1 /*	$OpenBSD: regcomp.c,v 1.20 2010/11/21 00:02:30 tedu Exp $ */
2 /*-
3  * Copyright (c) 1992, 1993, 1994 Henry Spencer.
4  * Copyright (c) 1992, 1993, 1994
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * This code is derived from software contributed to Berkeley by
8  * Henry Spencer.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  * 3. Neither the name of the University nor the names of its contributors
19  *    may be used to endorse or promote products derived from this software
20  *    without specific prior written permission.
21  *
22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  *
34  *	@(#)regcomp.c	8.5 (Berkeley) 3/20/94
35  */
36 
37 #include <sys/types.h>
38 #include <stdio.h>
39 #include <string.h>
40 #include <ctype.h>
41 #include <limits.h>
42 #include <stdlib.h>
43 #include "r_regex.h"
44 #include "r_util/r_str.h"
45 #include "r_util/r_assert.h"
46 
47 #include "utils.h"
48 #include "regex2.h"
49 
50 #include "cclass.h"
51 #include "cname.h"
52 
53 /*
54  * parse structure, passed up and down to avoid global variables and
55  * other clumsinesses
56  */
57 struct parse {
58 	char *next;		/* next character in RE */
59 	char *end;		/* end of string (-> NUL normally) */
60 	int error;		/* has an error been seen? */
61 	sop *strip;		/* malloced strip */
62 	sopno ssize;		/* malloced strip size (allocated) */
63 	sopno slen;		/* malloced strip length (used) */
64 	int ncsalloc;		/* number of csets allocated */
65 	struct re_guts *g;
66 #	define	NPAREN	10	/* we need to remember () 1-9 for back refs */
67 	sopno pbegin[NPAREN];	/* -> ( ([0] unused) */
68 	sopno pend[NPAREN];	/* -> ) ([0] unused) */
69 };
70 
71 static void p_ere(struct parse *, int);
72 static void p_ere_exp(struct parse *);
73 static void p_str(struct parse *);
74 static void p_bre(struct parse *, int, int);
75 static int p_simp_re(struct parse *, int);
76 static int p_count(struct parse *);
77 static void p_bracket(struct parse *);
78 static void p_b_term(struct parse *, cset *);
79 static void p_b_cclass(struct parse *, cset *);
80 static void p_b_eclass(struct parse *, cset *);
81 static char p_b_symbol(struct parse *);
82 static char p_b_coll_elem(struct parse *, int);
83 static char othercase(int);
84 static void bothcases(struct parse *, int);
85 static void ordinary(struct parse *, int);
86 static void special(struct parse *, int);
87 static void nonnewline(struct parse *);
88 static void repeat(struct parse *, sopno, int, int);
89 static int seterr(struct parse *, int);
90 static cset *allocset(struct parse *);
91 static void freeset(struct parse *, cset *);
92 static int freezeset(struct parse *, cset *);
93 static int firstch(struct parse *, cset *);
94 static int nch(struct parse *, cset *);
95 static void mcadd(struct parse *, cset *, char *);
96 static void mcinvert(struct parse *, cset *);
97 static void mccase(struct parse *, cset *);
98 static int isinsets(struct re_guts *, int);
99 static int samesets(struct re_guts *, int, int);
100 static void categorize(struct parse *, struct re_guts *);
101 static sopno dupl(struct parse *, sopno, sopno);
102 static void doemit(struct parse *, sop, size_t);
103 static void doinsert(struct parse *, sop, size_t, sopno);
104 static void dofwd(struct parse *, sopno, sop);
105 static void enlarge(struct parse *, sopno);
106 static void stripsnug(struct parse *, struct re_guts *);
107 static void findmust(struct parse *, struct re_guts *);
108 static sopno pluscount(struct parse *, struct re_guts *);
109 
110 static char nuls[10];		/* place to point scanner in event of error */
111 
112 /*
113  * macros for use with parse structure
114  * BEWARE:  these know that the parse structure is named `p' !!!
115  */
116 #define	PEEK()	(*p->next)
117 #define	PEEK2()	(*(p->next+1))
118 #define	MORE()	(p->next < p->end)
119 #define	MORE2()	(p->next+1 < p->end)
120 #define	SEE(c)	(MORE() && PEEK() == (c))
121 #define	SEETWO(a, b)	(MORE() && MORE2() && PEEK() == (a) && PEEK2() == (b))
122 #define	EAT(c)	((SEE(c)) ? (NEXT(), 1) : 0)
123 #define	EATTWO(a, b)	((SEETWO(a, b)) ? (NEXT2(), 1) : 0)
124 #define	NEXT()	(p->next++)
125 #define	NEXT2()	(p->next += 2)
126 #define	NEXTn(n)	(p->next += (n))
127 #define	GETNEXT()	(*p->next++)
128 #define	SETERROR(e)	seterr(p, (e))
129 #define	REQUIRE(co, e)	(void)((co) || SETERROR(e))
130 #define	MUSTSEE(c, e)	(REQUIRE(MORE() && PEEK() == (c), e))
131 #define	MUSTEAT(c, e)	(REQUIRE(MORE() && GETNEXT() == (c), e))
132 #define	MUSTNOTSEE(c, e)	(REQUIRE(!MORE() || PEEK() != (c), e))
133 #define	EMIT(op, sopnd)	doemit(p, (sop)(op), (size_t)(sopnd))
134 #define	INSERT(op, pos)	doinsert(p, (sop)(op), HERE()-(pos)+1, pos)
135 #define	AHEAD(pos)		dofwd(p, pos, HERE()-(pos))
136 #define	ASTERN(sop, pos)	EMIT(sop, HERE()-(pos))
137 #define	HERE()		(p->slen)
138 #define	THERE()		(p->slen - 1)
139 #define	THERETHERE()	(p->slen - 2)
140 #define	DROP(n)	(p->slen -= (n))
141 
142 
r_regex_match(const char * pattern,const char * flags,const char * text)143 R_API bool r_regex_match(const char *pattern, const char *flags, const char *text) {
144 	RRegex rx;
145 	int re_flags = r_regex_flags (flags);
146 	if (r_regex_init (&rx, pattern, re_flags)) {
147 		eprintf ("r_regex_match: /%s/ fails to compile.\n", pattern);
148 		return false;
149 	}
150 	int rc = r_regex_exec (&rx, text, 0, 0, re_flags);
151 	r_regex_fini (&rx);
152 	return rc == 0;
153 }
154 
r_regex_match_list(RRegex * rx,const char * text)155 R_API RList *r_regex_match_list(RRegex *rx, const char *text) {
156 	RList *list = r_list_newf (free);
157 	RRegexMatch match;
158 
159 	/* Initialize the boundaries for R_REGEX_STARTEND */
160 	match.rm_so = 0;
161 	match.rm_eo = strlen (text);
162 	while (!r_regex_exec (rx, text, 1, &match, rx->re_flags | R_REGEX_STARTEND)) {
163 		size_t entry_len = match.rm_eo - match.rm_so + 1;
164 		char *entry = r_str_ndup (text + match.rm_so, entry_len);
165 		r_list_append (list, entry);
166 		/* Update the boundaries for R_REGEX_STARTEND */
167 		match.rm_so = match.rm_eo;
168 		match.rm_eo = strlen (text);
169 	}
170 	return list;
171 }
172 
r_regex_new(const char * pattern,const char * flags)173 R_API RRegex *r_regex_new(const char *pattern, const char *flags) {
174 	r_return_val_if_fail (pattern, NULL);
175 	RRegex *r, rx = {0};
176 	if (r_regex_init (&rx, pattern, r_regex_flags (flags))) {
177 		return NULL;
178 	}
179 	r = R_NEW (RRegex);
180 	if (!r) {
181 		return NULL;
182 	}
183 	memcpy (r, &rx, sizeof (RRegex));
184 	return r;
185 }
186 
r_regex_flags(const char * f)187 R_API int r_regex_flags(const char *f) {
188 	int flags = 0;
189 	if (!f || !*f) {
190 		return 0;
191 	}
192 	if (strchr (f, 'e')) {
193 		flags |= R_REGEX_EXTENDED;
194 	}
195 	if (strchr (f, 'i')) {
196 		flags |= R_REGEX_ICASE;
197 	}
198 	if (strchr (f, 's')) {
199 		flags |= R_REGEX_NOSUB;
200 	}
201 	if (strchr (f, 'n')) {
202 		flags |= R_REGEX_NEWLINE;
203 	}
204 	if (strchr (f, 'N')) {
205 		flags |= R_REGEX_NOSPEC;
206 	}
207 	if (strchr (f, 'p')) {
208 		flags |= R_REGEX_PEND;
209 	}
210 	if (strchr (f, 'd')) {
211 		flags |= R_REGEX_DUMP;
212 	}
213 	return flags;
214 }
215 
r_regex_fini(RRegex * preg)216 R_API void r_regex_fini(RRegex *preg) {
217 	struct re_guts *g;
218 	if (!preg) {
219 		return;
220 	}
221 	if (preg->re_magic != MAGIC1) {	/* oops */
222 		return;			/* nice to complain, but hard */
223 	}
224 
225 	g = preg->re_g;
226 	if (!g || g->magic != MAGIC2) { /* oops again */
227 		return;
228 	}
229 	preg->re_magic = 0;		/* mark it invalid */
230 	g->magic = 0;			/* mark it invalid */
231 
232 	free (g->strip);
233 	free (g->sets);
234 	free (g->setbits);
235 	free (g->must);
236 	free (g);
237 }
238 
r_regex_free(RRegex * preg)239 R_API void r_regex_free(RRegex *preg) {
240 	r_regex_fini (preg);
241 	free (preg);
242 }
243 
244 /*
245  - regcomp - interface for parser and compilation
246  - 0 success, otherwise R_REGEX_something
247  */
r_regex_init(RRegex * preg,const char * pattern,int cflags)248 R_API int r_regex_init(RRegex *preg, const char *pattern, int cflags) {
249 	struct parse pa;
250 	struct parse *p = &pa;
251 	int i;
252 	size_t len;
253 	cflags &= ~R_REGEX_DUMP;
254 	if (!preg || ((cflags & R_REGEX_EXTENDED) && (cflags & R_REGEX_NOSPEC))) {
255 		return R_REGEX_INVARG;
256 	}
257 	if (cflags & R_REGEX_PEND) {
258 		if (preg->re_endp < pattern) {
259 			return R_REGEX_INVARG;
260 		}
261 		len = preg->re_endp - pattern;
262 	} else {
263 		len = strlen ((char *)pattern);
264 	}
265 	/* do the mallocs early so failure handling is easy */
266 	struct re_guts *g = (struct re_guts *)calloc (sizeof (struct re_guts) + (NC - 1), sizeof (cat_t));
267 	if (!g) {
268 		return R_REGEX_ESPACE;
269 	}
270 	/*
271 	 * Limit the pattern space to avoid a 32-bit overflow on buffer
272 	 * extension.  Also avoid any signed overflow in case of conversion
273 	 * so make the real limit based on a 31-bit overflow.
274 	 *
275 	 * Likely not applicable on 64-bit systems but handle the case
276 	 * generically (who are we to stop people from using ~715MB+
277 	 * patterns?).
278 	 */
279 	size_t maxlen = ((size_t) - 1 >> 1) / sizeof (sop) * 2 / 3;
280 	if (len >= maxlen) {
281 		free (g);
282 		return R_REGEX_ESPACE;
283 	}
284 	preg->re_flags = cflags;
285 	p->ssize = len / (size_t)2 * (size_t)3 + (size_t)1;	/* ugh */
286 	if (p->ssize < len) {
287 		free (g);
288 		return R_REGEX_ESPACE;
289 	}
290 
291 	p->strip = (sop *)calloc (p->ssize, sizeof (sop));
292 	if (!p->strip) {
293 		free (g);
294 		return R_REGEX_ESPACE;
295 	}
296 	p->slen = 0;
297 	if (!p->strip) {
298 		free (g);
299 		return R_REGEX_ESPACE;
300 	}
301 
302 	/* set things up */
303 	p->g = g;
304 	p->next = (char *)pattern;	/* convenience; we do not modify it */
305 	p->end = p->next + len;
306 	p->error = 0;
307 	p->ncsalloc = 0;
308 	for (i = 0; i < NPAREN; i++) {
309 		p->pbegin[i] = 0;
310 		p->pend[i] = 0;
311 	}
312 	g->csetsize = NC;
313 	g->sets = NULL;
314 	g->setbits = NULL;
315 	g->ncsets = 0;
316 	g->cflags = cflags;
317 	g->iflags = 0;
318 	g->nbol = 0;
319 	g->neol = 0;
320 	g->must = NULL;
321 	g->mlen = 0;
322 	g->nsub = 0;
323 	g->ncategories = 1;	/* category 0 is "everything else" */
324 	g->categories = &g->catspace[-(CHAR_MIN)]; // WTF
325 	(void) memset ((char *)g->catspace, 0, NC*sizeof(cat_t));
326 	g->backrefs = 0;
327 
328 	/* do it */
329 	EMIT (OEND, 0);
330 	g->firststate = THERE ();
331 	if (cflags & R_REGEX_EXTENDED) {
332 		p_ere(p, OUT);
333 	} else if (cflags & R_REGEX_NOSPEC) {
334 		p_str (p);
335 	} else {
336 		p_bre (p, OUT, OUT);
337 	}
338 	EMIT (OEND, 0);
339 	g->laststate = THERE ();
340 
341 	/* tidy up loose ends and fill things in */
342 	categorize (p, g);
343 	stripsnug (p, g);
344 	findmust (p, g);
345 	g->nplus = pluscount (p, g);
346 	g->magic = MAGIC2;
347 	preg->re_nsub = g->nsub;
348 	preg->re_g = g;
349 	preg->re_magic = MAGIC1;
350 	/* not debugging, so can't rely on the asssert() in regexec() */
351 	if (g->iflags & BAD) {
352 		SETERROR (R_REGEX_ASSERT);
353 	}
354 	if (p->error) {
355 		r_regex_fini (preg);
356 	}
357 	return p->error;
358 }
359 
360 /*
361  - p_ere - ERE parser top level, concatenation and alternation
362  */
p_ere(struct parse * p,int stop)363 static void p_ere(struct parse *p, int stop) { /* character this ERE should end at */
364 	bool isFirst = true;
365 	sopno prevback = 0;
366 	sopno prevfwd = 0;
367 	sopno conc = 0;
368 	char c;
369 
370 	for (;;) {
371 		/* do a bunch of concatenated expressions */
372 		conc = HERE();
373 		while (MORE() && (c = PEEK()) != '|' && c != stop) {
374 			p_ere_exp (p);
375 		}
376 		REQUIRE (HERE () != conc, R_REGEX_EMPTY); /* require nonempty */
377 
378 		if (!EAT('|')) {
379 			break;		/* NOTE BREAK OUT */
380 		}
381 		if (isFirst) {
382 			INSERT (OCH_, conc);	/* offset is wrong */
383 			prevfwd = conc;
384 			prevback = conc;
385 			isFirst = false;
386 		}
387 		ASTERN(OOR1, prevback);
388 		prevback = THERE();
389 		AHEAD(prevfwd);			/* fix previous offset */
390 		prevfwd = HERE();
391 		EMIT(OOR2, 0);			/* offset is very wrong */
392 	}
393 
394 	if (!isFirst) {		/* tail-end fixups */
395 		AHEAD(prevfwd);
396 		ASTERN(O_CH, prevback);
397 	}
398 	//asert(!MORE() || SEE(stop));
399 }
400 
401 /*
402  - p_ere_exp - parse one subERE, an atom possibly followed by a repetition op
403  */
p_ere_exp(struct parse * p)404 static void p_ere_exp(struct parse *p) {
405 	char c;
406 	sopno pos;
407 	int count;
408 	int count2;
409 	sopno subno;
410 	int wascaret = 0;
411 
412 	if (!MORE()) {		/* caller should have ensured this */
413 		return;
414 	}
415 	c = GETNEXT();
416 
417 	pos = HERE();
418 	switch (c) {
419 	case '(':
420 		REQUIRE(MORE(), R_REGEX_EPAREN);
421 		p->g->nsub++;
422 		subno = p->g->nsub;
423 		if (subno < NPAREN) {
424 			p->pbegin[subno] = HERE ();
425 		}
426 		EMIT(OLPAREN, subno);
427 		if (!SEE (')')) {
428 			p_ere (p, ')');
429 		}
430 		if (subno < NPAREN) {
431 			p->pend[subno] = HERE();
432 			if (!p->pend[subno]) {
433 				break;
434 			}
435 		}
436 		EMIT(ORPAREN, subno);
437 		MUSTEAT(')', R_REGEX_EPAREN);
438 		break;
439 	case '^':
440 		EMIT(OBOL, 0);
441 		p->g->iflags |= USEBOL;
442 		p->g->nbol++;
443 		wascaret = 1;
444 		break;
445 	case '$':
446 		EMIT(OEOL, 0);
447 		p->g->iflags |= USEEOL;
448 		p->g->neol++;
449 		break;
450 	case '|':
451 		SETERROR(R_REGEX_EMPTY);
452 		break;
453 	case '*':
454 	case '+':
455 	case '?':
456 		SETERROR(R_REGEX_BADRPT);
457 		break;
458 	case '.':
459 		if (p->g->cflags & R_REGEX_NEWLINE) {
460 			nonnewline(p);
461 		} else {
462 			EMIT (OANY, 0);
463 		}
464 		break;
465 	case '[':
466 		p_bracket(p);
467 		break;
468 	case '\\':
469 		REQUIRE(MORE(), R_REGEX_EESCAPE);
470 		c = GETNEXT();
471 		if (!isalpha(c)) {
472 			ordinary(p, c);
473 		} else {
474 			special(p, c);
475 		}
476 		break;
477 	case '{':		/* okay as ordinary except if digit follows */
478 		REQUIRE(!MORE() || !isdigit((ut8)PEEK()), R_REGEX_BADRPT);
479 		/* FALLTHROUGH */
480 	default:
481 		ordinary(p, c);
482 		break;
483 	}
484 
485 	if (!MORE ()) {
486 		return;
487 	}
488 	c = PEEK();
489 	/* we call { a repetition if followed by a digit */
490 	if (!(c == '*' || c == '+' || c == '?' ||
491 		    (c == '{' && MORE2 () && isdigit ((ut8)PEEK2 ())))) {
492 		return; /* no repetition, we're done */
493 	}
494 	NEXT();
495 
496 	REQUIRE(!wascaret, R_REGEX_BADRPT);
497 	switch (c) {
498 	case '*':	/* implemented as +? */
499 		/* this case does not require the (y|) trick, noKLUDGE */
500 		INSERT(OPLUS_, pos);
501 		ASTERN(O_PLUS, pos);
502 		INSERT(OQUEST_, pos);
503 		ASTERN(O_QUEST, pos);
504 		break;
505 	case '+':
506 		INSERT(OPLUS_, pos);
507 		ASTERN(O_PLUS, pos);
508 		break;
509 	case '?':
510 		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
511 		INSERT(OCH_, pos);		/* offset slightly wrong */
512 		ASTERN(OOR1, pos);		/* this one's right */
513 		AHEAD(pos);			/* fix the OCH_ */
514 		EMIT(OOR2, 0);			/* offset very wrong... */
515 		AHEAD(THERE());			/* ...so fix it */
516 		ASTERN(O_CH, THERETHERE());
517 		break;
518 	case '{':
519 		count = p_count(p);
520 		if (EAT(',')) {
521 			if (isdigit((ut8)PEEK())) {
522 				count2 = p_count(p);
523 				REQUIRE(count <= count2, R_REGEX_BADBR);
524 			} else { /* single number with comma */
525 				count2 = INTFINITY;
526 			}
527 		} else { /* just a single number */
528 			count2 = count;
529 		}
530 		repeat(p, pos, count, count2);
531 		if (!EAT('}')) {	/* error heuristics */
532 			while (MORE () && PEEK () != '}') {
533 				NEXT ();
534 			}
535 			REQUIRE(MORE(), R_REGEX_EBRACE);
536 			SETERROR(R_REGEX_BADBR);
537 		}
538 		break;
539 	}
540 
541 	if (!MORE ()) {
542 		return;
543 	}
544 	c = PEEK();
545 	if (!(c == '*' || c == '+' || c == '?' ||
546 		    (c == '{' && MORE2 () && isdigit ((ut8)PEEK2 ())))) {
547 		return;
548 	}
549 	SETERROR(R_REGEX_BADRPT);
550 }
551 
552 /*
553  - p_str - string (no metacharacters) "parser"
554  */
p_str(struct parse * p)555 static void p_str(struct parse *p) {
556 	REQUIRE(MORE(), R_REGEX_EMPTY);
557 	while (MORE()) {
558 		ordinary(p, GETNEXT());
559 	}
560 }
561 
562 /*
563  - p_bre - BRE parser top level, anchoring and concatenation
564  * Giving end1 as OUT essentially eliminates the end1/end2 check.
565  *
566  * This implementation is a bit of a kludge, in that a trailing $ is first
567  * taken as an ordinary character and then revised to be an anchor.  The
568  * only undesirable side effect is that '$' gets included as a character
569  * category in such cases.  This is fairly harmless; not worth fixing.
570  * The amount of lookahead needed to avoid this kludge is excessive.
571  */
p_bre(struct parse * p,int end1,int end2)572 static void p_bre(struct parse *p,
573     int end1,		/* first terminating character */
574     int end2)		/* second terminating character */
575 {
576 	sopno start = HERE();
577 	int first = 1;			/* first subexpression? */
578 	int wasdollar = 0;
579 
580 	if (EAT('^')) {
581 		EMIT(OBOL, 0);
582 		p->g->iflags |= USEBOL;
583 		p->g->nbol++;
584 	}
585 	while (MORE() && !SEETWO(end1, end2)) {
586 		wasdollar = p_simp_re(p, first);
587 		first = 0;
588 	}
589 	if (wasdollar) {	/* oops, that was a trailing anchor */
590 		DROP(1);
591 		EMIT(OEOL, 0);
592 		p->g->iflags |= USEEOL;
593 		p->g->neol++;
594 	}
595 
596 	REQUIRE(HERE() != start, R_REGEX_EMPTY);	/* require nonempty */
597 }
598 
599 /*
600  - p_simp_re - parse a simple RE, an atom possibly followed by a repetition
601  */
602 static int			/* was the simple RE an unbackslashed $? */
p_simp_re(struct parse * p,int starordinary)603 p_simp_re(struct parse *p,
604     int starordinary)		/* is a leading * an ordinary character? */
605 {
606 	int c;
607 	int count;
608 	int count2;
609 	sopno pos;
610 	int i;
611 	sopno subno;
612 #	define	BACKSL	(1<<CHAR_BIT)
613 
614 	pos = HERE();		/* repetion op, if any, covers from here */
615 
616 	if (!MORE()) {		/* caller should have ensured this */
617 		return 0;
618 	}
619 	c = GETNEXT();
620 	if (c == '\\') {
621 		REQUIRE(MORE(), R_REGEX_EESCAPE);
622 		c = BACKSL | GETNEXT();
623 	}
624 	switch (c) {
625 	case '.':
626 		if (p->g->cflags & R_REGEX_NEWLINE) {
627 			nonnewline(p);
628 		} else {
629 			EMIT (OANY, 0);
630 		}
631 		break;
632 	case '[':
633 		p_bracket(p);
634 		break;
635 	case BACKSL|'{':
636 		SETERROR(R_REGEX_BADRPT);
637 		break;
638 	case BACKSL|'(':
639 		p->g->nsub++;
640 		subno = p->g->nsub;
641 		if (subno < NPAREN) {
642 			p->pbegin[subno] = HERE ();
643 		}
644 		EMIT(OLPAREN, subno);
645 		/* the MORE here is an error heuristic */
646 		if (MORE () && !SEETWO ('\\', ')')) {
647 			p_bre (p, '\\', ')');
648 		}
649 		if (subno < NPAREN) {
650 			p->pend[subno] = HERE();
651 			if (!p->pend[subno]) {
652 				break;
653 			}
654 		}
655 		EMIT(ORPAREN, subno);
656 		REQUIRE(EATTWO('\\', ')'), R_REGEX_EPAREN);
657 		break;
658 	case BACKSL|')':	/* should not get here -- must be user */
659 	case BACKSL|'}':
660 		SETERROR(R_REGEX_EPAREN);
661 		break;
662 	case BACKSL|'1':
663 	case BACKSL|'2':
664 	case BACKSL|'3':
665 	case BACKSL|'4':
666 	case BACKSL|'5':
667 	case BACKSL|'6':
668 	case BACKSL|'7':
669 	case BACKSL|'8':
670 	case BACKSL|'9':
671 		i = (c&~BACKSL) - '0';
672 		if (p->pend[i] != 0) {
673 			if (i <= p->g->nsub) {
674 				EMIT(OBACK_, i);
675 				if (p->pbegin[i] != 0 && OP(p->strip[p->pbegin[i]]) == OLPAREN &&
676 				  OP(p->strip[p->pend[i]]) == ORPAREN) {
677 					(void) dupl(p, p->pbegin[i]+1, p->pend[i]);
678 					EMIT(O_BACK, i);
679 				}
680 			}
681 		} else {
682 			SETERROR(R_REGEX_ESUBREG);
683 		}
684 		p->g->backrefs = 1;
685 		break;
686 	case '*':
687 		REQUIRE(starordinary, R_REGEX_BADRPT);
688 		/* FALLTHROUGH */
689 	default:
690 		ordinary(p, (char)c);
691 		break;
692 	}
693 
694 	if (EAT('*')) {		/* implemented as +? */
695 		/* this case does not require the (y|) trick, noKLUDGE */
696 		INSERT(OPLUS_, pos);
697 		ASTERN(O_PLUS, pos);
698 		INSERT(OQUEST_, pos);
699 		ASTERN(O_QUEST, pos);
700 	} else if (EATTWO('\\', '{')) {
701 		count = p_count(p);
702 		if (EAT(',')) {
703 			if (MORE() && isdigit((ut8)PEEK())) {
704 				count2 = p_count(p);
705 				REQUIRE(count <= count2, R_REGEX_BADBR);
706 			} else { /* single number with comma */
707 				count2 = INTFINITY;
708 			}
709 		} else { /* just a single number */
710 			count2 = count;
711 		}
712 		repeat(p, pos, count, count2);
713 		if (!EATTWO('\\', '}')) {	/* error heuristics */
714 			while (MORE () && !SEETWO ('\\', '}')) {
715 				NEXT ();
716 			}
717 			REQUIRE(MORE(), R_REGEX_EBRACE);
718 			SETERROR(R_REGEX_BADBR);
719 		}
720 	} else if (c == '$') { /* $ (but not \$) ends it */
721 		return (1);
722 	}
723 
724 	return(0);
725 }
726 
727 /*
728  - p_count - parse a repetition count
729  */
730 static int			/* the value */
p_count(struct parse * p)731 p_count(struct parse *p)
732 {
733 	int count = 0;
734 	int ndigits = 0;
735 
736 	while (MORE() && isdigit((ut8)PEEK()) && count <= DUPMAX) {
737 		count = count*10 + (GETNEXT() - '0');
738 		ndigits++;
739 	}
740 
741 	REQUIRE(ndigits > 0 && count <= DUPMAX, R_REGEX_BADBR);
742 	return(count);
743 }
744 
745 /*
746  - p_bracket - parse a bracketed character list
747  *
748  * Note a significant property of this code:  if the allocset() did SETERROR,
749  * no set operations are done.
750  */
p_bracket(struct parse * p)751 static void p_bracket(struct parse *p) {
752 	cset *cs;
753 	int invert = 0;
754 
755 	/* Dept of Truly Sickening Special-Case Kludges */
756 	if (p->next + 5 < p->end && strncmp(p->next, "[:<:]]", 6) == 0) {
757 		EMIT(OBOW, 0);
758 		NEXTn(6);
759 		return;
760 	}
761 	if (p->next + 5 < p->end && strncmp(p->next, "[:>:]]", 6) == 0) {
762 		EMIT(OEOW, 0);
763 		NEXTn(6);
764 		return;
765 	}
766 
767 	if (!(cs = allocset(p))) {
768 		/* allocset did set error status in p */
769 		return;
770 	}
771 
772 	if (EAT ('^')) {
773 		invert++; /* make note to invert set at end */
774 	}
775 	if (EAT (']')) {
776 		CHadd(cs, ']');
777 	} else if (EAT ('-')) {
778 		CHadd (cs, '-');
779 	}
780 	while (MORE () && PEEK () != ']' && !SEETWO ('-', ']')) {
781 		p_b_term (p, cs);
782 	}
783 	if (EAT ('-')) {
784 		CHadd (cs, '-');
785 	}
786 	MUSTEAT(']', R_REGEX_EBRACK);
787 
788 	if (p->error != 0) {	/* don't mess things up further */
789 		freeset(p, cs);
790 		return;
791 	}
792 
793 	if (p->g->cflags&R_REGEX_ICASE) {
794 		int i;
795 		int ci;
796 
797 		for (i = p->g->csetsize - 1; i >= 0; i--) {
798 			if (CHIN(cs, i) && isalpha(i)) {
799 				ci = othercase(i);
800 				if (ci != i) {
801 					CHadd (cs, ci);
802 				}
803 			}
804 		}
805 		if (cs->multis != NULL) {
806 			mccase (p, cs);
807 		}
808 	}
809 	if (invert) {
810 		int i;
811 
812 		for (i = p->g->csetsize - 1; i >= 0; i--) {
813 			if (CHIN (cs, i)) {
814 				CHsub(cs, i);
815 			} else {
816 				CHadd (cs, i);
817 			}
818 		}
819 		if (p->g->cflags & R_REGEX_NEWLINE) {
820 			CHsub (cs, '\n');
821 		}
822 		if (cs->multis != NULL) {
823 			mcinvert (p, cs);
824 		}
825 	}
826 
827 	if (cs->multis) {		/* xxx */
828 		return;
829 	}
830 
831 	if (nch(p, cs) == 1) {		/* optimize singleton sets */
832 		ordinary(p, firstch(p, cs));
833 		freeset(p, cs);
834 	} else {
835 		EMIT(OANYOF, freezeset(p, cs));
836 	}
837 }
838 
839 /*
840  - p_b_term - parse one term of a bracketed character list
841  */
p_b_term(struct parse * p,cset * cs)842 static void p_b_term(struct parse *p, cset *cs) {
843 	char c;
844 	char start = 0, finish;
845 	int i;
846 
847 	/* classify what we've got */
848 	switch ((MORE()) ? PEEK() : '\0') {
849 	case '[':
850 		c = (MORE2()) ? PEEK2() : '\0';
851 		break;
852 	case '-':
853 		SETERROR(R_REGEX_ERANGE);
854 		return;			/* NOTE RETURN */
855 		break;
856 	default:
857 		c = '\0';
858 		break;
859 	}
860 
861 	switch (c) {
862 	case ':':		/* character class */
863 		NEXT2();
864 		REQUIRE(MORE(), R_REGEX_EBRACK);
865 		c = PEEK();
866 		REQUIRE(c != '-' && c != ']', R_REGEX_ECTYPE);
867 		p_b_cclass(p, cs);
868 		REQUIRE(MORE(), R_REGEX_EBRACK);
869 		REQUIRE(EATTWO(':', ']'), R_REGEX_ECTYPE);
870 		break;
871 	case '=':		/* equivalence class */
872 		NEXT2();
873 		REQUIRE(MORE(), R_REGEX_EBRACK);
874 		c = PEEK();
875 		REQUIRE(c != '-' && c != ']', R_REGEX_ECOLLATE);
876 		p_b_eclass(p, cs);
877 		REQUIRE(MORE(), R_REGEX_EBRACK);
878 		REQUIRE(EATTWO('=', ']'), R_REGEX_ECOLLATE);
879 		break;
880 	default:		/* symbol, ordinary character, or range */
881 /* xxx revision needed for multichar stuff */
882 		start = p_b_symbol(p);
883 		if (SEE('-') && MORE2() && PEEK2() != ']') {
884 			/* range */
885 			NEXT();
886 			if (EAT ('-')) {
887 				finish = '-';
888 			} else {
889 				finish = p_b_symbol (p);
890 			}
891 		} else {
892 			finish = start;
893 		}
894 		/* xxx what about signed chars here... */
895 		REQUIRE(start <= finish, R_REGEX_ERANGE);
896 		for (i = start; i <= finish; i++) {
897 			CHadd (cs, i);
898 		}
899 		break;
900 	}
901 }
902 
903 /*
904  - p_b_cclass - parse a character-class name and deal with it
905  */
p_b_cclass(struct parse * p,cset * cs)906 static void p_b_cclass(struct parse *p, cset *cs) {
907 	char *sp = p->next;
908 	struct cclass *cp;
909 	size_t len;
910 	char *u;
911 	char c;
912 
913 	while (MORE () && isalpha ((unsigned char)PEEK ())) {
914 		NEXT ();
915 	}
916 	len = p->next - sp;
917 	for (cp = cclasses; cp->name != NULL; cp++) {
918 		if (strncmp (cp->name, sp, len) == 0 && cp->name[len] == '\0') {
919 			break;
920 		}
921 	}
922 	if (!cp->name) {
923 		/* oops, didn't find it */
924 		SETERROR(R_REGEX_ECTYPE);
925 		return;
926 	}
927 
928 	u = cp->chars;
929 	while ((c = *u++) != '\0') {
930 		CHadd (cs, c);
931 	}
932 	for (u = cp->multis; *u != '\0'; u += strlen (u) + 1) {
933 		MCadd (p, cs, u);
934 	}
935 }
936 
937 /*
938  - p_b_eclass - parse an equivalence-class name and deal with it
939  *
940  * This implementation is incomplete. xxx
941  */
p_b_eclass(struct parse * p,cset * cs)942 static void p_b_eclass(struct parse *p, cset *cs) {
943 	char c;
944 
945 	c = p_b_coll_elem(p, '=');
946 	CHadd(cs, c);
947 }
948 
949 /*
950  - p_b_symbol - parse a character or [..]ed multicharacter collating symbol
951  */
952 static char			/* value of symbol */
p_b_symbol(struct parse * p)953 p_b_symbol(struct parse *p)
954 {
955 	char value;
956 
957 	REQUIRE(MORE(), R_REGEX_EBRACK);
958 	if (!EATTWO ('[', '.')) {
959 		return (GETNEXT ());
960 	}
961 
962 	/* collating symbol */
963 	value = p_b_coll_elem(p, '.');
964 	REQUIRE(EATTWO('.', ']'), R_REGEX_ECOLLATE);
965 	return(value);
966 }
967 
968 /*
969  - p_b_coll_elem - parse a collating-element name and look it up
970  */
971 static char			/* value of collating element */
p_b_coll_elem(struct parse * p,int endc)972 p_b_coll_elem(struct parse *p,
973     int endc)			/* name ended by endc,']' */
974 {
975 	char *sp = p->next;
976 	struct cname *cp;
977 	int len;
978 
979 	while (MORE () && !SEETWO (endc, ']')) {
980 		NEXT ();
981 	}
982 	if (!MORE()) {
983 		SETERROR(R_REGEX_EBRACK);
984 		return(0);
985 	}
986 	len = p->next - sp;
987 	for (cp = cnames; cp->name != NULL; cp++) {
988 		if (strncmp (cp->name, sp, len) == 0 && cp->name[len] == '\0') {
989 			return (cp->code); /* known name */
990 		}
991 	}
992 	if (len == 1) {
993 		return (*sp); /* single character */
994 	}
995 	SETERROR(R_REGEX_ECOLLATE);			/* neither */
996 	return(0);
997 }
998 
999 /*
1000  - othercase - return the case counterpart of an alphabetic
1001  */
1002 static char			/* if no counterpart, return ch */
othercase(int ch)1003 othercase(int ch)
1004 {
1005 	ch = (ut8)ch;
1006 	if (isalpha(ch)) {
1007 		if (isupper (ch)) {
1008 			return ((ut8)tolower(ch));
1009 		} else if (islower (ch)) {
1010 			return ((ut8)toupper(ch));
1011 		} else { /* peculiar, but could happen */
1012 			return (ch);
1013 		}
1014 	}
1015 	return ch;
1016 }
1017 
1018 /*
1019  - bothcases - emit a dualcase version of a two-case character
1020  *
1021  * Boy, is this implementation ever a kludge...
1022  */
bothcases(struct parse * p,int ch)1023 static void bothcases(struct parse *p, int ch) {
1024 	char *oldnext = p->next;
1025 	char *oldend = p->end;
1026 	char bracket[3];
1027 
1028 	ch = (ut8)ch;
1029 	if (othercase(ch) != ch) {	/* p_bracket() would recurse */
1030 		p->next = bracket;
1031 		p->end = bracket+2;
1032 		bracket[0] = ch;
1033 		bracket[1] = ']';
1034 		bracket[2] = '\0';
1035 		p_bracket(p);
1036 		if (p->next == bracket+2) {
1037 			p->next = oldnext;
1038 			p->end = oldend;
1039 		}
1040 	}
1041 }
1042 
1043 /*
1044  - ordinary - emit an ordinary character
1045  */
1046 static void
ordinary(struct parse * p,int ch)1047 ordinary(struct parse *p, int ch)
1048 {
1049 	cat_t *cap = p->g->categories;
1050 
1051 	if ((p->g->cflags & R_REGEX_ICASE) && isalpha ((ut8)ch) && othercase (ch) != ch) {
1052 		bothcases(p, ch);
1053 	} else {
1054 		EMIT(OCHAR, (ut8)ch);
1055 		if (cap[ch] == 0) {
1056 			cap[ch] = p->g->ncategories++;
1057 		}
1058 	}
1059 }
1060 
1061 static void
special(struct parse * p,int ch)1062 special(struct parse *p, int ch) {
1063 	char *oldnext = p->next;
1064 	char *oldend = p->end;
1065 	char bracket[16] = {0};
1066 	char digits[3] = {0};
1067 	char c;
1068 	int num = 0;
1069 	switch (ch) {
1070 	case 'x':
1071 		digits[0] = GETNEXT ();
1072 		digits[1] = GETNEXT ();
1073 		c = (char)strtol (digits, NULL, 16);
1074 		ordinary (p, c);
1075 		return;
1076 	case 'n':
1077 		ordinary (p, '\n');
1078 		return;
1079 	case 't':
1080 		ordinary (p, '\t');
1081 		return;
1082 	case 'r':
1083 		ordinary (p, '\r');
1084 		return;
1085 	case 's':
1086 		num = 5;
1087 		memcpy (bracket, "\t\r\n ]", num);
1088 		break;
1089 	case 'd':
1090 		num = 4;
1091 		memcpy (bracket, "0-9]", num);
1092 		break;
1093 	case 'w':
1094 		num = 4;
1095 		memcpy (bracket, "a-z]", num);
1096 		break;
1097 	default:
1098 		SETERROR (R_REGEX_INVARG);
1099 		return;
1100 	}
1101 
1102 	p->next = bracket;
1103 	p->end = bracket + num;
1104 
1105 	p_bracket (p);
1106 
1107 	if (p->next == bracket + num) {
1108 		p->next = oldnext;
1109 		p->end = oldend;
1110 	}
1111 }
1112 
1113 /*
1114  - nonnewline - emit R_REGEX_NEWLINE version of OANY
1115  *
1116  * Boy, is this implementation ever a kludge...
1117  */
1118 static void
nonnewline(struct parse * p)1119 nonnewline(struct parse *p)
1120 {
1121 	char *oldnext = p->next;
1122 	char *oldend = p->end;
1123 	char bracket[4];
1124 
1125 	p->next = bracket;
1126 	p->end = bracket+3;
1127 	bracket[0] = '^';
1128 	bracket[1] = '\n';
1129 	bracket[2] = ']';
1130 	bracket[3] = '\0';
1131 	p_bracket(p);
1132 	if (p->next == bracket+3) {
1133 		p->next = oldnext;
1134 		p->end = oldend;
1135 	}
1136 }
1137 
1138 /*
1139  - repeat - generate code for a bounded repetition, recursively if needed
1140  */
1141 static void
repeat(struct parse * p,sopno start,int from,int to)1142 repeat(struct parse *p,
1143     sopno start,		/* operand from here to end of strip */
1144     int from,			/* repeated from this number */
1145     int to)			/* to this number of times (maybe INTFINITY) */
1146 {
1147 	sopno finish = HERE();
1148 #	define	N	2
1149 #	define	INF	3
1150 #	define	REP(f, t)	((f)*8 + (t))
1151 #	define	MAP(n)	(((n) <= 1) ? (n) : ((n) == INTFINITY) ? INF : N)
1152 	sopno copy;
1153 
1154 	if (p->error != 0) { /* head off possible runaway recursion */
1155 		return;
1156 	}
1157 
1158 	if (from > to) {
1159 		return;
1160 	}
1161 
1162 	switch (REP(MAP(from), MAP(to))) {
1163 	case REP(0, 0):			/* must be user doing this */
1164 		DROP(finish-start);	/* drop the operand */
1165 		break;
1166 	case REP(0, 1):			/* as x{1,1}? */
1167 	case REP(0, N):			/* as x{1,n}? */
1168 	case REP(0, INF):		/* as x{1,}? */
1169 		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1170 		INSERT(OCH_, start);		/* offset is wrong... */
1171 		repeat(p, start+1, 1, to);
1172 		ASTERN(OOR1, start);
1173 		AHEAD(start);			/* ... fix it */
1174 		EMIT(OOR2, 0);
1175 		AHEAD(THERE());
1176 		ASTERN(O_CH, THERETHERE());
1177 		break;
1178 	case REP(1, 1):			/* trivial case */
1179 		/* done */
1180 		break;
1181 	case REP(1, N):			/* as x?x{1,n-1} */
1182 		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1183 		INSERT(OCH_, start);
1184 		ASTERN(OOR1, start);
1185 		AHEAD(start);
1186 		EMIT(OOR2, 0);			/* offset very wrong... */
1187 		AHEAD(THERE());			/* ...so fix it */
1188 		ASTERN(O_CH, THERETHERE());
1189 		copy = dupl(p, start+1, finish+1);
1190 		if (copy == finish+4) {
1191 			repeat(p, copy, 1, to-1);
1192 		}
1193 		break;
1194 	case REP(1, INF):		/* as x+ */
1195 		INSERT(OPLUS_, start);
1196 		ASTERN(O_PLUS, start);
1197 		break;
1198 	case REP(N, N):			/* as xx{m-1,n-1} */
1199 		copy = dupl(p, start, finish);
1200 		repeat(p, copy, from-1, to-1);
1201 		break;
1202 	case REP(N, INF):		/* as xx{n-1,INF} */
1203 		copy = dupl(p, start, finish);
1204 		repeat(p, copy, from-1, to);
1205 		break;
1206 	default:			/* "can't happen" */
1207 		SETERROR(R_REGEX_ASSERT);	/* just in case */
1208 		break;
1209 	}
1210 }
1211 
1212 /*
1213  - seterr - set an error condition
1214  */
1215 static int			/* useless but makes type checking happy */
seterr(struct parse * p,int e)1216 seterr(struct parse *p, int e)
1217 {
1218 	if (p->error == 0) { /* keep earliest error condition */
1219 		p->error = e;
1220 	}
1221 	p->next = nuls;		/* try to bring things to a halt */
1222 	p->end = nuls;
1223 	return(0);		/* make the return value well-defined */
1224 }
1225 
1226 /*
1227  - allocset - allocate a set of characters for []
1228  */
allocset(struct parse * p)1229 static cset * allocset(struct parse *p) {
1230 	int no = p->g->ncsets++;
1231 	size_t nc;
1232 	size_t nbytes;
1233 	cset *cs;
1234 	size_t css = (size_t)p->g->csetsize;
1235 	int i;
1236 
1237 	if (no >= p->ncsalloc) {	/* need another column of space */
1238 		void *ptr;
1239 
1240 		p->ncsalloc += CHAR_BIT;
1241 		nc = p->ncsalloc;
1242 		if (nc % CHAR_BIT) {
1243 			goto nomem;
1244 		}
1245 		nbytes = nc / CHAR_BIT * css;
1246 
1247 		ptr = (cset *)realloc((char *)p->g->sets, nc * sizeof(cset));
1248 		if (!ptr) {
1249 			goto nomem;
1250 		}
1251 		p->g->sets = ptr;
1252 
1253 		ptr = (ut8 *)realloc((char *)p->g->setbits, nbytes);
1254 		if (!ptr) {
1255 			goto nomem;
1256 		}
1257 		p->g->setbits = ptr;
1258 
1259 		for (i = 0; i < no; i++) {
1260 			p->g->sets[i].ptr = p->g->setbits + css * (i / CHAR_BIT);
1261 		}
1262 
1263 		(void) memset((char *)p->g->setbits + (nbytes - css), 0, css);
1264 	}
1265 	/* XXX should not happen */
1266 	if (!p->g->sets || !p->g->setbits) {
1267 		goto nomem;
1268 	}
1269 
1270 	cs = &p->g->sets[no];
1271 	cs->ptr = p->g->setbits + css*((no)/CHAR_BIT);
1272 	cs->mask = 1 << ((no) % CHAR_BIT);
1273 	cs->hash = 0;
1274 	cs->smultis = 0;
1275 	cs->multis = NULL;
1276 
1277 	return(cs);
1278 nomem:
1279 	R_FREE(p->g->sets);
1280 	R_FREE(p->g->setbits);
1281 
1282 	SETERROR(R_REGEX_ESPACE);
1283 	/* caller's responsibility not to do set ops */
1284 	return(NULL);
1285 }
1286 
1287 /*
1288  - freeset - free a now-unused set
1289  */
freeset(struct parse * p,cset * cs)1290 static void freeset(struct parse *p, cset *cs) {
1291 	int i;
1292 	cset *top = &p->g->sets[p->g->ncsets];
1293 	size_t css = (size_t)p->g->csetsize;
1294 
1295 	for (i = 0; i < css; i++) {
1296 		CHsub (cs, i);
1297 	}
1298 	if (cs == top - 1) { /* recover only the easy case */
1299 		p->g->ncsets--;
1300 	}
1301 }
1302 
1303 /*
1304  - freezeset - final processing on a set of characters
1305  *
1306  * The main task here is merging identical sets.  This is usually a waste
1307  * of time (although the hash code minimizes the overhead), but can win
1308  * big if R_REGEX_ICASE is being used.  R_REGEX_ICASE, by the way, is why the hash
1309  * is done using addition rather than xor -- all ASCII [aA] sets xor to
1310  * the same value!
1311  */
1312 static int			/* set number */
freezeset(struct parse * p,cset * cs)1313 freezeset(struct parse *p, cset *cs)
1314 {
1315 	ut8 h = cs->hash;
1316 	int i;
1317 	cset *top = &p->g->sets[p->g->ncsets];
1318 	cset *cs2;
1319 	size_t css = (size_t)p->g->csetsize;
1320 
1321 	/* look for an earlier one which is the same */
1322 	for (cs2 = &p->g->sets[0]; cs2 < top; cs2++) {
1323 		if (cs2->hash == h && cs2 != cs) {
1324 			/* maybe */
1325 			for (i = 0; i < css; i++) {
1326 				if (!!CHIN (cs2, i) != !!CHIN (cs, i)) {
1327 					break; /* no */
1328 				}
1329 			}
1330 			if (i == css) {
1331 				break; /* yes */
1332 			}
1333 		}
1334 	}
1335 
1336 	if (cs2 < top) {	/* found one */
1337 		freeset(p, cs);
1338 		cs = cs2;
1339 	}
1340 
1341 	return((int)(cs - p->g->sets));
1342 }
1343 
1344 /*
1345  - firstch - return first character in a set (which must have at least one)
1346  */
1347 static int			/* character; there is no "none" value */
firstch(struct parse * p,cset * cs)1348 firstch(struct parse *p, cset *cs)
1349 {
1350 	int i;
1351 	size_t css = (size_t)p->g->csetsize;
1352 
1353 	for (i = 0; i < css; i++) {
1354 		if (CHIN (cs, i)) {
1355 			return ((char)i);
1356 		}
1357 	}
1358 	return(0);		/* arbitrary */
1359 }
1360 
1361 /*
1362  - nch - number of characters in a set
1363  */
nch(struct parse * p,cset * cs)1364 static int nch(struct parse *p, cset *cs) {
1365 	int i;
1366 	size_t css = (size_t)p->g->csetsize;
1367 	int n = 0;
1368 
1369 	for (i = 0; i < css; i++) {
1370 		if (CHIN (cs, i)) {
1371 			n++;
1372 		}
1373 	}
1374 	return(n);
1375 }
1376 
1377 /*
1378  - mcadd - add a collating element to a cset
1379  */
mcadd(struct parse * p,cset * cs,char * cp)1380 static void mcadd( struct parse *p, cset *cs, char *cp) {
1381 	size_t oldend = cs->smultis;
1382 	void *np;
1383 
1384 	cs->smultis += strlen(cp) + 1;
1385 	np = realloc(cs->multis, cs->smultis);
1386 	if (!np) {
1387 		if (cs->multis) {
1388 			free (cs->multis);
1389 		}
1390 		cs->multis = NULL;
1391 		SETERROR(R_REGEX_ESPACE);
1392 		return;
1393 	}
1394 	cs->multis = np;
1395 
1396 	r_str_ncpy (cs->multis + oldend - 1, cp, cs->smultis - oldend + 1);
1397 }
1398 
1399 /*
1400  - mcinvert - invert the list of collating elements in a cset
1401  *
1402  * This would have to know the set of possibilities.  Implementation
1403  * is deferred.
1404  */
1405 /* ARGSUSED */
mcinvert(struct parse * p,cset * cs)1406 static void mcinvert(struct parse *p, cset *cs) {
1407 	//asert(!cs->multis);	/* xxx */
1408 	return;
1409 }
1410 
1411 /*
1412  - mccase - add case counterparts of the list of collating elements in a cset
1413  *
1414  * This would have to know the set of possibilities.  Implementation
1415  * is deferred.
1416  */
1417 /* ARGSUSED */
mccase(struct parse * p,cset * cs)1418 static void mccase(struct parse *p, cset *cs) {
1419 	//asert(!cs->multis);	/* xxx */
1420 	return;
1421 }
1422 
1423 /*
1424  - isinsets - is this character in any sets?
1425  */
1426 static int			/* predicate */
isinsets(struct re_guts * g,int c)1427 isinsets(struct re_guts *g, int c)
1428 {
1429 	ut8 *col;
1430 	int i;
1431 	int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1432 	unsigned uc = (ut8)c;
1433 
1434 	for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize) {
1435 		if (col[uc] != 0) {
1436 			return (1);
1437 		}
1438 	}
1439 	return(0);
1440 }
1441 
1442 /*
1443  - samesets - are these two characters in exactly the same sets?
1444  */
1445 static int			/* predicate */
samesets(struct re_guts * g,int c1,int c2)1446 samesets(struct re_guts *g, int c1, int c2)
1447 {
1448 	ut8 *col;
1449 	int i;
1450 	int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1451 	unsigned uc1 = (ut8)c1;
1452 	unsigned uc2 = (ut8)c2;
1453 
1454 	for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize) {
1455 		if (col[uc1] != col[uc2]) {
1456 			return (0);
1457 		}
1458 	}
1459 	return(1);
1460 }
1461 
1462 /*
1463  - categorize - sort out character categories
1464  */
1465 static void
categorize(struct parse * p,struct re_guts * g)1466 categorize(struct parse *p, struct re_guts *g)
1467 {
1468 	cat_t *cats = g? g->categories : NULL;
1469 	unsigned int c;
1470 	unsigned int c2;
1471 	cat_t cat;
1472 
1473 	/* avoid making error situations worse */
1474 	if (!p || p->error != 0 || !cats) {
1475 		return;
1476 	}
1477 
1478 	for (c = CHAR_MIN; c <= CHAR_MAX; c++) {
1479 		if ( *(cats+c) && isinsets(g, c)) {
1480 			cat = g->ncategories++;
1481 			cats[c] = cat;
1482 			for (c2 = c + 1; c2 <= CHAR_MAX; c2++) {
1483 				if (cats[c2] == 0 && samesets (g, c, c2)) {
1484 					cats[c2] = cat;
1485 				}
1486 			}
1487 		}
1488 	}
1489 }
1490 
1491 /*
1492  - dupl - emit a duplicate of a bunch of sops
1493  */
1494 static sopno			/* start of duplicate */
dupl(struct parse * p,sopno start,sopno finish)1495 dupl(struct parse *p,
1496     sopno start,		/* from here */
1497     sopno finish)		/* to this less one */
1498 {
1499 	sopno ret = HERE();
1500 	sopno len = finish - start;
1501 
1502 	if (finish >= start) {
1503 		if (len == 0) {
1504 			return (ret);
1505 		}
1506 		enlarge(p, p->ssize + len);	/* this many unexpected additions */
1507 		if (p->ssize >= p->slen + len) {
1508 			(void) memcpy((char *)(p->strip + p->slen),
1509 			  (char *)(p->strip + start), (size_t)len*sizeof(sop));
1510 			p->slen += len;
1511 			return(ret);
1512 		}
1513 	}
1514 	return ret;
1515 }
1516 
1517 /*
1518  - doemit - emit a strip operator
1519  *
1520  * It might seem better to implement this as a macro with a function as
1521  * hard-case backup, but it's just too big and messy unless there are
1522  * some changes to the data structures.  Maybe later.
1523  */
1524 static void
doemit(struct parse * p,sop op,size_t opnd)1525 doemit(struct parse *p, sop op, size_t opnd)
1526 {
1527 	/* avoid making error situations worse */
1528 	if (p->error != 0) {
1529 		return;
1530 	}
1531 
1532 	/* deal with oversize operands ("can't happen", more or less) */
1533 	if (opnd < 1<<OPSHIFT) {
1534 
1535 		/* deal with undersized strip */
1536 		if (p->slen >= p->ssize) {
1537 			enlarge (p, (p->ssize + 1) / 2 * 3); /* +50% */
1538 		}
1539 		if (p->slen < p->ssize) {
1540 			/* finally, it's all reduced to the easy case */
1541 			p->strip[p->slen++] = SOP(op, opnd);
1542 		}
1543 	}
1544 }
1545 
1546 /*
1547  - doinsert - insert a sop into the strip
1548  */
1549 static void
doinsert(struct parse * p,sop op,size_t opnd,sopno pos)1550 doinsert(struct parse *p, sop op, size_t opnd, sopno pos)
1551 {
1552 	sopno sn;
1553 	sop s;
1554 	int i;
1555 
1556 	/* avoid making error situations worse */
1557 	if (p->error != 0) {
1558 		return;
1559 	}
1560 
1561 	sn = HERE();
1562 	EMIT(op, opnd);		/* do checks, ensure space */
1563 	if (HERE() != sn+1) {
1564 		return;
1565 	}
1566 	s = p->strip[sn];
1567 
1568 	/* adjust paren pointers */
1569 	if (pos > 0) {
1570 		for (i = 1; i < NPAREN; i++) {
1571 			if (p->pbegin[i] >= pos) {
1572 				p->pbegin[i]++;
1573 			}
1574 			if (p->pend[i] >= pos) {
1575 				p->pend[i]++;
1576 			}
1577 		}
1578 	}
1579 
1580 	memmove((char *)&p->strip[pos+1], (char *)&p->strip[pos],
1581 						(HERE()-pos-1)*sizeof(sop));
1582 	p->strip[pos] = s;
1583 }
1584 
1585 /*
1586  - dofwd - complete a forward reference
1587  */
1588 static void
dofwd(struct parse * p,sopno pos,sop value)1589 dofwd(struct parse *p, sopno pos, sop value)
1590 {
1591 	/* avoid making error situations worse */
1592 	if (p->error != 0) {
1593 		return;
1594 	}
1595 
1596 	if (value < 1<<OPSHIFT) {
1597 		p->strip[pos] = OP(p->strip[pos]) | value;
1598 	}
1599 }
1600 
1601 /*
1602  - enlarge - enlarge the strip
1603  */
1604 static void
enlarge(struct parse * p,sopno size)1605 enlarge(struct parse *p, sopno size)
1606 {
1607 	sop *sp;
1608 
1609 	if (p->ssize >= size) {
1610 		return;
1611 	}
1612 
1613 	sp = (sop *)realloc(p->strip, size*sizeof(sop));
1614 	if (!sp) {
1615 		SETERROR(R_REGEX_ESPACE);
1616 		return;
1617 	}
1618 	p->strip = sp;
1619 	p->ssize = size;
1620 }
1621 
1622 /*
1623  - stripsnug - compact the strip
1624  */
1625 static void
stripsnug(struct parse * p,struct re_guts * g)1626 stripsnug(struct parse *p, struct re_guts *g)
1627 {
1628 	g->nstates = p->slen;
1629 	g->strip = (sop *)realloc((char *)p->strip, p->slen * sizeof(sop));
1630 	if (!g->strip) {
1631 		SETERROR(R_REGEX_ESPACE);
1632 		g->strip = p->strip;
1633 	}
1634 }
1635 
1636 /*
1637  - findmust - fill in must and mlen with longest mandatory literal string
1638  *
1639  * This algorithm could do fancy things like analyzing the operands of |
1640  * for common subsequences.  Someday.  This code is simple and finds most
1641  * of the interesting cases.
1642  *
1643  * Note that must and mlen got initialized during setup.
1644  */
1645 static void
findmust(struct parse * p,struct re_guts * g)1646 findmust(struct parse *p, struct re_guts *g)
1647 {
1648 	sop *scan;
1649 	sop *start = NULL;    /* start initialized in the default case, after that */
1650 	sop *newstart = NULL; /* newstart was initialized in the OCHAR case */
1651 	sopno newlen;
1652 	sop s;
1653 	char *cp;
1654 	sopno i;
1655 
1656 	/* avoid making error situations worse */
1657 	if (p->error != 0) {
1658 		return;
1659 	}
1660 
1661 	/* find the longest OCHAR sequence in strip */
1662 	newlen = 0;
1663 	start = scan = g->strip + 1;
1664 	do {
1665 		s = *scan++;
1666 		switch (OP(s)) {
1667 		case OCHAR:		/* sequence member */
1668 			if (newlen == 0) { /* new sequence */
1669 				newstart = scan - 1;
1670 			}
1671 			newlen++;
1672 			break;
1673 		case OPLUS_:		/* things that don't break one */
1674 		case OLPAREN:
1675 		case ORPAREN:
1676 			break;
1677 		case OQUEST_:		/* things that must be skipped */
1678 		case OCH_:
1679 			scan--;
1680 			do {
1681 				scan += OPND(s);
1682 				s = *scan;
1683 				/* asert() interferes w debug printouts */
1684 				if (OP(s) != O_QUEST && OP(s) != O_CH &&
1685 							OP(s) != OOR2) {
1686 					g->iflags |= BAD;
1687 					return;
1688 				}
1689 			} while (OP(s) != O_QUEST && OP(s) != O_CH);
1690 			/* fallthrough */
1691 		default:		/* things that break a sequence */
1692 			if (newlen > g->mlen) {		/* ends one */
1693 				start = newstart;
1694 				g->mlen = newlen;
1695 			}
1696 			newlen = 0;
1697 			break;
1698 		}
1699 	} while (OP(s) != OEND);
1700 
1701 	if (g->mlen == 0) { /* there isn't one */
1702 		return;
1703 	}
1704 
1705 	/* turn it into a character string */
1706 	g->must = malloc((size_t)g->mlen + 1);
1707 	if (!g->must) {		/* argh; just forget it */
1708 		g->mlen = 0;
1709 		return;
1710 	}
1711 	cp = g->must;
1712 	scan = start;
1713 	for (i = g->mlen; i > 0; i--) {
1714 		while (OP (s = *scan++) != OCHAR) {
1715 			continue;
1716 		}
1717 		if (cp < g->must + g->mlen) {
1718 			*cp++ = (char)OPND(s);
1719 		}
1720 	}
1721 	if (cp == g->must + g->mlen) {
1722 		*cp++ = '\0';		/* just on general principles */
1723 	}
1724 }
1725 
1726 /*
1727  - pluscount - count + nesting
1728  */
1729 static sopno			/* nesting depth */
pluscount(struct parse * p,struct re_guts * g)1730 pluscount(struct parse *p, struct re_guts *g)
1731 {
1732 	sop *scan;
1733 	sop s;
1734 	sopno plusnest = 0;
1735 	sopno maxnest = 0;
1736 
1737 	if (p->error != 0) {
1738 		return (0); /* there may not be an OEND */
1739 	}
1740 
1741 	scan = g->strip + 1;
1742 	do {
1743 		s = *scan++;
1744 		switch (OP(s)) {
1745 		case OPLUS_:
1746 			plusnest++;
1747 			break;
1748 		case O_PLUS:
1749 			if (plusnest > maxnest) {
1750 				maxnest = plusnest;
1751 			}
1752 			plusnest--;
1753 			break;
1754 		}
1755 	} while (OP(s) != OEND);
1756 	if (plusnest != 0) {
1757 		g->iflags |= BAD;
1758 	}
1759 	return(maxnest);
1760 }
1761