1 /* $OpenBSD: engine.c,v 1.9 2004/04/02 18:34:33 otto 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. 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 * @(#)engine.c 8.5 (Berkeley) 3/20/94 36 */ 37 38 #if defined(SNAMES) && defined(LIBC_SCCS) && !defined(lint) 39 static char enginercsid[] = "$OpenBSD: engine.c,v 1.9 2004/04/02 18:34:33 otto Exp $"; 40 #endif /* SNAMES and LIBC_SCCS and not lint */ 41 42 /* 43 * The matching engine and friends. This file is #included by regexec.c 44 * after suitable #defines of a variety of macros used herein, so that 45 * different state representations can be used without duplicating masses 46 * of code. 47 */ 48 49 #ifdef SNAMES 50 #define matcher smatcher 51 #define fast sfast 52 #define slow sslow 53 #define dissect sdissect 54 #define backref sbackref 55 #define step sstep 56 #define print sprint 57 #define at sat 58 #define match smat 59 #define nope snope 60 #endif 61 #ifdef LNAMES 62 #define matcher lmatcher 63 #define fast lfast 64 #define slow lslow 65 #define dissect ldissect 66 #define backref lbackref 67 #define step lstep 68 #define print lprint 69 #define at lat 70 #define match lmat 71 #define nope lnope 72 #endif 73 74 /* another structure passed up and down to avoid zillions of parameters */ 75 struct match { 76 struct re_guts *g; 77 int eflags; 78 regmatch_t *pmatch; /* [nsub+1] (0 element unused) */ 79 char *offp; /* offsets work from here */ 80 char *beginp; /* start of string -- virtual NUL precedes */ 81 char *endp; /* end of string -- virtual NUL here */ 82 char *coldp; /* can be no match starting before here */ 83 char **lastpos; /* [nplus+1] */ 84 STATEVARS; 85 states st; /* current states */ 86 states fresh; /* states for a fresh start */ 87 states tmp; /* temporary */ 88 states empty; /* empty set of states */ 89 }; 90 91 /* ========= begin header generated by ./mkh ========= */ 92 #ifdef __cplusplus 93 extern "C" { 94 #endif 95 96 /* === engine.c === */ 97 static int matcher(struct re_guts *g, char *string, size_t nmatch, regmatch_t pmatch[], int eflags); 98 static char *dissect(struct match *m, char *start, char *stop, sopno startst, sopno stopst); 99 static char *backref(struct match *m, char *start, char *stop, sopno startst, sopno stopst, sopno lev); 100 static char *fast(struct match *m, char *start, char *stop, sopno startst, sopno stopst); 101 static char *slow(struct match *m, char *start, char *stop, sopno startst, sopno stopst); 102 static states step(struct re_guts *g, sopno start, sopno stop, states bef, int ch, states aft); 103 #define BOL (OUT+1) 104 #define EOL (BOL+1) 105 #define BOLEOL (BOL+2) 106 #define NOTHING (BOL+3) 107 #define BOW (BOL+4) 108 #define EOW (BOL+5) 109 #define CODEMAX (BOL+5) /* highest code used */ 110 #define NONCHAR(c) ((c) > CHAR_MAX) 111 #define NNONCHAR (CODEMAX-CHAR_MAX) 112 #ifdef REDEBUG 113 static void print(struct match *m, char *caption, states st, int ch, FILE *d); 114 #endif 115 #ifdef REDEBUG 116 static void at(struct match *m, char *title, char *start, char *stop, sopno startst, sopno stopst); 117 #endif 118 #ifdef REDEBUG 119 static char *pchar(int ch); 120 #endif 121 122 #ifdef __cplusplus 123 } 124 #endif 125 /* ========= end header generated by ./mkh ========= */ 126 127 #ifdef REDEBUG 128 #define SP(t, s, c) print(m, t, s, c, stdout) 129 #define AT(t, p1, p2, s1, s2) at(m, t, p1, p2, s1, s2) 130 #define NOTE(str) { if (m->eflags®_TRACE) (void)printf("=%s\n", (str)); } 131 static int nope = 0; 132 #else 133 #define SP(t, s, c) /* nothing */ 134 #define AT(t, p1, p2, s1, s2) /* nothing */ 135 #define NOTE(s) /* nothing */ 136 #endif 137 138 /* 139 - matcher - the actual matching engine 140 == static int matcher(register struct re_guts *g, char *string, \ 141 == size_t nmatch, regmatch_t pmatch[], int eflags); 142 */ 143 static int /* 0 success, REG_NOMATCH failure */ 144 matcher(g, string, nmatch, pmatch, eflags) 145 register struct re_guts *g; 146 char *string; 147 size_t nmatch; 148 regmatch_t pmatch[]; 149 int eflags; 150 { 151 register char *endp; 152 register int i; 153 struct match mv; 154 register struct match *m = &mv; 155 register char *dp; 156 const register sopno gf = g->firststate+1; /* +1 for OEND */ 157 const register sopno gl = g->laststate; 158 char *start; 159 char *stop; 160 161 /* simplify the situation where possible */ 162 if (g->cflags®_NOSUB) 163 nmatch = 0; 164 if (eflags®_STARTEND) { 165 start = string + pmatch[0].rm_so; 166 stop = string + pmatch[0].rm_eo; 167 } else { 168 start = string; 169 stop = start + strlen(start); 170 } 171 if (stop < start) 172 return(REG_INVARG); 173 174 /* prescreening; this does wonders for this rather slow code */ 175 if (g->must != NULL) { 176 for (dp = start; dp < stop; dp++) 177 if (*dp == g->must[0] && stop - dp >= g->mlen && 178 memcmp(dp, g->must, (size_t)g->mlen) == 0) 179 break; 180 if (dp == stop) /* we didn't find g->must */ 181 return(REG_NOMATCH); 182 } 183 184 /* match struct setup */ 185 m->g = g; 186 m->eflags = eflags; 187 m->pmatch = NULL; 188 m->lastpos = NULL; 189 m->offp = string; 190 m->beginp = start; 191 m->endp = stop; 192 STATESETUP(m, 4); 193 SETUP(m->st); 194 SETUP(m->fresh); 195 SETUP(m->tmp); 196 SETUP(m->empty); 197 CLEAR(m->empty); 198 199 /* this loop does only one repetition except for backrefs */ 200 for (;;) { 201 endp = fast(m, start, stop, gf, gl); 202 if (endp == NULL) { /* a miss */ 203 STATETEARDOWN(m); 204 return(REG_NOMATCH); 205 } 206 if (nmatch == 0 && !g->backrefs) 207 break; /* no further info needed */ 208 209 /* where? */ 210 assert(m->coldp != NULL); 211 for (;;) { 212 NOTE("finding start"); 213 endp = slow(m, m->coldp, stop, gf, gl); 214 if (endp != NULL) 215 break; 216 assert(m->coldp < m->endp); 217 m->coldp++; 218 } 219 if (nmatch == 1 && !g->backrefs) 220 break; /* no further info needed */ 221 222 /* oh my, he wants the subexpressions... */ 223 if (m->pmatch == NULL) 224 m->pmatch = (regmatch_t *)malloc((m->g->nsub + 1) * 225 sizeof(regmatch_t)); 226 if (m->pmatch == NULL) { 227 STATETEARDOWN(m); 228 return(REG_ESPACE); 229 } 230 for (i = 1; i <= m->g->nsub; i++) 231 m->pmatch[i].rm_so = m->pmatch[i].rm_eo = -1; 232 if (!g->backrefs && !(m->eflags®_BACKR)) { 233 NOTE("dissecting"); 234 dp = dissect(m, m->coldp, endp, gf, gl); 235 } else { 236 if (g->nplus > 0 && m->lastpos == NULL) 237 m->lastpos = (char **)malloc((g->nplus+1) * 238 sizeof(char *)); 239 if (g->nplus > 0 && m->lastpos == NULL) { 240 free(m->pmatch); 241 STATETEARDOWN(m); 242 return(REG_ESPACE); 243 } 244 NOTE("backref dissect"); 245 dp = backref(m, m->coldp, endp, gf, gl, (sopno)0); 246 } 247 if (dp != NULL) 248 break; 249 250 /* uh-oh... we couldn't find a subexpression-level match */ 251 assert(g->backrefs); /* must be back references doing it */ 252 assert(g->nplus == 0 || m->lastpos != NULL); 253 for (;;) { 254 if (dp != NULL || endp <= m->coldp) 255 break; /* defeat */ 256 NOTE("backoff"); 257 endp = slow(m, m->coldp, endp-1, gf, gl); 258 if (endp == NULL) 259 break; /* defeat */ 260 /* try it on a shorter possibility */ 261 #ifndef NDEBUG 262 for (i = 1; i <= m->g->nsub; i++) { 263 assert(m->pmatch[i].rm_so == -1); 264 assert(m->pmatch[i].rm_eo == -1); 265 } 266 #endif 267 NOTE("backoff dissect"); 268 dp = backref(m, m->coldp, endp, gf, gl, (sopno)0); 269 } 270 assert(dp == NULL || dp == endp); 271 if (dp != NULL) /* found a shorter one */ 272 break; 273 274 /* despite initial appearances, there is no match here */ 275 NOTE("false alarm"); 276 start = m->coldp + 1; /* recycle starting later */ 277 assert(start <= stop); 278 } 279 280 /* fill in the details if requested */ 281 if (nmatch > 0) { 282 pmatch[0].rm_so = m->coldp - m->offp; 283 pmatch[0].rm_eo = endp - m->offp; 284 } 285 if (nmatch > 1) { 286 assert(m->pmatch != NULL); 287 for (i = 1; i < nmatch; i++) 288 if (i <= m->g->nsub) 289 pmatch[i] = m->pmatch[i]; 290 else { 291 pmatch[i].rm_so = -1; 292 pmatch[i].rm_eo = -1; 293 } 294 } 295 296 if (m->pmatch != NULL) 297 free((char *)m->pmatch); 298 if (m->lastpos != NULL) 299 free((char *)m->lastpos); 300 STATETEARDOWN(m); 301 return(0); 302 } 303 304 /* 305 - dissect - figure out what matched what, no back references 306 == static char *dissect(register struct match *m, char *start, \ 307 == char *stop, sopno startst, sopno stopst); 308 */ 309 static char * /* == stop (success) always */ 310 dissect(m, start, stop, startst, stopst) 311 register struct match *m; 312 char *start; 313 char *stop; 314 sopno startst; 315 sopno stopst; 316 { 317 register int i; 318 register sopno ss; /* start sop of current subRE */ 319 register sopno es; /* end sop of current subRE */ 320 register char *sp; /* start of string matched by it */ 321 register char *stp; /* string matched by it cannot pass here */ 322 register char *rest; /* start of rest of string */ 323 register char *tail; /* string unmatched by rest of RE */ 324 register sopno ssub; /* start sop of subsubRE */ 325 register sopno esub; /* end sop of subsubRE */ 326 register char *ssp; /* start of string matched by subsubRE */ 327 register char *sep; /* end of string matched by subsubRE */ 328 register char *oldssp; /* previous ssp */ 329 register char *dp; 330 331 AT("diss", start, stop, startst, stopst); 332 sp = start; 333 for (ss = startst; ss < stopst; ss = es) { 334 /* identify end of subRE */ 335 es = ss; 336 switch (OP(m->g->strip[es])) { 337 case OPLUS_: 338 case OQUEST_: 339 es += OPND(m->g->strip[es]); 340 break; 341 case OCH_: 342 while (OP(m->g->strip[es]) != O_CH) 343 es += OPND(m->g->strip[es]); 344 break; 345 } 346 es++; 347 348 /* figure out what it matched */ 349 switch (OP(m->g->strip[ss])) { 350 case OEND: 351 assert(nope); 352 break; 353 case OCHAR: 354 sp++; 355 break; 356 case OBOL: 357 case OEOL: 358 case OBOW: 359 case OEOW: 360 break; 361 case OANY: 362 case OANYOF: 363 sp++; 364 break; 365 case OBACK_: 366 case O_BACK: 367 assert(nope); 368 break; 369 /* cases where length of match is hard to find */ 370 case OQUEST_: 371 stp = stop; 372 for (;;) { 373 /* how long could this one be? */ 374 rest = slow(m, sp, stp, ss, es); 375 assert(rest != NULL); /* it did match */ 376 /* could the rest match the rest? */ 377 tail = slow(m, rest, stop, es, stopst); 378 if (tail == stop) 379 break; /* yes! */ 380 /* no -- try a shorter match for this one */ 381 stp = rest - 1; 382 assert(stp >= sp); /* it did work */ 383 } 384 ssub = ss + 1; 385 esub = es - 1; 386 /* did innards match? */ 387 if (slow(m, sp, rest, ssub, esub) != NULL) { 388 dp = dissect(m, sp, rest, ssub, esub); 389 assert(dp == rest); 390 } else /* no */ 391 assert(sp == rest); 392 sp = rest; 393 break; 394 case OPLUS_: 395 stp = stop; 396 for (;;) { 397 /* how long could this one be? */ 398 rest = slow(m, sp, stp, ss, es); 399 assert(rest != NULL); /* it did match */ 400 /* could the rest match the rest? */ 401 tail = slow(m, rest, stop, es, stopst); 402 if (tail == stop) 403 break; /* yes! */ 404 /* no -- try a shorter match for this one */ 405 stp = rest - 1; 406 assert(stp >= sp); /* it did work */ 407 } 408 ssub = ss + 1; 409 esub = es - 1; 410 ssp = sp; 411 oldssp = ssp; 412 for (;;) { /* find last match of innards */ 413 sep = slow(m, ssp, rest, ssub, esub); 414 if (sep == NULL || sep == ssp) 415 break; /* failed or matched null */ 416 oldssp = ssp; /* on to next try */ 417 ssp = sep; 418 } 419 if (sep == NULL) { 420 /* last successful match */ 421 sep = ssp; 422 ssp = oldssp; 423 } 424 assert(sep == rest); /* must exhaust substring */ 425 assert(slow(m, ssp, sep, ssub, esub) == rest); 426 dp = dissect(m, ssp, sep, ssub, esub); 427 assert(dp == sep); 428 sp = rest; 429 break; 430 case OCH_: 431 stp = stop; 432 for (;;) { 433 /* how long could this one be? */ 434 rest = slow(m, sp, stp, ss, es); 435 assert(rest != NULL); /* it did match */ 436 /* could the rest match the rest? */ 437 tail = slow(m, rest, stop, es, stopst); 438 if (tail == stop) 439 break; /* yes! */ 440 /* no -- try a shorter match for this one */ 441 stp = rest - 1; 442 assert(stp >= sp); /* it did work */ 443 } 444 ssub = ss + 1; 445 esub = ss + OPND(m->g->strip[ss]) - 1; 446 assert(OP(m->g->strip[esub]) == OOR1); 447 for (;;) { /* find first matching branch */ 448 if (slow(m, sp, rest, ssub, esub) == rest) 449 break; /* it matched all of it */ 450 /* that one missed, try next one */ 451 assert(OP(m->g->strip[esub]) == OOR1); 452 esub++; 453 assert(OP(m->g->strip[esub]) == OOR2); 454 ssub = esub + 1; 455 esub += OPND(m->g->strip[esub]); 456 if (OP(m->g->strip[esub]) == OOR2) 457 esub--; 458 else 459 assert(OP(m->g->strip[esub]) == O_CH); 460 } 461 dp = dissect(m, sp, rest, ssub, esub); 462 assert(dp == rest); 463 sp = rest; 464 break; 465 case O_PLUS: 466 case O_QUEST: 467 case OOR1: 468 case OOR2: 469 case O_CH: 470 assert(nope); 471 break; 472 case OLPAREN: 473 i = OPND(m->g->strip[ss]); 474 assert(0 < i && i <= m->g->nsub); 475 m->pmatch[i].rm_so = sp - m->offp; 476 break; 477 case ORPAREN: 478 i = OPND(m->g->strip[ss]); 479 assert(0 < i && i <= m->g->nsub); 480 m->pmatch[i].rm_eo = sp - m->offp; 481 break; 482 default: /* uh oh */ 483 assert(nope); 484 break; 485 } 486 } 487 488 assert(sp == stop); 489 return(sp); 490 } 491 492 /* 493 - backref - figure out what matched what, figuring in back references 494 == static char *backref(register struct match *m, char *start, \ 495 == char *stop, sopno startst, sopno stopst, sopno lev); 496 */ 497 static char * /* == stop (success) or NULL (failure) */ 498 backref(m, start, stop, startst, stopst, lev) 499 register struct match *m; 500 char *start; 501 char *stop; 502 sopno startst; 503 sopno stopst; 504 sopno lev; /* PLUS nesting level */ 505 { 506 register int i; 507 register sopno ss; /* start sop of current subRE */ 508 register char *sp; /* start of string matched by it */ 509 register sopno ssub; /* start sop of subsubRE */ 510 register sopno esub; /* end sop of subsubRE */ 511 register char *ssp; /* start of string matched by subsubRE */ 512 register char *dp; 513 register size_t len; 514 register int hard; 515 register sop s; 516 register regoff_t offsave; 517 register cset *cs; 518 519 AT("back", start, stop, startst, stopst); 520 sp = start; 521 522 /* get as far as we can with easy stuff */ 523 hard = 0; 524 for (ss = startst; !hard && ss < stopst; ss++) 525 switch (OP(s = m->g->strip[ss])) { 526 case OCHAR: 527 if (sp == stop || *sp++ != (char)OPND(s)) 528 return(NULL); 529 break; 530 case OANY: 531 if (sp == stop) 532 return(NULL); 533 sp++; 534 break; 535 case OANYOF: 536 cs = &m->g->sets[OPND(s)]; 537 if (sp == stop || !CHIN(cs, *sp++)) 538 return(NULL); 539 break; 540 case OBOL: 541 if ( (sp == m->beginp && !(m->eflags®_NOTBOL)) || 542 (sp < m->endp && *(sp-1) == '\n' && 543 (m->g->cflags®_NEWLINE)) ) 544 { /* yes */ } 545 else 546 return(NULL); 547 break; 548 case OEOL: 549 if ( (sp == m->endp && !(m->eflags®_NOTEOL)) || 550 (sp < m->endp && *sp == '\n' && 551 (m->g->cflags®_NEWLINE)) ) 552 { /* yes */ } 553 else 554 return(NULL); 555 break; 556 case OBOW: 557 if (( (sp == m->beginp && !(m->eflags®_NOTBOL)) || 558 (sp < m->endp && *(sp-1) == '\n' && 559 (m->g->cflags®_NEWLINE)) || 560 (sp > m->beginp && 561 !ISWORD(*(sp-1))) ) && 562 (sp < m->endp && ISWORD(*sp)) ) 563 { /* yes */ } 564 else 565 return(NULL); 566 break; 567 case OEOW: 568 if (( (sp == m->endp && !(m->eflags®_NOTEOL)) || 569 (sp < m->endp && *sp == '\n' && 570 (m->g->cflags®_NEWLINE)) || 571 (sp < m->endp && !ISWORD(*sp)) ) && 572 (sp > m->beginp && ISWORD(*(sp-1))) ) 573 { /* yes */ } 574 else 575 return(NULL); 576 break; 577 case O_QUEST: 578 break; 579 case OOR1: /* matches null but needs to skip */ 580 ss++; 581 s = m->g->strip[ss]; 582 do { 583 assert(OP(s) == OOR2); 584 ss += OPND(s); 585 } while (OP(s = m->g->strip[ss]) != O_CH); 586 /* note that the ss++ gets us past the O_CH */ 587 break; 588 default: /* have to make a choice */ 589 hard = 1; 590 break; 591 } 592 if (!hard) { /* that was it! */ 593 if (sp != stop) 594 return(NULL); 595 return(sp); 596 } 597 ss--; /* adjust for the for's final increment */ 598 599 /* the hard stuff */ 600 AT("hard", sp, stop, ss, stopst); 601 s = m->g->strip[ss]; 602 switch (OP(s)) { 603 case OBACK_: /* the vilest depths */ 604 i = OPND(s); 605 assert(0 < i && i <= m->g->nsub); 606 if (m->pmatch[i].rm_eo == -1) 607 return(NULL); 608 assert(m->pmatch[i].rm_so != -1); 609 len = m->pmatch[i].rm_eo - m->pmatch[i].rm_so; 610 if (len == 0) 611 return(NULL); 612 assert(stop - m->beginp >= len); 613 if (sp > stop - len) 614 return(NULL); /* not enough left to match */ 615 ssp = m->offp + m->pmatch[i].rm_so; 616 if (memcmp(sp, ssp, len) != 0) 617 return(NULL); 618 while (m->g->strip[ss] != SOP(O_BACK, i)) 619 ss++; 620 return(backref(m, sp+len, stop, ss+1, stopst, lev)); 621 break; 622 case OQUEST_: /* to null or not */ 623 dp = backref(m, sp, stop, ss+1, stopst, lev); 624 if (dp != NULL) 625 return(dp); /* not */ 626 return(backref(m, sp, stop, ss+OPND(s)+1, stopst, lev)); 627 break; 628 case OPLUS_: 629 assert(m->lastpos != NULL); 630 assert(lev+1 <= m->g->nplus); 631 m->lastpos[lev+1] = sp; 632 return(backref(m, sp, stop, ss+1, stopst, lev+1)); 633 break; 634 case O_PLUS: 635 if (sp == m->lastpos[lev]) /* last pass matched null */ 636 return(backref(m, sp, stop, ss+1, stopst, lev-1)); 637 /* try another pass */ 638 m->lastpos[lev] = sp; 639 dp = backref(m, sp, stop, ss-OPND(s)+1, stopst, lev); 640 if (dp == NULL) 641 return(backref(m, sp, stop, ss+1, stopst, lev-1)); 642 else 643 return(dp); 644 break; 645 case OCH_: /* find the right one, if any */ 646 ssub = ss + 1; 647 esub = ss + OPND(s) - 1; 648 assert(OP(m->g->strip[esub]) == OOR1); 649 for (;;) { /* find first matching branch */ 650 dp = backref(m, sp, stop, ssub, esub, lev); 651 if (dp != NULL) 652 return(dp); 653 /* that one missed, try next one */ 654 if (OP(m->g->strip[esub]) == O_CH) 655 return(NULL); /* there is none */ 656 esub++; 657 assert(OP(m->g->strip[esub]) == OOR2); 658 ssub = esub + 1; 659 esub += OPND(m->g->strip[esub]); 660 if (OP(m->g->strip[esub]) == OOR2) 661 esub--; 662 else 663 assert(OP(m->g->strip[esub]) == O_CH); 664 } 665 break; 666 case OLPAREN: /* must undo assignment if rest fails */ 667 i = OPND(s); 668 assert(0 < i && i <= m->g->nsub); 669 offsave = m->pmatch[i].rm_so; 670 m->pmatch[i].rm_so = sp - m->offp; 671 dp = backref(m, sp, stop, ss+1, stopst, lev); 672 if (dp != NULL) 673 return(dp); 674 m->pmatch[i].rm_so = offsave; 675 return(NULL); 676 break; 677 case ORPAREN: /* must undo assignment if rest fails */ 678 i = OPND(s); 679 assert(0 < i && i <= m->g->nsub); 680 offsave = m->pmatch[i].rm_eo; 681 m->pmatch[i].rm_eo = sp - m->offp; 682 dp = backref(m, sp, stop, ss+1, stopst, lev); 683 if (dp != NULL) 684 return(dp); 685 m->pmatch[i].rm_eo = offsave; 686 return(NULL); 687 break; 688 default: /* uh oh */ 689 assert(nope); 690 break; 691 } 692 693 /* "can't happen" */ 694 assert(nope); 695 /* NOTREACHED */ 696 } 697 698 /* 699 - fast - step through the string at top speed 700 == static char *fast(register struct match *m, char *start, \ 701 == char *stop, sopno startst, sopno stopst); 702 */ 703 static char * /* where tentative match ended, or NULL */ 704 fast(m, start, stop, startst, stopst) 705 register struct match *m; 706 char *start; 707 char *stop; 708 sopno startst; 709 sopno stopst; 710 { 711 register states st = m->st; 712 register states fresh = m->fresh; 713 register states tmp = m->tmp; 714 register char *p = start; 715 register int c = (start == m->beginp) ? OUT : *(start-1); 716 register int lastc; /* previous c */ 717 register int flagch; 718 register int i; 719 register char *coldp; /* last p after which no match was underway */ 720 721 CLEAR(st); 722 SET1(st, startst); 723 st = step(m->g, startst, stopst, st, NOTHING, st); 724 ASSIGN(fresh, st); 725 SP("start", st, *p); 726 coldp = NULL; 727 for (;;) { 728 /* next character */ 729 lastc = c; 730 c = (p == m->endp) ? OUT : *p; 731 if (EQ(st, fresh)) 732 coldp = p; 733 734 /* is there an EOL and/or BOL between lastc and c? */ 735 flagch = '\0'; 736 i = 0; 737 if ( (lastc == '\n' && m->g->cflags®_NEWLINE) || 738 (lastc == OUT && !(m->eflags®_NOTBOL)) ) { 739 flagch = BOL; 740 i = m->g->nbol; 741 } 742 if ( (c == '\n' && m->g->cflags®_NEWLINE) || 743 (c == OUT && !(m->eflags®_NOTEOL)) ) { 744 flagch = (flagch == BOL) ? BOLEOL : EOL; 745 i += m->g->neol; 746 } 747 if (i != 0) { 748 for (; i > 0; i--) 749 st = step(m->g, startst, stopst, st, flagch, st); 750 SP("boleol", st, c); 751 } 752 753 /* how about a word boundary? */ 754 if ( (flagch == BOL || (lastc != OUT && !ISWORD(lastc))) && 755 (c != OUT && ISWORD(c)) ) { 756 flagch = BOW; 757 } 758 if ( (lastc != OUT && ISWORD(lastc)) && 759 (flagch == EOL || (c != OUT && !ISWORD(c))) ) { 760 flagch = EOW; 761 } 762 if (flagch == BOW || flagch == EOW) { 763 st = step(m->g, startst, stopst, st, flagch, st); 764 SP("boweow", st, c); 765 } 766 767 /* are we done? */ 768 if (ISSET(st, stopst) || p == stop) 769 break; /* NOTE BREAK OUT */ 770 771 /* no, we must deal with this character */ 772 ASSIGN(tmp, st); 773 ASSIGN(st, fresh); 774 assert(c != OUT); 775 st = step(m->g, startst, stopst, tmp, c, st); 776 SP("aft", st, c); 777 assert(EQ(step(m->g, startst, stopst, st, NOTHING, st), st)); 778 p++; 779 } 780 781 assert(coldp != NULL); 782 m->coldp = coldp; 783 if (ISSET(st, stopst)) 784 return(p+1); 785 else 786 return(NULL); 787 } 788 789 /* 790 - slow - step through the string more deliberately 791 == static char *slow(register struct match *m, char *start, \ 792 == char *stop, sopno startst, sopno stopst); 793 */ 794 static char * /* where it ended */ 795 slow(m, start, stop, startst, stopst) 796 register struct match *m; 797 char *start; 798 char *stop; 799 sopno startst; 800 sopno stopst; 801 { 802 register states st = m->st; 803 register states empty = m->empty; 804 register states tmp = m->tmp; 805 register char *p = start; 806 register int c = (start == m->beginp) ? OUT : *(start-1); 807 register int lastc; /* previous c */ 808 register int flagch; 809 register int i; 810 register char *matchp; /* last p at which a match ended */ 811 812 AT("slow", start, stop, startst, stopst); 813 CLEAR(st); 814 SET1(st, startst); 815 SP("sstart", st, *p); 816 st = step(m->g, startst, stopst, st, NOTHING, st); 817 matchp = NULL; 818 for (;;) { 819 /* next character */ 820 lastc = c; 821 c = (p == m->endp) ? OUT : *p; 822 823 /* is there an EOL and/or BOL between lastc and c? */ 824 flagch = '\0'; 825 i = 0; 826 if ( (lastc == '\n' && m->g->cflags®_NEWLINE) || 827 (lastc == OUT && !(m->eflags®_NOTBOL)) ) { 828 flagch = BOL; 829 i = m->g->nbol; 830 } 831 if ( (c == '\n' && m->g->cflags®_NEWLINE) || 832 (c == OUT && !(m->eflags®_NOTEOL)) ) { 833 flagch = (flagch == BOL) ? BOLEOL : EOL; 834 i += m->g->neol; 835 } 836 if (i != 0) { 837 for (; i > 0; i--) 838 st = step(m->g, startst, stopst, st, flagch, st); 839 SP("sboleol", st, c); 840 } 841 842 /* how about a word boundary? */ 843 if ( (flagch == BOL || (lastc != OUT && !ISWORD(lastc))) && 844 (c != OUT && ISWORD(c)) ) { 845 flagch = BOW; 846 } 847 if ( (lastc != OUT && ISWORD(lastc)) && 848 (flagch == EOL || (c != OUT && !ISWORD(c))) ) { 849 flagch = EOW; 850 } 851 if (flagch == BOW || flagch == EOW) { 852 st = step(m->g, startst, stopst, st, flagch, st); 853 SP("sboweow", st, c); 854 } 855 856 /* are we done? */ 857 if (ISSET(st, stopst)) 858 matchp = p; 859 if (EQ(st, empty) || p == stop) 860 break; /* NOTE BREAK OUT */ 861 862 /* no, we must deal with this character */ 863 ASSIGN(tmp, st); 864 ASSIGN(st, empty); 865 assert(c != OUT); 866 st = step(m->g, startst, stopst, tmp, c, st); 867 SP("saft", st, c); 868 assert(EQ(step(m->g, startst, stopst, st, NOTHING, st), st)); 869 p++; 870 } 871 872 return(matchp); 873 } 874 875 876 /* 877 - step - map set of states reachable before char to set reachable after 878 == static states step(register struct re_guts *g, sopno start, sopno stop, \ 879 == register states bef, int ch, register states aft); 880 == #define BOL (OUT+1) 881 == #define EOL (BOL+1) 882 == #define BOLEOL (BOL+2) 883 == #define NOTHING (BOL+3) 884 == #define BOW (BOL+4) 885 == #define EOW (BOL+5) 886 == #define CODEMAX (BOL+5) // highest code used 887 == #define NONCHAR(c) ((c) > CHAR_MAX) 888 == #define NNONCHAR (CODEMAX-CHAR_MAX) 889 */ 890 static states 891 step(g, start, stop, bef, ch, aft) 892 register struct re_guts *g; 893 sopno start; /* start state within strip */ 894 sopno stop; /* state after stop state within strip */ 895 register states bef; /* states reachable before */ 896 int ch; /* character or NONCHAR code */ 897 register states aft; /* states already known reachable after */ 898 { 899 register cset *cs; 900 register sop s; 901 register sopno pc; 902 register onestate here; /* note, macros know this name */ 903 register sopno look; 904 register int i; 905 906 for (pc = start, INIT(here, pc); pc != stop; pc++, INC(here)) { 907 s = g->strip[pc]; 908 switch (OP(s)) { 909 case OEND: 910 assert(pc == stop-1); 911 break; 912 case OCHAR: 913 /* only characters can match */ 914 assert(!NONCHAR(ch) || ch != (char)OPND(s)); 915 if (ch == (char)OPND(s)) 916 FWD(aft, bef, 1); 917 break; 918 case OBOL: 919 if (ch == BOL || ch == BOLEOL) 920 FWD(aft, bef, 1); 921 break; 922 case OEOL: 923 if (ch == EOL || ch == BOLEOL) 924 FWD(aft, bef, 1); 925 break; 926 case OBOW: 927 if (ch == BOW) 928 FWD(aft, bef, 1); 929 break; 930 case OEOW: 931 if (ch == EOW) 932 FWD(aft, bef, 1); 933 break; 934 case OANY: 935 if (!NONCHAR(ch)) 936 FWD(aft, bef, 1); 937 break; 938 case OANYOF: 939 cs = &g->sets[OPND(s)]; 940 if (!NONCHAR(ch) && CHIN(cs, ch)) 941 FWD(aft, bef, 1); 942 break; 943 case OBACK_: /* ignored here */ 944 case O_BACK: 945 FWD(aft, aft, 1); 946 break; 947 case OPLUS_: /* forward, this is just an empty */ 948 FWD(aft, aft, 1); 949 break; 950 case O_PLUS: /* both forward and back */ 951 FWD(aft, aft, 1); 952 i = ISSETBACK(aft, OPND(s)); 953 BACK(aft, aft, OPND(s)); 954 if (!i && ISSETBACK(aft, OPND(s))) { 955 /* oho, must reconsider loop body */ 956 pc -= OPND(s) + 1; 957 INIT(here, pc); 958 } 959 break; 960 case OQUEST_: /* two branches, both forward */ 961 FWD(aft, aft, 1); 962 FWD(aft, aft, OPND(s)); 963 break; 964 case O_QUEST: /* just an empty */ 965 FWD(aft, aft, 1); 966 break; 967 case OLPAREN: /* not significant here */ 968 case ORPAREN: 969 FWD(aft, aft, 1); 970 break; 971 case OCH_: /* mark the first two branches */ 972 FWD(aft, aft, 1); 973 assert(OP(g->strip[pc+OPND(s)]) == OOR2); 974 FWD(aft, aft, OPND(s)); 975 break; 976 case OOR1: /* done a branch, find the O_CH */ 977 if (ISSTATEIN(aft, here)) { 978 for (look = 1; 979 OP(s = g->strip[pc+look]) != O_CH; 980 look += OPND(s)) 981 assert(OP(s) == OOR2); 982 FWD(aft, aft, look); 983 } 984 break; 985 case OOR2: /* propagate OCH_'s marking */ 986 FWD(aft, aft, 1); 987 if (OP(g->strip[pc+OPND(s)]) != O_CH) { 988 assert(OP(g->strip[pc+OPND(s)]) == OOR2); 989 FWD(aft, aft, OPND(s)); 990 } 991 break; 992 case O_CH: /* just empty */ 993 FWD(aft, aft, 1); 994 break; 995 default: /* ooooops... */ 996 assert(nope); 997 break; 998 } 999 } 1000 1001 return(aft); 1002 } 1003 1004 #ifdef REDEBUG 1005 /* 1006 - print - print a set of states 1007 == #ifdef REDEBUG 1008 == static void print(struct match *m, char *caption, states st, \ 1009 == int ch, FILE *d); 1010 == #endif 1011 */ 1012 static void 1013 print(m, caption, st, ch, d) 1014 struct match *m; 1015 char *caption; 1016 states st; 1017 int ch; 1018 FILE *d; 1019 { 1020 register struct re_guts *g = m->g; 1021 register int i; 1022 register int first = 1; 1023 1024 if (!(m->eflags®_TRACE)) 1025 return; 1026 1027 (void)fprintf(d, "%s", caption); 1028 if (ch != '\0') 1029 (void)fprintf(d, " %s", pchar(ch)); 1030 for (i = 0; i < g->nstates; i++) 1031 if (ISSET(st, i)) { 1032 (void)fprintf(d, "%s%d", (first) ? "\t" : ", ", i); 1033 first = 0; 1034 } 1035 (void)fprintf(d, "\n"); 1036 } 1037 1038 /* 1039 - at - print current situation 1040 == #ifdef REDEBUG 1041 == static void at(struct match *m, char *title, char *start, char *stop, \ 1042 == sopno startst, sopno stopst); 1043 == #endif 1044 */ 1045 static void 1046 at(m, title, start, stop, startst, stopst) 1047 struct match *m; 1048 char *title; 1049 char *start; 1050 char *stop; 1051 sopno startst; 1052 sopno stopst; 1053 { 1054 if (!(m->eflags®_TRACE)) 1055 return; 1056 1057 (void)printf("%s %s-", title, pchar(*start)); 1058 (void)printf("%s ", pchar(*stop)); 1059 (void)printf("%ld-%ld\n", (long)startst, (long)stopst); 1060 } 1061 1062 #ifndef PCHARDONE 1063 #define PCHARDONE /* never again */ 1064 /* 1065 - pchar - make a character printable 1066 == #ifdef REDEBUG 1067 == static char *pchar(int ch); 1068 == #endif 1069 * 1070 * Is this identical to regchar() over in debug.c? Well, yes. But a 1071 * duplicate here avoids having a debugging-capable regexec.o tied to 1072 * a matching debug.o, and this is convenient. It all disappears in 1073 * the non-debug compilation anyway, so it doesn't matter much. 1074 */ 1075 static char * /* -> representation */ 1076 pchar(ch) 1077 int ch; 1078 { 1079 static char pbuf[10]; 1080 1081 if (isprint(ch) || ch == ' ') 1082 (void)snprintf(pbuf, sizeof pbuf, "%c", ch); 1083 else 1084 (void)snprintf(pbuf, sizeof pbuf, "\\%o", ch); 1085 return(pbuf); 1086 } 1087 #endif 1088 #endif 1089 1090 #undef matcher 1091 #undef fast 1092 #undef slow 1093 #undef dissect 1094 #undef backref 1095 #undef step 1096 #undef print 1097 #undef at 1098 #undef match 1099 #undef nope 1100