xref: /freebsd/lib/libc/regex/regcomp.c (revision d74e86d9e30043893d6b308468008b65640ddcae)
1 /*-
2  * Copyright (c) 1992, 1993, 1994 Henry Spencer.
3  * Copyright (c) 1992, 1993, 1994
4  *	The Regents of the University of California.  All rights reserved.
5  *
6  * This code is derived from software contributed to Berkeley by
7  * Henry Spencer.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed by the University of
20  *	California, Berkeley and its contributors.
21  * 4. Neither the name of the University nor the names of its contributors
22  *    may be used to endorse or promote products derived from this software
23  *    without specific prior written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35  * SUCH DAMAGE.
36  *
37  *	@(#)regcomp.c	8.5 (Berkeley) 3/20/94
38  */
39 
40 #if defined(LIBC_SCCS) && !defined(lint)
41 static char sccsid[] = "@(#)regcomp.c	8.5 (Berkeley) 3/20/94";
42 #endif /* LIBC_SCCS and not lint */
43 
44 #include <sys/types.h>
45 #include <stdio.h>
46 #include <string.h>
47 #include <ctype.h>
48 #include <limits.h>
49 #include <stdlib.h>
50 #include <regex.h>
51 
52 #include "collate.h"
53 
54 #include "utils.h"
55 #include "regex2.h"
56 
57 #include "cclass.h"
58 #include "cname.h"
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 /* ========= begin header generated by ./mkh ========= */
79 #ifdef __cplusplus
80 extern "C" {
81 #endif
82 
83 /* === regcomp.c === */
84 static void p_ere __P((struct parse *p, int stop));
85 static void p_ere_exp __P((struct parse *p));
86 static void p_str __P((struct parse *p));
87 static void p_bre __P((struct parse *p, int end1, int end2));
88 static int p_simp_re __P((struct parse *p, int starordinary));
89 static int p_count __P((struct parse *p));
90 static void p_bracket __P((struct parse *p));
91 static void p_b_term __P((struct parse *p, cset *cs));
92 static void p_b_cclass __P((struct parse *p, cset *cs));
93 static void p_b_eclass __P((struct parse *p, cset *cs));
94 static char p_b_symbol __P((struct parse *p));
95 static char p_b_coll_elem __P((struct parse *p, int endc));
96 static char othercase __P((int ch));
97 static void bothcases __P((struct parse *p, int ch));
98 static void ordinary __P((struct parse *p, int ch));
99 static void nonnewline __P((struct parse *p));
100 static void repeat __P((struct parse *p, sopno start, int from, int to));
101 static int seterr __P((struct parse *p, int e));
102 static cset *allocset __P((struct parse *p));
103 static void freeset __P((struct parse *p, cset *cs));
104 static int freezeset __P((struct parse *p, cset *cs));
105 static int firstch __P((struct parse *p, cset *cs));
106 static int nch __P((struct parse *p, cset *cs));
107 static void mcadd __P((struct parse *p, cset *cs, char *cp));
108 #if used
109 static void mcsub __P((cset *cs, char *cp));
110 static int mcin __P((cset *cs, char *cp));
111 static char *mcfind __P((cset *cs, char *cp));
112 #endif
113 static void mcinvert __P((struct parse *p, cset *cs));
114 static void mccase __P((struct parse *p, cset *cs));
115 static int isinsets __P((struct re_guts *g, int c));
116 static int samesets __P((struct re_guts *g, int c1, int c2));
117 static void categorize __P((struct parse *p, struct re_guts *g));
118 static sopno dupl __P((struct parse *p, sopno start, sopno finish));
119 static void doemit __P((struct parse *p, sop op, size_t opnd));
120 static void doinsert __P((struct parse *p, sop op, size_t opnd, sopno pos));
121 static void dofwd __P((struct parse *p, sopno pos, sop value));
122 static void enlarge __P((struct parse *p, sopno size));
123 static void stripsnug __P((struct parse *p, struct re_guts *g));
124 static void findmust __P((struct parse *p, struct re_guts *g));
125 static sopno pluscount __P((struct parse *p, struct re_guts *g));
126 
127 #ifdef __cplusplus
128 }
129 #endif
130 /* ========= end header generated by ./mkh ========= */
131 
132 static char nuls[10];		/* place to point scanner in event of error */
133 
134 /*
135  * macros for use with parse structure
136  * BEWARE:  these know that the parse structure is named `p' !!!
137  */
138 #define	PEEK()	(*p->next)
139 #define	PEEK2()	(*(p->next+1))
140 #define	MORE()	(p->next < p->end)
141 #define	MORE2()	(p->next+1 < p->end)
142 #define	SEE(c)	(MORE() && PEEK() == (c))
143 #define	SEETWO(a, b)	(MORE() && MORE2() && PEEK() == (a) && PEEK2() == (b))
144 #define	EAT(c)	((SEE(c)) ? (NEXT(), 1) : 0)
145 #define	EATTWO(a, b)	((SEETWO(a, b)) ? (NEXT2(), 1) : 0)
146 #define	NEXT()	(p->next++)
147 #define	NEXT2()	(p->next += 2)
148 #define	NEXTn(n)	(p->next += (n))
149 #define	GETNEXT()	(*p->next++)
150 #define	SETERROR(e)	seterr(p, (e))
151 #define	REQUIRE(co, e)	((co) || SETERROR(e))
152 #define	MUSTSEE(c, e)	(REQUIRE(MORE() && PEEK() == (c), e))
153 #define	MUSTEAT(c, e)	(REQUIRE(MORE() && GETNEXT() == (c), e))
154 #define	MUSTNOTSEE(c, e)	(REQUIRE(!MORE() || PEEK() != (c), e))
155 #define	EMIT(op, sopnd)	doemit(p, (sop)(op), (size_t)(sopnd))
156 #define	INSERT(op, pos)	doinsert(p, (sop)(op), HERE()-(pos)+1, pos)
157 #define	AHEAD(pos)		dofwd(p, pos, HERE()-(pos))
158 #define	ASTERN(sop, pos)	EMIT(sop, HERE()-pos)
159 #define	HERE()		(p->slen)
160 #define	THERE()		(p->slen - 1)
161 #define	THERETHERE()	(p->slen - 2)
162 #define	DROP(n)	(p->slen -= (n))
163 
164 #ifndef NDEBUG
165 static int never = 0;		/* for use in asserts; shuts lint up */
166 #else
167 #define	never	0		/* some <assert.h>s have bugs too */
168 #endif
169 
170 /*
171  - regcomp - interface for parser and compilation
172  = extern int regcomp(regex_t *, const char *, int);
173  = #define	REG_BASIC	0000
174  = #define	REG_EXTENDED	0001
175  = #define	REG_ICASE	0002
176  = #define	REG_NOSUB	0004
177  = #define	REG_NEWLINE	0010
178  = #define	REG_NOSPEC	0020
179  = #define	REG_PEND	0040
180  = #define	REG_DUMP	0200
181  */
182 int				/* 0 success, otherwise REG_something */
183 regcomp(preg, pattern, cflags)
184 regex_t *preg;
185 const char *pattern;
186 int cflags;
187 {
188 	struct parse pa;
189 	register struct re_guts *g;
190 	register struct parse *p = &pa;
191 	register int i;
192 	register size_t len;
193 #ifdef REDEBUG
194 #	define	GOODFLAGS(f)	(f)
195 #else
196 #	define	GOODFLAGS(f)	((f)&~REG_DUMP)
197 #endif
198 
199 	cflags = GOODFLAGS(cflags);
200 	if ((cflags&REG_EXTENDED) && (cflags&REG_NOSPEC))
201 		return(REG_INVARG);
202 
203 	if (cflags&REG_PEND) {
204 		if (preg->re_endp < pattern)
205 			return(REG_INVARG);
206 		len = preg->re_endp - pattern;
207 	} else
208 		len = strlen((char *)pattern);
209 
210 	/* do the mallocs early so failure handling is easy */
211 	g = (struct re_guts *)malloc(sizeof(struct re_guts) +
212 							(NC-1)*sizeof(cat_t));
213 	if (g == NULL)
214 		return(REG_ESPACE);
215 	p->ssize = len/(size_t)2*(size_t)3 + (size_t)1;	/* ugh */
216 	p->strip = (sop *)malloc(p->ssize * sizeof(sop));
217 	p->slen = 0;
218 	if (p->strip == NULL) {
219 		free((char *)g);
220 		return(REG_ESPACE);
221 	}
222 
223 	/* set things up */
224 	p->g = g;
225 	p->next = (char *)pattern;	/* convenience; we do not modify it */
226 	p->end = p->next + len;
227 	p->error = 0;
228 	p->ncsalloc = 0;
229 	for (i = 0; i < NPAREN; i++) {
230 		p->pbegin[i] = 0;
231 		p->pend[i] = 0;
232 	}
233 	g->csetsize = NC;
234 	g->sets = NULL;
235 	g->setbits = NULL;
236 	g->ncsets = 0;
237 	g->cflags = cflags;
238 	g->iflags = 0;
239 	g->nbol = 0;
240 	g->neol = 0;
241 	g->must = NULL;
242 	g->mlen = 0;
243 	g->nsub = 0;
244 	g->ncategories = 1;	/* category 0 is "everything else" */
245 	g->categories = &g->catspace[-(CHAR_MIN)];
246 	(void) memset((char *)g->catspace, 0, NC*sizeof(cat_t));
247 	g->backrefs = 0;
248 
249 	/* do it */
250 	EMIT(OEND, 0);
251 	g->firststate = THERE();
252 	if (cflags&REG_EXTENDED)
253 		p_ere(p, OUT);
254 	else if (cflags&REG_NOSPEC)
255 		p_str(p);
256 	else
257 		p_bre(p, OUT, OUT);
258 	EMIT(OEND, 0);
259 	g->laststate = THERE();
260 
261 	/* tidy up loose ends and fill things in */
262 	categorize(p, g);
263 	stripsnug(p, g);
264 	findmust(p, g);
265 	g->nplus = pluscount(p, g);
266 	g->magic = MAGIC2;
267 	preg->re_nsub = g->nsub;
268 	preg->re_g = g;
269 	preg->re_magic = MAGIC1;
270 #ifndef REDEBUG
271 	/* not debugging, so can't rely on the assert() in regexec() */
272 	if (g->iflags&BAD)
273 		SETERROR(REG_ASSERT);
274 #endif
275 
276 	/* win or lose, we're done */
277 	if (p->error != 0)	/* lose */
278 		regfree(preg);
279 	return(p->error);
280 }
281 
282 /*
283  - p_ere - ERE parser top level, concatenation and alternation
284  == static void p_ere(register struct parse *p, int stop);
285  */
286 static void
287 p_ere(p, stop)
288 register struct parse *p;
289 int stop;			/* character this ERE should end at */
290 {
291 	register char c;
292 	register sopno prevback;
293 	register sopno prevfwd;
294 	register sopno conc;
295 	register int first = 1;		/* is this the first alternative? */
296 
297 	for (;;) {
298 		/* do a bunch of concatenated expressions */
299 		conc = HERE();
300 		while (MORE() && (c = PEEK()) != '|' && c != stop)
301 			p_ere_exp(p);
302 		(void)REQUIRE(HERE() != conc, REG_EMPTY);	/* require nonempty */
303 
304 		if (!EAT('|'))
305 			break;		/* NOTE BREAK OUT */
306 
307 		if (first) {
308 			INSERT(OCH_, conc);	/* offset is wrong */
309 			prevfwd = conc;
310 			prevback = conc;
311 			first = 0;
312 		}
313 		ASTERN(OOR1, prevback);
314 		prevback = THERE();
315 		AHEAD(prevfwd);			/* fix previous offset */
316 		prevfwd = HERE();
317 		EMIT(OOR2, 0);			/* offset is very wrong */
318 	}
319 
320 	if (!first) {		/* tail-end fixups */
321 		AHEAD(prevfwd);
322 		ASTERN(O_CH, prevback);
323 	}
324 
325 	assert(!MORE() || SEE(stop));
326 }
327 
328 /*
329  - p_ere_exp - parse one subERE, an atom possibly followed by a repetition op
330  == static void p_ere_exp(register struct parse *p);
331  */
332 static void
333 p_ere_exp(p)
334 register struct parse *p;
335 {
336 	register char c;
337 	register sopno pos;
338 	register int count;
339 	register int count2;
340 	register sopno subno;
341 	int wascaret = 0;
342 
343 	assert(MORE());		/* caller should have ensured this */
344 	c = GETNEXT();
345 
346 	pos = HERE();
347 	switch (c) {
348 	case '(':
349 		(void)REQUIRE(MORE(), REG_EPAREN);
350 		p->g->nsub++;
351 		subno = p->g->nsub;
352 		if (subno < NPAREN)
353 			p->pbegin[subno] = HERE();
354 		EMIT(OLPAREN, subno);
355 		if (!SEE(')'))
356 			p_ere(p, ')');
357 		if (subno < NPAREN) {
358 			p->pend[subno] = HERE();
359 			assert(p->pend[subno] != 0);
360 		}
361 		EMIT(ORPAREN, subno);
362 		(void)MUSTEAT(')', REG_EPAREN);
363 		break;
364 #ifndef POSIX_MISTAKE
365 	case ')':		/* happens only if no current unmatched ( */
366 		/*
367 		 * You may ask, why the ifndef?  Because I didn't notice
368 		 * this until slightly too late for 1003.2, and none of the
369 		 * other 1003.2 regular-expression reviewers noticed it at
370 		 * all.  So an unmatched ) is legal POSIX, at least until
371 		 * we can get it fixed.
372 		 */
373 		SETERROR(REG_EPAREN);
374 		break;
375 #endif
376 	case '^':
377 		EMIT(OBOL, 0);
378 		p->g->iflags |= USEBOL;
379 		p->g->nbol++;
380 		wascaret = 1;
381 		break;
382 	case '$':
383 		EMIT(OEOL, 0);
384 		p->g->iflags |= USEEOL;
385 		p->g->neol++;
386 		break;
387 	case '|':
388 		SETERROR(REG_EMPTY);
389 		break;
390 	case '*':
391 	case '+':
392 	case '?':
393 		SETERROR(REG_BADRPT);
394 		break;
395 	case '.':
396 		if (p->g->cflags&REG_NEWLINE)
397 			nonnewline(p);
398 		else
399 			EMIT(OANY, 0);
400 		break;
401 	case '[':
402 		p_bracket(p);
403 		break;
404 	case '\\':
405 		(void)REQUIRE(MORE(), REG_EESCAPE);
406 		c = GETNEXT();
407 		ordinary(p, c);
408 		break;
409 	case '{':		/* okay as ordinary except if digit follows */
410 		(void)REQUIRE(!MORE() || !isdigit(PEEK()), REG_BADRPT);
411 		/* FALLTHROUGH */
412 	default:
413 		ordinary(p, c);
414 		break;
415 	}
416 
417 	if (!MORE())
418 		return;
419 	c = PEEK();
420 	/* we call { a repetition if followed by a digit */
421 	if (!( c == '*' || c == '+' || c == '?' ||
422 				(c == '{' && MORE2() && isdigit(PEEK2())) ))
423 		return;		/* no repetition, we're done */
424 	NEXT();
425 
426 	(void)REQUIRE(!wascaret, REG_BADRPT);
427 	switch (c) {
428 	case '*':	/* implemented as +? */
429 		/* this case does not require the (y|) trick, noKLUDGE */
430 		INSERT(OPLUS_, pos);
431 		ASTERN(O_PLUS, pos);
432 		INSERT(OQUEST_, pos);
433 		ASTERN(O_QUEST, pos);
434 		break;
435 	case '+':
436 		INSERT(OPLUS_, pos);
437 		ASTERN(O_PLUS, pos);
438 		break;
439 	case '?':
440 		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
441 		INSERT(OCH_, pos);		/* offset slightly wrong */
442 		ASTERN(OOR1, pos);		/* this one's right */
443 		AHEAD(pos);			/* fix the OCH_ */
444 		EMIT(OOR2, 0);			/* offset very wrong... */
445 		AHEAD(THERE());			/* ...so fix it */
446 		ASTERN(O_CH, THERETHERE());
447 		break;
448 	case '{':
449 		count = p_count(p);
450 		if (EAT(',')) {
451 			if (isdigit(PEEK())) {
452 				count2 = p_count(p);
453 				(void)REQUIRE(count <= count2, REG_BADBR);
454 			} else		/* single number with comma */
455 				count2 = INFINITY;
456 		} else		/* just a single number */
457 			count2 = count;
458 		repeat(p, pos, count, count2);
459 		if (!EAT('}')) {	/* error heuristics */
460 			while (MORE() && PEEK() != '}')
461 				NEXT();
462 			(void)REQUIRE(MORE(), REG_EBRACE);
463 			SETERROR(REG_BADBR);
464 		}
465 		break;
466 	}
467 
468 	if (!MORE())
469 		return;
470 	c = PEEK();
471 	if (!( c == '*' || c == '+' || c == '?' ||
472 				(c == '{' && MORE2() && isdigit(PEEK2())) ) )
473 		return;
474 	SETERROR(REG_BADRPT);
475 }
476 
477 /*
478  - p_str - string (no metacharacters) "parser"
479  == static void p_str(register struct parse *p);
480  */
481 static void
482 p_str(p)
483 register struct parse *p;
484 {
485 	(void)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  == static void p_bre(register struct parse *p, register int end1, \
493  ==	register int end2);
494  * Giving end1 as OUT essentially eliminates the end1/end2 check.
495  *
496  * This implementation is a bit of a kludge, in that a trailing $ is first
497  * taken as an ordinary character and then revised to be an anchor.  The
498  * only undesirable side effect is that '$' gets included as a character
499  * category in such cases.  This is fairly harmless; not worth fixing.
500  * The amount of lookahead needed to avoid this kludge is excessive.
501  */
502 static void
503 p_bre(p, end1, end2)
504 register struct parse *p;
505 register int end1;		/* first terminating character */
506 register int end2;		/* second terminating character */
507 {
508 	register sopno start = HERE();
509 	register int first = 1;			/* first subexpression? */
510 	register int wasdollar = 0;
511 
512 	if (EAT('^')) {
513 		EMIT(OBOL, 0);
514 		p->g->iflags |= USEBOL;
515 		p->g->nbol++;
516 	}
517 	while (MORE() && !SEETWO(end1, end2)) {
518 		wasdollar = p_simp_re(p, first);
519 		first = 0;
520 	}
521 	if (wasdollar) {	/* oops, that was a trailing anchor */
522 		DROP(1);
523 		EMIT(OEOL, 0);
524 		p->g->iflags |= USEEOL;
525 		p->g->neol++;
526 	}
527 
528 	(void)REQUIRE(HERE() != start, REG_EMPTY);	/* require nonempty */
529 }
530 
531 /*
532  - p_simp_re - parse a simple RE, an atom possibly followed by a repetition
533  == static int p_simp_re(register struct parse *p, int starordinary);
534  */
535 static int			/* was the simple RE an unbackslashed $? */
536 p_simp_re(p, starordinary)
537 register struct parse *p;
538 int starordinary;		/* is a leading * an ordinary character? */
539 {
540 	register int c;
541 	register int count;
542 	register int count2;
543 	register sopno pos;
544 	register int i;
545 	register sopno subno;
546 #	define	BACKSL	(1<<CHAR_BIT)
547 
548 	pos = HERE();		/* repetion op, if any, covers from here */
549 
550 	assert(MORE());		/* caller should have ensured this */
551 	c = GETNEXT();
552 	if (c == '\\') {
553 		(void)REQUIRE(MORE(), REG_EESCAPE);
554 		c = BACKSL | (unsigned char)GETNEXT();
555 	}
556 	switch (c) {
557 	case '.':
558 		if (p->g->cflags&REG_NEWLINE)
559 			nonnewline(p);
560 		else
561 			EMIT(OANY, 0);
562 		break;
563 	case '[':
564 		p_bracket(p);
565 		break;
566 	case BACKSL|'{':
567 		SETERROR(REG_BADRPT);
568 		break;
569 	case BACKSL|'(':
570 		p->g->nsub++;
571 		subno = p->g->nsub;
572 		if (subno < NPAREN)
573 			p->pbegin[subno] = HERE();
574 		EMIT(OLPAREN, subno);
575 		/* the MORE here is an error heuristic */
576 		if (MORE() && !SEETWO('\\', ')'))
577 			p_bre(p, '\\', ')');
578 		if (subno < NPAREN) {
579 			p->pend[subno] = HERE();
580 			assert(p->pend[subno] != 0);
581 		}
582 		EMIT(ORPAREN, subno);
583 		(void)REQUIRE(EATTWO('\\', ')'), REG_EPAREN);
584 		break;
585 	case BACKSL|')':	/* should not get here -- must be user */
586 	case BACKSL|'}':
587 		SETERROR(REG_EPAREN);
588 		break;
589 	case BACKSL|'1':
590 	case BACKSL|'2':
591 	case BACKSL|'3':
592 	case BACKSL|'4':
593 	case BACKSL|'5':
594 	case BACKSL|'6':
595 	case BACKSL|'7':
596 	case BACKSL|'8':
597 	case BACKSL|'9':
598 		i = (c&~BACKSL) - '0';
599 		assert(i < NPAREN);
600 		if (p->pend[i] != 0) {
601 			assert(i <= p->g->nsub);
602 			EMIT(OBACK_, i);
603 			assert(p->pbegin[i] != 0);
604 			assert(OP(p->strip[p->pbegin[i]]) == OLPAREN);
605 			assert(OP(p->strip[p->pend[i]]) == ORPAREN);
606 			(void) dupl(p, p->pbegin[i]+1, p->pend[i]);
607 			EMIT(O_BACK, i);
608 		} else
609 			SETERROR(REG_ESUBREG);
610 		p->g->backrefs = 1;
611 		break;
612 	case '*':
613 		(void)REQUIRE(starordinary, REG_BADRPT);
614 		/* FALLTHROUGH */
615 	default:
616 		ordinary(p, c &~ BACKSL);
617 		break;
618 	}
619 
620 	if (EAT('*')) {		/* implemented as +? */
621 		/* this case does not require the (y|) trick, noKLUDGE */
622 		INSERT(OPLUS_, pos);
623 		ASTERN(O_PLUS, pos);
624 		INSERT(OQUEST_, pos);
625 		ASTERN(O_QUEST, pos);
626 	} else if (EATTWO('\\', '{')) {
627 		count = p_count(p);
628 		if (EAT(',')) {
629 			if (MORE() && isdigit(PEEK())) {
630 				count2 = p_count(p);
631 				(void)REQUIRE(count <= count2, REG_BADBR);
632 			} else		/* single number with comma */
633 				count2 = INFINITY;
634 		} else		/* just a single number */
635 			count2 = count;
636 		repeat(p, pos, count, count2);
637 		if (!EATTWO('\\', '}')) {	/* error heuristics */
638 			while (MORE() && !SEETWO('\\', '}'))
639 				NEXT();
640 			(void)REQUIRE(MORE(), REG_EBRACE);
641 			SETERROR(REG_BADBR);
642 		}
643 	} else if (c == (unsigned char)'$')	/* $ (but not \$) ends it */
644 		return(1);
645 
646 	return(0);
647 }
648 
649 /*
650  - p_count - parse a repetition count
651  == static int p_count(register struct parse *p);
652  */
653 static int			/* the value */
654 p_count(p)
655 register struct parse *p;
656 {
657 	register int count = 0;
658 	register int ndigits = 0;
659 
660 	while (MORE() && isdigit(PEEK()) && count <= DUPMAX) {
661 		count = count*10 + (GETNEXT() - '0');
662 		ndigits++;
663 	}
664 
665 	(void)REQUIRE(ndigits > 0 && count <= DUPMAX, REG_BADBR);
666 	return(count);
667 }
668 
669 /*
670  - p_bracket - parse a bracketed character list
671  == static void p_bracket(register struct parse *p);
672  *
673  * Note a significant property of this code:  if the allocset() did SETERROR,
674  * no set operations are done.
675  */
676 static void
677 p_bracket(p)
678 register struct parse *p;
679 {
680 	register cset *cs = allocset(p);
681 	register int invert = 0;
682 
683 	/* Dept of Truly Sickening Special-Case Kludges */
684 	if (p->next + 5 < p->end && strncmp(p->next, "[:<:]]", 6) == 0) {
685 		EMIT(OBOW, 0);
686 		NEXTn(6);
687 		return;
688 	}
689 	if (p->next + 5 < p->end && strncmp(p->next, "[:>:]]", 6) == 0) {
690 		EMIT(OEOW, 0);
691 		NEXTn(6);
692 		return;
693 	}
694 
695 	if (EAT('^'))
696 		invert++;	/* make note to invert set at end */
697 	if (EAT(']'))
698 		CHadd(cs, ']');
699 	else if (EAT('-'))
700 		CHadd(cs, '-');
701 	while (MORE() && PEEK() != ']' && !SEETWO('-', ']'))
702 		p_b_term(p, cs);
703 	if (EAT('-'))
704 		CHadd(cs, '-');
705 	(void)MUSTEAT(']', REG_EBRACK);
706 
707 	if (p->error != 0)	/* don't mess things up further */
708 		return;
709 
710 	if (p->g->cflags&REG_ICASE) {
711 		register int i;
712 		register int ci;
713 
714 		for (i = p->g->csetsize - 1; i >= 0; i--)
715 			if (CHIN(cs, i) && isalpha(i)) {
716 				ci = othercase(i);
717 				if (ci != i)
718 					CHadd(cs, ci);
719 			}
720 		if (cs->multis != NULL)
721 			mccase(p, cs);
722 	}
723 	if (invert) {
724 		register int i;
725 
726 		for (i = p->g->csetsize - 1; i >= 0; i--)
727 			if (CHIN(cs, i))
728 				CHsub(cs, i);
729 			else
730 				CHadd(cs, i);
731 		if (p->g->cflags&REG_NEWLINE)
732 			CHsub(cs, '\n');
733 		if (cs->multis != NULL)
734 			mcinvert(p, cs);
735 	}
736 
737 	assert(cs->multis == NULL);		/* xxx */
738 
739 	if (nch(p, cs) == 1) {		/* optimize singleton sets */
740 		ordinary(p, firstch(p, cs));
741 		freeset(p, cs);
742 	} else
743 		EMIT(OANYOF, freezeset(p, cs));
744 }
745 
746 /*
747  - p_b_term - parse one term of a bracketed character list
748  == static void p_b_term(register struct parse *p, register cset *cs);
749  */
750 static void
751 p_b_term(p, cs)
752 register struct parse *p;
753 register cset *cs;
754 {
755 	register char c;
756 	register char start, finish;
757 	register int i;
758 
759 	/* classify what we've got */
760 	switch ((MORE()) ? PEEK() : '\0') {
761 	case '[':
762 		c = (MORE2()) ? PEEK2() : '\0';
763 		break;
764 	case '-':
765 		SETERROR(REG_ERANGE);
766 		return;			/* NOTE RETURN */
767 		break;
768 	default:
769 		c = '\0';
770 		break;
771 	}
772 
773 	switch (c) {
774 	case ':':		/* character class */
775 		NEXT2();
776 		(void)REQUIRE(MORE(), REG_EBRACK);
777 		c = PEEK();
778 		(void)REQUIRE(c != '-' && c != ']', REG_ECTYPE);
779 		p_b_cclass(p, cs);
780 		(void)REQUIRE(MORE(), REG_EBRACK);
781 		(void)REQUIRE(EATTWO(':', ']'), REG_ECTYPE);
782 		break;
783 	case '=':		/* equivalence class */
784 		NEXT2();
785 		(void)REQUIRE(MORE(), REG_EBRACK);
786 		c = PEEK();
787 		(void)REQUIRE(c != '-' && c != ']', REG_ECOLLATE);
788 		p_b_eclass(p, cs);
789 		(void)REQUIRE(MORE(), REG_EBRACK);
790 		(void)REQUIRE(EATTWO('=', ']'), REG_ECOLLATE);
791 		break;
792 	default:		/* symbol, ordinary character, or range */
793 /* xxx revision needed for multichar stuff */
794 		start = p_b_symbol(p);
795 		if (SEE('-') && MORE2() && PEEK2() != ']') {
796 			/* range */
797 			NEXT();
798 			if (EAT('-'))
799 				finish = '-';
800 			else
801 				finish = p_b_symbol(p);
802 		} else
803 			finish = start;
804 		if (start == finish)
805 			CHadd(cs, start);
806 		else {
807 			if (__collate_load_error) {
808 				(void)REQUIRE((uch)start <= (uch)finish, REG_ERANGE);
809 				for (i = (uch)start; i <= (uch)finish; i++)
810 					CHadd(cs, i);
811 			} else {
812 				(void)REQUIRE(__collate_range_cmp(start, finish) <= 0, REG_ERANGE);
813 				for (i = CHAR_MIN; i <= CHAR_MAX; i++) {
814 					if (   __collate_range_cmp(start, i) <= 0
815 					    && __collate_range_cmp(i, finish) <= 0
816 					   )
817 						CHadd(cs, i);
818 				}
819 			}
820 		}
821 		break;
822 	}
823 }
824 
825 /*
826  - p_b_cclass - parse a character-class name and deal with it
827  == static void p_b_cclass(register struct parse *p, register cset *cs);
828  */
829 static void
830 p_b_cclass(p, cs)
831 register struct parse *p;
832 register cset *cs;
833 {
834 	register int c;
835 	register char *sp = p->next;
836 	register struct cclass *cp;
837 	register size_t len;
838 
839 	while (MORE() && isalpha((uch)PEEK()))
840 		NEXT();
841 	len = p->next - sp;
842 	for (cp = cclasses; cp->name != NULL; cp++)
843 		if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0')
844 			break;
845 	if (cp->name == NULL) {
846 		/* oops, didn't find it */
847 		SETERROR(REG_ECTYPE);
848 		return;
849 	}
850 
851 	switch (cp->fidx) {
852 	case CALNUM:
853 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
854 			if (isalnum((uch)c))
855 				CHadd(cs, c);
856 		break;
857 	case CALPHA:
858 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
859 			if (isalpha((uch)c))
860 				CHadd(cs, c);
861 		break;
862 	case CBLANK:
863 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
864 			if (isblank((uch)c))
865 				CHadd(cs, c);
866 		break;
867 	case CCNTRL:
868 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
869 			if (iscntrl((uch)c))
870 				CHadd(cs, c);
871 		break;
872 	case CDIGIT:
873 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
874 			if (isdigit((uch)c))
875 				CHadd(cs, c);
876 		break;
877 	case CGRAPH:
878 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
879 			if (isgraph((uch)c))
880 				CHadd(cs, c);
881 		break;
882 	case CLOWER:
883 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
884 			if (islower((uch)c))
885 				CHadd(cs, c);
886 		break;
887 	case CPRINT:
888 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
889 			if (isprint((uch)c))
890 				CHadd(cs, c);
891 		break;
892 	case CPUNCT:
893 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
894 			if (ispunct((uch)c))
895 				CHadd(cs, c);
896 		break;
897 	case CSPACE:
898 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
899 			if (isspace((uch)c))
900 				CHadd(cs, c);
901 		break;
902 	case CUPPER:
903 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
904 			if (isupper((uch)c))
905 				CHadd(cs, c);
906 		break;
907 	case CXDIGIT:
908 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
909 			if (isxdigit((uch)c))
910 				CHadd(cs, c);
911 		break;
912 	}
913 #if 0
914 	for (u = cp->multis; *u != '\0'; u += strlen(u) + 1)
915 		MCadd(p, cs, u);
916 #endif
917 }
918 
919 /*
920  - p_b_eclass - parse an equivalence-class name and deal with it
921  == static void p_b_eclass(register struct parse *p, register cset *cs);
922  *
923  * This implementation is incomplete. xxx
924  */
925 static void
926 p_b_eclass(p, cs)
927 register struct parse *p;
928 register cset *cs;
929 {
930 	register char c;
931 
932 	c = p_b_coll_elem(p, '=');
933 	CHadd(cs, c);
934 }
935 
936 /*
937  - p_b_symbol - parse a character or [..]ed multicharacter collating symbol
938  == static char p_b_symbol(register struct parse *p);
939  */
940 static char			/* value of symbol */
941 p_b_symbol(p)
942 register struct parse *p;
943 {
944 	register char value;
945 
946 	(void)REQUIRE(MORE(), REG_EBRACK);
947 	if (!EATTWO('[', '.'))
948 		return(GETNEXT());
949 
950 	/* collating symbol */
951 	value = p_b_coll_elem(p, '.');
952 	(void)REQUIRE(EATTWO('.', ']'), REG_ECOLLATE);
953 	return(value);
954 }
955 
956 /*
957  - p_b_coll_elem - parse a collating-element name and look it up
958  == static char p_b_coll_elem(register struct parse *p, int endc);
959  */
960 static char			/* value of collating element */
961 p_b_coll_elem(p, endc)
962 register struct parse *p;
963 int endc;			/* name ended by endc,']' */
964 {
965 	register char *sp = p->next;
966 	register struct cname *cp;
967 	register int len;
968 
969 	while (MORE() && !SEETWO(endc, ']'))
970 		NEXT();
971 	if (!MORE()) {
972 		SETERROR(REG_EBRACK);
973 		return(0);
974 	}
975 	len = p->next - sp;
976 	for (cp = cnames; cp->name != NULL; cp++)
977 		if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0')
978 			return(cp->code);	/* known name */
979 	if (len == 1)
980 		return(*sp);	/* single character */
981 	SETERROR(REG_ECOLLATE);			/* neither */
982 	return(0);
983 }
984 
985 /*
986  - othercase - return the case counterpart of an alphabetic
987  == static char othercase(int ch);
988  */
989 static char			/* if no counterpart, return ch */
990 othercase(ch)
991 int ch;
992 {
993 	ch = (unsigned char)ch;
994 	assert(isalpha(ch));
995 	if (isupper(ch))
996 		return(tolower(ch));
997 	else if (islower(ch))
998 		return(toupper(ch));
999 	else			/* peculiar, but could happen */
1000 		return(ch);
1001 }
1002 
1003 /*
1004  - bothcases - emit a dualcase version of a two-case character
1005  == static void bothcases(register struct parse *p, int ch);
1006  *
1007  * Boy, is this implementation ever a kludge...
1008  */
1009 static void
1010 bothcases(p, ch)
1011 register struct parse *p;
1012 int ch;
1013 {
1014 	register char *oldnext = p->next;
1015 	register char *oldend = p->end;
1016 	char bracket[3];
1017 
1018 	ch = (unsigned char)ch;
1019 	assert(othercase(ch) != ch);	/* p_bracket() would recurse */
1020 	p->next = bracket;
1021 	p->end = bracket+2;
1022 	bracket[0] = ch;
1023 	bracket[1] = ']';
1024 	bracket[2] = '\0';
1025 	p_bracket(p);
1026 	assert(p->next == bracket+2);
1027 	p->next = oldnext;
1028 	p->end = oldend;
1029 }
1030 
1031 /*
1032  - ordinary - emit an ordinary character
1033  == static void ordinary(register struct parse *p, register int ch);
1034  */
1035 static void
1036 ordinary(p, ch)
1037 register struct parse *p;
1038 register int ch;
1039 {
1040 	register cat_t *cap = p->g->categories;
1041 
1042 	if ((p->g->cflags&REG_ICASE) && isalpha((unsigned char)ch) && othercase(ch) != ch)
1043 		bothcases(p, ch);
1044 	else {
1045 		EMIT(OCHAR, (unsigned char)ch);
1046 		if (cap[ch] == 0)
1047 			cap[ch] = p->g->ncategories++;
1048 	}
1049 }
1050 
1051 /*
1052  - nonnewline - emit REG_NEWLINE version of OANY
1053  == static void nonnewline(register struct parse *p);
1054  *
1055  * Boy, is this implementation ever a kludge...
1056  */
1057 static void
1058 nonnewline(p)
1059 register struct parse *p;
1060 {
1061 	register char *oldnext = p->next;
1062 	register char *oldend = p->end;
1063 	char bracket[4];
1064 
1065 	p->next = bracket;
1066 	p->end = bracket+3;
1067 	bracket[0] = '^';
1068 	bracket[1] = '\n';
1069 	bracket[2] = ']';
1070 	bracket[3] = '\0';
1071 	p_bracket(p);
1072 	assert(p->next == bracket+3);
1073 	p->next = oldnext;
1074 	p->end = oldend;
1075 }
1076 
1077 /*
1078  - repeat - generate code for a bounded repetition, recursively if needed
1079  == static void repeat(register struct parse *p, sopno start, int from, int to);
1080  */
1081 static void
1082 repeat(p, start, from, to)
1083 register struct parse *p;
1084 sopno start;			/* operand from here to end of strip */
1085 int from;			/* repeated from this number */
1086 int to;				/* to this number of times (maybe INFINITY) */
1087 {
1088 	register sopno finish = HERE();
1089 #	define	N	2
1090 #	define	INF	3
1091 #	define	REP(f, t)	((f)*8 + (t))
1092 #	define	MAP(n)	(((n) <= 1) ? (n) : ((n) == INFINITY) ? INF : N)
1093 	register sopno copy;
1094 
1095 	if (p->error != 0)	/* head off possible runaway recursion */
1096 		return;
1097 
1098 	assert(from <= to);
1099 
1100 	switch (REP(MAP(from), MAP(to))) {
1101 	case REP(0, 0):			/* must be user doing this */
1102 		DROP(finish-start);	/* drop the operand */
1103 		break;
1104 	case REP(0, 1):			/* as x{1,1}? */
1105 	case REP(0, N):			/* as x{1,n}? */
1106 	case REP(0, INF):		/* as x{1,}? */
1107 		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1108 		INSERT(OCH_, start);		/* offset is wrong... */
1109 		repeat(p, start+1, 1, to);
1110 		ASTERN(OOR1, start);
1111 		AHEAD(start);			/* ... fix it */
1112 		EMIT(OOR2, 0);
1113 		AHEAD(THERE());
1114 		ASTERN(O_CH, THERETHERE());
1115 		break;
1116 	case REP(1, 1):			/* trivial case */
1117 		/* done */
1118 		break;
1119 	case REP(1, N):			/* as x?x{1,n-1} */
1120 		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1121 		INSERT(OCH_, start);
1122 		ASTERN(OOR1, start);
1123 		AHEAD(start);
1124 		EMIT(OOR2, 0);			/* offset very wrong... */
1125 		AHEAD(THERE());			/* ...so fix it */
1126 		ASTERN(O_CH, THERETHERE());
1127 		copy = dupl(p, start+1, finish+1);
1128 		assert(copy == finish+4);
1129 		repeat(p, copy, 1, to-1);
1130 		break;
1131 	case REP(1, INF):		/* as x+ */
1132 		INSERT(OPLUS_, start);
1133 		ASTERN(O_PLUS, start);
1134 		break;
1135 	case REP(N, N):			/* as xx{m-1,n-1} */
1136 		copy = dupl(p, start, finish);
1137 		repeat(p, copy, from-1, to-1);
1138 		break;
1139 	case REP(N, INF):		/* as xx{n-1,INF} */
1140 		copy = dupl(p, start, finish);
1141 		repeat(p, copy, from-1, to);
1142 		break;
1143 	default:			/* "can't happen" */
1144 		SETERROR(REG_ASSERT);	/* just in case */
1145 		break;
1146 	}
1147 }
1148 
1149 /*
1150  - seterr - set an error condition
1151  == static int seterr(register struct parse *p, int e);
1152  */
1153 static int			/* useless but makes type checking happy */
1154 seterr(p, e)
1155 register struct parse *p;
1156 int e;
1157 {
1158 	if (p->error == 0)	/* keep earliest error condition */
1159 		p->error = e;
1160 	p->next = nuls;		/* try to bring things to a halt */
1161 	p->end = nuls;
1162 	return(0);		/* make the return value well-defined */
1163 }
1164 
1165 /*
1166  - allocset - allocate a set of characters for []
1167  == static cset *allocset(register struct parse *p);
1168  */
1169 static cset *
1170 allocset(p)
1171 register struct parse *p;
1172 {
1173 	register int no = p->g->ncsets++;
1174 	register size_t nc;
1175 	register size_t nbytes;
1176 	register cset *cs;
1177 	register size_t css = (size_t)p->g->csetsize;
1178 	register int i;
1179 
1180 	if (no >= p->ncsalloc) {	/* need another column of space */
1181 		p->ncsalloc += CHAR_BIT;
1182 		nc = p->ncsalloc;
1183 		assert(nc % CHAR_BIT == 0);
1184 		nbytes = nc / CHAR_BIT * css;
1185 		if (p->g->sets == NULL)
1186 			p->g->sets = (cset *)malloc(nc * sizeof(cset));
1187 		else
1188 			p->g->sets = (cset *)reallocf((char *)p->g->sets,
1189 							nc * sizeof(cset));
1190 		if (p->g->setbits == NULL)
1191 			p->g->setbits = (uch *)malloc(nbytes);
1192 		else {
1193 			p->g->setbits = (uch *)reallocf((char *)p->g->setbits,
1194 								nbytes);
1195 			/* xxx this isn't right if setbits is now NULL */
1196 			for (i = 0; i < no; i++)
1197 				p->g->sets[i].ptr = p->g->setbits + css*(i/CHAR_BIT);
1198 		}
1199 		if (p->g->sets != NULL && p->g->setbits != NULL)
1200 			(void) memset((char *)p->g->setbits + (nbytes - css),
1201 								0, css);
1202 		else {
1203 			no = 0;
1204 			SETERROR(REG_ESPACE);
1205 			/* caller's responsibility not to do set ops */
1206 		}
1207 	}
1208 
1209 	assert(p->g->sets != NULL);	/* xxx */
1210 	cs = &p->g->sets[no];
1211 	cs->ptr = p->g->setbits + css*((no)/CHAR_BIT);
1212 	cs->mask = 1 << ((no) % CHAR_BIT);
1213 	cs->hash = 0;
1214 	cs->smultis = 0;
1215 	cs->multis = NULL;
1216 
1217 	return(cs);
1218 }
1219 
1220 /*
1221  - freeset - free a now-unused set
1222  == static void freeset(register struct parse *p, register cset *cs);
1223  */
1224 static void
1225 freeset(p, cs)
1226 register struct parse *p;
1227 register cset *cs;
1228 {
1229 	register int i;
1230 	register cset *top = &p->g->sets[p->g->ncsets];
1231 	register size_t css = (size_t)p->g->csetsize;
1232 
1233 	for (i = 0; i < css; i++)
1234 		CHsub(cs, i);
1235 	if (cs == top-1)	/* recover only the easy case */
1236 		p->g->ncsets--;
1237 }
1238 
1239 /*
1240  - freezeset - final processing on a set of characters
1241  == static int freezeset(register struct parse *p, register cset *cs);
1242  *
1243  * The main task here is merging identical sets.  This is usually a waste
1244  * of time (although the hash code minimizes the overhead), but can win
1245  * big if REG_ICASE is being used.  REG_ICASE, by the way, is why the hash
1246  * is done using addition rather than xor -- all ASCII [aA] sets xor to
1247  * the same value!
1248  */
1249 static int			/* set number */
1250 freezeset(p, cs)
1251 register struct parse *p;
1252 register cset *cs;
1253 {
1254 	register short h = cs->hash;
1255 	register int i;
1256 	register cset *top = &p->g->sets[p->g->ncsets];
1257 	register cset *cs2;
1258 	register size_t css = (size_t)p->g->csetsize;
1259 
1260 	/* look for an earlier one which is the same */
1261 	for (cs2 = &p->g->sets[0]; cs2 < top; cs2++)
1262 		if (cs2->hash == h && cs2 != cs) {
1263 			/* maybe */
1264 			for (i = 0; i < css; i++)
1265 				if (!!CHIN(cs2, i) != !!CHIN(cs, i))
1266 					break;		/* no */
1267 			if (i == css)
1268 				break;			/* yes */
1269 		}
1270 
1271 	if (cs2 < top) {	/* found one */
1272 		freeset(p, cs);
1273 		cs = cs2;
1274 	}
1275 
1276 	return((int)(cs - p->g->sets));
1277 }
1278 
1279 /*
1280  - firstch - return first character in a set (which must have at least one)
1281  == static int firstch(register struct parse *p, register cset *cs);
1282  */
1283 static int			/* character; there is no "none" value */
1284 firstch(p, cs)
1285 register struct parse *p;
1286 register cset *cs;
1287 {
1288 	register int i;
1289 	register size_t css = (size_t)p->g->csetsize;
1290 
1291 	for (i = 0; i < css; i++)
1292 		if (CHIN(cs, i))
1293 			return((unsigned char)i);
1294 	assert(never);
1295 	return(0);		/* arbitrary */
1296 }
1297 
1298 /*
1299  - nch - number of characters in a set
1300  == static int nch(register struct parse *p, register cset *cs);
1301  */
1302 static int
1303 nch(p, cs)
1304 register struct parse *p;
1305 register cset *cs;
1306 {
1307 	register int i;
1308 	register size_t css = (size_t)p->g->csetsize;
1309 	register int n = 0;
1310 
1311 	for (i = 0; i < css; i++)
1312 		if (CHIN(cs, i))
1313 			n++;
1314 	return(n);
1315 }
1316 
1317 /*
1318  - mcadd - add a collating element to a cset
1319  == static void mcadd(register struct parse *p, register cset *cs, \
1320  ==	register char *cp);
1321  */
1322 static void
1323 mcadd(p, cs, cp)
1324 register struct parse *p;
1325 register cset *cs;
1326 register char *cp;
1327 {
1328 	register size_t oldend = cs->smultis;
1329 
1330 	cs->smultis += strlen(cp) + 1;
1331 	if (cs->multis == NULL)
1332 		cs->multis = malloc(cs->smultis);
1333 	else
1334 		cs->multis = reallocf(cs->multis, cs->smultis);
1335 	if (cs->multis == NULL) {
1336 		SETERROR(REG_ESPACE);
1337 		return;
1338 	}
1339 
1340 	(void) strcpy(cs->multis + oldend - 1, cp);
1341 	cs->multis[cs->smultis - 1] = '\0';
1342 }
1343 
1344 #if used
1345 /*
1346  - mcsub - subtract a collating element from a cset
1347  == static void mcsub(register cset *cs, register char *cp);
1348  */
1349 static void
1350 mcsub(cs, cp)
1351 register cset *cs;
1352 register char *cp;
1353 {
1354 	register char *fp = mcfind(cs, cp);
1355 	register size_t len = strlen(fp);
1356 
1357 	assert(fp != NULL);
1358 	(void) memmove(fp, fp + len + 1,
1359 				cs->smultis - (fp + len + 1 - cs->multis));
1360 	cs->smultis -= len;
1361 
1362 	if (cs->smultis == 0) {
1363 		free(cs->multis);
1364 		cs->multis = NULL;
1365 		return;
1366 	}
1367 
1368 	cs->multis = reallocf(cs->multis, cs->smultis);
1369 	assert(cs->multis != NULL);
1370 }
1371 
1372 /*
1373  - mcin - is a collating element in a cset?
1374  == static int mcin(register cset *cs, register char *cp);
1375  */
1376 static int
1377 mcin(cs, cp)
1378 register cset *cs;
1379 register char *cp;
1380 {
1381 	return(mcfind(cs, cp) != NULL);
1382 }
1383 
1384 /*
1385  - mcfind - find a collating element in a cset
1386  == static char *mcfind(register cset *cs, register char *cp);
1387  */
1388 static char *
1389 mcfind(cs, cp)
1390 register cset *cs;
1391 register char *cp;
1392 {
1393 	register char *p;
1394 
1395 	if (cs->multis == NULL)
1396 		return(NULL);
1397 	for (p = cs->multis; *p != '\0'; p += strlen(p) + 1)
1398 		if (strcmp(cp, p) == 0)
1399 			return(p);
1400 	return(NULL);
1401 }
1402 #endif
1403 
1404 /*
1405  - mcinvert - invert the list of collating elements in a cset
1406  == static void mcinvert(register struct parse *p, register cset *cs);
1407  *
1408  * This would have to know the set of possibilities.  Implementation
1409  * is deferred.
1410  */
1411 static void
1412 mcinvert(p, cs)
1413 register struct parse *p;
1414 register cset *cs;
1415 {
1416 	assert(cs->multis == NULL);	/* xxx */
1417 }
1418 
1419 /*
1420  - mccase - add case counterparts of the list of collating elements in a cset
1421  == static void mccase(register struct parse *p, register cset *cs);
1422  *
1423  * This would have to know the set of possibilities.  Implementation
1424  * is deferred.
1425  */
1426 static void
1427 mccase(p, cs)
1428 register struct parse *p;
1429 register cset *cs;
1430 {
1431 	assert(cs->multis == NULL);	/* xxx */
1432 }
1433 
1434 /*
1435  - isinsets - is this character in any sets?
1436  == static int isinsets(register struct re_guts *g, int c);
1437  */
1438 static int			/* predicate */
1439 isinsets(g, c)
1440 register struct re_guts *g;
1441 int c;
1442 {
1443 	register uch *col;
1444 	register int i;
1445 	register int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1446 	register unsigned uc = (unsigned char)c;
1447 
1448 	for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1449 		if (col[uc] != 0)
1450 			return(1);
1451 	return(0);
1452 }
1453 
1454 /*
1455  - samesets - are these two characters in exactly the same sets?
1456  == static int samesets(register struct re_guts *g, int c1, int c2);
1457  */
1458 static int			/* predicate */
1459 samesets(g, c1, c2)
1460 register struct re_guts *g;
1461 int c1;
1462 int c2;
1463 {
1464 	register uch *col;
1465 	register int i;
1466 	register int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1467 	register unsigned uc1 = (unsigned char)c1;
1468 	register unsigned uc2 = (unsigned char)c2;
1469 
1470 	for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1471 		if (col[uc1] != col[uc2])
1472 			return(0);
1473 	return(1);
1474 }
1475 
1476 /*
1477  - categorize - sort out character categories
1478  == static void categorize(struct parse *p, register struct re_guts *g);
1479  */
1480 static void
1481 categorize(p, g)
1482 struct parse *p;
1483 register struct re_guts *g;
1484 {
1485 	register cat_t *cats = g->categories;
1486 	register int c;
1487 	register int c2;
1488 	register cat_t cat;
1489 
1490 	/* avoid making error situations worse */
1491 	if (p->error != 0)
1492 		return;
1493 
1494 	for (c = CHAR_MIN; c <= CHAR_MAX; c++)
1495 		if (cats[c] == 0 && isinsets(g, c)) {
1496 			cat = g->ncategories++;
1497 			cats[c] = cat;
1498 			for (c2 = c+1; c2 <= CHAR_MAX; c2++)
1499 				if (cats[c2] == 0 && samesets(g, c, c2))
1500 					cats[c2] = cat;
1501 		}
1502 }
1503 
1504 /*
1505  - dupl - emit a duplicate of a bunch of sops
1506  == static sopno dupl(register struct parse *p, sopno start, sopno finish);
1507  */
1508 static sopno			/* start of duplicate */
1509 dupl(p, start, finish)
1510 register struct parse *p;
1511 sopno start;			/* from here */
1512 sopno finish;			/* to this less one */
1513 {
1514 	register sopno ret = HERE();
1515 	register sopno len = finish - start;
1516 
1517 	assert(finish >= start);
1518 	if (len == 0)
1519 		return(ret);
1520 	enlarge(p, p->ssize + len);	/* this many unexpected additions */
1521 	assert(p->ssize >= p->slen + len);
1522 	(void) memcpy((char *)(p->strip + p->slen),
1523 		(char *)(p->strip + start), (size_t)len*sizeof(sop));
1524 	p->slen += len;
1525 	return(ret);
1526 }
1527 
1528 /*
1529  - doemit - emit a strip operator
1530  == static void doemit(register struct parse *p, sop op, size_t opnd);
1531  *
1532  * It might seem better to implement this as a macro with a function as
1533  * hard-case backup, but it's just too big and messy unless there are
1534  * some changes to the data structures.  Maybe later.
1535  */
1536 static void
1537 doemit(p, op, opnd)
1538 register struct parse *p;
1539 sop op;
1540 size_t opnd;
1541 {
1542 	/* avoid making error situations worse */
1543 	if (p->error != 0)
1544 		return;
1545 
1546 	/* deal with oversize operands ("can't happen", more or less) */
1547 	assert(opnd < 1<<OPSHIFT);
1548 
1549 	/* deal with undersized strip */
1550 	if (p->slen >= p->ssize)
1551 		enlarge(p, (p->ssize+1) / 2 * 3);	/* +50% */
1552 	assert(p->slen < p->ssize);
1553 
1554 	/* finally, it's all reduced to the easy case */
1555 	p->strip[p->slen++] = SOP(op, opnd);
1556 }
1557 
1558 /*
1559  - doinsert - insert a sop into the strip
1560  == static void doinsert(register struct parse *p, sop op, size_t opnd, sopno pos);
1561  */
1562 static void
1563 doinsert(p, op, opnd, pos)
1564 register struct parse *p;
1565 sop op;
1566 size_t opnd;
1567 sopno pos;
1568 {
1569 	register sopno sn;
1570 	register sop s;
1571 	register int i;
1572 
1573 	/* avoid making error situations worse */
1574 	if (p->error != 0)
1575 		return;
1576 
1577 	sn = HERE();
1578 	EMIT(op, opnd);		/* do checks, ensure space */
1579 	assert(HERE() == sn+1);
1580 	s = p->strip[sn];
1581 
1582 	/* adjust paren pointers */
1583 	assert(pos > 0);
1584 	for (i = 1; i < NPAREN; i++) {
1585 		if (p->pbegin[i] >= pos) {
1586 			p->pbegin[i]++;
1587 		}
1588 		if (p->pend[i] >= pos) {
1589 			p->pend[i]++;
1590 		}
1591 	}
1592 
1593 	memmove((char *)&p->strip[pos+1], (char *)&p->strip[pos],
1594 						(HERE()-pos-1)*sizeof(sop));
1595 	p->strip[pos] = s;
1596 }
1597 
1598 /*
1599  - dofwd - complete a forward reference
1600  == static void dofwd(register struct parse *p, sopno pos, sop value);
1601  */
1602 static void
1603 dofwd(p, pos, value)
1604 register struct parse *p;
1605 register sopno pos;
1606 sop value;
1607 {
1608 	/* avoid making error situations worse */
1609 	if (p->error != 0)
1610 		return;
1611 
1612 	assert(value < 1<<OPSHIFT);
1613 	p->strip[pos] = OP(p->strip[pos]) | value;
1614 }
1615 
1616 /*
1617  - enlarge - enlarge the strip
1618  == static void enlarge(register struct parse *p, sopno size);
1619  */
1620 static void
1621 enlarge(p, size)
1622 register struct parse *p;
1623 register sopno size;
1624 {
1625 	register sop *sp;
1626 
1627 	if (p->ssize >= size)
1628 		return;
1629 
1630 	sp = (sop *)realloc(p->strip, size*sizeof(sop));
1631 	if (sp == NULL) {
1632 		SETERROR(REG_ESPACE);
1633 		return;
1634 	}
1635 	p->strip = sp;
1636 	p->ssize = size;
1637 }
1638 
1639 /*
1640  - stripsnug - compact the strip
1641  == static void stripsnug(register struct parse *p, register struct re_guts *g);
1642  */
1643 static void
1644 stripsnug(p, g)
1645 register struct parse *p;
1646 register struct re_guts *g;
1647 {
1648 	g->nstates = p->slen;
1649 	g->strip = (sop *)realloc((char *)p->strip, p->slen * sizeof(sop));
1650 	if (g->strip == NULL) {
1651 		SETERROR(REG_ESPACE);
1652 		g->strip = p->strip;
1653 	}
1654 }
1655 
1656 /*
1657  - findmust - fill in must and mlen with longest mandatory literal string
1658  == static void findmust(register struct parse *p, register struct re_guts *g);
1659  *
1660  * This algorithm could do fancy things like analyzing the operands of |
1661  * for common subsequences.  Someday.  This code is simple and finds most
1662  * of the interesting cases.
1663  *
1664  * Note that must and mlen got initialized during setup.
1665  */
1666 static void
1667 findmust(p, g)
1668 struct parse *p;
1669 register struct re_guts *g;
1670 {
1671 	register sop *scan;
1672 	sop *start;
1673 	register sop *newstart;
1674 	register sopno newlen;
1675 	register sop s;
1676 	register char *cp;
1677 	register sopno i;
1678 
1679 	/* avoid making error situations worse */
1680 	if (p->error != 0)
1681 		return;
1682 
1683 	/* find the longest OCHAR sequence in strip */
1684 	newlen = 0;
1685 	scan = g->strip + 1;
1686 	do {
1687 		s = *scan++;
1688 		switch (OP(s)) {
1689 		case OCHAR:		/* sequence member */
1690 			if (newlen == 0)		/* new sequence */
1691 				newstart = scan - 1;
1692 			newlen++;
1693 			break;
1694 		case OPLUS_:		/* things that don't break one */
1695 		case OLPAREN:
1696 		case ORPAREN:
1697 			break;
1698 		case OQUEST_:		/* things that must be skipped */
1699 		case OCH_:
1700 			scan--;
1701 			do {
1702 				scan += OPND(s);
1703 				s = *scan;
1704 				/* assert() interferes w debug printouts */
1705 				if (OP(s) != O_QUEST && OP(s) != O_CH &&
1706 							OP(s) != OOR2) {
1707 					g->iflags |= BAD;
1708 					return;
1709 				}
1710 			} while (OP(s) != O_QUEST && OP(s) != O_CH);
1711 			/* fallthrough */
1712 		default:		/* things that break a sequence */
1713 			if (newlen > g->mlen) {		/* ends one */
1714 				start = newstart;
1715 				g->mlen = newlen;
1716 			}
1717 			newlen = 0;
1718 			break;
1719 		}
1720 	} while (OP(s) != OEND);
1721 
1722 	if (g->mlen == 0)		/* there isn't one */
1723 		return;
1724 
1725 	/* turn it into a character string */
1726 	g->must = malloc((size_t)g->mlen + 1);
1727 	if (g->must == NULL) {		/* argh; just forget it */
1728 		g->mlen = 0;
1729 		return;
1730 	}
1731 	cp = g->must;
1732 	scan = start;
1733 	for (i = g->mlen; i > 0; i--) {
1734 		while (OP(s = *scan++) != OCHAR)
1735 			continue;
1736 		assert(cp < g->must + g->mlen);
1737 		*cp++ = (char)OPND(s);
1738 	}
1739 	assert(cp == g->must + g->mlen);
1740 	*cp++ = '\0';		/* just on general principles */
1741 }
1742 
1743 /*
1744  - pluscount - count + nesting
1745  == static sopno pluscount(register struct parse *p, register struct re_guts *g);
1746  */
1747 static sopno			/* nesting depth */
1748 pluscount(p, g)
1749 struct parse *p;
1750 register struct re_guts *g;
1751 {
1752 	register sop *scan;
1753 	register sop s;
1754 	register sopno plusnest = 0;
1755 	register sopno maxnest = 0;
1756 
1757 	if (p->error != 0)
1758 		return(0);	/* there may not be an OEND */
1759 
1760 	scan = g->strip + 1;
1761 	do {
1762 		s = *scan++;
1763 		switch (OP(s)) {
1764 		case OPLUS_:
1765 			plusnest++;
1766 			break;
1767 		case O_PLUS:
1768 			if (plusnest > maxnest)
1769 				maxnest = plusnest;
1770 			plusnest--;
1771 			break;
1772 		}
1773 	} while (OP(s) != OEND);
1774 	if (plusnest != 0)
1775 		g->iflags |= BAD;
1776 	return(maxnest);
1777 }
1778