lib/libc/regex/regcomp.c

   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 */
 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
 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
 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
 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
 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 $? */
 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 */
 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
 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
 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
 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
 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 */
 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 */
 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 */
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
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
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
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
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 */
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 *
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
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 */
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 */
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
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
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
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
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 */
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 */
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
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 */
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
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
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
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
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
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
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 */
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 }