regex/regcomp.c

   1 /*-
   2  * Copyright (c) 1992, 1993, 1994 Henry Spencer.
   3  * Copyright (c) 1992, 1993, 1994
   4  *      The Regents of the University of California.  All rights reserved.
   5  *
   6  * This code is derived from software contributed to Berkeley by
   7  * Henry Spencer.
   8  *
   9  * Redistribution and use in source and binary forms, with or without
  10  * modification, are permitted provided that the following conditions
  11  * are met:
  12  * 1. Redistributions of source code must retain the above copyright
  13  *    notice, this list of conditions and the following disclaimer.
  14  * 2. Redistributions in binary form must reproduce the above copyright
  15  *    notice, this list of conditions and the following disclaimer in the
  16  *    documentation and/or other materials provided with the distribution.
  17  * 3. All advertising materials mentioning features or use of this software
  18  *    must display the following acknowledgement:
  19  *      This product includes software developed by the University of
  20  *      California, Berkeley and its contributors.
  21  * 4. Neither the name of the University nor the names of its contributors
  22  *    may be used to endorse or promote products derived from this software
  23  *    without specific prior written permission.
  24  *
  25  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  28  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  29  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  33  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  34  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  35  * SUCH DAMAGE.
  36  *
  37  *      @(#)regcomp.c   8.5 (Berkeley) 3/20/94
  38  */
  39
  40 #if defined(LIBC_SCCS) && !defined(lint)
  41 static char sccsid[] = "@(#)regcomp.c   8.5 (Berkeley) 3/20/94";
  42 #endif /* LIBC_SCCS and not lint */
  43
  44 #include <sys/types.h>
  45 #include <stdio.h>
  46 #include <string.h>
  47 #include <ctype.h>
  48 #include <limits.h>
  49 #include <stdlib.h>
  50 #include "regex.h"
  51
  52 #include "utils.h"
  53 #include "regex2.h"
  54
  55 #include "cclass.h"
  56 #include "cname.h"
  57
  58 /*
  59  * parse structure, passed up and down to avoid global variables and
  60  * other clumsinesses
  61  */
  62 struct parse {
  63         char *next;             /* next character in RE */
  64         char *end;              /* end of string (-> NUL normally) */
  65         int error;              /* has an error been seen? */
  66         sop *strip;             /* malloced strip */
  67         sopno ssize;            /* malloced strip size (allocated) */
  68         sopno slen;             /* malloced strip length (used) */
  69         int ncsalloc;           /* number of csets allocated */
  70         struct re_guts *g;
  71 #       define  NPAREN  10      /* we need to remember () 1-9 for back refs */
  72         sopno pbegin[NPAREN];   /* -> ( ([0] unused) */
  73         sopno pend[NPAREN];     /* -> ) ([0] unused) */
  74 };
  75
  76 /* ========= begin header generated by ./mkh ========= */
  77 #ifdef __cplusplus
  78 extern "C" {
  79 #endif
  80
  81 /* === regcomp.c === */
  82 static void p_ere __P((struct parse *p, int stop));
  83 static void p_ere_exp __P((struct parse *p));
  84 static void p_str __P((struct parse *p));
  85 static void p_bre __P((struct parse *p, int end1, int end2));
  86 static int p_simp_re __P((struct parse *p, int starordinary));
  87 static int p_count __P((struct parse *p));
  88 static void p_bracket __P((struct parse *p));
  89 static void p_b_term __P((struct parse *p, cset *cs));
  90 static void p_b_cclass __P((struct parse *p, cset *cs));
  91 static void p_b_eclass __P((struct parse *p, cset *cs));
  92 static char p_b_symbol __P((struct parse *p));
  93 static char p_b_coll_elem __P((struct parse *p, int endc));
  94 static char othercase __P((int ch));
  95 static void bothcases __P((struct parse *p, int ch));
  96 static void ordinary __P((struct parse *p, int ch));
  97 static void nonnewline __P((struct parse *p));
  98 static void repeat __P((struct parse *p, sopno start, int from, int to));
  99 static int seterr __P((struct parse *p, int e));
 100 static cset *allocset __P((struct parse *p));
 101 static void freeset __P((struct parse *p, cset *cs));
 102 static int freezeset __P((struct parse *p, cset *cs));
 103 static int firstch __P((struct parse *p, cset *cs));
 104 static int nch __P((struct parse *p, cset *cs));
 105 static void mcadd __P((struct parse *p, cset *cs, char *cp));
 106 static void mcsub __P((cset *cs, char *cp));
 107 static int mcin __P((cset *cs, char *cp));
 108 static char *mcfind __P((cset *cs, char *cp));
 109 static void mcinvert __P((struct parse *p, cset *cs));
 110 static void mccase __P((struct parse *p, cset *cs));
 111 static int isinsets __P((struct re_guts *g, int c));
 112 static int samesets __P((struct re_guts *g, int c1, int c2));
 113 static void categorize __P((struct parse *p, struct re_guts *g));
 114 static sopno dupl __P((struct parse *p, sopno start, sopno finish));
 115 static void doemit __P((struct parse *p, sop op, size_t opnd));
 116 static void doinsert __P((struct parse *p, sop op, size_t opnd, sopno pos));
 117 static void dofwd __P((struct parse *p, sopno pos, sop value));
 118 static void enlarge __P((struct parse *p, sopno size));
 119 static void stripsnug __P((struct parse *p, struct re_guts *g));
 120 static void findmust __P((struct parse *p, struct re_guts *g));
 121 static sopno pluscount __P((struct parse *p, struct re_guts *g));
 122
 123 #ifdef __cplusplus
 124 }
 125 #endif
 126 /* ========= end header generated by ./mkh ========= */
 127
 128 static char nuls[10];           /* place to point scanner in event of error */
 129
 130 /*
 131  * macros for use with parse structure
 132  * BEWARE:  these know that the parse structure is named `p' !!!
 133  */
 134 #define PEEK()  (*p->next)
 135 #define PEEK2() (*(p->next+1))
 136 #define MORE()  (p->next < p->end)
 137 #define MORE2() (p->next+1 < p->end)
 138 #define SEE(c)  (MORE() && PEEK() == (c))
 139 #define SEETWO(a, b)    (MORE() && MORE2() && PEEK() == (a) && PEEK2() == (b))
 140 #define EAT(c)  ((SEE(c)) ? (NEXT(), 1) : 0)
 141 #define EATTWO(a, b)    ((SEETWO(a, b)) ? (NEXT2(), 1) : 0)
 142 #define NEXT()  (p->next++)
 143 #define NEXT2() (p->next += 2)
 144 #define NEXTn(n)        (p->next += (n))
 145 #define GETNEXT()       (*p->next++)
 146 #define SETERROR(e)     seterr(p, (e))
 147 #define REQUIRE(co, e)  ((void) ((co) || SETERROR(e)))
 148 #define MUSTSEE(c, e)   (REQUIRE(MORE() && PEEK() == (c), e))
 149 #define MUSTEAT(c, e)   (REQUIRE(MORE() && GETNEXT() == (c), e))
 150 #define MUSTNOTSEE(c, e)        (REQUIRE(!MORE() || PEEK() != (c), e))
 151 #define EMIT(op, sopnd) doemit(p, (sop)(op), (size_t)(sopnd))
 152 #define INSERT(op, pos) doinsert(p, (sop)(op), HERE()-(pos)+1, pos)
 153 #define AHEAD(pos)              dofwd(p, pos, HERE()-(pos))
 154 #define ASTERN(sop, pos)        EMIT(sop, HERE()-pos)
 155 #define HERE()          (p->slen)
 156 #define THERE()         (p->slen - 1)
 157 #define THERETHERE()    (p->slen - 2)
 158 #define DROP(n) (p->slen -= (n))
 159
 160 #ifndef NDEBUG
 161 static int never = 0;           /* for use in asserts; shuts lint up */
 162 #else
 163 #define never   0               /* some <assert.h>s have bugs too */
 164 #endif
 165
 166 /*
 167  - regcomp - interface for parser and compilation
 168  = extern int regcomp(regex_t *, const char *, int);
 169  = #define      REG_BASIC       0000
 170  = #define      REG_EXTENDED    0001
 171  = #define      REG_ICASE       0002
 172  = #define      REG_NOSUB       0004
 173  = #define      REG_NEWLINE     0010
 174  = #define      REG_NOSPEC      0020
 175  = #define      REG_PEND        0040
 176  = #define      REG_DUMP        0200
 177  */
 178 int     __stdcall                       /* 0 success, otherwise REG_something */
 179 regcomp(regex_t *preg, const char *pattern, int cflags)
 180 {
 181         struct parse pa;
 182         register struct re_guts *g;
 183         register struct parse *p = &pa;
 184         register int i;
 185         register size_t len;
 186         register size_t maxlen;
 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((char *)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         maxlen = ((size_t)-1 >> 1) / sizeof(sop) * 2 / 3;
 210         if (len >= maxlen) {
 211                 free((char *)g);
 212                 return(REG_ESPACE);
 213         }
 214         p->ssize = len/(size_t)2*(size_t)3 + (size_t)1; /* ugh */
 215         if(p->ssize < len || p->ssize > LONG_MAX / sizeof(sop)){
 216           free((char *)g);
 217           return REG_INVARG;
 218         }
 219         p->strip = (sop *)malloc(p->ssize * sizeof(sop));
 220         p->slen = 0;
 221         if (p->strip == NULL) {
 222                 free((char *)g);
 223                 return(REG_ESPACE);
 224         }
 225
 226         /* set things up */
 227         p->g = g;
 228         p->next = (char *)pattern;      /* convenience; we do not modify it */
 229         p->end = p->next + len;
 230         p->error = 0;
 231         p->ncsalloc = 0;
 232         for (i = 0; i < NPAREN; i++) {
 233                 p->pbegin[i] = 0;
 234                 p->pend[i] = 0;
 235         }
 236         g->csetsize = NC;
 237         g->sets = NULL;
 238         g->setbits = NULL;
 239         g->ncsets = 0;
 240         g->cflags = cflags;
 241         g->iflags = 0;
 242         g->nbol = 0;
 243         g->neol = 0;
 244         g->must = NULL;
 245         g->mlen = 0;
 246         g->nsub = 0;
 247         g->ncategories = 1;     /* category 0 is "everything else" */
 248         g->categories = &g->catspace[-(CHAR_MIN)];
 249         (void) memset((char *)g->catspace, 0, NC*sizeof(cat_t));
 250         g->backrefs = 0;
 251
 252         /* do it */
 253         EMIT(OEND, 0);
 254         g->firststate = THERE();
 255         if (cflags&REG_EXTENDED)
 256                 p_ere(p, OUT);
 257         else if (cflags&REG_NOSPEC)
 258                 p_str(p);
 259         else
 260                 p_bre(p, OUT, OUT);
 261         EMIT(OEND, 0);
 262         g->laststate = THERE();
 263
 264         /* tidy up loose ends and fill things in */
 265         categorize(p, g);
 266         stripsnug(p, g);
 267         findmust(p, g);
 268         g->nplus = pluscount(p, g);
 269         g->magic = MAGIC2;
 270         preg->re_nsub = g->nsub;
 271         preg->re_g = g;
 272         preg->re_magic = MAGIC1;
 273 #ifndef REDEBUG
 274         /* not debugging, so can't rely on the assert() in regexec() */
 275         if (g->iflags&BAD)
 276                 SETERROR(REG_ASSERT);
 277 #endif
 278
 279         /* win or lose, we're done */
 280         if (p->error != 0)      /* lose */
 281                 regfree(preg);
 282         return(p->error);
 283 }
 284
 285 /*
 286  - p_ere - ERE parser top level, concatenation and alternation
 287  == static void p_ere(register struct parse *p, int stop);
 288  */
 289 static void
 290 p_ere(struct parse *p,
 291       int stop)                 /* character this ERE should end at */
 292 {
 293         register char c;
 294         register sopno prevback;
 295         register sopno prevfwd;
 296         register sopno conc;
 297         register int first = 1;         /* is this the first alternative? */
 298
 299         for (;;) {
 300                 /* do a bunch of concatenated expressions */
 301                 conc = HERE();
 302                 while (MORE() && (c = PEEK()) != '|' && c != stop)
 303                         p_ere_exp(p);
 304                 REQUIRE(HERE() != conc, REG_EMPTY);     /* require nonempty */
 305
 306                 if (!EAT('|'))
 307                         break;          /* NOTE BREAK OUT */
 308
 309                 if (first) {
 310                         INSERT(OCH_, conc);     /* offset is wrong */
 311                         prevfwd = conc;
 312                         prevback = conc;
 313                         first = 0;
 314                 }
 315                 ASTERN(OOR1, prevback);
 316                 prevback = THERE();
 317                 AHEAD(prevfwd);                 /* fix previous offset */
 318                 prevfwd = HERE();
 319                 EMIT(OOR2, 0);                  /* offset is very wrong */
 320         }
 321
 322         if (!first) {           /* tail-end fixups */
 323                 AHEAD(prevfwd);
 324                 ASTERN(O_CH, prevback);
 325         }
 326
 327         assert(!MORE() || SEE(stop));
 328 }
 329
 330 /*
 331  - p_ere_exp - parse one subERE, an atom possibly followed by a repetition op
 332  == static void p_ere_exp(register struct parse *p);
 333  */
 334 static void
 335 p_ere_exp(struct parse *p)
 336 {
 337         register char c;
 338         register sopno pos;
 339         register int count;
 340         register int count2;
 341         register sopno subno;
 342         int wascaret = 0;
 343
 344         assert(MORE());         /* caller should have ensured this */
 345         c = GETNEXT();
 346
 347         pos = HERE();
 348         switch (c) {
 349         case '(':
 350                 REQUIRE(MORE(), REG_EPAREN);
 351                 p->g->nsub++;
 352                 subno = p->g->nsub;
 353                 if (subno < NPAREN)
 354                         p->pbegin[subno] = HERE();
 355                 EMIT(OLPAREN, subno);
 356                 if (!SEE(')'))
 357                         p_ere(p, ')');
 358                 if (subno < NPAREN) {
 359                         p->pend[subno] = HERE();
 360                         assert(p->pend[subno] != 0);
 361                 }
 362                 EMIT(ORPAREN, subno);
 363                 MUSTEAT(')', REG_EPAREN);
 364                 break;
 365 #ifndef POSIX_MISTAKE
 366         case ')':               /* happens only if no current unmatched ( */
 367                 /*
 368                  * You may ask, why the ifndef?  Because I didn't notice
 369                  * this until slightly too late for 1003.2, and none of the
 370                  * other 1003.2 regular-expression reviewers noticed it at
 371                  * all.  So an unmatched ) is legal POSIX, at least until
 372                  * we can get it fixed.
 373                  */
 374                 SETERROR(REG_EPAREN);
 375                 break;
 376 #endif
 377         case '^':
 378                 EMIT(OBOL, 0);
 379                 p->g->iflags |= USEBOL;
 380                 p->g->nbol++;
 381                 wascaret = 1;
 382                 break;
 383         case '$':
 384                 EMIT(OEOL, 0);
 385                 p->g->iflags |= USEEOL;
 386                 p->g->neol++;
 387                 break;
 388         case '|':
 389                 SETERROR(REG_EMPTY);
 390                 break;
 391         case '*':
 392         case '+':
 393         case '?':
 394                 SETERROR(REG_BADRPT);
 395                 break;
 396         case '.':
 397                 if (p->g->cflags&REG_NEWLINE)
 398                         nonnewline(p);
 399                 else
 400                         EMIT(OANY, 0);
 401                 break;
 402         case '[':
 403                 p_bracket(p);
 404                 break;
 405         case '\\':
 406                 REQUIRE(MORE(), REG_EESCAPE);
 407                 c = GETNEXT();
 408                 ordinary(p, c);
 409                 break;
 410         case '{':               /* okay as ordinary except if digit follows */
 411                 REQUIRE(!MORE() || !isdigit(PEEK()), REG_BADRPT);
 412                 /* FALLTHROUGH */
 413         default:
 414                 ordinary(p, c);
 415                 break;
 416         }
 417
 418         if (!MORE())
 419                 return;
 420         c = PEEK();
 421         /* we call { a repetition if followed by a digit */
 422         if (!( c == '*' || c == '+' || c == '?' ||
 423                                 (c == '{' && MORE2() && isdigit(PEEK2())) ))
 424                 return;         /* no repetition, we're done */
 425         NEXT();
 426
 427         REQUIRE(!wascaret, REG_BADRPT);
 428         switch (c) {
 429         case '*':       /* implemented as +? */
 430                 /* this case does not require the (y|) trick, noKLUDGE */
 431                 INSERT(OPLUS_, pos);
 432                 ASTERN(O_PLUS, pos);
 433                 INSERT(OQUEST_, pos);
 434                 ASTERN(O_QUEST, pos);
 435                 break;
 436         case '+':
 437                 INSERT(OPLUS_, pos);
 438                 ASTERN(O_PLUS, pos);
 439                 break;
 440         case '?':
 441                 /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
 442                 INSERT(OCH_, pos);              /* offset slightly wrong */
 443                 ASTERN(OOR1, pos);              /* this one's right */
 444                 AHEAD(pos);                     /* fix the OCH_ */
 445                 EMIT(OOR2, 0);                  /* offset very wrong... */
 446                 AHEAD(THERE());                 /* ...so fix it */
 447                 ASTERN(O_CH, THERETHERE());
 448                 break;
 449         case '{':
 450                 count = p_count(p);
 451                 if (EAT(',')) {
 452                         if (isdigit(PEEK())) {
 453                                 count2 = p_count(p);
 454                                 REQUIRE(count <= count2, REG_BADBR);
 455                         } else          /* single number with comma */
 456                                 count2 = INFINITY;
 457                 } else          /* just a single number */
 458                         count2 = count;
 459                 repeat(p, pos, count, count2);
 460                 if (!EAT('}')) {        /* error heuristics */
 461                         while (MORE() && PEEK() != '}')
 462                                 NEXT();
 463                         REQUIRE(MORE(), REG_EBRACE);
 464                         SETERROR(REG_BADBR);
 465                 }
 466                 break;
 467         }
 468
 469         if (!MORE())
 470                 return;
 471         c = PEEK();
 472         if (!( c == '*' || c == '+' || c == '?' ||
 473                                 (c == '{' && MORE2() && isdigit(PEEK2())) ) )
 474                 return;
 475         SETERROR(REG_BADRPT);
 476 }
 477
 478 /*
 479  - p_str - string (no metacharacters) "parser"
 480  == static void p_str(register struct parse *p);
 481  */
 482 static void
 483 p_str(struct parse *p)
 484 {
 485         REQUIRE(MORE(), REG_EMPTY);
 486         while (MORE())
 487                 ordinary(p, GETNEXT());
 488 }
 489
 490 /*
 491  - p_bre - BRE parser top level, anchoring and concatenation
 492  == static void p_bre(register struct parse *p, register int end1, \
 493  ==     register int end2);
 494  * Giving end1 as OUT essentially eliminates the end1/end2 check.
 495  *
 496  * This implementation is a bit of a kludge, in that a trailing $ is first
 497  * taken as an ordinary character and then revised to be an anchor.  The
 498  * only undesirable side effect is that '$' gets included as a character
 499  * category in such cases.  This is fairly harmless; not worth fixing.
 500  * The amount of lookahead needed to avoid this kludge is excessive.
 501  */
 502 static void
 503 p_bre(struct parse *p,
 504       int end1,                 /* first terminating character */
 505       int end2)                 /* second terminating character */
 506 {
 507         register sopno start = HERE();
 508         register int first = 1;                 /* first subexpression? */
 509         register int wasdollar = 0;
 510
 511         if (EAT('^')) {
 512                 EMIT(OBOL, 0);
 513                 p->g->iflags |= USEBOL;
 514                 p->g->nbol++;
 515         }
 516         while (MORE() && !SEETWO(end1, end2)) {
 517                 wasdollar = p_simp_re(p, first);
 518                 first = 0;
 519         }
 520         if (wasdollar) {        /* oops, that was a trailing anchor */
 521                 DROP(1);
 522                 EMIT(OEOL, 0);
 523                 p->g->iflags |= USEEOL;
 524                 p->g->neol++;
 525         }
 526
 527         REQUIRE(HERE() != start, REG_EMPTY);    /* require nonempty */
 528 }
 529
 530 /*
 531  - p_simp_re - parse a simple RE, an atom possibly followed by a repetition
 532  == static int p_simp_re(register struct parse *p, int starordinary);
 533  */
 534 static int                      /* was the simple RE an unbackslashed $? */
 535 p_simp_re(struct parse *p,
 536           int starordinary)     /* is a leading * an ordinary character? */
 537 {
 538         register int c;
 539         register int count;
 540         register int count2;
 541         register sopno pos;
 542         register int i;
 543         register sopno subno;
 544 #       define  BACKSL  (1<<CHAR_BIT)
 545
 546         pos = HERE();           /* repetion op, if any, covers from here */
 547
 548         assert(MORE());         /* caller should have ensured this */
 549         c = GETNEXT();
 550         if (c == '\\') {
 551                 REQUIRE(MORE(), REG_EESCAPE);
 552                 c = BACKSL | (unsigned char)GETNEXT();
 553         }
 554         switch (c) {
 555         case '.':
 556                 if (p->g->cflags&REG_NEWLINE)
 557                         nonnewline(p);
 558                 else
 559                         EMIT(OANY, 0);
 560                 break;
 561         case '[':
 562                 p_bracket(p);
 563                 break;
 564         case BACKSL|'{':
 565                 SETERROR(REG_BADRPT);
 566                 break;
 567         case BACKSL|'(':
 568                 p->g->nsub++;
 569                 subno = p->g->nsub;
 570                 if (subno < NPAREN)
 571                         p->pbegin[subno] = HERE();
 572                 EMIT(OLPAREN, subno);
 573                 /* the MORE here is an error heuristic */
 574                 if (MORE() && !SEETWO('\\', ')'))
 575                         p_bre(p, '\\', ')');
 576                 if (subno < NPAREN) {
 577                         p->pend[subno] = HERE();
 578                         assert(p->pend[subno] != 0);
 579                 }
 580                 EMIT(ORPAREN, subno);
 581                 REQUIRE(EATTWO('\\', ')'), REG_EPAREN);
 582                 break;
 583         case BACKSL|')':        /* should not get here -- must be user */
 584         case BACKSL|'}':
 585                 SETERROR(REG_EPAREN);
 586                 break;
 587         case BACKSL|'1':
 588         case BACKSL|'2':
 589         case BACKSL|'3':
 590         case BACKSL|'4':
 591         case BACKSL|'5':
 592         case BACKSL|'6':
 593         case BACKSL|'7':
 594         case BACKSL|'8':
 595         case BACKSL|'9':
 596                 i = (c&~BACKSL) - '0';
 597                 assert(i < NPAREN);
 598                 if (p->pend[i] != 0) {
 599                         assert(i <= p->g->nsub);
 600                         EMIT(OBACK_, i);
 601                         assert(p->pbegin[i] != 0);
 602                         assert(OP(p->strip[p->pbegin[i]]) == OLPAREN);
 603                         assert(OP(p->strip[p->pend[i]]) == ORPAREN);
 604                         (void) dupl(p, p->pbegin[i]+1, p->pend[i]);
 605                         EMIT(O_BACK, i);
 606                 } else
 607                         SETERROR(REG_ESUBREG);
 608                 p->g->backrefs = 1;
 609                 break;
 610         case '*':
 611                 REQUIRE(starordinary, REG_BADRPT);
 612                 /* FALLTHROUGH */
 613         default:
 614                 ordinary(p, c &~ BACKSL);
 615                 break;
 616         }
 617
 618         if (EAT('*')) {         /* implemented as +? */
 619                 /* this case does not require the (y|) trick, noKLUDGE */
 620                 INSERT(OPLUS_, pos);
 621                 ASTERN(O_PLUS, pos);
 622                 INSERT(OQUEST_, pos);
 623                 ASTERN(O_QUEST, pos);
 624         } else if (EATTWO('\\', '{')) {
 625                 count = p_count(p);
 626                 if (EAT(',')) {
 627                         if (MORE() && isdigit(PEEK())) {
 628                                 count2 = p_count(p);
 629                                 REQUIRE(count <= count2, REG_BADBR);
 630                         } else          /* single number with comma */
 631                                 count2 = INFINITY;
 632                 } else          /* just a single number */
 633                         count2 = count;
 634                 repeat(p, pos, count, count2);
 635                 if (!EATTWO('\\', '}')) {       /* error heuristics */
 636                         while (MORE() && !SEETWO('\\', '}'))
 637                                 NEXT();
 638                         REQUIRE(MORE(), REG_EBRACE);
 639                         SETERROR(REG_BADBR);
 640                 }
 641         } else if (c == (unsigned char)'$')     /* $ (but not \$) ends it */
 642                 return(1);
 643
 644         return(0);
 645 }
 646
 647 /*
 648  - p_count - parse a repetition count
 649  == static int p_count(register struct parse *p);
 650  */
 651 static int                      /* the value */
 652 p_count(struct parse *p)
 653 {
 654         register int count = 0;
 655         register int ndigits = 0;
 656
 657         while (MORE() && isdigit(PEEK()) && count <= DUPMAX) {
 658                 count = count*10 + (GETNEXT() - '0');
 659                 ndigits++;
 660         }
 661
 662         REQUIRE(ndigits > 0 && count <= DUPMAX, REG_BADBR);
 663         return(count);
 664 }
 665
 666 /*
 667  - p_bracket - parse a bracketed character list
 668  == static void p_bracket(register struct parse *p);
 669  *
 670  * Note a significant property of this code:  if the allocset() did SETERROR,
 671  * no set operations are done.
 672  */
 673 static void
 674 p_bracket(struct parse *p)
 675 {
 676         register cset *cs = allocset(p);
 677         register int invert = 0;
 678
 679         /* Dept of Truly Sickening Special-Case Kludges */
 680         if (p->next + 5 < p->end && strncmp(p->next, "[:<:]]", 6) == 0) {
 681                 EMIT(OBOW, 0);
 682                 NEXTn(6);
 683                 return;
 684         }
 685         if (p->next + 5 < p->end && strncmp(p->next, "[:>:]]", 6) == 0) {
 686                 EMIT(OEOW, 0);
 687                 NEXTn(6);
 688                 return;
 689         }
 690
 691         if (EAT('^'))
 692                 invert++;       /* make note to invert set at end */
 693         if (EAT(']'))
 694                 CHadd(cs, ']');
 695         else if (EAT('-'))
 696                 CHadd(cs, '-');
 697         while (MORE() && PEEK() != ']' && !SEETWO('-', ']'))
 698                 p_b_term(p, cs);
 699         if (EAT('-'))
 700                 CHadd(cs, '-');
 701         MUSTEAT(']', REG_EBRACK);
 702
 703         if (p->error != 0)      /* don't mess things up further */
 704                 return;
 705
 706         if (p->g->cflags&REG_ICASE) {
 707                 register int i;
 708                 register int ci;
 709
 710                 for (i = p->g->csetsize - 1; i >= 0; i--)
 711                         if (CHIN(cs, i) && isalpha(i)) {
 712                                 ci = othercase(i);
 713                                 if (ci != i)
 714                                         CHadd(cs, ci);
 715                         }
 716                 if (cs->multis != NULL)
 717                         mccase(p, cs);
 718         }
 719         if (invert) {
 720                 register int i;
 721
 722                 for (i = p->g->csetsize - 1; i >= 0; i--)
 723                         if (CHIN(cs, i))
 724                                 CHsub(cs, i);
 725                         else
 726                                 CHadd(cs, i);
 727                 if (p->g->cflags&REG_NEWLINE)
 728                         CHsub(cs, '\n');
 729                 if (cs->multis != NULL)
 730                         mcinvert(p, cs);
 731         }
 732
 733         assert(cs->multis == NULL);             /* xxx */
 734
 735         if (nch(p, cs) == 1) {          /* optimize singleton sets */
 736                 ordinary(p, firstch(p, cs));
 737                 freeset(p, cs);
 738         } else
 739                 EMIT(OANYOF, freezeset(p, cs));
 740 }
 741
 742 /*
 743  - p_b_term - parse one term of a bracketed character list
 744  == static void p_b_term(register struct parse *p, register cset *cs);
 745  */
 746 static void
 747 p_b_term(struct parse *p, cset *cs)
 748 {
 749         register char c;
 750         register char start, finish;
 751         register int i;
 752
 753         /* classify what we've got */
 754         switch ((MORE()) ? PEEK() : '\0') {
 755         case '[':
 756                 c = (MORE2()) ? PEEK2() : '\0';
 757                 break;
 758         case '-':
 759                 SETERROR(REG_ERANGE);
 760                 return;                 /* NOTE RETURN */
 761                 break;
 762         default:
 763                 c = '\0';
 764                 break;
 765         }
 766
 767         switch (c) {
 768         case ':':               /* character class */
 769                 NEXT2();
 770                 REQUIRE(MORE(), REG_EBRACK);
 771                 c = PEEK();
 772                 REQUIRE(c != '-' && c != ']', REG_ECTYPE);
 773                 p_b_cclass(p, cs);
 774                 REQUIRE(MORE(), REG_EBRACK);
 775                 REQUIRE(EATTWO(':', ']'), REG_ECTYPE);
 776                 break;
 777         case '=':               /* equivalence class */
 778                 NEXT2();
 779                 REQUIRE(MORE(), REG_EBRACK);
 780                 c = PEEK();
 781                 REQUIRE(c != '-' && c != ']', REG_ECOLLATE);
 782                 p_b_eclass(p, cs);
 783                 REQUIRE(MORE(), REG_EBRACK);
 784                 REQUIRE(EATTWO('=', ']'), REG_ECOLLATE);
 785                 break;
 786         default:                /* symbol, ordinary character, or range */
 787 /* xxx revision needed for multichar stuff */
 788                 start = p_b_symbol(p);
 789                 if (SEE('-') && MORE2() && PEEK2() != ']') {
 790                         /* range */
 791                         NEXT();
 792                         if (EAT('-'))
 793                                 finish = '-';
 794                         else
 795                                 finish = p_b_symbol(p);
 796                 } else
 797                         finish = start;
 798 /* xxx what about signed chars here... */
 799                 REQUIRE(start <= finish, REG_ERANGE);
 800                 for (i = start; i <= finish; i++)
 801                         CHadd(cs, i);
 802                 break;
 803         }
 804 }
 805
 806 /*
 807  - p_b_cclass - parse a character-class name and deal with it
 808  == static void p_b_cclass(register struct parse *p, register cset *cs);
 809  */
 810 static void
 811 p_b_cclass(struct parse *p, cset *cs)
 812 {
 813         register char *sp = p->next;
 814         register struct cclass *cp;
 815         register size_t len;
 816         register char *u;
 817         register char c;
 818
 819         while (MORE() && isalpha(PEEK()))
 820                 NEXT();
 821         len = p->next - sp;
 822         for (cp = cclasses; cp->name != NULL; cp++)
 823                 if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0')
 824                         break;
 825         if (cp->name == NULL) {
 826                 /* oops, didn't find it */
 827                 SETERROR(REG_ECTYPE);
 828                 return;
 829         }
 830
 831         u = cp->chars;
 832         while ((c = *u++) != '\0')
 833                 CHadd(cs, c);
 834         for (u = cp->multis; *u != '\0'; u += strlen(u) + 1)
 835                 MCadd(p, cs, u);
 836 }
 837
 838 /*
 839  - p_b_eclass - parse an equivalence-class name and deal with it
 840  == static void p_b_eclass(register struct parse *p, register cset *cs);
 841  *
 842  * This implementation is incomplete. xxx
 843  */
 844 static void
 845 p_b_eclass(struct parse *p, cset *cs)
 846 {
 847         register char c;
 848
 849         c = p_b_coll_elem(p, '=');
 850         CHadd(cs, c);
 851 }
 852
 853 /*
 854  - p_b_symbol - parse a character or [..]ed multicharacter collating symbol
 855  == static char p_b_symbol(register struct parse *p);
 856  */
 857 static char                     /* value of symbol */
 858 p_b_symbol(struct parse *p)
 859 {
 860         register char value;
 861
 862         REQUIRE(MORE(), REG_EBRACK);
 863         if (!EATTWO('[', '.'))
 864                 return(GETNEXT());
 865
 866         /* collating symbol */
 867         value = p_b_coll_elem(p, '.');
 868         REQUIRE(EATTWO('.', ']'), REG_ECOLLATE);
 869         return(value);
 870 }
 871
 872 /*
 873  - p_b_coll_elem - parse a collating-element name and look it up
 874  == static char p_b_coll_elem(register struct parse *p, int endc);
 875  */
 876 static char                     /* value of collating element */
 877 p_b_coll_elem(struct parse *p,
 878               int endc)         /* name ended by endc,']' */
 879 {
 880         register char *sp = p->next;
 881         register struct cname *cp;
 882         register int len;
 883
 884         while (MORE() && !SEETWO(endc, ']'))
 885                 NEXT();
 886         if (!MORE()) {
 887                 SETERROR(REG_EBRACK);
 888                 return(0);
 889         }
 890         len = p->next - sp;
 891         for (cp = cnames; cp->name != NULL; cp++)
 892                 if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0')
 893                         return(cp->code);       /* known name */
 894         if (len == 1)
 895                 return(*sp);    /* single character */
 896         SETERROR(REG_ECOLLATE);                 /* neither */
 897         return(0);
 898 }
 899
 900 /*
 901  - othercase - return the case counterpart of an alphabetic
 902  == static char othercase(int ch);
 903  */
 904 static char                     /* if no counterpart, return ch */
 905 othercase(int ch)
 906 {
 907         assert(isalpha(ch));
 908         if (isupper(ch))
 909                 return(tolower(ch));
 910         else if (islower(ch))
 911                 return(toupper(ch));
 912         else                    /* peculiar, but could happen */
 913                 return(ch);
 914 }
 915
 916 /*
 917  - bothcases - emit a dualcase version of a two-case character
 918  == static void bothcases(register struct parse *p, int ch);
 919  *
 920  * Boy, is this implementation ever a kludge...
 921  */
 922 static void
 923 bothcases(struct parse *p, int ch)
 924 {
 925         register char *oldnext = p->next;
 926         register char *oldend = p->end;
 927         char bracket[3];
 928
 929         assert(othercase(ch) != ch);    /* p_bracket() would recurse */
 930         p->next = bracket;
 931         p->end = bracket+2;
 932         bracket[0] = ch;
 933         bracket[1] = ']';
 934         bracket[2] = '\0';
 935         p_bracket(p);
 936         assert(p->next == bracket+2);
 937         p->next = oldnext;
 938         p->end = oldend;
 939 }
 940
 941 /*
 942  - ordinary - emit an ordinary character
 943  == static void ordinary(register struct parse *p, register int ch);
 944  */
 945 static void
 946 ordinary(struct parse *p, int ch)
 947 {
 948         register cat_t *cap = p->g->categories;
 949
 950         if ((p->g->cflags&REG_ICASE) && isalpha(ch) && othercase(ch) != ch)
 951                 bothcases(p, ch);
 952         else {
 953                 EMIT(OCHAR, (unsigned char)ch);
 954                 if (cap[ch] == 0)
 955                         cap[ch] = p->g->ncategories++;
 956         }
 957 }
 958
 959 /*
 960  - nonnewline - emit REG_NEWLINE version of OANY
 961  == static void nonnewline(register struct parse *p);
 962  *
 963  * Boy, is this implementation ever a kludge...
 964  */
 965 static void
 966 nonnewline(struct parse *p)
 967 {
 968         register char *oldnext = p->next;
 969         register char *oldend = p->end;
 970         char bracket[4];
 971
 972         p->next = bracket;
 973         p->end = bracket+3;
 974         bracket[0] = '^';
 975         bracket[1] = '\n';
 976         bracket[2] = ']';
 977         bracket[3] = '\0';
 978         p_bracket(p);
 979         assert(p->next == bracket+3);
 980         p->next = oldnext;
 981         p->end = oldend;
 982 }
 983
 984 /*
 985  - repeat - generate code for a bounded repetition, recursively if needed
 986  == static void repeat(register struct parse *p, sopno start, int from, int to);
 987  */
 988 static void
 989 repeat(struct parse *p,
 990        sopno start,             /* operand from here to end of strip */
 991        int from,                /* repeated from this number */
 992        int to)                  /* to this number of times (maybe INFINITY) */
 993 {
 994         register sopno finish = HERE();
 995 #       define  N       2
 996 #       define  INF     3
 997 #       define  REP(f, t)       ((f)*8 + (t))
 998 #       define  MAP(n)  (((n) <= 1) ? (n) : ((n) == INFINITY) ? INF : N)
 999         register sopno copy;
1000
1001         if (p->error != 0)      /* head off possible runaway recursion */
1002                 return;
1003
1004         assert(from <= to);
1005
1006         switch (REP(MAP(from), MAP(to))) {
1007         case REP(0, 0):                 /* must be user doing this */
1008                 DROP(finish-start);     /* drop the operand */
1009                 break;
1010         case REP(0, 1):                 /* as x{1,1}? */
1011         case REP(0, N):                 /* as x{1,n}? */
1012         case REP(0, INF):               /* as x{1,}? */
1013                 /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1014                 INSERT(OCH_, start);            /* offset is wrong... */
1015                 repeat(p, start+1, 1, to);
1016                 ASTERN(OOR1, start);
1017                 AHEAD(start);                   /* ... fix it */
1018                 EMIT(OOR2, 0);
1019                 AHEAD(THERE());
1020                 ASTERN(O_CH, THERETHERE());
1021                 break;
1022         case REP(1, 1):                 /* trivial case */
1023                 /* done */
1024                 break;
1025         case REP(1, N):                 /* as x?x{1,n-1} */
1026                 /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
1027                 INSERT(OCH_, start);
1028                 ASTERN(OOR1, start);
1029                 AHEAD(start);
1030                 EMIT(OOR2, 0);                  /* offset very wrong... */
1031                 AHEAD(THERE());                 /* ...so fix it */
1032                 ASTERN(O_CH, THERETHERE());
1033                 copy = dupl(p, start+1, finish+1);
1034                 assert(copy == finish+4);
1035                 repeat(p, copy, 1, to-1);
1036                 break;
1037         case REP(1, INF):               /* as x+ */
1038                 INSERT(OPLUS_, start);
1039                 ASTERN(O_PLUS, start);
1040                 break;
1041         case REP(N, N):                 /* as xx{m-1,n-1} */
1042                 copy = dupl(p, start, finish);
1043                 repeat(p, copy, from-1, to-1);
1044                 break;
1045         case REP(N, INF):               /* as xx{n-1,INF} */
1046                 copy = dupl(p, start, finish);
1047                 repeat(p, copy, from-1, to);
1048                 break;
1049         default:                        /* "can't happen" */
1050                 SETERROR(REG_ASSERT);   /* just in case */
1051                 break;
1052         }
1053 }
1054
1055 /*
1056  - seterr - set an error condition
1057  == static int seterr(register struct parse *p, int e);
1058  */
1059 static int                      /* useless but makes type checking happy */
1060 seterr(struct parse *p, int e)
1061 {
1062         if (p->error == 0)      /* keep earliest error condition */
1063                 p->error = e;
1064         p->next = nuls;         /* try to bring things to a halt */
1065         p->end = nuls;
1066         return(0);              /* make the return value well-defined */
1067 }
1068
1069 /*
1070  - allocset - allocate a set of characters for []
1071  == static cset *allocset(register struct parse *p);
1072  */
1073 static cset *
1074 allocset(struct parse *p)
1075 {
1076         register int no = p->g->ncsets++;
1077         register size_t nc;
1078         register size_t nbytes;
1079         register cset *cs;
1080         register size_t css = (size_t)p->g->csetsize;
1081         register int i;
1082
1083         if (no >= p->ncsalloc) {        /* need another column of space */
1084                 p->ncsalloc += CHAR_BIT;
1085                 nc = p->ncsalloc;
1086                 assert(nc % CHAR_BIT == 0);
1087                 nbytes = nc / CHAR_BIT * css;
1088                 if (p->g->sets == NULL)
1089                         p->g->sets = (cset *)malloc(nc * sizeof(cset));
1090                 else
1091                         p->g->sets = (cset *)realloc((char *)p->g->sets,
1092                                                         nc * sizeof(cset));
1093                 if (p->g->setbits == NULL)
1094                         p->g->setbits = (uch *)malloc(nbytes);
1095                 else {
1096                         p->g->setbits = (uch *)realloc((char *)p->g->setbits,
1097                                                                 nbytes);
1098                         /* xxx this isn't right if setbits is now NULL */
1099                         for (i = 0; i < no; i++)
1100                                 p->g->sets[i].ptr = p->g->setbits + css*(i/CHAR_BIT);
1101                 }
1102                 if (p->g->sets != NULL && p->g->setbits != NULL)
1103                         (void) memset((char *)p->g->setbits + (nbytes - css),
1104                                                                 0, css);
1105                 else {
1106                         no = 0;
1107                         SETERROR(REG_ESPACE);
1108                         /* caller's responsibility not to do set ops */
1109                 }
1110         }
1111
1112         assert(p->g->sets != NULL);     /* xxx */
1113         cs = &p->g->sets[no];
1114         cs->ptr = p->g->setbits + css*((no)/CHAR_BIT);
1115         cs->mask = 1 << ((no) % CHAR_BIT);
1116         cs->hash = 0;
1117         cs->smultis = 0;
1118         cs->multis = NULL;
1119
1120         return(cs);
1121 }
1122
1123 /*
1124  - freeset - free a now-unused set
1125  == static void freeset(register struct parse *p, register cset *cs);
1126  */
1127 static void
1128 freeset(struct parse *p, cset *cs)
1129 {
1130         register int i;
1131         register cset *top = &p->g->sets[p->g->ncsets];
1132         register size_t css = (size_t)p->g->csetsize;
1133
1134         for (i = 0; i < css; i++)
1135                 CHsub(cs, i);
1136         if (cs == top-1)        /* recover only the easy case */
1137                 p->g->ncsets--;
1138 }
1139
1140 /*
1141  - freezeset - final processing on a set of characters
1142  == static int freezeset(register struct parse *p, register cset *cs);
1143  *
1144  * The main task here is merging identical sets.  This is usually a waste
1145  * of time (although the hash code minimizes the overhead), but can win
1146  * big if REG_ICASE is being used.  REG_ICASE, by the way, is why the hash
1147  * is done using addition rather than xor -- all ASCII [aA] sets xor to
1148  * the same value!
1149  */
1150 static int                      /* set number */
1151 freezeset(struct parse *p, cset *cs)
1152 {
1153         register uch h = cs->hash;
1154         register int i;
1155         register cset *top = &p->g->sets[p->g->ncsets];
1156         register cset *cs2;
1157         register size_t css = (size_t)p->g->csetsize;
1158
1159         /* look for an earlier one which is the same */
1160         for (cs2 = &p->g->sets[0]; cs2 < top; cs2++)
1161                 if (cs2->hash == h && cs2 != cs) {
1162                         /* maybe */
1163                         for (i = 0; i < css; i++)
1164                                 if (!!CHIN(cs2, i) != !!CHIN(cs, i))
1165                                         break;          /* no */
1166                         if (i == css)
1167                                 break;                  /* yes */
1168                 }
1169
1170         if (cs2 < top) {        /* found one */
1171                 freeset(p, cs);
1172                 cs = cs2;
1173         }
1174
1175         return((int)(cs - p->g->sets));
1176 }
1177
1178 /*
1179  - firstch - return first character in a set (which must have at least one)
1180  == static int firstch(register struct parse *p, register cset *cs);
1181  */
1182 static int                      /* character; there is no "none" value */
1183 firstch(struct parse *p, cset *cs)
1184 {
1185         register int i;
1186         register size_t css = (size_t)p->g->csetsize;
1187
1188         for (i = 0; i < css; i++)
1189                 if (CHIN(cs, i))
1190                         return((char)i);
1191         assert(never);
1192         return(0);              /* arbitrary */
1193 }
1194
1195 /*
1196  - nch - number of characters in a set
1197  == static int nch(register struct parse *p, register cset *cs);
1198  */
1199 static int
1200 nch(struct parse *p, cset *cs)
1201 {
1202         register int i;
1203         register size_t css = (size_t)p->g->csetsize;
1204         register int n = 0;
1205
1206         for (i = 0; i < css; i++)
1207                 if (CHIN(cs, i))
1208                         n++;
1209         return(n);
1210 }
1211
1212 /*
1213  - mcadd - add a collating element to a cset
1214  == static void mcadd(register struct parse *p, register cset *cs, \
1215  ==     register char *cp);
1216  */
1217 static void
1218 mcadd(struct parse *p, cset *cs, char *cp)
1219 {
1220         register size_t oldend = cs->smultis;
1221
1222         cs->smultis += strlen(cp) + 1;
1223         if (cs->multis == NULL)
1224                 cs->multis = malloc(cs->smultis);
1225         else
1226                 cs->multis = realloc(cs->multis, cs->smultis);
1227         if (cs->multis == NULL) {
1228                 SETERROR(REG_ESPACE);
1229                 return;
1230         }
1231
1232         (void) strcpy(cs->multis + oldend - 1, cp);
1233         cs->multis[cs->smultis - 1] = '\0';
1234 }
1235
1236 /*
1237  - mcsub - subtract a collating element from a cset
1238  == static void mcsub(register cset *cs, register char *cp);
1239  */
1240 static void
1241 mcsub(cset *cs, char *cp)
1242 {
1243         register char *fp = mcfind(cs, cp);
1244         register size_t len = strlen(fp);
1245
1246         assert(fp != NULL);
1247         (void) memmove(fp, fp + len + 1,
1248                                 cs->smultis - (fp + len + 1 - cs->multis));
1249         cs->smultis -= len;
1250
1251         if (cs->smultis == 0) {
1252                 free(cs->multis);
1253                 cs->multis = NULL;
1254                 return;
1255         }
1256
1257         cs->multis = realloc(cs->multis, cs->smultis);
1258         assert(cs->multis != NULL);
1259 }
1260
1261 /*
1262  - mcin - is a collating element in a cset?
1263  == static int mcin(register cset *cs, register char *cp);
1264  */
1265 static int
1266 mcin(cset *cs, char *cp)
1267 {
1268         return(mcfind(cs, cp) != NULL);
1269 }
1270
1271 /*
1272  - mcfind - find a collating element in a cset
1273  == static char *mcfind(register cset *cs, register char *cp);
1274  */
1275 static char *
1276 mcfind(cset *cs, char *cp)
1277 {
1278         register char *p;
1279
1280         if (cs->multis == NULL)
1281                 return(NULL);
1282         for (p = cs->multis; *p != '\0'; p += strlen(p) + 1)
1283                 if (strcmp(cp, p) == 0)
1284                         return(p);
1285         return(NULL);
1286 }
1287
1288 /*
1289  - mcinvert - invert the list of collating elements in a cset
1290  == static void mcinvert(register struct parse *p, register cset *cs);
1291  *
1292  * This would have to know the set of possibilities.  Implementation
1293  * is deferred.
1294  */
1295 static void
1296 mcinvert(struct parse *p, cset *cs)
1297 {
1298         assert(cs->multis == NULL);     /* xxx */
1299 }
1300
1301 /*
1302  - mccase - add case counterparts of the list of collating elements in a cset
1303  == static void mccase(register struct parse *p, register cset *cs);
1304  *
1305  * This would have to know the set of possibilities.  Implementation
1306  * is deferred.
1307  */
1308 static void
1309 mccase(struct parse *p, cset *cs)
1310 {
1311         assert(cs->multis == NULL);     /* xxx */
1312 }
1313
1314 /*
1315  - isinsets - is this character in any sets?
1316  == static int isinsets(register struct re_guts *g, int c);
1317  */
1318 static int                      /* predicate */
1319 isinsets(struct re_guts *g, int c)
1320 {
1321         register uch *col;
1322         register int i;
1323         register int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1324         register unsigned uc = (unsigned char)c;
1325
1326         for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1327                 if (col[uc] != 0)
1328                         return(1);
1329         return(0);
1330 }
1331
1332 /*
1333  - samesets - are these two characters in exactly the same sets?
1334  == static int samesets(register struct re_guts *g, int c1, int c2);
1335  */
1336 static int                      /* predicate */
1337 samesets(struct re_guts *g, int c1, int c2)
1338 {
1339         register uch *col;
1340         register int i;
1341         register int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT;
1342         register unsigned uc1 = (unsigned char)c1;
1343         register unsigned uc2 = (unsigned char)c2;
1344
1345         for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize)
1346                 if (col[uc1] != col[uc2])
1347                         return(0);
1348         return(1);
1349 }
1350
1351 /*
1352  - categorize - sort out character categories
1353  == static void categorize(struct parse *p, register struct re_guts *g);
1354  */
1355 static void
1356 categorize(struct parse *p, struct re_guts *g)
1357 {
1358         register cat_t *cats = g->categories;
1359         register int c;
1360         register int c2;
1361         register cat_t cat;
1362
1363         /* avoid making error situations worse */
1364         if (p->error != 0)
1365                 return;
1366
1367         for (c = CHAR_MIN; c <= CHAR_MAX; c++)
1368                 if (cats[c] == 0 && isinsets(g, c)) {
1369                         cat = g->ncategories++;
1370                         cats[c] = cat;
1371                         for (c2 = c+1; c2 <= CHAR_MAX; c2++)
1372                                 if (cats[c2] == 0 && samesets(g, c, c2))
1373                                         cats[c2] = cat;
1374                 }
1375 }
1376
1377 /*
1378  - dupl - emit a duplicate of a bunch of sops
1379  == static sopno dupl(register struct parse *p, sopno start, sopno finish);
1380  */
1381 static sopno                    /* start of duplicate */
1382 dupl(struct parse *p,
1383      sopno start,               /* from here */
1384      sopno finish)              /* to this less one */
1385 {
1386         register sopno ret = HERE();
1387         register sopno len = finish - start;
1388
1389         assert(finish >= start);
1390         if (len == 0)
1391                 return(ret);
1392         enlarge(p, p->ssize + len);     /* this many unexpected additions */
1393         assert(p->ssize >= p->slen + len);
1394         (void) memcpy((char *)(p->strip + p->slen),
1395                 (char *)(p->strip + start), (size_t)len*sizeof(sop));
1396         p->slen += len;
1397         return(ret);
1398 }
1399
1400 /*
1401  - doemit - emit a strip operator
1402  == static void doemit(register struct parse *p, sop op, size_t opnd);
1403  *
1404  * It might seem better to implement this as a macro with a function as
1405  * hard-case backup, but it's just too big and messy unless there are
1406  * some changes to the data structures.  Maybe later.
1407  */
1408 static void
1409 doemit(struct parse *p, sop op, size_t opnd)
1410 {
1411         /* avoid making error situations worse */
1412         if (p->error != 0)
1413                 return;
1414
1415         /* deal with oversize operands ("can't happen", more or less) */
1416         assert(opnd < 1<<OPSHIFT);
1417
1418         /* deal with undersized strip */
1419         if (p->slen >= p->ssize)
1420                 enlarge(p, (p->ssize+1) / 2 * 3);       /* +50% */
1421         assert(p->slen < p->ssize);
1422
1423         /* finally, it's all reduced to the easy case */
1424         p->strip[p->slen++] = SOP(op, opnd);
1425 }
1426
1427 /*
1428  - doinsert - insert a sop into the strip
1429  == static void doinsert(register struct parse *p, sop op, size_t opnd, sopno pos);
1430  */
1431 static void
1432 doinsert(struct parse *p, sop op, size_t opnd, sopno pos)
1433 {
1434         register sopno sn;
1435         register sop s;
1436         register int i;
1437
1438         /* avoid making error situations worse */
1439         if (p->error != 0)
1440                 return;
1441
1442         sn = HERE();
1443         EMIT(op, opnd);         /* do checks, ensure space */
1444         assert(HERE() == sn+1);
1445         s = p->strip[sn];
1446
1447         /* adjust paren pointers */
1448         assert(pos > 0);
1449         for (i = 1; i < NPAREN; i++) {
1450                 if (p->pbegin[i] >= pos) {
1451                         p->pbegin[i]++;
1452                 }
1453                 if (p->pend[i] >= pos) {
1454                         p->pend[i]++;
1455                 }
1456         }
1457
1458         memmove((char *)&p->strip[pos+1], (char *)&p->strip[pos],
1459                                                 (HERE()-pos-1)*sizeof(sop));
1460         p->strip[pos] = s;
1461 }
1462
1463 /*
1464  - dofwd - complete a forward reference
1465  == static void dofwd(register struct parse *p, sopno pos, sop value);
1466  */
1467 static void
1468 dofwd(struct parse *p, sopno pos, sop value)
1469 {
1470         /* avoid making error situations worse */
1471         if (p->error != 0)
1472                 return;
1473
1474         assert(value < 1<<OPSHIFT);
1475         p->strip[pos] = OP(p->strip[pos]) | value;
1476 }
1477
1478 /*
1479  - enlarge - enlarge the strip
1480  == static void enlarge(register struct parse *p, sopno size);
1481  */
1482 static void
1483 enlarge(struct parse *p, sopno size)
1484 {
1485         register sop *sp;
1486
1487         if (p->ssize >= size)
1488                 return;
1489
1490         sp = (sop *)realloc(p->strip, size*sizeof(sop));
1491         if (sp == NULL) {
1492                 SETERROR(REG_ESPACE);
1493                 return;
1494         }
1495         p->strip = sp;
1496         p->ssize = size;
1497 }
1498
1499 /*
1500  - stripsnug - compact the strip
1501  == static void stripsnug(register struct parse *p, register struct re_guts *g);
1502  */
1503 static void
1504 stripsnug(struct parse *p, struct re_guts *g)
1505 {
1506         g->nstates = p->slen;
1507         g->strip = (sop *)realloc((char *)p->strip, p->slen * sizeof(sop));
1508         if (g->strip == NULL) {
1509                 SETERROR(REG_ESPACE);
1510                 g->strip = p->strip;
1511         }
1512 }
1513
1514 /*
1515  - findmust - fill in must and mlen with longest mandatory literal string
1516  == static void findmust(register struct parse *p, register struct re_guts *g);
1517  *
1518  * This algorithm could do fancy things like analyzing the operands of |
1519  * for common subsequences.  Someday.  This code is simple and finds most
1520  * of the interesting cases.
1521  *
1522  * Note that must and mlen got initialized during setup.
1523  */
1524 static void
1525 findmust(struct parse *p, struct re_guts *g)
1526 {
1527         register sop *scan;
1528         sop *start;
1529         register sop *newstart;
1530         register sopno newlen;
1531         register sop s;
1532         register char *cp;
1533         register sopno i;
1534
1535         /* avoid making error situations worse */
1536         if (p->error != 0)
1537                 return;
1538
1539         /* find the longest OCHAR sequence in strip */
1540         newlen = 0;
1541         scan = g->strip + 1;
1542         do {
1543                 s = *scan++;
1544                 switch (OP(s)) {
1545                 case OCHAR:             /* sequence member */
1546                         if (newlen == 0)                /* new sequence */
1547                                 newstart = scan - 1;
1548                         newlen++;
1549                         break;
1550                 case OPLUS_:            /* things that don't break one */
1551                 case OLPAREN:
1552                 case ORPAREN:
1553                         break;
1554                 case OQUEST_:           /* things that must be skipped */
1555                 case OCH_:
1556                         scan--;
1557                         do {
1558                                 scan += OPND(s);
1559                                 s = *scan;
1560                                 /* assert() interferes w debug printouts */
1561                                 if (OP(s) != O_QUEST && OP(s) != O_CH &&
1562                                                         OP(s) != OOR2) {
1563                                         g->iflags |= BAD;
1564                                         return;
1565                                 }
1566                         } while (OP(s) != O_QUEST && OP(s) != O_CH);
1567                         /* fallthrough */
1568                 default:                /* things that break a sequence */
1569                         if (newlen > g->mlen) {         /* ends one */
1570                                 start = newstart;
1571                                 g->mlen = newlen;
1572                         }
1573                         newlen = 0;
1574                         break;
1575                 }
1576         } while (OP(s) != OEND);
1577
1578         if (g->mlen == 0)               /* there isn't one */
1579                 return;
1580
1581         /* turn it into a character string */
1582         g->must = malloc((size_t)g->mlen + 1);
1583         if (g->must == NULL) {          /* argh; just forget it */
1584                 g->mlen = 0;
1585                 return;
1586         }
1587         cp = g->must;
1588         scan = start;
1589         for (i = g->mlen; i > 0; i--) {
1590                 while (OP(s = *scan++) != OCHAR)
1591                         continue;
1592                 assert(cp < g->must + g->mlen);
1593                 *cp++ = (char)OPND(s);
1594         }
1595         assert(cp == g->must + g->mlen);
1596         *cp++ = '\0';           /* just on general principles */
1597 }
1598
1599 /*
1600  - pluscount - count + nesting
1601  == static sopno pluscount(register struct parse *p, register struct re_guts *g);
1602  */
1603 static sopno                    /* nesting depth */
1604 pluscount(struct parse *p, struct re_guts *g)
1605 {
1606         register sop *scan;
1607         register sop s;
1608         register sopno plusnest = 0;
1609         register sopno maxnest = 0;
1610
1611         if (p->error != 0)
1612                 return(0);      /* there may not be an OEND */
1613
1614         scan = g->strip + 1;
1615         do {
1616                 s = *scan++;
1617                 switch (OP(s)) {
1618                 case OPLUS_:
1619                         plusnest++;
1620                         break;
1621                 case O_PLUS:
1622                         if (plusnest > maxnest)
1623                                 maxnest = plusnest;
1624                         plusnest--;
1625                         break;
1626                 }
1627         } while (OP(s) != OEND);
1628         if (plusnest != 0)
1629                 g->iflags |= BAD;
1630         return(maxnest);
1631 }