414 /*
1415  - enlarge - enlarge the strip
1416  */
1417 static void
enlarge(struct parse * p,sopno size)1418 enlarge(struct parse *p, sopno size)
1419 {
1420 	sop *sp;
1421 
1422 	if (p->ssize >= size)
1423 		return;
1424 
1425 	if ((uintptr_t)size > SIZE_MAX / sizeof(sop)) {
1426 		SETERROR(REG_ESPACE);
1427 		return;
1428 	}
1429 
1430 	sp = (sop *)realloc(p->strip, size*sizeof(sop));
1431 	if (sp == NULL) {
1432 		SETERROR(REG_ESPACE);
1433 		return;
1434 	}
1435 	p->strip = sp;
1436 	p->ssize = size;
1437 }
1438 
1439 /*
1440  - stripsnug - compact the strip
1441  */
1442 static void
stripsnug(struct parse * p,struct re_guts * g)1443 stripsnug(struct parse *p, struct re_guts *g)
1444 {
1445 	g->nstates = p->slen;
1446 	if ((uintptr_t)p->slen > SIZE_MAX / sizeof(sop)) {
1447 		g->strip = p->strip;
1448 		SETERROR(REG_ESPACE);
1449 		return;
1450 	}
1451 
1452 	g->strip = (sop *)realloc((char *)p->strip, p->slen * sizeof(sop));
1453 	if (g->strip == NULL) {
1454 		SETERROR(REG_ESPACE);
1455 		g->strip = p->strip;
1456 	}
1457 }
1458 
1459 /*
1460  - findmust - fill in must and mlen with longest mandatory literal string
1461  *
1462  * This algorithm could do fancy things like analyzing the operands of |
1463  * for common subsequences.  Someday.  This code is simple and finds most
1464  * of the interesting cases.
1465  *
1466  * Note that must and mlen got initialized during setup.
1467  */
1468 static void
findmust(struct parse * p,struct re_guts * g)1469 findmust(struct parse *p, struct re_guts *g)
1470 {
1471 	sop *scan;
1472 	sop *start = 0; /* start initialized in the default case, after that */
1473 	sop *newstart = 0; /* newstart was initialized in the OCHAR case */
1474 	sopno newlen;
1475 	sop s;
1476 	char *cp;
1477 	sopno i;
1478 
1479 	/* avoid making error situations worse */
1480 	if (p->error != 0)
1481 		return;
1482 
1483 	/* find the longest OCHAR sequence in strip */
1484 	newlen = 0;
1485 	scan = g->strip + 1;
1486 	do {
1487 		s = *scan++;
1488 		switch (OP(s)) {
1489 		case OCHAR:		/* sequence member */
1490 			if (newlen == 0)		/* new sequence */
1491 				newstart = scan - 1;
1492 			newlen++;
1493 			break;
1494 		case OPLUS_:		/* things that don't break one */
1495 		case OLPAREN:
1496 		case ORPAREN:
1497 			break;
1498 		case OQUEST_:		/* things that must be skipped */
1499 		case OCH_:
1500 			scan--;
1501 			do {
1502 				scan += OPND(s);
1503 				s = *scan;
1504 				/* assert() interferes w debug printouts */
1505 				if (OP(s) != O_QUEST && OP(s) != O_CH &&
1506 							OP(s) != OOR2) {
1507 					g->iflags |= REGEX_BAD;
1508 					return;
1509 				}
1510 			} while (OP(s) != O_QUEST && OP(s) != O_CH);
1511 			/* fallthrough */
1512 		default:		/* things that break a sequence */
1513 			if (newlen > g->mlen) {		/* ends one */
1514 				start = newstart;
1515 				g->mlen = newlen;
1516 			}
1517 			newlen = 0;
1518 			break;
1519 		}
1520 	} while (OP(s) != OEND);
1521 
1522 	if (g->mlen == 0)		/* there isn't one */
1523 		return;
1524 
1525 	/* turn it into a character string */
1526 	g->must = malloc((size_t)g->mlen + 1);
1527 	if (g->must == NULL) {		/* argh; just forget it */
1528 		g->mlen = 0;
1529 		return;
1530 	}
1531 	cp = g->must;
1532 	scan = start;
1533 	for (i = g->mlen; i > 0; i--) {
1534 		while (OP(s = *scan++) != OCHAR)
1535 			continue;
1536 		assert(cp < g->must + g->mlen);
1537 		*cp++ = (char)OPND(s);
1538 	}
1539 	assert(cp == g->must + g->mlen);
1540 	*cp++ = '\0';		/* just on general principles */
1541 }
1542 
1543 /*
1544  - pluscount - count + nesting
1545  */
1546 static sopno			/* nesting depth */
pluscount(struct parse * p,struct re_guts * g)1547 pluscount(struct parse *p, struct re_guts *g)
1548 {
1549 	sop *scan;
1550 	sop s;
1551 	sopno plusnest = 0;
1552 	sopno maxnest = 0;
1553 
1554 	if (p->error != 0)
1555 		return(0);	/* there may not be an OEND */
1556 
1557 	scan = g->strip + 1;
1558 	do {
1559 		s = *scan++;
1560 		switch (OP(s)) {
1561 		case OPLUS_:
1562 			plusnest++;
1563 			break;
1564 		case O_PLUS:
1565 			if (plusnest > maxnest)
1566 				maxnest = plusnest;
1567 			plusnest--;
1568 			break;
1569 		}
1570 	} while (OP(s) != OEND);
1571 	if (plusnest != 0)
1572 		g->iflags |= REGEX_BAD;
1573 	return(maxnest);
1574 }
1575