contrib/nvi2/regex/regcomp.c

   1 /*      $NetBSD: regcomp.c,v 1.7 2011/11/19 17:45:11 tnozaki Exp $ */
   2
   3 /*-
   4  * Copyright (c) 1992, 1993, 1994 Henry Spencer.
   5  * Copyright (c) 1992, 1993, 1994
   6  *      The Regents of the University of California.  All rights reserved.
   7  *
   8  * This code is derived from software contributed to Berkeley by
   9  * Henry Spencer of the University of Toronto.
  10  *
  11  * Redistribution and use in source and binary forms, with or without
  12  * modification, are permitted provided that the following conditions
  13  * are met:
  14  * 1. Redistributions of source code must retain the above copyright
  15  *    notice, this list of conditions and the following disclaimer.
  16  * 2. Redistributions in binary form must reproduce the above copyright
  17  *    notice, this list of conditions and the following disclaimer in the
  18  *    documentation and/or other materials provided with the distribution.
  19  * 3. Neither the name of the University nor the names of its contributors
  20  *    may be used to endorse or promote products derived from this software
  21  *    without specific prior written permission.
  22  *
  23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  33  * SUCH DAMAGE.
  34  *
  35  *      @(#)regcomp.c   8.4 (Berkeley) 3/19/94
  36  */
  37
  38 #if defined(LIBC_SCCS) && !defined(lint)
  39 static char sccsid[] = "@(#)regcomp.c   8.4 (Berkeley) 3/19/94";
  40 #endif /* LIBC_SCCS and not lint */
  41
  42 #include <sys/types.h>
  43 #include <stdio.h>
  44 #include <string.h>
  45 #include <ctype.h>
  46 #include <limits.h>
  47 #include <stdlib.h>
  48 #include <regex.h>
  49
  50 #include "utils.h"
  51 #include "regex2.h"
  52
  53 #include "cclass.h"
  54 #include "cname.h"
  55
  56 /*
  57  * parse structure, passed up and down to avoid global variables and
  58  * other clumsinesses
  59  */
  60 struct parse {
  61         RCHAR_T *next;          /* next character in RE */
  62         RCHAR_T *end;           /* end of string (-> NUL normally) */
  63         int error;              /* has an error been seen? */
  64         sop *strip;             /* malloced strip */
  65         RCHAR_T *stripdata;     /* malloced stripdata */
  66         sopno ssize;            /* malloced strip size (allocated) */
  67         sopno slen;             /* malloced strip length (used) */
  68         int ncsalloc;           /* number of csets allocated */
  69         struct re_guts *g;
  70 #       define  NPAREN  10      /* we need to remember () 1-9 for back refs */
  71         sopno pbegin[NPAREN];   /* -> ( ([0] unused) */
  72         sopno pend[NPAREN];     /* -> ) ([0] unused) */
  73 };
  74
  75 /* ========= begin header generated by ./mkh ========= */
  76 #ifdef __cplusplus
  77 extern "C" {
  78 #endif
  79
  80 /* === regcomp.c === */
  81 static void p_ere(struct parse *p, int stop, size_t reclimit);
  82 static void p_ere_exp(struct parse *p, size_t reclimit);
  83 static void p_str(struct parse *p);
  84 static void p_bre(struct parse *p, int end1, int end2, size_t reclimit);
  85 static int p_simp_re(struct parse *p, int starordinary, size_t reclimit);
  86 static int p_count(struct parse *p);
  87 static void p_bracket(struct parse *p);
  88 static void p_b_term(struct parse *p, cset *cs);
  89 static void p_b_cclass(struct parse *p, cset *cs);
  90 static void p_b_eclass(struct parse *p, cset *cs);
  91 static char p_b_symbol(struct parse *p);
  92 static char p_b_coll_elem(struct parse *p, int endc);
  93 static char othercase(int ch);
  94 static void bothcases(struct parse *p, int ch);
  95 static void ordinary(struct parse *p, int ch);
  96 static void nonnewline(struct parse *p);
  97 static void repeat(struct parse *p, sopno start, int from, int to, size_t reclimit);
  98 static int seterr(struct parse *p, int e);
  99 static cset *allocset(struct parse *p);
 100 static void freeset(struct parse *p, cset *cs);
 101 static int freezeset(struct parse *p, cset *cs);
 102 static int firstch(struct parse *p, cset *cs);
 103 static int nch(struct parse *p, cset *cs);
 104 static void mcadd(struct parse *p, cset *cs, const char *cp);
 105 #ifdef notdef
 106 static void mcsub(cset *cs, char *cp);
 107 static int mcin(cset *cs, char *cp);
 108 static char *mcfind(cset *cs, char *cp);
 109 #endif
 110 static void mcinvert(struct parse *p, cset *cs);
 111 static void mccase(struct parse *p, cset *cs);
 112 #ifdef notdef
 113 static int isinsets(struct re_guts *g, int c);
 114 static int samesets(struct re_guts *g, int c1, int c2);
 115 #endif
 116 static void categorize(struct parse *p, struct re_guts *g);
 117 static sopno dupl(struct parse *p, sopno start, sopno finish);
 118 static void doemit(struct parse *p, sop op, size_t opnd);
 119 static void doinsert(struct parse *p, sop op, size_t opnd, sopno pos);
 120 static void dofwd(struct parse *p, sopno pos, sop value);
 121 static int enlarge(struct parse *p, sopno size);
 122 static void stripsnug(struct parse *p, struct re_guts *g);
 123 static void findmust(struct parse *p, struct re_guts *g);
 124 static sopno pluscount(struct parse *p, struct re_guts *g);
 125
 126 #ifdef __cplusplus
 127 }
 128 #endif
 129 /* ========= end header generated by ./mkh ========= */
 130
 131 static RCHAR_T nuls[10];                /* place to point scanner in event of error */
 132
 133 /*
 134  * macros for use with parse structure
 135  * BEWARE:  these know that the parse structure is named `p' !!!
 136  */
 137 #define PEEK()  (*p->next)
 138 #define PEEK2() (*(p->next+1))
 139 #define MORE()  (p->next < p->end)
 140 #define MORE2() (p->next+1 < p->end)
 141 #define SEE(c)  (MORE() && PEEK() == (c))
 142 #define SEETWO(a, b)    (MORE() && MORE2() && PEEK() == (a) && PEEK2() == (b))
 143 #define EAT(c)  ((SEE(c)) ? (NEXT(), 1) : 0)
 144 #define EATTWO(a, b)    ((SEETWO(a, b)) ? (NEXT2(), 1) : 0)
 145 #define NEXT()  (p->next++)
 146 #define NEXT2() (p->next += 2)
 147 #define NEXTn(n)        (p->next += (n))
 148 #define GETNEXT()       (*p->next++)
 149 #define SETERROR(e)     seterr(p, (e))
 150 #define REQUIRE(co, e)  ((co) || SETERROR(e))
 151 #define MUSTSEE(c, e)   (REQUIRE(MORE() && PEEK() == (c), e))
 152 #define MUSTEAT(c, e)   (REQUIRE(MORE() && GETNEXT() == (c), e))
 153 #define MUSTNOTSEE(c, e)        (REQUIRE(!MORE() || PEEK() != (c), e))
 154 #define EMIT(op, sopnd) doemit(p, (sop)(op), (size_t)(sopnd))
 155 #define INSERT(op, pos) doinsert(p, (sop)(op), HERE()-(pos)+1, pos)
 156 #define AHEAD(pos)              dofwd(p, pos, HERE()-(pos))
 157 #define ASTERN(sop, pos)        EMIT(sop, HERE()-pos)
 158 #define HERE()          (p->slen)
 159 #define THERE()         (p->slen - 1)
 160 #define THERETHERE()    (p->slen - 2)
 161 #define DROP(n) (p->slen -= (n))
 162
 163 #ifndef NDEBUG
 164 static int never = 0;           /* for use in asserts; shuts lint up */
 165 #else
 166 #define never   0               /* some <assert.h>s have bugs too */
 167 #endif
 168
 169 #define MEMLIMIT        0x8000000
 170 #define MEMSIZE(p) \
 171         ((p)->ncsalloc / CHAR_BIT * (p)->g->csetsize + \
 172         (p)->ncsalloc * sizeof(cset) + \
 173         (p)->ssize * sizeof(sop))
 174 #define RECLIMIT        256
 175
 176 /*
 177  - regcomp - interface for parser and compilation
 178  */
 179 int                             /* 0 success, otherwise REG_something */
 180 regcomp(regex_t *preg, const RCHAR_T *pattern, int cflags)
 181 {
 182         struct parse pa;
 183         struct re_guts *g;
 184         struct parse *p = &pa;
 185         int i;
 186         size_t len;
 187 #ifdef REDEBUG
 188 #       define  GOODFLAGS(f)    (f)
 189 #else
 190 #       define  GOODFLAGS(f)    ((f)&~REG_DUMP)
 191 #endif
 192
 193         cflags = GOODFLAGS(cflags);
 194         if ((cflags&REG_EXTENDED) && (cflags&REG_NOSPEC))
 195                 return(REG_INVARG);
 196
 197         if (cflags&REG_PEND) {
 198                 if (preg->re_endp < pattern)
 199                         return(REG_INVARG);
 200                 len = preg->re_endp - pattern;
 201         } else
 202                 len = STRLEN(pattern);
 203
 204         /* do the mallocs early so failure handling is easy */
 205         g = (struct re_guts *)malloc(sizeof(struct re_guts) +
 206                                                         (NC-1)*sizeof(cat_t));
 207         if (g == NULL)
 208                 return(REG_ESPACE);
 209         p->ssize = len/(size_t)2*(size_t)3 + (size_t)1; /* ugh */
 210         p->strip = (sop *)malloc(p->ssize * sizeof(sop));
 211         if (p->strip == NULL) {
 212                 free((char *)g);
 213                 return(REG_ESPACE);
 214         }
 215         p->stripdata = (RCHAR_T *)malloc(p->ssize * sizeof(RCHAR_T));
 216         if (p->stripdata == NULL) {
 217                 free((char *)p->strip);
 218                 free((char *)g);
 219                 return(REG_ESPACE);
 220         }
 221         p->slen = 0;
 222
 223         /* set things up */
 224         p->g = g;
 225         p->next = (RCHAR_T *)pattern;   /* convenience; we do not modify it */
 226         p->end = p->next + len;
 227         p->error = 0;
 228         p->ncsalloc = 0;
 229         for (i = 0; i < NPAREN; i++) {
 230                 p->pbegin[i] = 0;
 231                 p->pend[i] = 0;
 232         }
 233         g->csetsize = NC;
 234         g->sets = NULL;
 235         g->setbits = NULL;
 236         g->ncsets = 0;
 237         g->cflags = cflags;
 238         g->iflags = 0;
 239         g->nbol = 0;
 240         g->neol = 0;
 241         g->must = NULL;
 242         g->mlen = 0;
 243         g->nsub = 0;
 244 #if 0
 245         g->ncategories = 1;     /* category 0 is "everything else" */
 246         g->categories = &g->catspace[-(CHAR_MIN)];
 247         memset((char *)g->catspace, 0, NC*sizeof(cat_t));
 248 #endif
 249         g->backrefs = 0;
 250
 251         /* do it */
 252         EMIT(OEND, 0);
 253         g->firststate = THERE();
 254         if (cflags&REG_EXTENDED)
 255                 p_ere(p, OUT, 0);
 256         else if (cflags&REG_NOSPEC)
 257                 p_str(p);
 258         else
 259                 p_bre(p, OUT, OUT, 0);
 260         EMIT(OEND, 0);
 261         g->laststate = THERE();
 262
 263         /* tidy up loose ends and fill things in */
 264         categorize(p, g);
 265         stripsnug(p, g);
 266         findmust(p, g);
 267         g->nplus = pluscount(p, g);
 268         g->magic = MAGIC2;
 269         preg->re_nsub = g->nsub;
 270         preg->re_g = g;
 271         preg->re_magic = MAGIC1;
 272 #ifndef REDEBUG
 273         /* not debugging, so can't rely on the assert() in regexec() */
 274         if (g->iflags&BAD)
 275                 SETERROR(REG_ASSERT);
 276 #endif
 277
 278         /* win or lose, we're done */
 279         if (p->error != 0)      /* lose */
 280                 regfree(preg);
 281         return(p->error);
 282 }
 283
 284 /*
 285  - p_ere - ERE parser top level, concatenation and alternation
 286  */
 287 static void
 288 p_ere(struct parse *p, int stop, size_t reclimit)
 289                                 /* character this ERE should end at */
 290 {
 291         char c;
 292         sopno prevback = 0;
 293         sopno prevfwd = 0;
 294         sopno conc;
 295         int first = 1;          /* is this the first alternative? */
 296
 297         if (reclimit++ > RECLIMIT || p->error == REG_ESPACE) {
 298                 p->error = REG_ESPACE;
 299                 return;
 300         }
 301
 302         for (;;) {
 303                 /* do a bunch of concatenated expressions */
 304                 conc = HERE();
 305                 while (MORE() && (c = PEEK()) != '|' && c != stop)
 306                         p_ere_exp(p, reclimit);
 307                 (void)REQUIRE(HERE() != conc, REG_EMPTY);       /* require nonempty */
 308
 309                 if (!EAT('|'))
 310                         break;          /* NOTE BREAK OUT */
 311
 312                 if (first) {
 313                         INSERT(OCH_, conc);     /* offset is wrong */
 314                         prevfwd = conc;
 315                         prevback = conc;
 316                         first = 0;
 317                 }
 318                 ASTERN(OOR1, prevback);
 319                 prevback = THERE();
 320                 AHEAD(prevfwd);                 /* fix previous offset */
 321                 prevfwd = HERE();
 322                 EMIT(OOR2, 0);                  /* offset is very wrong */
 323         }
 324
 325         if (!first) {           /* tail-end fixups */
 326                 AHEAD(prevfwd);
 327                 ASTERN(O_CH, prevback);
 328         }
 329
 330         assert(!MORE() || SEE(stop));
 331 }
 332
 333 /*
 334  - p_ere_exp - parse one subERE, an atom possibly followed by a repetition op
 335  */
 336 static void
 337 p_ere_exp(struct parse *p, size_t reclimit)
 338 {
 339         char c;
 340         sopno pos;
 341         int count;
 342         int count2;
 343         sopno subno;
 344         int wascaret = 0;
 345
 346         assert(MORE());         /* caller should have ensured this */
 347         c = GETNEXT();
 348
 349         pos = HERE();
 350         switch (c) {
 351         case '(':
 352                 (void)REQUIRE(MORE(), REG_EPAREN);
 353                 p->g->nsub++;
 354                 subno = p->g->nsub;
 355                 if (subno < NPAREN)
 356                         p->pbegin[subno] = HERE();
 357                 EMIT(OLPAREN, subno);
 358                 if (!SEE(')'))
 359                         p_ere(p, ')', reclimit);
 360                 if (subno < NPAREN) {
 361                         p->pend[subno] = HERE();
 362                         assert(p->pend[subno] != 0);
 363                 }
 364                 EMIT(ORPAREN, subno);
 365                 (void)MUSTEAT(')', REG_EPAREN);
 366                 break;
 367 #ifndef POSIX_MISTAKE
 368         case ')':               /* happens only if no current unmatched ( */
 369                 /*
 370                  * You may ask, why the ifndef?  Because I didn't notice
 371                  * this until slightly too late for 1003.2, and none of the
 372                  * other 1003.2 regular-expression reviewers noticed it at
 373                  * all.  So an unmatched ) is legal POSIX, at least until
 374                  * we can get it fixed.
 375                  */
 376                 SETERROR(REG_EPAREN);
 377                 break;
 378 #endif
 379         case '^':
 380                 EMIT(OBOL, 0);
 381                 p->g->iflags |= USEBOL;
 382                 p->g->nbol++;
 383                 wascaret = 1;
 384                 break;
 385         case '$':
 386                 EMIT(OEOL, 0);
 387                 p->g->iflags |= USEEOL;
 388                 p->g->neol++;
 389                 break;
 390         case '|':
 391                 SETERROR(REG_EMPTY);
 392                 break;
 393         case '*':
 394         case '+':
 395         case '?':
 396                 SETERROR(REG_BADRPT);
 397                 break;
 398         case '.':
 399                 if (p->g->cflags&REG_NEWLINE)
 400                         nonnewline(p);
 401                 else
 402                         EMIT(OANY, 0);
 403                 break;
 404         case '[':
 405                 p_bracket(p);
 406                 break;
 407         case '\\':
 408                 (void)REQUIRE(MORE(), REG_EESCAPE);
 409                 c = GETNEXT();
 410                 ordinary(p, c);
 411                 break;
 412         case '{':               /* okay as ordinary except if digit follows */
 413                 (void)REQUIRE(!MORE() || !ISDIGIT((UCHAR_T)PEEK()), REG_BADRPT);
 414                 /* FALLTHROUGH */
 415         default:
 416                 ordinary(p, c);
 417                 break;
 418         }
 419
 420         if (!MORE())
 421                 return;
 422         c = PEEK();
 423         /* we call { a repetition if followed by a digit */
 424         if (!( c == '*' || c == '+' || c == '?' ||
 425                                 (c == '{' && MORE2() && ISDIGIT((UCHAR_T)PEEK2())) ))
 426                 return;         /* no repetition, we're done */
 427         NEXT();
 428
 429         (void)REQUIRE(!wascaret, REG_BADRPT);
 430         switch (c) {
 431         case '*':       /* implemented as +? */
 432                 /* this case does not require the (y|) trick, noKLUDGE */
 433                 INSERT(OPLUS_, pos);
 434                 ASTERN(O_PLUS, pos);
 435                 INSERT(OQUEST_, pos);
 436                 ASTERN(O_QUEST, pos);
 437                 break;
 438         case '+':
 439                 INSERT(OPLUS_, pos);
 440                 ASTERN(O_PLUS, pos);
 441                 break;
 442         case '?':
 443                 /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
 444                 INSERT(OCH_, pos);              /* offset slightly wrong */
 445                 ASTERN(OOR1, pos);              /* this one's right */
 446                 AHEAD(pos);                     /* fix the OCH_ */
 447                 EMIT(OOR2, 0);                  /* offset very wrong... */
 448                 AHEAD(THERE());                 /* ...so fix it */
 449                 ASTERN(O_CH, THERETHERE());
 450                 break;
 451         case '{':
 452                 count = p_count(p);
 453                 if (EAT(',')) {
 454                         if (ISDIGIT((UCHAR_T)PEEK())) {
 455                                 count2 = p_count(p);
 456                                 (void)REQUIRE(count <= count2, REG_BADBR);
 457                         } else          /* single number with comma */
 458                                 count2 = INFINITY;
 459                 } else          /* just a single number */
 460                         count2 = count;
 461                 repeat(p, pos, count, count2, 0);
 462                 if (!EAT('}')) {        /* error heuristics */
 463                         while (MORE() && PEEK() != '}')
 464                                 NEXT();
 465                         (void)REQUIRE(MORE(), REG_EBRACE);
 466                         SETERROR(REG_BADBR);
 467                 }
 468                 break;
 469         }
 470
 471         if (!MORE())
 472                 return;
 473         c = PEEK();
 474         if (!( c == '*' || c == '+' || c == '?' ||
 475                                 (c == '{' && MORE2() && ISDIGIT((UCHAR_T)PEEK2())) ) )
 476                 return;
 477         SETERROR(REG_BADRPT);
 478 }
 479
 480 /*
 481  - p_str - string (no metacharacters) "parser"
 482  */
 483 static void
 484 p_str(struct parse *p)
 485 {
 486         (void)REQUIRE(MORE(), REG_EMPTY);
 487         while (MORE())
 488                 ordinary(p, GETNEXT());
 489 }
 490
 491 /*
 492  - p_bre - BRE parser top level, anchoring and concatenation
 493  * Giving end1 as OUT essentially eliminates the end1/end2 check.
 494  *
 495  * This implementation is a bit of a kludge, in that a trailing $ is first
 496  * taken as an ordinary character and then revised to be an anchor.  The
 497  * only undesirable side effect is that '$' gets included as a character
 498  * category in such cases.  This is fairly harmless; not worth fixing.
 499  * The amount of lookahead needed to avoid this kludge is excessive.
 500  */
 501 static void
 502 p_bre(struct parse *p,
 503     int end1,           /* first terminating character */
 504     int end2,           /* second terminating character */
 505     size_t reclimit)
 506 {
 507         sopno start;
 508         int first = 1;                  /* first subexpression? */
 509         int wasdollar = 0;
 510
 511         if (reclimit++ > RECLIMIT || p->error == REG_ESPACE) {
 512                 p->error = REG_ESPACE;
 513                 return;
 514         }
 515
 516         start = HERE();
 517
 518         if (EAT('^')) {
 519                 EMIT(OBOL, 0);
 520                 p->g->iflags |= USEBOL;
 521                 p->g->nbol++;
 522         }
 523         while (MORE() && !SEETWO(end1, end2)) {
 524                 wasdollar = p_simp_re(p, first, reclimit);
 525                 first = 0;
 526         }
 527         if (wasdollar) {        /* oops, that was a trailing anchor */
 528                 DROP(1);
 529                 EMIT(OEOL, 0);
 530                 p->g->iflags |= USEEOL;
 531                 p->g->neol++;
 532         }
 533
 534         (void)REQUIRE(HERE() != start, REG_EMPTY);      /* require nonempty */
 535 }
 536
 537 /*
 538  - p_simp_re - parse a simple RE, an atom possibly followed by a repetition
 539  */
 540 static int                      /* was the simple RE an unbackslashed $? */
 541 p_simp_re(struct parse *p,
 542     int starordinary,           /* is a leading * an ordinary character? */
 543     size_t reclimit)
 544 {
 545         int c;
 546         int count;
 547         int count2;
 548         sopno pos;
 549         int i;
 550         sopno subno;
 551         int backsl;
 552
 553         pos = HERE();           /* repetion op, if any, covers from here */
 554
 555         assert(MORE());         /* caller should have ensured this */
 556         c = GETNEXT();
 557         backsl = c == '\\';
 558         if (backsl) {
 559                 (void)REQUIRE(MORE(), REG_EESCAPE);
 560                 c = (unsigned char)GETNEXT();
 561                 switch (c) {
 562                 case '{':
 563                         SETERROR(REG_BADRPT);
 564                         break;
 565                 case '(':
 566                         p->g->nsub++;
 567                         subno = p->g->nsub;
 568                         if (subno < NPAREN)
 569                                 p->pbegin[subno] = HERE();
 570                         EMIT(OLPAREN, subno);
 571                         /* the MORE here is an error heuristic */
 572                         if (MORE() && !SEETWO('\\', ')'))
 573                                 p_bre(p, '\\', ')', reclimit);
 574                         if (subno < NPAREN) {
 575                                 p->pend[subno] = HERE();
 576                                 assert(p->pend[subno] != 0);
 577                         }
 578                         EMIT(ORPAREN, subno);
 579                         (void)REQUIRE(EATTWO('\\', ')'), REG_EPAREN);
 580                         break;
 581                 case ')':       /* should not get here -- must be user */
 582                 case '}':
 583                         SETERROR(REG_EPAREN);
 584                         break;
 585                 case '1':
 586                 case '2':
 587                 case '3':
 588                 case '4':
 589                 case '5':
 590                 case '6':
 591                 case '7':
 592                 case '8':
 593                 case '9':
 594                         i = c - '0';
 595                         assert(i < NPAREN);
 596                         if (p->pend[i] != 0) {
 597                                 assert(i <= p->g->nsub);
 598                                 EMIT(OBACK_, i);
 599                                 assert(p->pbegin[i] != 0);
 600                                 assert(p->strip[p->pbegin[i]] == OLPAREN);
 601                                 assert(p->strip[p->pend[i]] == ORPAREN);
 602                                 (void) dupl(p, p->pbegin[i]+1, p->pend[i]);
 603                                 EMIT(O_BACK, i);
 604                         } else
 605                                 SETERROR(REG_ESUBREG);
 606                         p->g->backrefs = 1;
 607                         break;
 608                 default:
 609                         ordinary(p, c);
 610                         break;
 611                 }
 612         } else {
 613                 switch (c) {
 614                 case '.':
 615                         if (p->g->cflags&REG_NEWLINE)
 616                                 nonnewline(p);
 617                         else
 618                                 EMIT(OANY, 0);
 619                         break;
 620                 case '[':
 621                         p_bracket(p);
 622                         break;
 623                 case '*':
 624                         (void)REQUIRE(starordinary, REG_BADRPT);
 625                         /* FALLTHROUGH */
 626                 default:
 627                         ordinary(p, c);
 628                         break;
 629                 }
 630         }
 631
 632         if (EAT('*')) {         /* implemented as +? */
 633                 /* this case does not require the (y|) trick, noKLUDGE */
 634                 INSERT(OPLUS_, pos);
 635                 ASTERN(O_PLUS, pos);
 636                 INSERT(OQUEST_, pos);
 637                 ASTERN(O_QUEST, pos);
 638         } else if (EATTWO('\\', '{')) {
 639                 count = p_count(p);
 640                 if (EAT(',')) {
 641                         if (MORE() && ISDIGIT((UCHAR_T)PEEK())) {
 642                                 count2 = p_count(p);
 643                                 (void)REQUIRE(count <= count2, REG_BADBR);
 644                         } else          /* single number with comma */
 645                                 count2 = INFINITY;
 646                 } else          /* just a single number */
 647                         count2 = count;
 648                 repeat(p, pos, count, count2, reclimit);
 649                 if (!EATTWO('\\', '}')) {       /* error heuristics */
 650                         while (MORE() && !SEETWO('\\', '}'))
 651                                 NEXT();
 652                         (void)REQUIRE(MORE(), REG_EBRACE);
 653                         SETERROR(REG_BADBR);
 654                 }
 655         } else if (!backsl && c == (unsigned char)'$')  /* $ (but not \$) ends it */
 656                 return(1);
 657
 658         return(0);
 659 }
 660
 661 /*
 662  - p_count - parse a repetition count
 663  */
 664 static int                      /* the value */
 665 p_count(struct parse *p)
 666 {
 667         int count = 0;
 668         int ndigits = 0;
 669
 670         while (MORE() && ISDIGIT((UCHAR_T)PEEK()) && count <= DUPMAX) {
 671                 count = count*10 + (GETNEXT() - '0');
 672                 ndigits++;
 673         }
 674
 675         (void)REQUIRE(ndigits > 0 && count <= DUPMAX, REG_BADBR);
 676         return(count);
 677 }
 678
 679 /*
 680  - p_bracket - parse a bracketed character list
 681  *
 682  * Note a significant property of this code:  if the allocset() did SETERROR,
 683  * no set operations are done.
 684  */
 685 static void
 686 p_bracket(struct parse *p)
 687 {
 688         cset *cs;
 689         int invert = 0;
 690         static RCHAR_T bow[] = { '[', ':', '<', ':', ']', ']' };
 691         static RCHAR_T eow[] = { '[', ':', '>', ':', ']', ']' };
 692
 693         cs = allocset(p);
 694         if (cs == NULL)
 695                 return;
 696
 697         /* Dept of Truly Sickening Special-Case Kludges */
 698         if (p->next + 5 < p->end && MEMCMP(p->next, bow, 6) == 0) {
 699                 EMIT(OBOW, 0);
 700                 NEXTn(6);
 701                 return;
 702         }
 703         if (p->next + 5 < p->end && MEMCMP(p->next, eow, 6) == 0) {
 704                 EMIT(OEOW, 0);
 705                 NEXTn(6);
 706                 return;
 707         }
 708
 709         if (EAT('^'))
 710                 invert++;       /* make note to invert set at end */
 711         if (EAT(']'))
 712                 CHadd(cs, ']');
 713         else if (EAT('-'))
 714                 CHadd(cs, '-');
 715         while (MORE() && PEEK() != ']' && !SEETWO('-', ']'))
 716                 p_b_term(p, cs);
 717         if (EAT('-'))
 718                 CHadd(cs, '-');
 719         (void)MUSTEAT(']', REG_EBRACK);
 720
 721         if (p->error != 0)      /* don't mess things up further */
 722                 return;
 723
 724         if (p->g->cflags&REG_ICASE) {
 725                 int i;
 726                 int ci;
 727
 728                 for (i = p->g->csetsize - 1; i >= 0; i--)
 729                         if (CHIN(cs, i) && isalpha(i)) {
 730                                 ci = othercase(i);
 731                                 if (ci != i)
 732                                         CHadd(cs, ci);
 733                         }
 734                 if (cs->multis != NULL)
 735                         mccase(p, cs);
 736         }
 737         if (invert) {
 738                 int i;
 739
 740                 for (i = p->g->csetsize - 1; i >= 0; i--)
 741                         if (CHIN(cs, i))
 742                                 CHsub(cs, i);
 743                         else
 744                                 CHadd(cs, i);
 745                 if (p->g->cflags&REG_NEWLINE)
 746                         CHsub(cs, '\n');
 747                 if (cs->multis != NULL)
 748                         mcinvert(p, cs);
 749         }
 750
 751         assert(cs->multis == NULL);             /* xxx */
 752
 753         if (nch(p, cs) == 1) {          /* optimize singleton sets */
 754                 ordinary(p, firstch(p, cs));
 755                 freeset(p, cs);
 756         } else
 757                 EMIT(OANYOF, freezeset(p, cs));
 758 }
 759
 760 /*
 761  - p_b_term - parse one term of a bracketed character list
 762  */
 763 static void
 764 p_b_term(struct parse *p, cset *cs)
 765 {
 766         char c;
 767         char start, finish;
 768         int i;
 769
 770         /* classify what we've got */
 771         switch ((MORE()) ? PEEK() : '\0') {
 772         case '[':
 773                 c = (MORE2()) ? PEEK2() : '\0';
 774                 break;
 775         case '-':
 776                 SETERROR(REG_ERANGE);
 777                 return;                 /* NOTE RETURN */
 778                 break;
 779         default:
 780                 c = '\0';
 781                 break;
 782         }
 783
 784         switch (c) {
 785         case ':':               /* character class */
 786                 NEXT2();
 787                 (void)REQUIRE(MORE(), REG_EBRACK);
 788                 c = PEEK();
 789                 (void)REQUIRE(c != '-' && c != ']', REG_ECTYPE);
 790                 p_b_cclass(p, cs);
 791                 (void)REQUIRE(MORE(), REG_EBRACK);
 792                 (void)REQUIRE(EATTWO(':', ']'), REG_ECTYPE);
 793                 break;
 794         case '=':               /* equivalence class */
 795                 NEXT2();
 796                 (void)REQUIRE(MORE(), REG_EBRACK);
 797                 c = PEEK();
 798                 (void)REQUIRE(c != '-' && c != ']', REG_ECOLLATE);
 799                 p_b_eclass(p, cs);
 800                 (void)REQUIRE(MORE(), REG_EBRACK);
 801                 (void)REQUIRE(EATTWO('=', ']'), REG_ECOLLATE);
 802                 break;
 803         default:                /* symbol, ordinary character, or range */
 804 /* xxx revision needed for multichar stuff */
 805                 start = p_b_symbol(p);
 806                 if (SEE('-') && MORE2() && PEEK2() != ']') {
 807                         /* range */
 808                         NEXT();
 809                         if (EAT('-'))
 810                                 finish = '-';
 811                         else
 812                                 finish = p_b_symbol(p);
 813                 } else
 814                         finish = start;
 815 /* xxx what about signed chars here... */
 816                 (void)REQUIRE(start <= finish, REG_ERANGE);
 817                 for (i = start; i <= finish; i++)
 818                         CHadd(cs, i);
 819                 break;
 820         }
 821 }
 822
 823 /*
 824  - p_b_cclass - parse a character-class name and deal with it
 825  */
 826 static void
 827 p_b_cclass(struct parse *p, cset *cs)
 828 {
 829         RCHAR_T *sp = p->next;
 830         struct cclass *cp;
 831         size_t len;
 832         const char *u;
 833         char c;
 834
 835         while (MORE() && isalpha(PEEK()))
 836                 NEXT();
 837         len = p->next - sp;
 838         for (cp = cclasses; cp->name != NULL; cp++)
 839                 if (STRLEN(cp->name) == len && !MEMCMP(cp->name, sp, len))
 840                         break;
 841         if (cp->name == NULL) {
 842                 /* oops, didn't find it */
 843                 SETERROR(REG_ECTYPE);
 844                 return;
 845         }
 846
 847         u = cp->chars;
 848         while ((c = *u++) != '\0')
 849                 CHadd(cs, c);
 850         for (u = cp->multis; *u != '\0'; u += strlen(u) + 1)
 851                 MCadd(p, cs, u);
 852 }
 853
 854 /*
 855  - p_b_eclass - parse an equivalence-class name and deal with it
 856  *
 857  * This implementation is incomplete. xxx
 858  */
 859 static void
 860 p_b_eclass(struct parse *p, cset *cs)
 861 {
 862         char c;
 863
 864         c = p_b_coll_elem(p, '=');
 865         CHadd(cs, c);
 866 }
 867
 868 /*
 869  - p_b_symbol - parse a character or [..]ed multicharacter collating symbol
 870  */
 871 static char                     /* value of symbol */
 872 p_b_symbol(struct parse *p)
 873 {
 874         char value;
 875
 876         (void)REQUIRE(MORE(), REG_EBRACK);
 877         if (!EATTWO('[', '.'))
 878                 return(GETNEXT());
 879
 880         /* collating symbol */
 881         value = p_b_coll_elem(p, '.');
 882         (void)REQUIRE(EATTWO('.', ']'), REG_ECOLLATE);
 883         return(value);
 884 }
 885
 886 /*
 887  - p_b_coll_elem - parse a collating-element name and look it up
 888  */
 889 static char                     /* value of collating element */
 890 p_b_coll_elem(struct parse *p, int endc)
 891
 892                                 /* name ended by endc,']' */
 893 {
 894         RCHAR_T *sp = p->next;
 895         struct cname *cp;
 896         size_t len;
 897
 898         while (MORE() && !SEETWO(endc, ']'))
 899                 NEXT();
 900         if (!MORE()) {
 901                 SETERROR(REG_EBRACK);
 902                 return(0);
 903         }
 904         len = p->next - sp;
 905         for (cp = cnames; cp->name != NULL; cp++)
 906                 if (STRLEN(cp->name) == len && MEMCMP(cp->name, sp, len))
 907                         return(cp->code);       /* known name */
 908         if (len == 1)
 909                 return(*sp);    /* single character */
 910         SETERROR(REG_ECOLLATE);                 /* neither */
 911         return(0);
 912 }
 913
 914 /*
 915  - othercase - return the case counterpart of an alphabetic
 916  */
 917 static char                     /* if no counterpart, return ch */
 918 othercase(int ch)
 919 {
 920         assert(isalpha(ch));
 921         if (isupper(ch))
 922                 return(tolower(ch));
 923         else if (islower(ch))
 924                 return(toupper(ch));
 925         else                    /* peculiar, but could happen */
 926                 return(ch);
 927 }
 928
 929 /*
 930  - bothcases - emit a dualcase version of a two-case character
 931  *
 932  * Boy, is this implementation ever a kludge...
 933  */
 934 static void
 935 bothcases(struct parse *p, int ch)
 936 {
 937         RCHAR_T *oldnext = p->next;
 938         RCHAR_T *oldend = p->end;
 939         RCHAR_T bracket[3];
 940
 941         assert(othercase(ch) != ch);    /* p_bracket() would recurse */
 942         p->next = bracket;
 943         p->end = bracket+2;
 944         bracket[0] = ch;
 945         bracket[1] = ']';
 946         bracket[2] = '\0';
 947         p_bracket(p);
 948         assert(p->next == bracket+2);
 949         p->next = oldnext;
 950         p->end = oldend;
 951 }
 952
 953 /*
 954  - ordinary - emit an ordinary character
 955  */
 956 static void
 957 ordinary(struct parse *p, int ch)
 958 {
 959 /*
 960         cat_t *cap = p->g->categories;
 961 */
 962
 963         if ((p->g->cflags&REG_ICASE) && isalpha(ch) && othercase(ch) != ch)
 964                 bothcases(p, ch);
 965         else {
 966                 EMIT(OCHAR, (UCHAR_T)ch);
 967 /*
 968                 if (cap[ch] == 0)
 969                         cap[ch] = p->g->ncategories++;
 970 */
 971         }
 972 }
 973
 974 /*
 975  - nonnewline - emit REG_NEWLINE version of OANY
 976  *
 977  * Boy, is this implementation ever a kludge...
 978  */
 979 static void
 980 nonnewline(struct parse *p)
 981 {
 982         RCHAR_T *oldnext = p->next;
 983         RCHAR_T *oldend = p->end;
 984         RCHAR_T bracket[4];
 985
 986         p->next = bracket;
 987         p->end = bracket+3;
 988         bracket[0] = '^';
 989         bracket[1] = '\n';
 990         bracket[2] = ']';
 991         bracket[3] = '\0';
 992         p_bracket(p);
 993         assert(p->next == bracket+3);
 994         p->next = oldnext;
 995         p->end = oldend;
 996 }
 997
 998 /*
 999  - repeat - generate code for a bounded repetition, recursively if needed
1000  */
1001 static void
1002 repeat(struct parse *p,
1003     sopno start,                /* operand from here to end of strip */
1004     int from,                   /* repeated from this number */
1005     int to,                     /* to this number of times (maybe INFINITY) */
1006     size_t reclimit)
1007 {
1008         sopno finish;
1009 #       define  N       2
1010 #       define  INF     3
1011 #       define  REP(f, t)       ((f)*8 + (t))
1012 #       define  MAP(n)  (((n) <= 1) ? (n) : ((n) == INFINITY) ? INF : N)
1013         sopno copy;
1014
1015         if (reclimit++ > RECLIMIT)
1016                 p->error = REG_ESPACE;
1017         if (p->error)
1018                 return;
1019
1020         finish = HERE();
1021
1022         assert(from <= to);
1023
1024         switch (REP(MAP(from), MAP(to))) {
1025         case REP(0, 0):                 /* must be user doing this */
1026                 DROP(finish-start);     /* drop the operand */
1027                 break;
1028         case REP(0, 1):                 /* as x{1,1}? */
1029         case REP(0, N):                 /* as x{1,n}? */
1030         case REP(0, INF):               /* as x{1,}? */
1031                 /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1032                 INSERT(OCH_, start);            /* offset is wrong... */
1033                 repeat(p, start+1, 1, to, reclimit);
1034                 ASTERN(OOR1, start);
1035                 AHEAD(start);                   /* ... fix it */
1036                 EMIT(OOR2, 0);
1037                 AHEAD(THERE());
1038                 ASTERN(O_CH, THERETHERE());
1039                 break;
1040         case REP(1, 1):                 /* trivial case */
1041                 /* done */
1042                 break;
1043         case REP(1, N):                 /* as x?x{1,n-1} */
1044                 /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1045                 INSERT(OCH_, start);
1046                 ASTERN(OOR1, start);
1047                 AHEAD(start);
1048                 EMIT(OOR2, 0);                  /* offset very wrong... */
1049                 AHEAD(THERE());                 /* ...so fix it */
1050                 ASTERN(O_CH, THERETHERE());
1051                 copy = dupl(p, start+1, finish+1);
1052                 assert(copy == finish+4);
1053                 repeat(p, copy, 1, to-1, reclimit);
1054                 break;
1055         case REP(1, INF):               /* as x+ */
1056                 INSERT(OPLUS_, start);
1057                 ASTERN(O_PLUS, start);
1058                 break;
1059         case REP(N, N):                 /* as xx{m-1,n-1} */
1060                 copy = dupl(p, start, finish);
1061                 repeat(p, copy, from-1, to-1, reclimit);
1062                 break;
1063         case REP(N, INF):               /* as xx{n-1,INF} */
1064                 copy = dupl(p, start, finish);
1065                 repeat(p, copy, from-1, to, reclimit);
1066                 break;
1067         default:                        /* "can't happen" */
1068                 SETERROR(REG_ASSERT);   /* just in case */
1069                 break;
1070         }
1071 }
1072
1073 /*
1074  - seterr - set an error condition
1075  */
1076 static int                      /* useless but makes type checking happy */
1077 seterr(struct parse *p, int e)
1078 {
1079         if (p->error == 0)      /* keep earliest error condition */
1080                 p->error = e;
1081         p->next = nuls;         /* try to bring things to a halt */
1082         p->end = nuls;
1083         return(0);              /* make the return value well-defined */
1084 }
1085
1086 /*
1087  - allocset - allocate a set of characters for []
1088  */
1089 static cset *
1090 allocset(struct parse *p)
1091 {
1092         int no = p->g->ncsets++;
1093         size_t nc;
1094         size_t nbytes;
1095         cset *cs;
1096         size_t css = (size_t)p->g->csetsize;
1097         int i;
1098
1099         if (no >= p->ncsalloc) {        /* need another column of space */
1100                 p->ncsalloc += CHAR_BIT;
1101                 nc = p->ncsalloc;
1102                 assert(nc % CHAR_BIT == 0);
1103                 nbytes = nc / CHAR_BIT * css;
1104                 if (MEMSIZE(p) > MEMLIMIT)
1105                         goto oomem;
1106                 if (p->g->sets == NULL)
1107                         p->g->sets = (cset *)malloc(nc * sizeof(cset));
1108                 else
1109                         p->g->sets = (cset *)realloc((char *)p->g->sets,
1110                                                         nc * sizeof(cset));
1111                 if (p->g->setbits == NULL)
1112                         p->g->setbits = (uch *)malloc(nbytes);
1113                 else {
1114                         p->g->setbits = (uch *)realloc((char *)p->g->setbits,
1115                                                                 nbytes);
1116                         /* xxx this isn't right if setbits is now NULL */
1117                         for (i = 0; i < no; i++)
1118                                 p->g->sets[i].ptr = p->g->setbits + css*(i/CHAR_BIT);
1119                 }
1120                 if (p->g->sets != NULL && p->g->setbits != NULL)
1121                         memset((char *)p->g->setbits + (nbytes - css),
1122                                                                 0, css);
1123                 else {
1124 oomem:
1125                         no = 0;
1126                         SETERROR(REG_ESPACE);
1127                         /* caller's responsibility not to do set ops */
1128                         return NULL;
1129                 }
1130         }
1131
1132         cs = &p->g->sets[no];
1133         cs->ptr = p->g->setbits + css*((no)/CHAR_BIT);
1134         cs->mask = 1 << ((no) % CHAR_BIT);
1135         cs->hash = 0;
1136         cs->smultis = 0;
1137         cs->multis = NULL;
1138
1139         return(cs);
1140 }
1141
1142 /*
1143  - freeset - free a now-unused set
1144  */
1145 static void
1146 freeset(struct parse *p, cset *cs)
1147 {
1148         size_t i;
1149         cset *top = &p->g->sets[p->g->ncsets];
1150         size_t css = (size_t)p->g->csetsize;
1151
1152         for (i = 0; i < css; i++)
1153                 CHsub(cs, i);
1154         if (cs == top-1)        /* recover only the easy case */
1155                 p->g->ncsets--;
1156 }
1157
1158 /*
1159  - freezeset - final processing on a set of characters
1160  *
1161  * The main task here is merging identical sets.  This is usually a waste
1162  * of time (although the hash code minimizes the overhead), but can win
1163  * big if REG_ICASE is being used.  REG_ICASE, by the way, is why the hash
1164  * is done using addition rather than xor -- all ASCII [aA] sets xor to
1165  * the same value!
1166  */
1167 static int                      /* set number */
1168 freezeset(struct parse *p, cset *cs)
1169 {
1170         uch h = cs->hash;
1171         size_t i;
1172         cset *top = &p->g->sets[p->g->ncsets];
1173         cset *cs2;
1174         size_t css = (size_t)p->g->csetsize;
1175
1176         /* look for an earlier one which is the same */
1177         for (cs2 = &p->g->sets[0]; cs2 < top; cs2++)
1178                 if (cs2->hash == h && cs2 != cs) {
1179                         /* maybe */
1180                         for (i = 0; i < css; i++)
1181                                 if (!!CHIN(cs2, i) != !!CHIN(cs, i))
1182                                         break;          /* no */
1183                         if (i == css)
1184                                 break;                  /* yes */
1185                 }
1186
1187         if (cs2 < top) {        /* found one */
1188                 freeset(p, cs);
1189                 cs = cs2;
1190         }
1191
1192         return((int)(cs - p->g->sets));
1193 }
1194
1195 /*
1196  - firstch - return first character in a set (which must have at least one)
1197  */
1198 static int                      /* character; there is no "none" value */
1199 firstch(struct parse *p, cset *cs)
1200 {
1201         size_t i;
1202         size_t css = (size_t)p->g->csetsize;
1203
1204         for (i = 0; i < css; i++)
1205                 if (CHIN(cs, i))
1206                         return((char)i);
1207         assert(never);
1208         return(0);              /* arbitrary */
1209 }
1210
1211 /*
1212  - nch - number of characters in a set
1213  */
1214 static int
1215 nch(struct parse *p, cset *cs)
1216 {
1217         size_t i;
1218         size_t css = (size_t)p->g->csetsize;
1219         int n = 0;
1220
1221         for (i = 0; i < css; i++)
1222                 if (CHIN(cs, i))
1223                         n++;
1224         return(n);
1225 }
1226
1227 /*
1228  - mcadd - add a collating element to a cset
1229  */
1230 static void
1231 mcadd(struct parse *p, cset *cs, const char *cp)
1232 {
1233         size_t oldend = cs->smultis;
1234         void *np;
1235
1236         cs->smultis += strlen(cp) + 1;
1237         np = realloc(cs->multis, cs->smultis);
1238         if (np == NULL) {
1239                 if (cs->multis)
1240                         free(cs->multis);
1241                 cs->multis = NULL;
1242                 SETERROR(REG_ESPACE);
1243                 return;
1244         }
1245         cs->multis = np;
1246
1247         strlcpy(cs->multis + oldend - 1, cp, cs->smultis - oldend + 1);
1248 }
1249
1250 /*
1251  - mcinvert - invert the list of collating elements in a cset
1252  *
1253  * This would have to know the set of possibilities.  Implementation
1254  * is deferred.
1255  */
1256 static void
1257 mcinvert(struct parse *p, cset *cs)
1258 {
1259         assert(cs->multis == NULL);     /* xxx */
1260 }
1261
1262 /*
1263  - mccase - add case counterparts of the list of collating elements in a cset
1264  *
1265  * This would have to know the set of possibilities.  Implementation
1266  * is deferred.
1267  */
1268 static void
1269 mccase(struct parse *p, cset *cs)
1270 {
1271         assert(cs->multis == NULL);     /* xxx */
1272 }
1273
1274 #ifdef notdef
1275 /*
1276  - isinsets - is this character in any sets?
1277  */
1278 static int                      /* predicate */
1279 isinsets(struct re_guts *g, int c)
1280 {
1281         uch *col;
1282         int i;
1283         int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1284         unsigned uc = (unsigned char)c;
1285
1286         for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1287                 if (col[uc] != 0)
1288                         return(1);
1289         return(0);
1290 }
1291
1292 /*
1293  - samesets - are these two characters in exactly the same sets?
1294  */
1295 static int                      /* predicate */
1296 samesets(struct re_guts *g, int c1, int c2)
1297 {
1298         uch *col;
1299         int i;
1300         int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1301         unsigned uc1 = (unsigned char)c1;
1302         unsigned uc2 = (unsigned char)c2;
1303
1304         for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1305                 if (col[uc1] != col[uc2])
1306                         return(0);
1307         return(1);
1308 }
1309 #endif
1310
1311 /*
1312  - categorize - sort out character categories
1313  */
1314 static void
1315 categorize(struct parse *p, struct re_guts *g)
1316 {
1317 #ifdef notdef
1318         cat_t *cats = g->categories;
1319         int c;
1320         int c2;
1321         cat_t cat;
1322
1323         /* avoid making error situations worse */
1324         if (p->error != 0)
1325                 return;
1326
1327         for (c = CHAR_MIN; c <= CHAR_MAX; c++)
1328                 if (cats[c] == 0 && isinsets(g, c)) {
1329                         cat = g->ncategories++;
1330                         cats[c] = cat;
1331                         for (c2 = c+1; c2 <= CHAR_MAX; c2++)
1332                                 if (cats[c2] == 0 && samesets(g, c, c2))
1333                                         cats[c2] = cat;
1334                 }
1335 #endif
1336 }
1337
1338 /*
1339  - dupl - emit a duplicate of a bunch of sops
1340  */
1341 static sopno                    /* start of duplicate */
1342 dupl(struct parse *p,
1343     sopno start,                /* from here */
1344     sopno finish)               /* to this less one */
1345 {
1346         sopno ret = HERE();
1347         sopno len = finish - start;
1348
1349         assert(finish >= start);
1350         if (len == 0)
1351                 return(ret);
1352         if (!enlarge(p, p->ssize + len))        /* this many unexpected additions */
1353                 return ret;
1354         assert(p->ssize >= p->slen + len);
1355         (void) memcpy((char *)(p->strip + p->slen),
1356                 (char *)(p->strip + start), (size_t)len*sizeof(sop));
1357         (void) memcpy((char *)(p->stripdata + p->slen),
1358                 (char *)(p->stripdata + start), (size_t)len*sizeof(RCHAR_T));
1359         p->slen += len;
1360         return(ret);
1361 }
1362
1363 /*
1364  - doemit - emit a strip operator
1365  *
1366  * It might seem better to implement this as a macro with a function as
1367  * hard-case backup, but it's just too big and messy unless there are
1368  * some changes to the data structures.  Maybe later.
1369  */
1370 static void
1371 doemit(struct parse *p, sop op, size_t opnd)
1372 {
1373         /* avoid making error situations worse */
1374         if (p->error != 0)
1375                 return;
1376
1377         /* deal with oversize operands ("can't happen", more or less) */
1378         assert(opnd < 1);
1379
1380         /* deal with undersized strip */
1381         if (p->slen >= p->ssize)
1382                 if (!enlarge(p, (p->ssize+1) / 2 * 3))  /* +50% */
1383                         return;
1384
1385         /* finally, it's all reduced to the easy case */
1386         p->strip[p->slen] = op;
1387         p->stripdata[p->slen] = opnd;
1388         p->slen++;
1389 }
1390
1391 /*
1392  - doinsert - insert a sop into the strip
1393  */
1394 static void
1395 doinsert(struct parse *p, sop op, size_t opnd, sopno pos)
1396 {
1397         sopno sn;
1398         sop s;
1399         RCHAR_T d;
1400         int i;
1401
1402         /* avoid making error situations worse */
1403         if (p->error != 0)
1404                 return;
1405
1406         sn = HERE();
1407         EMIT(op, opnd);         /* do checks, ensure space */
1408         assert(HERE() == sn+1);
1409         s = p->strip[sn];
1410         d = p->stripdata[sn];
1411
1412         /* adjust paren pointers */
1413         assert(pos > 0);
1414         for (i = 1; i < NPAREN; i++) {
1415                 if (p->pbegin[i] >= pos) {
1416                         p->pbegin[i]++;
1417                 }
1418                 if (p->pend[i] >= pos) {
1419                         p->pend[i]++;
1420                 }
1421         }
1422
1423         memmove((char *)&p->strip[pos+1], (char *)&p->strip[pos],
1424                                                 (HERE()-pos-1)*sizeof(sop));
1425         memmove((char *)&p->stripdata[pos+1], (char *)&p->stripdata[pos],
1426                                                 (HERE()-pos-1)*sizeof(RCHAR_T));
1427         p->strip[pos] = s;
1428         p->stripdata[pos] = d;
1429 }
1430
1431 /*
1432  - dofwd - complete a forward reference
1433  */
1434 static void
1435 dofwd(struct parse *p, sopno pos, sop value)
1436 {
1437         /* avoid making error situations worse */
1438         if (p->error != 0)
1439                 return;
1440
1441         assert(value < 1);
1442         p->stripdata[pos] = value;
1443 }
1444
1445 /*
1446  - enlarge - enlarge the strip
1447  */
1448 static int
1449 enlarge(struct parse *p, sopno size)
1450 {
1451         sop *sp;
1452         RCHAR_T *dp;
1453         sopno osize;
1454
1455         if (p->ssize >= size)
1456                 return 1;
1457
1458         osize = p->ssize;
1459         p->ssize = size;
1460         if (MEMSIZE(p) > MEMLIMIT)
1461                 goto oomem;
1462         sp = realloc(p->strip, p->ssize * sizeof(sop));
1463         if (sp == NULL)
1464                 goto oomem;
1465         p->strip = sp;
1466         dp = realloc(p->stripdata, p->ssize * sizeof(RCHAR_T));
1467         if (dp == NULL) {
1468 oomem:
1469                 p->ssize = osize;
1470                 SETERROR(REG_ESPACE);
1471                 return 0;
1472         }
1473         p->stripdata = dp;
1474         return 1;
1475 }
1476
1477 /*
1478  - stripsnug - compact the strip
1479  */
1480 static void
1481 stripsnug(struct parse *p, struct re_guts *g)
1482 {
1483         g->nstates = p->slen;
1484         g->strip = (sop *)realloc((char *)p->strip,
1485             p->slen * sizeof(sop));
1486         if (g->strip == NULL) {
1487                 SETERROR(REG_ESPACE);
1488                 g->strip = p->strip;
1489         }
1490         g->stripdata = (RCHAR_T *)realloc((char *)p->stripdata,
1491             p->slen * sizeof(RCHAR_T));
1492         if (g->stripdata == NULL) {
1493                 SETERROR(REG_ESPACE);
1494                 g->stripdata = p->stripdata;
1495         }
1496 }
1497
1498 /*
1499  - findmust - fill in must and mlen with longest mandatory literal string
1500  *
1501  * This algorithm could do fancy things like analyzing the operands of |
1502  * for common subsequences.  Someday.  This code is simple and finds most
1503  * of the interesting cases.
1504  *
1505  * Note that must and mlen got initialized during setup.
1506  */
1507 static void
1508 findmust(struct parse *p, struct re_guts *g)
1509 {
1510         sop *scans;
1511         RCHAR_T *scand;
1512         sop *starts = 0;
1513         RCHAR_T *startd = NULL;
1514         sop *newstarts = 0;
1515         RCHAR_T *newstartd = NULL;
1516         sopno newlen;
1517         sop s;
1518         RCHAR_T d;
1519         RCHAR_T *cp;
1520         sopno i;
1521
1522         /* avoid making error situations worse */
1523         if (p->error != 0)
1524                 return;
1525
1526         /* find the longest OCHAR sequence in strip */
1527         newlen = 0;
1528         scans = g->strip + 1;
1529         scand = g->stripdata + 1;
1530         do {
1531                 s = *scans++;
1532                 d = *scand++;
1533                 switch (s) {
1534                 case OCHAR:             /* sequence member */
1535                         if (newlen == 0) {              /* new sequence */
1536                                 newstarts = scans - 1;
1537                                 newstartd = scand - 1;
1538                         }
1539                         newlen++;
1540                         break;
1541                 case OPLUS_:            /* things that don't break one */
1542                 case OLPAREN:
1543                 case ORPAREN:
1544                         break;
1545                 case OQUEST_:           /* things that must be skipped */
1546                 case OCH_:
1547                         scans--;
1548                         scand--;
1549                         do {
1550                                 scans += d;
1551                                 scand += d;
1552                                 s = *scans;
1553                                 d = *scand;
1554                                 /* assert() interferes w debug printouts */
1555                                 if (s != O_QUEST && s != O_CH && s != OOR2) {
1556                                         g->iflags |= BAD;
1557                                         return;
1558                                 }
1559                         } while (s != O_QUEST && s != O_CH);
1560                         /* fallthrough */
1561                 default:                /* things that break a sequence */
1562                         if (newlen > g->mlen) {         /* ends one */
1563                                 starts = newstarts;
1564                                 startd = newstartd;
1565                                 g->mlen = newlen;
1566                         }
1567                         newlen = 0;
1568                         break;
1569                 }
1570         } while (s != OEND);
1571
1572         if (g->mlen == 0)               /* there isn't one */
1573                 return;
1574
1575         /* turn it into a character string */
1576         g->must = malloc(((size_t)g->mlen + 1) * sizeof(RCHAR_T));
1577         if (g->must == NULL) {          /* argh; just forget it */
1578                 g->mlen = 0;
1579                 return;
1580         }
1581         cp = g->must;
1582         scans = starts;
1583         scand = startd;
1584         for (i = g->mlen; i > 0; i--) {
1585                 for (;;) {
1586                         s = *scans++;
1587                         d = *scand++;
1588                         if (s == OCHAR)
1589                                 break;
1590                 }
1591                 assert(cp < g->must + g->mlen);
1592                 *cp++ = d;
1593         }
1594         assert(cp == g->must + g->mlen);
1595         *cp++ = '\0';           /* just on general principles */
1596 }
1597
1598 /*
1599  - pluscount - count + nesting
1600  */
1601 static sopno                    /* nesting depth */
1602 pluscount(struct parse *p, struct re_guts *g)
1603 {
1604         sop *scan;
1605         sop s;
1606         sopno plusnest = 0;
1607         sopno maxnest = 0;
1608
1609         if (p->error != 0)
1610                 return(0);      /* there may not be an OEND */
1611
1612         scan = g->strip + 1;
1613         do {
1614                 s = *scan++;
1615                 switch (s) {
1616                 case OPLUS_:
1617                         plusnest++;
1618                         break;
1619                 case O_PLUS:
1620                         if (plusnest > maxnest)
1621                                 maxnest = plusnest;
1622                         plusnest--;
1623                         break;
1624                 }
1625         } while (s != OEND);
1626         if (plusnest != 0)
1627                 g->iflags |= BAD;
1628         return(maxnest);
1629 }