• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1  /*	$NetBSD: regcomp.c,v 1.36 2015/09/12 19:08:47 christos Exp $	*/
2  
3  /*-
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  /*-
38   * Copyright (c) 1992, 1993, 1994 Henry Spencer.
39   *
40   * This code is derived from software contributed to Berkeley by
41   * Henry Spencer.
42   *
43   * Redistribution and use in source and binary forms, with or without
44   * modification, are permitted provided that the following conditions
45   * are met:
46   * 1. Redistributions of source code must retain the above copyright
47   *    notice, this list of conditions and the following disclaimer.
48   * 2. Redistributions in binary form must reproduce the above copyright
49   *    notice, this list of conditions and the following disclaimer in the
50   *    documentation and/or other materials provided with the distribution.
51   * 3. All advertising materials mentioning features or use of this software
52   *    must display the following acknowledgement:
53   *	This product includes software developed by the University of
54   *	California, Berkeley and its contributors.
55   * 4. Neither the name of the University nor the names of its contributors
56   *    may be used to endorse or promote products derived from this software
57   *    without specific prior written permission.
58   *
59   * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
60   * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
61   * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
62   * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
63   * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
64   * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
65   * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
66   * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
67   * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
68   * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
69   * SUCH DAMAGE.
70   *
71   *	@(#)regcomp.c	8.5 (Berkeley) 3/20/94
72   */
73  
74  #include <sys/cdefs.h>
75  #if defined(LIBC_SCCS) && !defined(lint)
76  #if 0
77  static char sccsid[] = "@(#)regcomp.c	8.5 (Berkeley) 3/20/94";
78  #else
79  __RCSID("$NetBSD: regcomp.c,v 1.36 2015/09/12 19:08:47 christos Exp $");
80  #endif
81  #endif /* LIBC_SCCS and not lint */
82  
83  #include "namespace.h"
84  #include <sys/types.h>
85  
86  #include <assert.h>
87  #include <ctype.h>
88  #include <limits.h>
89  #include <stdio.h>
90  #include <stdlib.h>
91  #include <string.h>
92  #include <regex.h>
93  
94  #ifdef __weak_alias
95  __weak_alias(regcomp,_regcomp)
96  #endif
97  
98  #include "utils.h"
99  #include "regex2.h"
100  
101  #include "cclass.h"
102  #include "cname.h"
103  
104  /*
105   * parse structure, passed up and down to avoid global variables and
106   * other clumsinesses
107   */
108  struct parse {
109  	const char *next;	/* next character in RE */
110  	const char *end;	/* end of string (-> NUL normally) */
111  	int error;		/* has an error been seen? */
112  	sop *strip;		/* malloced strip */
113  	sopno ssize;		/* malloced strip size (allocated) */
114  	sopno slen;		/* malloced strip length (used) */
115  	size_t ncsalloc;	/* number of csets allocated */
116  	struct re_guts *g;
117  #	define	NPAREN	10	/* we need to remember () 1-9 for back refs */
118  	sopno pbegin[NPAREN];	/* -> ( ([0] unused) */
119  	sopno pend[NPAREN];	/* -> ) ([0] unused) */
120  };
121  
122  /* ========= begin header generated by ./mkh ========= */
123  #ifdef __cplusplus
124  extern "C" {
125  #endif
126  
127  /* === regcomp.c === */
128  static void p_ere(struct parse *p, int stop, size_t reclimit);
129  static void p_ere_exp(struct parse *p, size_t reclimit);
130  static void p_str(struct parse *p);
131  static void p_bre(struct parse *p, int end1, int end2, size_t reclimit);
132  static int p_simp_re(struct parse *p, int starordinary, size_t reclimit);
133  static int p_count(struct parse *p);
134  static void p_bracket(struct parse *p);
135  static void p_b_term(struct parse *p, cset *cs);
136  static void p_b_cclass(struct parse *p, cset *cs);
137  static void p_b_eclass(struct parse *p, cset *cs);
138  static char p_b_symbol(struct parse *p);
139  static char p_b_coll_elem(struct parse *p, int endc);
140  static int othercase(int ch);
141  static void bothcases(struct parse *p, int ch);
142  static void ordinary(struct parse *p, int ch);
143  static void nonnewline(struct parse *p);
144  static void repeat(struct parse *p, sopno start, int from, int to, size_t reclimit);
145  static int seterr(struct parse *p, int e);
146  static cset *allocset(struct parse *p);
147  static void freeset(struct parse *p, cset *cs);
148  static sopno freezeset(struct parse *p, cset *cs);
149  static int firstch(struct parse *p, cset *cs);
150  static int nch(struct parse *p, cset *cs);
151  static void mcadd(struct parse *p, cset *cs, const char *cp);
152  #if 0
153  static void mcsub(cset *cs, char *cp);
154  static int mcin(cset *cs, char *cp);
155  static char *mcfind(cset *cs, char *cp);
156  #endif
157  static void mcinvert(struct parse *p, cset *cs);
158  static void mccase(struct parse *p, cset *cs);
159  static int isinsets(struct re_guts *g, int c);
160  static int samesets(struct re_guts *g, int c1, int c2);
161  static void categorize(struct parse *p, struct re_guts *g);
162  static sopno dupl(struct parse *p, sopno start, sopno finish);
163  static void doemit(struct parse *p, sop op, sopno opnd);
164  static void doinsert(struct parse *p, sop op, sopno opnd, sopno pos);
165  static void dofwd(struct parse *p, sopno pos, sopno value);
166  static int enlarge(struct parse *p, sopno size);
167  static void stripsnug(struct parse *p, struct re_guts *g);
168  static void findmust(struct parse *p, struct re_guts *g);
169  static sopno pluscount(struct parse *p, struct re_guts *g);
170  
171  #ifdef __cplusplus
172  }
173  #endif
174  /* ========= end header generated by ./mkh ========= */
175  
176  static char nuls[10];		/* place to point scanner in event of error */
177  
178  /*
179   * macros for use with parse structure
180   * BEWARE:  these know that the parse structure is named `p' !!!
181   */
182  #define	PEEK()	(*p->next)
183  #define	PEEK2()	(*(p->next+1))
184  #define	MORE()	(p->next < p->end)
185  #define	MORE2()	(p->next+1 < p->end)
186  #define	SEE(c)	(MORE() && PEEK() == (c))
187  #define	SEETWO(a, b)	(MORE() && MORE2() && PEEK() == (a) && PEEK2() == (b))
188  #define	EAT(c)	((SEE(c)) ? (NEXT(), 1) : 0)
189  #define	EATTWO(a, b)	((SEETWO(a, b)) ? (NEXT2(), 1) : 0)
190  #define	NEXT()	(p->next++)
191  #define	NEXT2()	(p->next += 2)
192  #define	NEXTn(n)	(p->next += (n))
193  #define	GETNEXT()	(*p->next++)
194  #define	SETERROR(e)	seterr(p, (e))
195  #define	REQUIRE(co, e)	(void) ((co) || SETERROR(e))
196  #define	MUSTSEE(c, e)	(REQUIRE(MORE() && PEEK() == (c), e))
197  #define	MUSTEAT(c, e)	(void) (REQUIRE(MORE() && GETNEXT() == (c), e))
198  #define	MUSTNOTSEE(c, e)	(REQUIRE(!MORE() || PEEK() != (c), e))
199  #define	EMIT(op, sopnd)	doemit(p, (sop)(op), sopnd)
200  #define	INSERT(op, pos)	doinsert(p, (sop)(op), HERE()-(pos)+1, pos)
201  #define	AHEAD(pos)		dofwd(p, pos, HERE()-(pos))
202  #define	ASTERN(sop, pos)	EMIT(sop, HERE()-pos)
203  #define	HERE()		(p->slen)
204  #define	THERE()		(p->slen - 1)
205  #define	THERETHERE()	(p->slen - 2)
206  #define	DROP(n)	(p->slen -= (n))
207  
208  #ifndef NDEBUG
209  static int never = 0;		/* for use in asserts; shuts lint up */
210  #else
211  #define	never	0		/* some <assert.h>s have bugs too */
212  #endif
213  
214  #define	MEMLIMIT	0x8000000
215  #define MEMSIZE(p) \
216  	((p)->ncsalloc / CHAR_BIT * (p)->g->csetsize + \
217  	(p)->ncsalloc * sizeof(cset) + \
218  	(p)->ssize * sizeof(sop))
219  #define	RECLIMIT	256
220  
221  /*
222   - regcomp - interface for parser and compilation
223   = extern int regcomp(regex_t *, const char *, int);
224   = #define	REG_BASIC	0000
225   = #define	REG_EXTENDED	0001
226   = #define	REG_ICASE	0002
227   = #define	REG_NOSUB	0004
228   = #define	REG_NEWLINE	0010
229   = #define	REG_NOSPEC	0020
230   = #define	REG_PEND	0040
231   = #define	REG_DUMP	0200
232   */
233  int				/* 0 success, otherwise REG_something */
regcomp(regex_t * preg,const char * pattern,int cflags)234  regcomp(
235      regex_t *preg,
236      const char *pattern,
237      int cflags)
238  {
239  	struct parse pa;
240  	struct re_guts *g;
241  	struct parse *p = &pa;
242  	int i;
243  	size_t len;
244  #ifdef REDEBUG
245  #	define	GOODFLAGS(f)	(f)
246  #else
247  #	define	GOODFLAGS(f)	((f)&~REG_DUMP)
248  #endif
249  
250  	_DIAGASSERT(preg != NULL);
251  	_DIAGASSERT(pattern != NULL);
252  
253  	cflags = GOODFLAGS(cflags);
254  	if ((cflags&REG_EXTENDED) && (cflags&REG_NOSPEC))
255  		return(REG_INVARG);
256  
257  	if (cflags&REG_PEND) {
258  		if (preg->re_endp < pattern)
259  			return(REG_INVARG);
260  		len = preg->re_endp - pattern;
261  	} else
262  		len = strlen(pattern);
263  
264  	/* do the mallocs early so failure handling is easy */
265  	g = malloc(sizeof(struct re_guts) + (NC - 1) * sizeof(cat_t));
266  	if (g == NULL)
267  		return(REG_ESPACE);
268  	p->ssize = len/(size_t)2*(size_t)3 + (size_t)1;	/* ugh */
269  	p->strip = calloc(p->ssize, sizeof(sop));
270  	p->slen = 0;
271  	if (p->strip == NULL) {
272  		free(g);
273  		return(REG_ESPACE);
274  	}
275  
276  	/* set things up */
277  	p->g = g;
278  	p->next = pattern;
279  	p->end = p->next + len;
280  	p->error = 0;
281  	p->ncsalloc = 0;
282  	for (i = 0; i < NPAREN; i++) {
283  		p->pbegin[i] = 0;
284  		p->pend[i] = 0;
285  	}
286  	g->csetsize = NC;
287  	g->sets = NULL;
288  	g->setbits = NULL;
289  	g->ncsets = 0;
290  	g->cflags = cflags;
291  	g->iflags = 0;
292  	g->nbol = 0;
293  	g->neol = 0;
294  	g->must = NULL;
295  	g->mlen = 0;
296  	g->nsub = 0;
297  	g->ncategories = 1;	/* category 0 is "everything else" */
298  	g->categories = &g->catspace[-(CHAR_MIN)];
299  	(void) memset((char *)g->catspace, 0, NC*sizeof(cat_t));
300  	g->backrefs = 0;
301  
302  	/* do it */
303  	EMIT(OEND, 0);
304  	g->firststate = THERE();
305  	if (cflags&REG_EXTENDED)
306  		p_ere(p, OUT, 0);
307  	else if (cflags&REG_NOSPEC)
308  		p_str(p);
309  	else
310  		p_bre(p, OUT, OUT, 0);
311  	EMIT(OEND, 0);
312  	g->laststate = THERE();
313  
314  	/* tidy up loose ends and fill things in */
315  	categorize(p, g);
316  	stripsnug(p, g);
317  	findmust(p, g);
318  	g->nplus = pluscount(p, g);
319  	g->magic = MAGIC2;
320  	preg->re_nsub = g->nsub;
321  	preg->re_g = g;
322  	preg->re_magic = MAGIC1;
323  #ifndef REDEBUG
324  	/* not debugging, so can't rely on the assert() in regexec() */
325  	if (g->iflags&BAD)
326  		SETERROR(REG_ASSERT);
327  #endif
328  
329  	/* win or lose, we're done */
330  	if (p->error != 0)	/* lose */
331  		regfree(preg);
332  	return(p->error);
333  }
334  
335  /*
336   - p_ere - ERE parser top level, concatenation and alternation
337   == static void p_ere(struct parse *p, int stop, size_t reclimit);
338   */
339  static void
p_ere(struct parse * p,int stop,size_t reclimit)340  p_ere(
341      struct parse *p,
342      int stop,			/* character this ERE should end at */
343      size_t reclimit)
344  {
345  	char c;
346  	sopno prevback = 0;	/* pacify gcc */
347  	sopno prevfwd = 0; 	/* pacify gcc */
348  	sopno conc;
349  	int first = 1;		/* is this the first alternative? */
350  
351  	_DIAGASSERT(p != NULL);
352  
353  	if (reclimit++ > RECLIMIT || p->error == REG_ESPACE) {
354  		p->error = REG_ESPACE;
355  		return;
356  	}
357  
358  	for (;;) {
359  		/* do a bunch of concatenated expressions */
360  		conc = HERE();
361  		while (MORE() && (c = PEEK()) != '|' && c != stop)
362  			p_ere_exp(p, reclimit);
363  		REQUIRE(HERE() != conc, REG_EMPTY);	/* require nonempty */
364  
365  		if (!EAT('|'))
366  			break;		/* NOTE BREAK OUT */
367  
368  		if (first) {
369  			INSERT(OCH_, conc);	/* offset is wrong */
370  			prevfwd = conc;
371  			prevback = conc;
372  			first = 0;
373  		}
374  		ASTERN(OOR1, prevback);
375  		prevback = THERE();
376  		AHEAD(prevfwd);			/* fix previous offset */
377  		prevfwd = HERE();
378  		EMIT(OOR2, 0);			/* offset is very wrong */
379  	}
380  
381  	if (!first) {		/* tail-end fixups */
382  		AHEAD(prevfwd);
383  		ASTERN(O_CH, prevback);
384  	}
385  
386  	assert(!MORE() || SEE(stop));
387  }
388  
389  /*
390   - p_ere_exp - parse one subERE, an atom possibly followed by a repetition op
391   == static void p_ere_exp(struct parse *p, size_t reclimit);
392   */
393  static void
p_ere_exp(struct parse * p,size_t reclimit)394  p_ere_exp(
395      struct parse *p,
396      size_t reclimit)
397  {
398  	char c;
399  	sopno pos;
400  	int count;
401  	int count2;
402  	sopno subno;
403  	int wascaret = 0;
404  
405  	_DIAGASSERT(p != NULL);
406  
407  	assert(MORE());		/* caller should have ensured this */
408  	c = GETNEXT();
409  
410  	pos = HERE();
411  	switch (c) {
412  	case '(':
413  		REQUIRE(MORE(), REG_EPAREN);
414  		p->g->nsub++;
415  		subno = p->g->nsub;
416  		if (subno < NPAREN)
417  			p->pbegin[subno] = HERE();
418  		EMIT(OLPAREN, subno);
419  		if (!SEE(')'))
420  			p_ere(p, ')', reclimit);
421  		if (subno < NPAREN) {
422  			p->pend[subno] = HERE();
423  			assert(p->pend[subno] != 0);
424  		}
425  		EMIT(ORPAREN, subno);
426  		MUSTEAT(')', REG_EPAREN);
427  		break;
428  #ifndef POSIX_MISTAKE
429  	case ')':		/* happens only if no current unmatched ( */
430  		/*
431  		 * You may ask, why the ifndef?  Because I didn't notice
432  		 * this until slightly too late for 1003.2, and none of the
433  		 * other 1003.2 regular-expression reviewers noticed it at
434  		 * all.  So an unmatched ) is legal POSIX, at least until
435  		 * we can get it fixed.
436  		 */
437  		SETERROR(REG_EPAREN);
438  		break;
439  #endif
440  	case '^':
441  		EMIT(OBOL, 0);
442  		p->g->iflags |= USEBOL;
443  		p->g->nbol++;
444  		wascaret = 1;
445  		break;
446  	case '$':
447  		EMIT(OEOL, 0);
448  		p->g->iflags |= USEEOL;
449  		p->g->neol++;
450  		break;
451  	case '|':
452  		SETERROR(REG_EMPTY);
453  		break;
454  	case '*':
455  	case '+':
456  	case '?':
457  		SETERROR(REG_BADRPT);
458  		break;
459  	case '.':
460  		if (p->g->cflags&REG_NEWLINE)
461  			nonnewline(p);
462  		else
463  			EMIT(OANY, 0);
464  		break;
465  	case '[':
466  		p_bracket(p);
467  		break;
468  	case '\\':
469  		REQUIRE(MORE(), REG_EESCAPE);
470  		c = GETNEXT();
471  		ordinary(p, c);
472  		break;
473  	case '{':		/* okay as ordinary except if digit follows */
474  		REQUIRE(!MORE() || !isdigit((unsigned char)PEEK()), REG_BADRPT);
475  		/* FALLTHROUGH */
476  	default:
477  		ordinary(p, c);
478  		break;
479  	}
480  
481  	if (!MORE())
482  		return;
483  	c = PEEK();
484  	/* we call { a repetition if followed by a digit */
485  	if (!( c == '*' || c == '+' || c == '?' ||
486  	    (c == '{' && MORE2() && isdigit((unsigned char)PEEK2())) ))
487  		return;		/* no repetition, we're done */
488  	NEXT();
489  
490  	REQUIRE(!wascaret, REG_BADRPT);
491  	switch (c) {
492  	case '*':	/* implemented as +? */
493  		/* this case does not require the (y|) trick, noKLUDGE */
494  		INSERT(OPLUS_, pos);
495  		ASTERN(O_PLUS, pos);
496  		INSERT(OQUEST_, pos);
497  		ASTERN(O_QUEST, pos);
498  		break;
499  	case '+':
500  		INSERT(OPLUS_, pos);
501  		ASTERN(O_PLUS, pos);
502  		break;
503  	case '?':
504  		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
505  		INSERT(OCH_, pos);		/* offset slightly wrong */
506  		ASTERN(OOR1, pos);		/* this one's right */
507  		AHEAD(pos);			/* fix the OCH_ */
508  		EMIT(OOR2, 0);			/* offset very wrong... */
509  		AHEAD(THERE());			/* ...so fix it */
510  		ASTERN(O_CH, THERETHERE());
511  		break;
512  	case '{':
513  		count = p_count(p);
514  		if (EAT(',')) {
515  			if (isdigit((unsigned char)PEEK())) {
516  				count2 = p_count(p);
517  				REQUIRE(count <= count2, REG_BADBR);
518  			} else		/* single number with comma */
519  				count2 = INFINITY;
520  		} else		/* just a single number */
521  			count2 = count;
522  		repeat(p, pos, count, count2, 0);
523  		if (!EAT('}')) {	/* error heuristics */
524  			while (MORE() && PEEK() != '}')
525  				NEXT();
526  			REQUIRE(MORE(), REG_EBRACE);
527  			SETERROR(REG_BADBR);
528  		}
529  		break;
530  	}
531  
532  	if (!MORE())
533  		return;
534  	c = PEEK();
535  	if (!( c == '*' || c == '+' || c == '?' ||
536  	    (c == '{' && MORE2() && isdigit((unsigned char)PEEK2())) ) )
537  		return;
538  	SETERROR(REG_BADRPT);
539  }
540  
541  /*
542   - p_str - string (no metacharacters) "parser"
543   == static void p_str(struct parse *p);
544   */
545  static void
p_str(struct parse * p)546  p_str(
547      struct parse *p)
548  {
549  
550  	_DIAGASSERT(p != NULL);
551  
552  	REQUIRE(MORE(), REG_EMPTY);
553  	while (MORE())
554  		ordinary(p, GETNEXT());
555  }
556  
557  /*
558   - p_bre - BRE parser top level, anchoring and concatenation
559   == static void p_bre(struct parse *p, int end1, \
560   ==	int end2, size_t reclimit);
561   * Giving end1 as OUT essentially eliminates the end1/end2 check.
562   *
563   * This implementation is a bit of a kludge, in that a trailing $ is first
564   * taken as an ordinary character and then revised to be an anchor.  The
565   * only undesirable side effect is that '$' gets included as a character
566   * category in such cases.  This is fairly harmless; not worth fixing.
567   * The amount of lookahead needed to avoid this kludge is excessive.
568   */
569  static void
p_bre(struct parse * p,int end1,int end2,size_t reclimit)570  p_bre(
571      struct parse *p,
572      int end1,		/* first terminating character */
573      int end2,		/* second terminating character */
574      size_t reclimit)
575  {
576  	sopno start;
577  	int first = 1;			/* first subexpression? */
578  	int wasdollar = 0;
579  
580  	_DIAGASSERT(p != NULL);
581  
582  	if (reclimit++ > RECLIMIT || p->error == REG_ESPACE) {
583  		p->error = REG_ESPACE;
584  		return;
585  	}
586  
587  	start = HERE();
588  
589  	if (EAT('^')) {
590  		EMIT(OBOL, 0);
591  		p->g->iflags |= USEBOL;
592  		p->g->nbol++;
593  	}
594  	while (MORE() && !SEETWO(end1, end2)) {
595  		wasdollar = p_simp_re(p, first, reclimit);
596  		first = 0;
597  	}
598  	if (wasdollar) {	/* oops, that was a trailing anchor */
599  		DROP(1);
600  		EMIT(OEOL, 0);
601  		p->g->iflags |= USEEOL;
602  		p->g->neol++;
603  	}
604  
605  	REQUIRE(HERE() != start, REG_EMPTY);	/* require nonempty */
606  }
607  
608  /*
609   - p_simp_re - parse a simple RE, an atom possibly followed by a repetition
610   == static int p_simp_re(struct parse *p, int starordinary, size_t reclimit);
611   */
612  static int			/* was the simple RE an unbackslashed $? */
p_simp_re(struct parse * p,int starordinary,size_t reclimit)613  p_simp_re(
614      struct parse *p,
615      int starordinary,		/* is a leading * an ordinary character? */
616      size_t reclimit)
617  {
618  	int c;
619  	int count;
620  	int count2;
621  	sopno pos, i;
622  	sopno subno;
623  #	define	BACKSL	(1<<CHAR_BIT)
624  
625  	_DIAGASSERT(p != NULL);
626  
627  	pos = HERE();		/* repetion op, if any, covers from here */
628  
629  	assert(MORE());		/* caller should have ensured this */
630  	c = GETNEXT();
631  	if (c == '\\') {
632  		REQUIRE(MORE(), REG_EESCAPE);
633  		c = BACKSL | (unsigned char)GETNEXT();
634  	}
635  	switch (c) {
636  	case '.':
637  		if (p->g->cflags&REG_NEWLINE)
638  			nonnewline(p);
639  		else
640  			EMIT(OANY, 0);
641  		break;
642  	case '[':
643  		p_bracket(p);
644  		break;
645  	case BACKSL|'{':
646  		SETERROR(REG_BADRPT);
647  		break;
648  	case BACKSL|'(':
649  		p->g->nsub++;
650  		subno = p->g->nsub;
651  		if (subno < NPAREN)
652  			p->pbegin[subno] = HERE();
653  		EMIT(OLPAREN, subno);
654  		/* the MORE here is an error heuristic */
655  		if (MORE() && !SEETWO('\\', ')'))
656  			p_bre(p, '\\', ')', reclimit);
657  		if (subno < NPAREN) {
658  			p->pend[subno] = HERE();
659  			assert(p->pend[subno] != 0);
660  		}
661  		EMIT(ORPAREN, subno);
662  		REQUIRE(EATTWO('\\', ')'), REG_EPAREN);
663  		break;
664  	case BACKSL|')':	/* should not get here -- must be user */
665  	case BACKSL|'}':
666  		SETERROR(REG_EPAREN);
667  		break;
668  	case BACKSL|'1':
669  	case BACKSL|'2':
670  	case BACKSL|'3':
671  	case BACKSL|'4':
672  	case BACKSL|'5':
673  	case BACKSL|'6':
674  	case BACKSL|'7':
675  	case BACKSL|'8':
676  	case BACKSL|'9':
677  		i = (c&~BACKSL) - '0';
678  		assert(i < NPAREN);
679  		if (p->pend[i] != 0) {
680  			assert(i <= p->g->nsub);
681  			EMIT(OBACK_, i);
682  			assert(p->pbegin[i] != 0);
683  			assert(OP(p->strip[p->pbegin[i]]) == OLPAREN);
684  			assert(OP(p->strip[p->pend[i]]) == ORPAREN);
685  			(void) dupl(p, p->pbegin[i]+1, p->pend[i]);
686  			EMIT(O_BACK, i);
687  		} else
688  			SETERROR(REG_ESUBREG);
689  		p->g->backrefs = 1;
690  		break;
691  	case '*':
692  		REQUIRE(starordinary, REG_BADRPT);
693  		/* FALLTHROUGH */
694  	default:
695  		ordinary(p, c &~ BACKSL);
696  		break;
697  	}
698  
699  	if (EAT('*')) {		/* implemented as +? */
700  		/* this case does not require the (y|) trick, noKLUDGE */
701  		INSERT(OPLUS_, pos);
702  		ASTERN(O_PLUS, pos);
703  		INSERT(OQUEST_, pos);
704  		ASTERN(O_QUEST, pos);
705  	} else if (EATTWO('\\', '{')) {
706  		count = p_count(p);
707  		if (EAT(',')) {
708  			if (MORE() && isdigit((unsigned char)PEEK())) {
709  				count2 = p_count(p);
710  				REQUIRE(count <= count2, REG_BADBR);
711  			} else		/* single number with comma */
712  				count2 = INFINITY;
713  		} else		/* just a single number */
714  			count2 = count;
715  		repeat(p, pos, count, count2, 0);
716  		if (!EATTWO('\\', '}')) {	/* error heuristics */
717  			while (MORE() && !SEETWO('\\', '}'))
718  				NEXT();
719  			REQUIRE(MORE(), REG_EBRACE);
720  			SETERROR(REG_BADBR);
721  		}
722  	} else if (c == (unsigned char)'$')	/* $ (but not \$) ends it */
723  		return(1);
724  
725  	return(0);
726  }
727  
728  /*
729   - p_count - parse a repetition count
730   == static int p_count(struct parse *p);
731   */
732  static int			/* the value */
p_count(struct parse * p)733  p_count(
734      struct parse *p)
735  {
736  	int count = 0;
737  	int ndigits = 0;
738  
739  	_DIAGASSERT(p != NULL);
740  
741  	while (MORE() && isdigit((unsigned char)PEEK()) && count <= DUPMAX) {
742  		count = count*10 + (GETNEXT() - '0');
743  		ndigits++;
744  	}
745  
746  	REQUIRE(ndigits > 0 && count <= DUPMAX, REG_BADBR);
747  	return(count);
748  }
749  
750  /*
751   - p_bracket - parse a bracketed character list
752   == static void p_bracket(struct parse *p);
753   *
754   * Note a significant property of this code:  if the allocset() did SETERROR,
755   * no set operations are done.
756   */
757  static void
p_bracket(struct parse * p)758  p_bracket(
759      struct parse *p)
760  {
761  	cset *cs;
762  	int invert = 0;
763  	_DIAGASSERT(p != NULL);
764  
765  	cs = allocset(p);
766  	if (cs == NULL)
767  		return;
768  
769  	/* Dept of Truly Sickening Special-Case Kludges */
770  	if (p->next + 5 < p->end && strncmp(p->next, "[:<:]]",
771  					    (size_t)6) == 0) {
772  		EMIT(OBOW, 0);
773  		NEXTn(6);
774  		return;
775  	}
776  	if (p->next + 5 < p->end && strncmp(p->next, "[:>:]]",
777  					    (size_t)6) == 0) {
778  		EMIT(OEOW, 0);
779  		NEXTn(6);
780  		return;
781  	}
782  
783  	if (EAT('^'))
784  		invert++;	/* make note to invert set at end */
785  	if (EAT(']'))
786  		CHadd(cs, ']');
787  	else if (EAT('-'))
788  		CHadd(cs, '-');
789  	while (MORE() && PEEK() != ']' && !SEETWO('-', ']'))
790  		p_b_term(p, cs);
791  	if (EAT('-'))
792  		CHadd(cs, '-');
793  	MUSTEAT(']', REG_EBRACK);
794  
795  	if (p->error != 0)	/* don't mess things up further */
796  		return;
797  
798  	if (p->g->cflags&REG_ICASE) {
799  		ssize_t i;
800  		int ci;
801  
802  		for (i = p->g->csetsize - 1; i >= 0; i--)
803  			if (CHIN(cs, i) && isalpha(i)) {
804  				ci = othercase((int)i);
805  				if (ci != i)
806  					CHadd(cs, ci);
807  			}
808  		if (cs->multis != NULL)
809  			mccase(p, cs);
810  	}
811  	if (invert) {
812  		ssize_t i;
813  
814  		for (i = p->g->csetsize - 1; i >= 0; i--)
815  			if (CHIN(cs, i))
816  				CHsub(cs, (int)i);
817  			else
818  				CHadd(cs, (int)i);
819  		if (p->g->cflags&REG_NEWLINE)
820  			CHsub(cs, '\n');
821  		if (cs->multis != NULL)
822  			mcinvert(p, cs);
823  	}
824  
825  	assert(cs->multis == NULL);		/* xxx */
826  
827  	if (nch(p, cs) == 1) {		/* optimize singleton sets */
828  		ordinary(p, firstch(p, cs));
829  		freeset(p, cs);
830  	} else
831  		EMIT(OANYOF, freezeset(p, cs));
832  }
833  
834  /*
835   - p_b_term - parse one term of a bracketed character list
836   == static void p_b_term(struct parse *p, cset *cs);
837   */
838  static void
p_b_term(struct parse * p,cset * cs)839  p_b_term(
840      struct parse *p,
841      cset *cs)
842  {
843  	char c;
844  	char start, finish;
845  	int i;
846  
847  	_DIAGASSERT(p != NULL);
848  	_DIAGASSERT(cs != NULL);
849  
850  	/* classify what we've got */
851  	switch ((MORE()) ? PEEK() : '\0') {
852  	case '[':
853  		c = (MORE2()) ? PEEK2() : '\0';
854  		break;
855  
856  	case '-':
857  		SETERROR(REG_ERANGE);
858  		return;			/* NOTE RETURN */
859  
860  	default:
861  		c = '\0';
862  		break;
863  	}
864  
865  	switch (c) {
866  	case ':':		/* character class */
867  		NEXT2();
868  		REQUIRE(MORE(), REG_EBRACK);
869  		c = PEEK();
870  		REQUIRE(c != '-' && c != ']', REG_ECTYPE);
871  		p_b_cclass(p, cs);
872  		REQUIRE(MORE(), REG_EBRACK);
873  		REQUIRE(EATTWO(':', ']'), REG_ECTYPE);
874  		break;
875  	case '=':		/* equivalence class */
876  		NEXT2();
877  		REQUIRE(MORE(), REG_EBRACK);
878  		c = PEEK();
879  		REQUIRE(c != '-' && c != ']', REG_ECOLLATE);
880  		p_b_eclass(p, cs);
881  		REQUIRE(MORE(), REG_EBRACK);
882  		REQUIRE(EATTWO('=', ']'), REG_ECOLLATE);
883  		break;
884  	default:		/* symbol, ordinary character, or range */
885  /* xxx revision needed for multichar stuff */
886  		start = p_b_symbol(p);
887  		if (SEE('-') && MORE2() && PEEK2() != ']') {
888  			/* range */
889  			NEXT();
890  			if (EAT('-'))
891  				finish = '-';
892  			else
893  				finish = p_b_symbol(p);
894  		} else
895  			finish = start;
896  /* xxx what about signed chars here... */
897  		REQUIRE(start <= finish, REG_ERANGE);
898  		for (i = start; i <= finish; i++)
899  			CHadd(cs, i);
900  		break;
901  	}
902  }
903  
904  /*
905   - p_b_cclass - parse a character-class name and deal with it
906   == static void p_b_cclass(struct parse *p, cset *cs);
907   */
908  static void
p_b_cclass(struct parse * p,cset * cs)909  p_b_cclass(
910      struct parse *p,
911      cset *cs)
912  {
913  	const char *sp;
914  	const struct cclass *cp;
915  	size_t len;
916  	const char *u;
917  	char c;
918  
919  	_DIAGASSERT(p != NULL);
920  	_DIAGASSERT(cs != NULL);
921  
922  	sp = p->next;
923  
924  	while (MORE() && isalpha((unsigned char)PEEK()))
925  		NEXT();
926  	len = p->next - sp;
927  	for (cp = cclasses; cp->name != NULL; cp++)
928  		if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0')
929  			break;
930  	if (cp->name == NULL) {
931  		/* oops, didn't find it */
932  		SETERROR(REG_ECTYPE);
933  		return;
934  	}
935  
936  	u = cp->chars;
937  	while ((c = *u++) != '\0')
938  		CHadd(cs, c);
939  	for (u = cp->multis; *u != '\0'; u += strlen(u) + 1)
940  		MCadd(p, cs, u);
941  }
942  
943  /*
944   - p_b_eclass - parse an equivalence-class name and deal with it
945   == static void p_b_eclass(struct parse *p, cset *cs);
946   *
947   * This implementation is incomplete. xxx
948   */
949  static void
p_b_eclass(struct parse * p,cset * cs)950  p_b_eclass(
951      struct parse *p,
952      cset *cs)
953  {
954  	char c;
955  
956  	_DIAGASSERT(p != NULL);
957  	_DIAGASSERT(cs != NULL);
958  
959  	c = p_b_coll_elem(p, '=');
960  	CHadd(cs, c);
961  }
962  
963  /*
964   - p_b_symbol - parse a character or [..]ed multicharacter collating symbol
965   == static char p_b_symbol(struct parse *p);
966   */
967  static char			/* value of symbol */
p_b_symbol(struct parse * p)968  p_b_symbol(
969      struct parse *p)
970  {
971  	char value;
972  
973  	_DIAGASSERT(p != NULL);
974  
975  	REQUIRE(MORE(), REG_EBRACK);
976  	if (!EATTWO('[', '.'))
977  		return(GETNEXT());
978  
979  	/* collating symbol */
980  	value = p_b_coll_elem(p, '.');
981  	REQUIRE(EATTWO('.', ']'), REG_ECOLLATE);
982  	return(value);
983  }
984  
985  /*
986   - p_b_coll_elem - parse a collating-element name and look it up
987   == static char p_b_coll_elem(struct parse *p, int endc);
988   */
989  static char			/* value of collating element */
p_b_coll_elem(struct parse * p,int endc)990  p_b_coll_elem(
991      struct parse *p,
992      int endc)			/* name ended by endc,']' */
993  {
994  	const char *sp;
995  	const struct cname *cp;
996  	size_t len;
997  
998  	_DIAGASSERT(p != NULL);
999  
1000  	sp = p->next;
1001  
1002  	while (MORE() && !SEETWO(endc, ']'))
1003  		NEXT();
1004  	if (!MORE()) {
1005  		SETERROR(REG_EBRACK);
1006  		return(0);
1007  	}
1008  	len = p->next - sp;
1009  	for (cp = cnames; cp->name != NULL; cp++)
1010  		if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0')
1011  			return(cp->code);	/* known name */
1012  	if (len == 1)
1013  		return(*sp);	/* single character */
1014  	SETERROR(REG_ECOLLATE);			/* neither */
1015  	return(0);
1016  }
1017  
1018  /*
1019   - othercase - return the case counterpart of an alphabetic
1020   == static int othercase(int ch);
1021   */
1022  static int			/* if no counterpart, return ch */
othercase(int ch)1023  othercase(
1024      int ch)
1025  {
1026  	assert(isalpha(ch));
1027  	if (isupper(ch))
1028  		return(tolower(ch));
1029  	else if (islower(ch))
1030  		return(toupper(ch));
1031  	else			/* peculiar, but could happen */
1032  		return(ch);
1033  }
1034  
1035  /*
1036   - bothcases - emit a dualcase version of a two-case character
1037   == static void bothcases(struct parse *p, int ch);
1038   *
1039   * Boy, is this implementation ever a kludge...
1040   */
1041  static void
bothcases(struct parse * p,int ch)1042  bothcases(
1043      struct parse *p,
1044      int ch)
1045  {
1046  	const char *oldnext;
1047  	const char *oldend;
1048  	char bracket[3];
1049  
1050  	_DIAGASSERT(p != NULL);
1051  
1052  	oldnext = p->next;
1053  	oldend = p->end;
1054  
1055  	assert(othercase(ch) != ch);	/* p_bracket() would recurse */
1056  	p->next = bracket;
1057  	p->end = bracket+2;
1058  	bracket[0] = ch;
1059  	bracket[1] = ']';
1060  	bracket[2] = '\0';
1061  	p_bracket(p);
1062  	assert(p->next == bracket+2);
1063  	p->next = oldnext;
1064  	p->end = oldend;
1065  }
1066  
1067  /*
1068   - ordinary - emit an ordinary character
1069   == static void ordinary(struct parse *p, int ch);
1070   */
1071  static void
ordinary(struct parse * p,int ch)1072  ordinary(
1073      struct parse *p,
1074      int ch)
1075  {
1076  	cat_t *cap;
1077  	unsigned char uc = (unsigned char)ch;
1078  
1079  	_DIAGASSERT(p != NULL);
1080  
1081  	cap = p->g->categories;
1082  	if ((p->g->cflags & REG_ICASE) && isalpha(uc) && othercase(uc) != uc)
1083  		bothcases(p, uc);
1084  	else {
1085  		EMIT(OCHAR, (sopno)uc);
1086  		if (cap[uc] == 0) {
1087  			_DIAGASSERT(__type_fit(unsigned char,
1088  			    p->g->ncategories + 1));
1089  			cap[uc] = (unsigned char)p->g->ncategories++;
1090  		}
1091  	}
1092  }
1093  
1094  /*
1095   - nonnewline - emit REG_NEWLINE version of OANY
1096   == static void nonnewline(struct parse *p);
1097   *
1098   * Boy, is this implementation ever a kludge...
1099   */
1100  static void
nonnewline(struct parse * p)1101  nonnewline(
1102      struct parse *p)
1103  {
1104  	const char *oldnext;
1105  	const char *oldend;
1106  	char bracket[4];
1107  
1108  	_DIAGASSERT(p != NULL);
1109  
1110  	oldnext = p->next;
1111  	oldend = p->end;
1112  
1113  	p->next = bracket;
1114  	p->end = bracket+3;
1115  	bracket[0] = '^';
1116  	bracket[1] = '\n';
1117  	bracket[2] = ']';
1118  	bracket[3] = '\0';
1119  	p_bracket(p);
1120  	assert(p->next == bracket+3);
1121  	p->next = oldnext;
1122  	p->end = oldend;
1123  }
1124  
1125  /*
1126   - repeat - generate code for a bounded repetition, recursively if needed
1127   == static void repeat(struct parse *p, sopno start, int from, int to,
1128   == size_t reclimit);
1129   */
1130  static void
repeat(struct parse * p,sopno start,int from,int to,size_t reclimit)1131  repeat(
1132      struct parse *p,
1133      sopno start,		/* operand from here to end of strip */
1134      int from,			/* repeated from this number */
1135      int to,			/* to this number of times (maybe INFINITY) */
1136      size_t reclimit)
1137  {
1138  	sopno finish;
1139  #	define	N	2
1140  #	define	INF	3
1141  #	define	REP(f, t)	((f)*8 + (t))
1142  #	define	MAP(n)	(((n) <= 1) ? (n) : ((n) == INFINITY) ? INF : N)
1143  	sopno copy;
1144  
1145  	_DIAGASSERT(p != NULL);
1146  
1147  	if (reclimit++ > RECLIMIT)
1148  		p->error = REG_ESPACE;
1149  	if (p->error)
1150  		return;
1151  
1152  	finish = HERE();
1153  
1154  	assert(from <= to);
1155  
1156  	switch (REP(MAP(from), MAP(to))) {
1157  	case REP(0, 0):			/* must be user doing this */
1158  		DROP(finish-start);	/* drop the operand */
1159  		break;
1160  	case REP(0, 1):			/* as x{1,1}? */
1161  	case REP(0, N):			/* as x{1,n}? */
1162  	case REP(0, INF):		/* as x{1,}? */
1163  		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1164  		INSERT(OCH_, start);		/* offset is wrong... */
1165  		repeat(p, start+1, 1, to, reclimit);
1166  		ASTERN(OOR1, start);
1167  		AHEAD(start);			/* ... fix it */
1168  		EMIT(OOR2, 0);
1169  		AHEAD(THERE());
1170  		ASTERN(O_CH, THERETHERE());
1171  		break;
1172  	case REP(1, 1):			/* trivial case */
1173  		/* done */
1174  		break;
1175  	case REP(1, N):			/* as x?x{1,n-1} */
1176  		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1177  		INSERT(OCH_, start);
1178  		ASTERN(OOR1, start);
1179  		AHEAD(start);
1180  		EMIT(OOR2, 0);			/* offset very wrong... */
1181  		AHEAD(THERE());			/* ...so fix it */
1182  		ASTERN(O_CH, THERETHERE());
1183  		copy = dupl(p, start+1, finish+1);
1184  		assert(copy == finish+4);
1185  		repeat(p, copy, 1, to-1, reclimit);
1186  		break;
1187  	case REP(1, INF):		/* as x+ */
1188  		INSERT(OPLUS_, start);
1189  		ASTERN(O_PLUS, start);
1190  		break;
1191  	case REP(N, N):			/* as xx{m-1,n-1} */
1192  		copy = dupl(p, start, finish);
1193  		repeat(p, copy, from-1, to-1, reclimit);
1194  		break;
1195  	case REP(N, INF):		/* as xx{n-1,INF} */
1196  		copy = dupl(p, start, finish);
1197  		repeat(p, copy, from-1, to, reclimit);
1198  		break;
1199  	default:			/* "can't happen" */
1200  		SETERROR(REG_ASSERT);	/* just in case */
1201  		break;
1202  	}
1203  }
1204  
1205  /*
1206   - seterr - set an error condition
1207   == static int seterr(struct parse *p, int e);
1208   */
1209  static int			/* useless but makes type checking happy */
seterr(struct parse * p,int e)1210  seterr(
1211      struct parse *p,
1212      int e)
1213  {
1214  
1215  	_DIAGASSERT(p != NULL);
1216  
1217  	if (p->error == 0)	/* keep earliest error condition */
1218  		p->error = e;
1219  	p->next = nuls;		/* try to bring things to a halt */
1220  	p->end = nuls;
1221  	return(0);		/* make the return value well-defined */
1222  }
1223  
1224  /*
1225   - allocset - allocate a set of characters for []
1226   == static cset *allocset(struct parse *p);
1227   */
1228  static cset *
allocset(struct parse * p)1229  allocset(
1230      struct parse *p)
1231  {
1232  	size_t no;
1233  	size_t nc;
1234  	size_t nbytes;
1235  	cset *cs;
1236  	size_t css;
1237  	size_t i;
1238  	void *old_ptr;
1239  
1240  	_DIAGASSERT(p != NULL);
1241  
1242  	no = p->g->ncsets++;
1243  	css = (size_t)p->g->csetsize;
1244  	if (no >= p->ncsalloc) {	/* need another column of space */
1245  		p->ncsalloc += CHAR_BIT;
1246  		nc = p->ncsalloc;
1247  		assert(nc % CHAR_BIT == 0);
1248  		nbytes = nc / CHAR_BIT * css;
1249  		if (MEMSIZE(p) > MEMLIMIT)
1250  			goto oomem;
1251  		if (reallocarr(&p->g->sets, nc, sizeof(cset)))
1252  			goto oomem;
1253  		old_ptr = p->g->setbits;
1254  		if (reallocarr(&p->g->setbits, nc / CHAR_BIT, css)) {
1255  			free(old_ptr);
1256  			goto oomem;
1257  		}
1258  		if (old_ptr != p->g->setbits) {
1259  			for (i = 0; i < no; i++)
1260  				p->g->sets[i].ptr = p->g->setbits + css*(i/CHAR_BIT);
1261  		}
1262  		(void) memset((char *)p->g->setbits + (nbytes - css), 0, css);
1263  	}
1264  
1265  	cs = &p->g->sets[no];
1266  	cs->ptr = p->g->setbits + css*((no)/CHAR_BIT);
1267  	cs->mask = 1 << (unsigned int)((no) % CHAR_BIT);
1268  	cs->hash = 0;
1269  	cs->smultis = 0;
1270  	cs->multis = NULL;
1271  
1272  	return(cs);
1273  
1274  oomem:
1275  	SETERROR(REG_ESPACE);
1276  	/* caller's responsibility not to do set ops */
1277  	return NULL;
1278  }
1279  
1280  /*
1281   - freeset - free a now-unused set
1282   == static void freeset(struct parse *p, cset *cs);
1283   */
1284  static void
freeset(struct parse * p,cset * cs)1285  freeset(
1286      struct parse *p,
1287      cset *cs)
1288  {
1289  	size_t i;
1290  	cset *top;
1291  	size_t css;
1292  
1293  	_DIAGASSERT(p != NULL);
1294  	_DIAGASSERT(cs != NULL);
1295  
1296  	top = &p->g->sets[p->g->ncsets];
1297  	css = (size_t)p->g->csetsize;
1298  
1299  	for (i = 0; i < css; i++)
1300  		CHsub(cs, (int)i);
1301  	if (cs == top-1)	/* recover only the easy case */
1302  		p->g->ncsets--;
1303  }
1304  
1305  /*
1306   - freezeset - final processing on a set of characters
1307   == static int freezeset(struct parse *p, cset *cs);
1308   *
1309   * The main task here is merging identical sets.  This is usually a waste
1310   * of time (although the hash code minimizes the overhead), but can win
1311   * big if REG_ICASE is being used.  REG_ICASE, by the way, is why the hash
1312   * is done using addition rather than xor -- all ASCII [aA] sets xor to
1313   * the same value!
1314   */
1315  static sopno			/* set number */
freezeset(struct parse * p,cset * cs)1316  freezeset(
1317      struct parse *p,
1318      cset *cs)
1319  {
1320  	uch h;
1321  	size_t i;
1322  	cset *top;
1323  	cset *cs2;
1324  	size_t css;
1325  
1326  	_DIAGASSERT(p != NULL);
1327  	_DIAGASSERT(cs != NULL);
1328  
1329  	h = cs->hash;
1330  	top = &p->g->sets[p->g->ncsets];
1331  	css = (size_t)p->g->csetsize;
1332  
1333  	/* look for an earlier one which is the same */
1334  	for (cs2 = &p->g->sets[0]; cs2 < top; cs2++)
1335  		if (cs2->hash == h && cs2 != cs) {
1336  			/* maybe */
1337  			for (i = 0; i < css; i++)
1338  				if (!!CHIN(cs2, i) != !!CHIN(cs, i))
1339  					break;		/* no */
1340  			if (i == css)
1341  				break;			/* yes */
1342  		}
1343  
1344  	if (cs2 < top) {	/* found one */
1345  		freeset(p, cs);
1346  		cs = cs2;
1347  	}
1348  
1349  	return (sopno)(cs - p->g->sets);
1350  }
1351  
1352  /*
1353   - firstch - return first character in a set (which must have at least one)
1354   == static int firstch(struct parse *p, cset *cs);
1355   */
1356  static int			/* character; there is no "none" value */
firstch(struct parse * p,cset * cs)1357  firstch(
1358      struct parse *p,
1359      cset *cs)
1360  {
1361  	size_t i;
1362  	size_t css;
1363  
1364  	_DIAGASSERT(p != NULL);
1365  	_DIAGASSERT(cs != NULL);
1366  
1367  	css = (size_t)p->g->csetsize;
1368  
1369  	for (i = 0; i < css; i++)
1370  		if (CHIN(cs, i))
1371  			return((char)i);
1372  	assert(never);
1373  	return(0);		/* arbitrary */
1374  }
1375  
1376  /*
1377   - nch - number of characters in a set
1378   == static int nch(struct parse *p, cset *cs);
1379   */
1380  static int
nch(struct parse * p,cset * cs)1381  nch(
1382      struct parse *p,
1383      cset *cs)
1384  {
1385  	size_t i;
1386  	size_t css;
1387  	int n = 0;
1388  
1389  	_DIAGASSERT(p != NULL);
1390  	_DIAGASSERT(cs != NULL);
1391  
1392  	css = (size_t)p->g->csetsize;
1393  
1394  	for (i = 0; i < css; i++)
1395  		if (CHIN(cs, i))
1396  			n++;
1397  	return(n);
1398  }
1399  
1400  /*
1401   - mcadd - add a collating element to a cset
1402   == static void mcadd(struct parse *p, cset *cs, \
1403   ==	char *cp);
1404   */
1405  static void
mcadd(struct parse * p,cset * cs,const char * cp)1406  mcadd(
1407      struct parse *p,
1408      cset *cs,
1409      const char *cp)
1410  {
1411  	size_t oldend;
1412  
1413  	_DIAGASSERT(p != NULL);
1414  	_DIAGASSERT(cs != NULL);
1415  	_DIAGASSERT(cp != NULL);
1416  
1417  	oldend = cs->smultis;
1418  
1419  	cs->smultis += strlen(cp) + 1;
1420  	if (cs->multis == NULL)
1421  		cs->multis = malloc(cs->smultis);
1422  	else
1423  		cs->multis = realloc(cs->multis, cs->smultis);
1424  	if (cs->multis == NULL) {
1425  		SETERROR(REG_ESPACE);
1426  		return;
1427  	}
1428  
1429  	(void) strcpy(cs->multis + oldend - 1, cp);
1430  	cs->multis[cs->smultis - 1] = '\0';
1431  }
1432  
1433  #if 0
1434  /*
1435   - mcsub - subtract a collating element from a cset
1436   == static void mcsub(cset *cs, char *cp);
1437   */
1438  static void
1439  mcsub(
1440      cset *cs,
1441      char *cp)
1442  {
1443  	char *fp;
1444  	size_t len;
1445  
1446  	_DIAGASSERT(cs != NULL);
1447  	_DIAGASSERT(cp != NULL);
1448  
1449  	fp = mcfind(cs, cp);
1450  	len = strlen(fp);
1451  
1452  	assert(fp != NULL);
1453  	(void) memmove(fp, fp + len + 1,
1454  				cs->smultis - (fp + len + 1 - cs->multis));
1455  	cs->smultis -= len;
1456  
1457  	if (cs->smultis == 0) {
1458  		free(cs->multis);
1459  		cs->multis = NULL;
1460  		return;
1461  	}
1462  
1463  	cs->multis = realloc(cs->multis, cs->smultis);
1464  	assert(cs->multis != NULL);
1465  }
1466  
1467  /*
1468   - mcin - is a collating element in a cset?
1469   == static int mcin(cset *cs, char *cp);
1470   */
1471  static int
1472  mcin(
1473      cset *cs,
1474      char *cp)
1475  {
1476  
1477  	_DIAGASSERT(cs != NULL);
1478  	_DIAGASSERT(cp != NULL);
1479  
1480  	return(mcfind(cs, cp) != NULL);
1481  }
1482  
1483  /*
1484   - mcfind - find a collating element in a cset
1485   == static char *mcfind(cset *cs, char *cp);
1486   */
1487  static char *
1488  mcfind(
1489      cset *cs,
1490      char *cp)
1491  {
1492  	char *p;
1493  
1494  	_DIAGASSERT(cs != NULL);
1495  	_DIAGASSERT(cp != NULL);
1496  
1497  	if (cs->multis == NULL)
1498  		return(NULL);
1499  	for (p = cs->multis; *p != '\0'; p += strlen(p) + 1)
1500  		if (strcmp(cp, p) == 0)
1501  			return(p);
1502  	return(NULL);
1503  }
1504  #endif
1505  
1506  /*
1507   - mcinvert - invert the list of collating elements in a cset
1508   == static void mcinvert(struct parse *p, cset *cs);
1509   *
1510   * This would have to know the set of possibilities.  Implementation
1511   * is deferred.
1512   */
1513  /* ARGSUSED */
1514  static void
mcinvert(struct parse * p,cset * cs)1515  mcinvert(
1516      struct parse *p,
1517      cset *cs)
1518  {
1519  
1520  	_DIAGASSERT(p != NULL);
1521  	_DIAGASSERT(cs != NULL);
1522  
1523  	assert(cs->multis == NULL);	/* xxx */
1524  }
1525  
1526  /*
1527   - mccase - add case counterparts of the list of collating elements in a cset
1528   == static void mccase(struct parse *p, cset *cs);
1529   *
1530   * This would have to know the set of possibilities.  Implementation
1531   * is deferred.
1532   */
1533  /* ARGSUSED */
1534  static void
mccase(struct parse * p,cset * cs)1535  mccase(
1536      struct parse *p,
1537      cset *cs)
1538  {
1539  
1540  	_DIAGASSERT(p != NULL);
1541  	_DIAGASSERT(cs != NULL);
1542  
1543  	assert(cs->multis == NULL);	/* xxx */
1544  }
1545  
1546  /*
1547   - isinsets - is this character in any sets?
1548   == static int isinsets(struct re_guts *g, int c);
1549   */
1550  static int			/* predicate */
isinsets(struct re_guts * g,int c)1551  isinsets(
1552      struct re_guts *g,
1553      int c)
1554  {
1555  	uch *col;
1556  	size_t i;
1557  	size_t ncols;
1558  	unsigned uc = (unsigned char)c;
1559  
1560  	_DIAGASSERT(g != NULL);
1561  
1562  	if (g->setbits == NULL)
1563  		return 0;
1564  
1565  	ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1566  
1567  	for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1568  		if (col[uc] != 0)
1569  			return(1);
1570  	return(0);
1571  }
1572  
1573  /*
1574   - samesets - are these two characters in exactly the same sets?
1575   == static int samesets(struct re_guts *g, int c1, int c2);
1576   */
1577  static int			/* predicate */
samesets(struct re_guts * g,int c1,int c2)1578  samesets(
1579      struct re_guts *g,
1580      int c1,
1581      int c2)
1582  {
1583  	uch *col;
1584  	size_t i;
1585  	size_t ncols;
1586  	unsigned uc1 = (unsigned char)c1;
1587  	unsigned uc2 = (unsigned char)c2;
1588  
1589  	_DIAGASSERT(g != NULL);
1590  
1591  	ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1592  
1593  	for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1594  		if (col[uc1] != col[uc2])
1595  			return(0);
1596  	return(1);
1597  }
1598  
1599  /*
1600   - categorize - sort out character categories
1601   == static void categorize(struct parse *p, struct re_guts *g);
1602   */
1603  static void
categorize(struct parse * p,struct re_guts * g)1604  categorize(
1605      struct parse *p,
1606      struct re_guts *g)
1607  {
1608  	cat_t *cats;
1609  	int c;
1610  	int c2;
1611  	cat_t cat;
1612  
1613  	_DIAGASSERT(p != NULL);
1614  	_DIAGASSERT(g != NULL);
1615  
1616  	cats = g->categories;
1617  
1618  	/* avoid making error situations worse */
1619  	if (p->error != 0)
1620  		return;
1621  
1622  	for (c = CHAR_MIN; c <= CHAR_MAX; c++)
1623  		if (cats[c] == 0 && isinsets(g, c)) {
1624  			_DIAGASSERT(__type_fit(unsigned char,
1625  			    g->ncategories + 1));
1626  			cat = g->ncategories++;
1627  			cats[c] = cat;
1628  			for (c2 = c+1; c2 <= CHAR_MAX; c2++)
1629  				if (cats[c2] == 0 && samesets(g, c, c2))
1630  					cats[c2] = cat;
1631  		}
1632  }
1633  
1634  /*
1635   - dupl - emit a duplicate of a bunch of sops
1636   == static sopno dupl(struct parse *p, sopno start, sopno finish);
1637   */
1638  static sopno			/* start of duplicate */
dupl(struct parse * p,sopno start,sopno finish)1639  dupl(
1640      struct parse *p,
1641      sopno start,			/* from here */
1642      sopno finish)			/* to this less one */
1643  {
1644  	sopno ret;
1645  	sopno len = finish - start;
1646  
1647  	_DIAGASSERT(p != NULL);
1648  
1649  	ret = HERE();
1650  
1651  	assert(finish >= start);
1652  	if (len == 0)
1653  		return(ret);
1654  	if (!enlarge(p, p->ssize + len))/* this many unexpected additions */
1655  		return ret;
1656  	(void)memcpy(p->strip + p->slen, p->strip + start,
1657  	    (size_t)len * sizeof(sop));
1658  	p->slen += len;
1659  	return(ret);
1660  }
1661  
1662  /*
1663   - doemit - emit a strip operator
1664   == static void doemit(struct parse *p, sop op, size_t opnd);
1665   *
1666   * It might seem better to implement this as a macro with a function as
1667   * hard-case backup, but it's just too big and messy unless there are
1668   * some changes to the data structures.  Maybe later.
1669   */
1670  static void
doemit(struct parse * p,sop op,sopno opnd)1671  doemit(
1672      struct parse *p,
1673      sop op,
1674      sopno opnd)
1675  {
1676  	_DIAGASSERT(p != NULL);
1677  
1678  	/* avoid making error situations worse */
1679  	if (p->error != 0)
1680  		return;
1681  
1682  	/* deal with oversize operands ("can't happen", more or less) */
1683  	assert(opnd < 1<<OPSHIFT);
1684  
1685  	/* deal with undersized strip */
1686  	if (p->slen >= p->ssize)
1687  		if (!enlarge(p, (p->ssize+1) / 2 * 3))	/* +50% */
1688  			return;
1689  
1690  	/* finally, it's all reduced to the easy case */
1691  	p->strip[p->slen++] = (sop)SOP(op, opnd);
1692  }
1693  
1694  /*
1695   - doinsert - insert a sop into the strip
1696   == static void doinsert(struct parse *p, sop op, size_t opnd, sopno pos);
1697   */
1698  static void
doinsert(struct parse * p,sop op,sopno opnd,sopno pos)1699  doinsert(
1700      struct parse *p,
1701      sop op,
1702      sopno opnd,
1703      sopno pos)
1704  {
1705  	sopno sn;
1706  	sop s;
1707  	int i;
1708  
1709  	_DIAGASSERT(p != NULL);
1710  
1711  	/* avoid making error situations worse */
1712  	if (p->error != 0)
1713  		return;
1714  
1715  	sn = HERE();
1716  	EMIT(op, opnd);		/* do checks, ensure space */
1717  	assert(HERE() == sn+1);
1718  	s = p->strip[sn];
1719  
1720  	/* adjust paren pointers */
1721  	assert(pos > 0);
1722  	for (i = 1; i < NPAREN; i++) {
1723  		if (p->pbegin[i] >= pos) {
1724  			p->pbegin[i]++;
1725  		}
1726  		if (p->pend[i] >= pos) {
1727  			p->pend[i]++;
1728  		}
1729  	}
1730  
1731  	memmove(&p->strip[pos+1], &p->strip[pos], (HERE()-pos-1)*sizeof(sop));
1732  	p->strip[pos] = s;
1733  }
1734  
1735  /*
1736   - dofwd - complete a forward reference
1737   == static void dofwd(struct parse *p, sopno pos, sop value);
1738   */
1739  static void
dofwd(struct parse * p,sopno pos,sopno value)1740  dofwd(
1741      struct parse *p,
1742      sopno pos,
1743      sopno value)
1744  {
1745  
1746  	_DIAGASSERT(p != NULL);
1747  
1748  	/* avoid making error situations worse */
1749  	if (p->error != 0)
1750  		return;
1751  
1752  	assert(value < 1<<OPSHIFT);
1753  	p->strip[pos] = (sop)(OP(p->strip[pos]) | value);
1754  }
1755  
1756  /*
1757   - enlarge - enlarge the strip
1758   == static void enlarge(struct parse *p, sopno size);
1759   */
1760  static int
enlarge(struct parse * p,sopno size)1761  enlarge(struct parse *p, sopno size)
1762  {
1763  	_DIAGASSERT(p != NULL);
1764  
1765  	if (p->ssize >= size)
1766  		return 1;
1767  
1768  	if (MEMSIZE(p) > MEMLIMIT || reallocarr(&p->strip, size, sizeof(sop))) {
1769  		SETERROR(REG_ESPACE);
1770  		return 0;
1771  	}
1772  	p->ssize = size;
1773  	return 1;
1774  }
1775  
1776  /*
1777   - stripsnug - compact the strip
1778   == static void stripsnug(struct parse *p, struct re_guts *g);
1779   */
1780  static void
stripsnug(struct parse * p,struct re_guts * g)1781  stripsnug(
1782      struct parse *p,
1783      struct re_guts *g)
1784  {
1785  
1786  	_DIAGASSERT(p != NULL);
1787  	_DIAGASSERT(g != NULL);
1788  
1789  	g->nstates = p->slen;
1790  	g->strip = p->strip;
1791  	reallocarr(&g->strip, p->slen, sizeof(sop));
1792  	/* Ignore error as tries to free memory only. */
1793  }
1794  
1795  /*
1796   - findmust - fill in must and mlen with longest mandatory literal string
1797   == static void findmust(struct parse *p, struct re_guts *g);
1798   *
1799   * This algorithm could do fancy things like analyzing the operands of |
1800   * for common subsequences.  Someday.  This code is simple and finds most
1801   * of the interesting cases.
1802   *
1803   * Note that must and mlen got initialized during setup.
1804   */
1805  static void
findmust(struct parse * p,struct re_guts * g)1806  findmust(
1807      struct parse *p,
1808      struct re_guts *g)
1809  {
1810  	sop *scan;
1811  	sop *start = NULL;
1812  	sop *newstart = NULL;
1813  	sopno newlen;
1814  	sop s;
1815  	char *cp;
1816  	sopno i;
1817  
1818  	_DIAGASSERT(p != NULL);
1819  	_DIAGASSERT(g != NULL);
1820  
1821  	/* avoid making error situations worse */
1822  	if (p->error != 0)
1823  		return;
1824  
1825  	/* find the longest OCHAR sequence in strip */
1826  	newlen = 0;
1827  	scan = g->strip + 1;
1828  	do {
1829  		s = *scan++;
1830  		switch (OP(s)) {
1831  		case OCHAR:		/* sequence member */
1832  			if (newlen == 0)		/* new sequence */
1833  				newstart = scan - 1;
1834  			newlen++;
1835  			break;
1836  		case OPLUS_:		/* things that don't break one */
1837  		case OLPAREN:
1838  		case ORPAREN:
1839  			break;
1840  		case OQUEST_:		/* things that must be skipped */
1841  		case OCH_:
1842  			scan--;
1843  			do {
1844  				scan += OPND(s);
1845  				s = *scan;
1846  				/* assert() interferes w debug printouts */
1847  				if (OP(s) != O_QUEST && OP(s) != O_CH &&
1848  							OP(s) != OOR2) {
1849  					g->iflags |= BAD;
1850  					return;
1851  				}
1852  			} while (OP(s) != O_QUEST && OP(s) != O_CH);
1853  			/* FALLTHROUGH */
1854  		default:		/* things that break a sequence */
1855  			if (newlen > g->mlen) {		/* ends one */
1856  				start = newstart;
1857  				g->mlen = newlen;
1858  			}
1859  			newlen = 0;
1860  			break;
1861  		}
1862  	} while (OP(s) != OEND);
1863  
1864  	if (start == NULL)
1865  		g->mlen = 0;
1866  
1867  	if (g->mlen == 0)	/* there isn't one */
1868  		return;
1869  
1870  	/* turn it into a character string */
1871  	g->must = malloc((size_t)g->mlen + 1);
1872  	if (g->must == NULL) {		/* argh; just forget it */
1873  		g->mlen = 0;
1874  		return;
1875  	}
1876  	cp = g->must;
1877  	scan = start;
1878  	for (i = g->mlen; i > 0; i--) {
1879  		while (OP(s = *scan++) != OCHAR)
1880  			continue;
1881  		assert(cp < g->must + g->mlen);
1882  		*cp++ = (char)OPND(s);
1883  	}
1884  	assert(cp == g->must + g->mlen);
1885  	*cp++ = '\0';		/* just on general principles */
1886  }
1887  
1888  /*
1889   - pluscount - count + nesting
1890   == static sopno pluscount(struct parse *p, struct re_guts *g);
1891   */
1892  static sopno			/* nesting depth */
pluscount(struct parse * p,struct re_guts * g)1893  pluscount(
1894      struct parse *p,
1895      struct re_guts *g)
1896  {
1897  	sop *scan;
1898  	sop s;
1899  	sopno plusnest = 0;
1900  	sopno maxnest = 0;
1901  
1902  	_DIAGASSERT(p != NULL);
1903  	_DIAGASSERT(g != NULL);
1904  
1905  	if (p->error != 0)
1906  		return(0);	/* there may not be an OEND */
1907  
1908  	scan = g->strip + 1;
1909  	do {
1910  		s = *scan++;
1911  		switch (OP(s)) {
1912  		case OPLUS_:
1913  			plusnest++;
1914  			break;
1915  		case O_PLUS:
1916  			if (plusnest > maxnest)
1917  				maxnest = plusnest;
1918  			plusnest--;
1919  			break;
1920  		}
1921  	} while (OP(s) != OEND);
1922  	if (plusnest != 0)
1923  		g->iflags |= BAD;
1924  	return(maxnest);
1925  }
1926