engine.c revision 1.1
1 /* 2 * The matching engine and friends. This file is #included by regexec.c 3 * after suitable #defines of a variety of macros used herein, so that 4 * different state representations can be used without duplicating masses 5 * of code. 6 */ 7 8 #ifdef SNAMES 9 #define matcher smatcher 10 #define fast sfast 11 #define slow sslow 12 #define dissect sdissect 13 #define backref sbackref 14 #define step sstep 15 #define print sprint 16 #define at sat 17 #define match smat 18 #endif 19 #ifdef LNAMES 20 #define matcher lmatcher 21 #define fast lfast 22 #define slow lslow 23 #define dissect ldissect 24 #define backref lbackref 25 #define step lstep 26 #define print lprint 27 #define at lat 28 #define match lmat 29 #endif 30 31 /* another structure passed up and down to avoid zillions of parameters */ 32 struct match { 33 struct re_guts *g; 34 int eflags; 35 regmatch_t *pmatch; /* [nsub+1] (0 element unused) */ 36 char *offp; /* offsets work from here */ 37 char *beginp; /* start of string -- virtual NUL precedes */ 38 char *endp; /* end of string -- virtual NUL here */ 39 char *coldp; /* can be no match starting before here */ 40 char **lastpos; /* [nplus+1] */ 41 STATEVARS; 42 states st; /* current states */ 43 states fresh; /* states for a fresh start */ 44 states tmp; /* temporary */ 45 states empty; /* empty set of states */ 46 }; 47 48 #include "engine.ih" 49 50 #ifdef REDEBUG 51 #define SP(t, s, c) print(m, t, s, c, stdout) 52 #define AT(t, p1, p2, s1, s2) at(m, t, p1, p2, s1, s2) 53 #define NOTE(str) { if (m->eflags®_TRACE) printf("=%s\n", (str)); } 54 #else 55 #define SP(t, s, c) /* nothing */ 56 #define AT(t, p1, p2, s1, s2) /* nothing */ 57 #define NOTE(s) /* nothing */ 58 #endif 59 60 /* 61 - matcher - the actual matching engine 62 == static int matcher(register struct re_guts *g, char *string, \ 63 == size_t nmatch, regmatch_t pmatch[], int eflags); 64 */ 65 static int /* 0 success, REG_NOMATCH failure */ 66 matcher(g, string, nmatch, pmatch, eflags) 67 register struct re_guts *g; 68 char *string; 69 size_t nmatch; 70 regmatch_t pmatch[]; 71 int eflags; 72 { 73 register char *endp; 74 register int i; 75 struct match mv; 76 register struct match *m = &mv; 77 register char *dp; 78 const register sopno gf = g->firststate+1; /* +1 for OEND */ 79 const register sopno gl = g->laststate; 80 char *start; 81 char *stop; 82 83 /* simplify the situation where possible */ 84 if (g->cflags®_NOSUB) 85 nmatch = 0; 86 if (eflags®_STARTEND) { 87 start = string + pmatch[0].rm_so; 88 stop = string + pmatch[0].rm_eo; 89 } else { 90 start = string; 91 stop = start + strlen(start); 92 } 93 if (stop < start) 94 return(REG_INVARG); 95 96 /* prescreening; this does wonders for this rather slow code */ 97 if (g->must != NULL) { 98 for (dp = start; dp < stop; dp++) 99 if (*dp == g->must[0] && stop - dp >= g->mlen && 100 memcmp(dp, g->must, (size_t)g->mlen) == 0) 101 break; 102 if (dp == stop) /* we didn't find g->must */ 103 return(REG_NOMATCH); 104 } 105 106 /* match struct setup */ 107 m->g = g; 108 m->eflags = eflags; 109 m->pmatch = NULL; 110 m->lastpos = NULL; 111 m->offp = string; 112 m->beginp = start; 113 m->endp = stop; 114 STATESETUP(m, 4); 115 SETUP(m->st); 116 SETUP(m->fresh); 117 SETUP(m->tmp); 118 SETUP(m->empty); 119 CLEAR(m->empty); 120 121 /* this loop does only one repetition except for backrefs */ 122 for (;;) { 123 endp = fast(m, start, stop, gf, gl); 124 if (endp == NULL) { /* a miss */ 125 STATETEARDOWN(m); 126 return(REG_NOMATCH); 127 } 128 if (nmatch == 0 && !g->backrefs) 129 break; /* no further info needed */ 130 131 /* where? */ 132 assert(m->coldp != NULL); 133 for (;;) { 134 NOTE("finding start"); 135 endp = slow(m, m->coldp, stop, gf, gl); 136 if (endp != NULL) 137 break; 138 assert(m->coldp < m->endp); 139 m->coldp++; 140 } 141 if (nmatch == 1 && !g->backrefs) 142 break; /* no further info needed */ 143 144 /* oh my, he wants the subexpressions... */ 145 if (m->pmatch == NULL) 146 m->pmatch = (regmatch_t *)malloc((m->g->nsub + 1) * 147 sizeof(regmatch_t)); 148 if (m->pmatch == NULL) { 149 STATETEARDOWN(m); 150 return(REG_ESPACE); 151 } 152 for (i = 1; i <= m->g->nsub; i++) 153 m->pmatch[i].rm_so = m->pmatch[i].rm_eo = -1; 154 if (!g->backrefs && !(m->eflags®_BACKR)) { 155 NOTE("dissecting"); 156 dp = dissect(m, m->coldp, endp, gf, gl); 157 } else { 158 if (g->nplus > 0 && m->lastpos == NULL) 159 m->lastpos = (char **)malloc((g->nplus+1) * 160 sizeof(char *)); 161 if (g->nplus > 0 && m->lastpos == NULL) { 162 free(m->pmatch); 163 STATETEARDOWN(m); 164 return(REG_ESPACE); 165 } 166 NOTE("backref dissect"); 167 dp = backref(m, m->coldp, endp, gf, gl, (sopno)0); 168 } 169 if (dp != NULL) 170 break; 171 172 /* uh-oh... we couldn't find a subexpression-level match */ 173 assert(g->backrefs); /* must be back references doing it */ 174 assert(g->nplus == 0 || m->lastpos != NULL); 175 for (;;) { 176 if (dp != NULL || endp <= m->coldp) 177 break; /* defeat */ 178 NOTE("backoff"); 179 endp = slow(m, m->coldp, endp-1, gf, gl); 180 if (endp == NULL) 181 break; /* defeat */ 182 /* try it on a shorter possibility */ 183 #ifndef NDEBUG 184 for (i = 1; i <= m->g->nsub; i++) { 185 assert(m->pmatch[i].rm_so == -1); 186 assert(m->pmatch[i].rm_eo == -1); 187 } 188 #endif 189 NOTE("backoff dissect"); 190 dp = backref(m, m->coldp, endp, gf, gl, (sopno)0); 191 } 192 assert(dp == NULL || dp == endp); 193 if (dp != NULL) /* found a shorter one */ 194 break; 195 196 /* despite initial appearances, there is no match here */ 197 NOTE("false alarm"); 198 start = m->coldp + 1; /* recycle starting later */ 199 assert(start <= stop); 200 } 201 202 /* fill in the details if requested */ 203 if (nmatch > 0) { 204 pmatch[0].rm_so = m->coldp - m->offp; 205 pmatch[0].rm_eo = endp - m->offp; 206 } 207 if (nmatch > 1) { 208 assert(m->pmatch != NULL); 209 for (i = 1; i < nmatch; i++) 210 if (i <= m->g->nsub) 211 pmatch[i] = m->pmatch[i]; 212 else { 213 pmatch[i].rm_so = -1; 214 pmatch[i].rm_eo = -1; 215 } 216 } 217 218 if (m->pmatch != NULL) 219 free((char *)m->pmatch); 220 if (m->lastpos != NULL) 221 free((char *)m->lastpos); 222 STATETEARDOWN(m); 223 return(0); 224 } 225 226 /* 227 - dissect - figure out what matched what, no back references 228 == static char *dissect(register struct match *m, char *start, \ 229 == char *stop, sopno startst, sopno stopst); 230 */ 231 static char * /* == stop (success) always */ 232 dissect(m, start, stop, startst, stopst) 233 register struct match *m; 234 char *start; 235 char *stop; 236 sopno startst; 237 sopno stopst; 238 { 239 register int i; 240 register sopno ss; /* start sop of current subRE */ 241 register sopno es; /* end sop of current subRE */ 242 register char *sp; /* start of string matched by it */ 243 register char *stp; /* string matched by it cannot pass here */ 244 register char *rest; /* start of rest of string */ 245 register char *tail; /* string unmatched by rest of RE */ 246 register sopno ssub; /* start sop of subsubRE */ 247 register sopno esub; /* end sop of subsubRE */ 248 register char *ssp; /* start of string matched by subsubRE */ 249 register char *sep; /* end of string matched by subsubRE */ 250 register char *oldssp; /* previous ssp */ 251 register char *dp; 252 253 AT("diss", start, stop, startst, stopst); 254 sp = start; 255 for (ss = startst; ss < stopst; ss = es) { 256 /* identify end of subRE */ 257 es = ss; 258 switch (OP(m->g->strip[es])) { 259 case OPLUS_: 260 case OQUEST_: 261 es += OPND(m->g->strip[es]); 262 break; 263 case OCH_: 264 while (OP(m->g->strip[es]) != O_CH) 265 es += OPND(m->g->strip[es]); 266 break; 267 } 268 es++; 269 270 /* figure out what it matched */ 271 switch (OP(m->g->strip[ss])) { 272 case OEND: 273 assert(nope); 274 break; 275 case OCHAR: 276 sp++; 277 break; 278 case OBOL: 279 case OEOL: 280 case OBOW: 281 case OEOW: 282 break; 283 case OANY: 284 case OANYOF: 285 sp++; 286 break; 287 case OBACK_: 288 case O_BACK: 289 assert(nope); 290 break; 291 /* cases where length of match is hard to find */ 292 case OQUEST_: 293 stp = stop; 294 for (;;) { 295 /* how long could this one be? */ 296 rest = slow(m, sp, stp, ss, es); 297 assert(rest != NULL); /* it did match */ 298 /* could the rest match the rest? */ 299 tail = slow(m, rest, stop, es, stopst); 300 if (tail == stop) 301 break; /* yes! */ 302 /* no -- try a shorter match for this one */ 303 stp = rest - 1; 304 assert(stp >= sp); /* it did work */ 305 } 306 ssub = ss + 1; 307 esub = es - 1; 308 /* did innards match? */ 309 if (slow(m, sp, rest, ssub, esub) != NULL) { 310 dp = dissect(m, sp, rest, ssub, esub); 311 assert(dp == rest); 312 } else /* no */ 313 assert(sp == rest); 314 sp = rest; 315 break; 316 case OPLUS_: 317 stp = stop; 318 for (;;) { 319 /* how long could this one be? */ 320 rest = slow(m, sp, stp, ss, es); 321 assert(rest != NULL); /* it did match */ 322 /* could the rest match the rest? */ 323 tail = slow(m, rest, stop, es, stopst); 324 if (tail == stop) 325 break; /* yes! */ 326 /* no -- try a shorter match for this one */ 327 stp = rest - 1; 328 assert(stp >= sp); /* it did work */ 329 } 330 ssub = ss + 1; 331 esub = es - 1; 332 ssp = sp; 333 oldssp = ssp; 334 for (;;) { /* find last match of innards */ 335 sep = slow(m, ssp, rest, ssub, esub); 336 if (sep == NULL || sep == ssp) 337 break; /* failed or matched null */ 338 oldssp = ssp; /* on to next try */ 339 ssp = sep; 340 } 341 if (sep == NULL) { 342 /* last successful match */ 343 sep = ssp; 344 ssp = oldssp; 345 } 346 assert(sep == rest); /* must exhaust substring */ 347 assert(slow(m, ssp, sep, ssub, esub) == rest); 348 dp = dissect(m, ssp, sep, ssub, esub); 349 assert(dp == sep); 350 sp = rest; 351 break; 352 case OCH_: 353 stp = stop; 354 for (;;) { 355 /* how long could this one be? */ 356 rest = slow(m, sp, stp, ss, es); 357 assert(rest != NULL); /* it did match */ 358 /* could the rest match the rest? */ 359 tail = slow(m, rest, stop, es, stopst); 360 if (tail == stop) 361 break; /* yes! */ 362 /* no -- try a shorter match for this one */ 363 stp = rest - 1; 364 assert(stp >= sp); /* it did work */ 365 } 366 ssub = ss + 1; 367 esub = ss + OPND(m->g->strip[ss]) - 1; 368 assert(OP(m->g->strip[esub]) == OOR1); 369 for (;;) { /* find first matching branch */ 370 if (slow(m, sp, rest, ssub, esub) == rest) 371 break; /* it matched all of it */ 372 /* that one missed, try next one */ 373 assert(OP(m->g->strip[esub]) == OOR1); 374 esub++; 375 assert(OP(m->g->strip[esub]) == OOR2); 376 ssub = esub + 1; 377 esub += OPND(m->g->strip[esub]); 378 if (OP(m->g->strip[esub]) == OOR2) 379 esub--; 380 else 381 assert(OP(m->g->strip[esub]) == O_CH); 382 } 383 dp = dissect(m, sp, rest, ssub, esub); 384 assert(dp == rest); 385 sp = rest; 386 break; 387 case O_PLUS: 388 case O_QUEST: 389 case OOR1: 390 case OOR2: 391 case O_CH: 392 assert(nope); 393 break; 394 case OLPAREN: 395 i = OPND(m->g->strip[ss]); 396 m->pmatch[i].rm_so = sp - m->offp; 397 break; 398 case ORPAREN: 399 i = OPND(m->g->strip[ss]); 400 m->pmatch[i].rm_eo = sp - m->offp; 401 break; 402 default: /* uh oh */ 403 assert(nope); 404 break; 405 } 406 } 407 408 assert(sp == stop); 409 return(sp); 410 } 411 412 /* 413 - backref - figure out what matched what, figuring in back references 414 == static char *backref(register struct match *m, char *start, \ 415 == char *stop, sopno startst, sopno stopst, sopno lev); 416 */ 417 static char * /* == stop (success) or NULL (failure) */ 418 backref(m, start, stop, startst, stopst, lev) 419 register struct match *m; 420 char *start; 421 char *stop; 422 sopno startst; 423 sopno stopst; 424 sopno lev; /* PLUS nesting level */ 425 { 426 register int i; 427 register sopno ss; /* start sop of current subRE */ 428 register char *sp; /* start of string matched by it */ 429 register sopno ssub; /* start sop of subsubRE */ 430 register sopno esub; /* end sop of subsubRE */ 431 register char *ssp; /* start of string matched by subsubRE */ 432 register char *dp; 433 register size_t len; 434 register int hard; 435 register sop s; 436 register regoff_t offsave; 437 register cset *cs; 438 439 AT("back", start, stop, startst, stopst); 440 sp = start; 441 442 /* get as far as we can with easy stuff */ 443 hard = 0; 444 for (ss = startst; !hard && ss < stopst; ss++) 445 switch (OP(s = m->g->strip[ss])) { 446 case OCHAR: 447 if (sp == stop || *sp++ != (char)OPND(s)) 448 return(NULL); 449 break; 450 case OANY: 451 if (sp == stop) 452 return(NULL); 453 sp++; 454 break; 455 case OANYOF: 456 cs = &m->g->sets[OPND(s)]; 457 if (sp == stop || !CHIN(cs, *sp++)) 458 return(NULL); 459 break; 460 case OBOL: 461 if ( (sp == m->beginp && !(m->eflags®_NOTBOL)) || 462 (sp < m->endp && *(sp-1) == '\n' && 463 (m->g->cflags®_NEWLINE)) ) 464 { /* yes */ } 465 else 466 return(NULL); 467 break; 468 case OEOL: 469 if ( (sp == m->endp && !(m->eflags®_NOTEOL)) || 470 (sp < m->endp && *sp == '\n' && 471 (m->g->cflags®_NEWLINE)) ) 472 { /* yes */ } 473 else 474 return(NULL); 475 break; 476 case OBOW: 477 if (( (sp == m->beginp && !(m->eflags®_NOTBOL)) || 478 (sp < m->endp && *(sp-1) == '\n' && 479 (m->g->cflags®_NEWLINE)) || 480 (sp > m->beginp && 481 !isalnum(*(sp-1))) ) && 482 (sp < m->endp && isalnum(*sp)) ) 483 { /* yes */ } 484 else 485 return(NULL); 486 break; 487 case OEOW: 488 if (( (sp == m->endp && !(m->eflags®_NOTEOL)) || 489 (sp < m->endp && *sp == '\n' && 490 (m->g->cflags®_NEWLINE)) || 491 (sp < m->endp && !isalnum(*sp)) ) && 492 (sp > m->beginp && isalnum(*(sp-1))) ) 493 { /* yes */ } 494 else 495 return(NULL); 496 break; 497 case O_QUEST: 498 break; 499 case OOR1: /* matches null but needs to skip */ 500 ss++; 501 s = m->g->strip[ss]; 502 do { 503 assert(OP(s) == OOR2); 504 ss += OPND(s); 505 } while (OP(s = m->g->strip[ss]) != O_CH); 506 /* note that the ss++ gets us past the O_CH */ 507 break; 508 default: /* have to make a choice */ 509 hard = 1; 510 break; 511 } 512 if (!hard) { /* that was it! */ 513 if (sp != stop) 514 return(NULL); 515 return(sp); 516 } 517 ss--; /* adjust for the for's final increment */ 518 519 /* the hard stuff */ 520 AT("hard", sp, stop, ss, stopst); 521 s = m->g->strip[ss]; 522 switch (OP(s)) { 523 case OBACK_: /* the vilest depths */ 524 i = OPND(s); 525 if (m->pmatch[i].rm_eo == -1) 526 return(NULL); 527 assert(m->pmatch[i].rm_so != -1); 528 len = m->pmatch[i].rm_eo - m->pmatch[i].rm_so; 529 assert(stop - m->beginp >= len); 530 if (sp > stop - len) 531 return(NULL); /* not enough left to match */ 532 ssp = m->offp + m->pmatch[i].rm_so; 533 if (memcmp(sp, ssp, len) != 0) 534 return(NULL); 535 while (m->g->strip[ss] != SOP(O_BACK, i)) 536 ss++; 537 return(backref(m, sp+len, stop, ss+1, stopst, lev)); 538 break; 539 case OQUEST_: /* to null or not */ 540 dp = backref(m, sp, stop, ss+1, stopst, lev); 541 if (dp != NULL) 542 return(dp); /* not */ 543 return(backref(m, sp, stop, ss+OPND(s)+1, stopst, lev)); 544 break; 545 case OPLUS_: 546 assert(m->lastpos != NULL); 547 assert(lev+1 <= m->g->nplus); 548 m->lastpos[lev+1] = sp; 549 return(backref(m, sp, stop, ss+1, stopst, lev+1)); 550 break; 551 case O_PLUS: 552 if (sp == m->lastpos[lev]) /* last pass matched null */ 553 return(backref(m, sp, stop, ss+1, stopst, lev-1)); 554 /* try another pass */ 555 m->lastpos[lev] = sp; 556 dp = backref(m, sp, stop, ss-OPND(s)+1, stopst, lev); 557 if (dp == NULL) 558 return(backref(m, sp, stop, ss+1, stopst, lev-1)); 559 else 560 return(dp); 561 break; 562 case OCH_: /* find the right one, if any */ 563 ssub = ss + 1; 564 esub = ss + OPND(s) - 1; 565 assert(OP(m->g->strip[esub]) == OOR1); 566 for (;;) { /* find first matching branch */ 567 dp = backref(m, sp, stop, ssub, esub, lev); 568 if (dp != NULL) 569 return(dp); 570 /* that one missed, try next one */ 571 if (OP(m->g->strip[esub]) == O_CH) 572 return(NULL); /* there is none */ 573 esub++; 574 assert(OP(m->g->strip[esub]) == OOR2); 575 ssub = esub + 1; 576 esub += OPND(m->g->strip[esub]); 577 if (OP(m->g->strip[esub]) == OOR2) 578 esub--; 579 else 580 assert(OP(m->g->strip[esub]) == O_CH); 581 } 582 break; 583 case OLPAREN: /* must undo assignment if rest fails */ 584 i = OPND(s); 585 offsave = m->pmatch[i].rm_so; 586 m->pmatch[i].rm_so = sp - m->offp; 587 dp = backref(m, sp, stop, ss+1, stopst, lev); 588 if (dp != NULL) 589 return(dp); 590 m->pmatch[i].rm_so = offsave; 591 return(NULL); 592 break; 593 case ORPAREN: /* must undo assignment if rest fails */ 594 i = OPND(s); 595 offsave = m->pmatch[i].rm_eo; 596 m->pmatch[i].rm_eo = sp - m->offp; 597 dp = backref(m, sp, stop, ss+1, stopst, lev); 598 if (dp != NULL) 599 return(dp); 600 m->pmatch[i].rm_eo = offsave; 601 return(NULL); 602 break; 603 default: /* uh oh */ 604 assert(nope); 605 break; 606 } 607 608 /* "can't happen" */ 609 assert(nope); 610 /* NOTREACHED */ 611 } 612 613 /* 614 - fast - step through the string at top speed 615 == static char *fast(register struct match *m, char *start, \ 616 == char *stop, sopno startst, sopno stopst); 617 */ 618 static char * /* where tentative match ended, or NULL */ 619 fast(m, start, stop, startst, stopst) 620 register struct match *m; 621 char *start; 622 char *stop; 623 sopno startst; 624 sopno stopst; 625 { 626 register states st = m->st; 627 register states fresh = m->fresh; 628 register states tmp = m->tmp; 629 register char *p = start; 630 register int c = (start == m->beginp) ? OUT : *(start-1); 631 register int lastc; /* previous c */ 632 register int flagch; 633 register int i; 634 register char *coldp; /* last p after which no match was underway */ 635 636 CLEAR(st); 637 SET1(st, startst); 638 st = step(m->g, startst, stopst, st, NOTHING, st); 639 ASSIGN(fresh, st); 640 SP("start", st, *p); 641 coldp = NULL; 642 for (;;) { 643 /* next character */ 644 lastc = c; 645 c = (p == m->endp) ? OUT : *p; 646 if (EQ(st, fresh)) 647 coldp = p; 648 649 /* is there an EOL and/or BOL between lastc and c? */ 650 flagch = '\0'; 651 i = 0; 652 if ( (lastc == '\n' && m->g->cflags®_NEWLINE) || 653 (lastc == OUT && !(m->eflags®_NOTBOL)) ) { 654 flagch = BOL; 655 i = m->g->nbol; 656 } 657 if ( (c == '\n' && m->g->cflags®_NEWLINE) || 658 (c == OUT && !(m->eflags®_NOTEOL)) ) { 659 flagch = (flagch == BOL) ? BOLEOL : EOL; 660 i += m->g->neol; 661 } 662 if (i != 0) { 663 for (; i > 0; i--) 664 st = step(m->g, startst, stopst, st, flagch, st); 665 SP("boleol", st, c); 666 } 667 668 /* how about a word boundary? */ 669 if ( (flagch == BOL || (lastc != OUT && !isalnum(lastc))) && 670 (c != OUT && isalnum(c)) ) { 671 flagch = BOW; 672 } 673 if ( (lastc != OUT && isalnum(lastc)) && 674 (flagch == EOL || (c != OUT && !isalnum(c))) ) { 675 flagch = EOW; 676 } 677 if (flagch == BOW || flagch == EOW) { 678 st = step(m->g, startst, stopst, st, flagch, st); 679 SP("boweow", st, c); 680 } 681 682 /* are we done? */ 683 if (ISSET(st, stopst) || p == stop) 684 break; /* NOTE BREAK OUT */ 685 686 /* no, we must deal with this character */ 687 ASSIGN(tmp, st); 688 ASSIGN(st, fresh); 689 assert(c != OUT); 690 st = step(m->g, startst, stopst, tmp, c, st); 691 SP("aft", st, c); 692 assert(EQ(step(m->g, startst, stopst, st, NOTHING, st), st)); 693 p++; 694 } 695 696 assert(coldp != NULL); 697 m->coldp = coldp; 698 if (ISSET(st, stopst)) 699 return(p+1); 700 else 701 return(NULL); 702 } 703 704 /* 705 - slow - step through the string more deliberately 706 == static char *slow(register struct match *m, char *start, \ 707 == char *stop, sopno startst, sopno stopst); 708 */ 709 static char * /* where it ended */ 710 slow(m, start, stop, startst, stopst) 711 register struct match *m; 712 char *start; 713 char *stop; 714 sopno startst; 715 sopno stopst; 716 { 717 register states st = m->st; 718 register states empty = m->empty; 719 register states tmp = m->tmp; 720 register char *p = start; 721 register int c = (start == m->beginp) ? OUT : *(start-1); 722 register int lastc; /* previous c */ 723 register int flagch; 724 register int i; 725 register char *matchp; /* last p at which a match ended */ 726 727 AT("slow", start, stop, startst, stopst); 728 CLEAR(st); 729 SET1(st, startst); 730 SP("sstart", st, *p); 731 st = step(m->g, startst, stopst, st, NOTHING, st); 732 matchp = NULL; 733 for (;;) { 734 /* next character */ 735 lastc = c; 736 c = (p == m->endp) ? OUT : *p; 737 738 /* is there an EOL and/or BOL between lastc and c? */ 739 flagch = '\0'; 740 i = 0; 741 if ( (lastc == '\n' && m->g->cflags®_NEWLINE) || 742 (lastc == OUT && !(m->eflags®_NOTBOL)) ) { 743 flagch = BOL; 744 i = m->g->nbol; 745 } 746 if ( (c == '\n' && m->g->cflags®_NEWLINE) || 747 (c == OUT && !(m->eflags®_NOTEOL)) ) { 748 flagch = (flagch == BOL) ? BOLEOL : EOL; 749 i += m->g->neol; 750 } 751 if (i != 0) { 752 for (; i > 0; i--) 753 st = step(m->g, startst, stopst, st, flagch, st); 754 SP("sboleol", st, c); 755 } 756 757 /* how about a word boundary? */ 758 if ( (flagch == BOL || (lastc != OUT && !isalnum(lastc))) && 759 (c != OUT && isalnum(c)) ) { 760 flagch = BOW; 761 } 762 if ( (lastc != OUT && isalnum(lastc)) && 763 (flagch == EOL || (c != OUT && !isalnum(c))) ) { 764 flagch = EOW; 765 } 766 if (flagch == BOW || flagch == EOW) { 767 st = step(m->g, startst, stopst, st, flagch, st); 768 SP("sboweow", st, c); 769 } 770 771 /* are we done? */ 772 if (ISSET(st, stopst)) 773 matchp = p; 774 if (EQ(st, empty) || p == stop) 775 break; /* NOTE BREAK OUT */ 776 777 /* no, we must deal with this character */ 778 ASSIGN(tmp, st); 779 ASSIGN(st, empty); 780 assert(c != OUT); 781 st = step(m->g, startst, stopst, tmp, c, st); 782 SP("saft", st, c); 783 assert(EQ(step(m->g, startst, stopst, st, NOTHING, st), st)); 784 p++; 785 } 786 787 return(matchp); 788 } 789 790 791 /* 792 - step - map set of states reachable before char to set reachable after 793 == static states step(register struct re_guts *g, int start, int stop, \ 794 == register states bef, int ch, register states aft); 795 == #define BOL (OUT+1) 796 == #define EOL (BOL+1) 797 == #define BOLEOL (BOL+2) 798 == #define NOTHING (BOL+3) 799 == #define BOW (BOL+4) 800 == #define EOW (BOL+5) 801 == #define CODEMAX (BOL+5) // highest code used 802 == #define NONCHAR(c) ((c) > CHAR_MAX) 803 == #define NNONCHAR (CODEMAX-CHAR_MAX) 804 */ 805 static states 806 step(g, start, stop, bef, ch, aft) 807 register struct re_guts *g; 808 int start; /* start state within strip */ 809 int stop; /* state after stop state within strip */ 810 register states bef; /* states reachable before */ 811 int ch; /* character or NONCHAR code */ 812 register states aft; /* states already known reachable after */ 813 { 814 register cset *cs; 815 register sop s; 816 register sopno pc; 817 register onestate here; /* note, macros know this name */ 818 register sopno look; 819 register int i; 820 821 for (pc = start, INIT(here, pc); pc != stop; pc++, INC(here)) { 822 s = g->strip[pc]; 823 switch (OP(s)) { 824 case OEND: 825 assert(pc == stop-1); 826 break; 827 case OCHAR: 828 /* only characters can match */ 829 assert(!NONCHAR(ch) || ch != (char)OPND(s)); 830 if (ch == (char)OPND(s)) 831 FWD(aft, bef, 1); 832 break; 833 case OBOL: 834 if (ch == BOL || ch == BOLEOL) 835 FWD(aft, bef, 1); 836 break; 837 case OEOL: 838 if (ch == EOL || ch == BOLEOL) 839 FWD(aft, bef, 1); 840 break; 841 case OBOW: 842 if (ch == BOW) 843 FWD(aft, bef, 1); 844 break; 845 case OEOW: 846 if (ch == EOW) 847 FWD(aft, bef, 1); 848 break; 849 case OANY: 850 if (!NONCHAR(ch)) 851 FWD(aft, bef, 1); 852 break; 853 case OANYOF: 854 cs = &g->sets[OPND(s)]; 855 if (!NONCHAR(ch) && CHIN(cs, ch)) 856 FWD(aft, bef, 1); 857 break; 858 case OBACK_: /* ignored here */ 859 case O_BACK: 860 FWD(aft, aft, 1); 861 break; 862 case OPLUS_: /* forward, this is just an empty */ 863 FWD(aft, aft, 1); 864 break; 865 case O_PLUS: /* both forward and back */ 866 FWD(aft, aft, 1); 867 i = ISSETBACK(aft, OPND(s)); 868 BACK(aft, aft, OPND(s)); 869 if (!i && ISSETBACK(aft, OPND(s))) { 870 /* oho, must reconsider loop body */ 871 pc -= OPND(s) + 1; 872 INIT(here, pc); 873 } 874 break; 875 case OQUEST_: /* two branches, both forward */ 876 FWD(aft, aft, 1); 877 FWD(aft, aft, OPND(s)); 878 break; 879 case O_QUEST: /* just an empty */ 880 FWD(aft, aft, 1); 881 break; 882 case OLPAREN: /* not significant here */ 883 case ORPAREN: 884 FWD(aft, aft, 1); 885 break; 886 case OCH_: /* mark the first two branches */ 887 FWD(aft, aft, 1); 888 assert(OP(g->strip[pc+OPND(s)]) == OOR2); 889 FWD(aft, aft, OPND(s)); 890 break; 891 case OOR1: /* done a branch, find the O_CH */ 892 if (ISSTATEIN(aft, here)) { 893 for (look = 1; 894 OP(s = g->strip[pc+look]) != O_CH; 895 look += OPND(s)) 896 assert(OP(s) == OOR2); 897 FWD(aft, aft, look); 898 } 899 break; 900 case OOR2: /* propagate OCH_'s marking */ 901 FWD(aft, aft, 1); 902 if (OP(g->strip[pc+OPND(s)]) != O_CH) { 903 assert(OP(g->strip[pc+OPND(s)]) == OOR2); 904 FWD(aft, aft, OPND(s)); 905 } 906 break; 907 case O_CH: /* just empty */ 908 FWD(aft, aft, 1); 909 break; 910 default: /* ooooops... */ 911 assert(nope); 912 break; 913 } 914 } 915 916 return(aft); 917 } 918 919 #ifdef REDEBUG 920 /* 921 - print - print a set of states 922 == #ifdef REDEBUG 923 == static void print(struct match *m, char *caption, states st, \ 924 == int ch, FILE *d); 925 == #endif 926 */ 927 static void 928 print(m, caption, st, ch, d) 929 struct match *m; 930 char *caption; 931 states st; 932 int ch; 933 FILE *d; 934 { 935 register struct re_guts *g = m->g; 936 register int i; 937 register int first = 1; 938 939 if (!(m->eflags®_TRACE)) 940 return; 941 942 fprintf(d, "%s", caption); 943 if (ch != '\0') 944 fprintf(d, " %s", pchar(ch)); 945 for (i = 0; i < g->nstates; i++) 946 if (ISSET(st, i)) { 947 fprintf(d, "%s%d", (first) ? "\t" : ", ", i); 948 first = 0; 949 } 950 fprintf(d, "\n"); 951 } 952 953 /* 954 - at - print current situation 955 == #ifdef REDEBUG 956 == static void at(struct match *m, char *title, char *start, char *stop, \ 957 == sopno startst, sopno stopst); 958 == #endif 959 */ 960 static void 961 at(m, title, start, stop, startst, stopst) 962 struct match *m; 963 char *title; 964 char *start; 965 char *stop; 966 sopno startst; 967 sopno stopst; 968 { 969 if (!(m->eflags®_TRACE)) 970 return; 971 972 printf("%s %s-", title, pchar(*start)); 973 printf("%s ", pchar(*stop)); 974 printf("%ld-%ld\n", (long)startst, (long)stopst); 975 } 976 977 #ifndef PCHARDONE 978 #define PCHARDONE /* never again */ 979 /* 980 - pchar - make a character printable 981 == #ifdef REDEBUG 982 == static char *pchar(int ch); 983 == #endif 984 * 985 * Is this identical to regchar() over in debug.c? Well, yes. But a 986 * duplicate here avoids having a debugging-capable regexec.o tied to 987 * a matching debug.o, and this is convenient. It all disappears in 988 * the non-debug compilation anyway, so it doesn't matter much. 989 */ 990 static char * /* -> representation */ 991 pchar(ch) 992 int ch; 993 { 994 static char pbuf[10]; 995 996 if (isprint(ch) || ch == ' ') 997 sprintf(pbuf, "%c", ch); 998 else 999 sprintf(pbuf, "\\%o", ch); 1000 return(pbuf); 1001 } 1002 #endif 1003 #endif 1004 1005 #undef matcher 1006 #undef fast 1007 #undef slow 1008 #undef dissect 1009 #undef backref 1010 #undef step 1011 #undef print 1012 #undef at 1013 #undef match 1014
Indexes created Sat Sep 20 22:09:52 GMT 2025