| 1 | /* regcomp.c |
| 2 | */ |
| 3 | |
| 4 | /* |
| 5 | * "A fair jaw-cracker dwarf-language must be." --Samwise Gamgee |
| 6 | */ |
| 7 | |
| 8 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
| 9 | * confused with the original package (see point 3 below). Thanks, Henry! |
| 10 | */ |
| 11 | |
| 12 | /* Additional note: this code is very heavily munged from Henry's version |
| 13 | * in places. In some spots I've traded clarity for efficiency, so don't |
| 14 | * blame Henry for some of the lack of readability. |
| 15 | */ |
| 16 | |
| 17 | /* The names of the functions have been changed from regcomp and |
| 18 | * regexec to pregcomp and pregexec in order to avoid conflicts |
| 19 | * with the POSIX routines of the same names. |
| 20 | */ |
| 21 | |
| 22 | /*SUPPRESS 112*/ |
| 23 | /* |
| 24 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
| 25 | * |
| 26 | * Copyright (c) 1986 by University of Toronto. |
| 27 | * Written by Henry Spencer. Not derived from licensed software. |
| 28 | * |
| 29 | * Permission is granted to anyone to use this software for any |
| 30 | * purpose on any computer system, and to redistribute it freely, |
| 31 | * subject to the following restrictions: |
| 32 | * |
| 33 | * 1. The author is not responsible for the consequences of use of |
| 34 | * this software, no matter how awful, even if they arise |
| 35 | * from defects in it. |
| 36 | * |
| 37 | * 2. The origin of this software must not be misrepresented, either |
| 38 | * by explicit claim or by omission. |
| 39 | * |
| 40 | * 3. Altered versions must be plainly marked as such, and must not |
| 41 | * be misrepresented as being the original software. |
| 42 | * |
| 43 | * |
| 44 | **** Alterations to Henry's code are... |
| 45 | **** |
| 46 | **** Copyright (c) 1991-1994, Larry Wall |
| 47 | **** |
| 48 | **** You may distribute under the terms of either the GNU General Public |
| 49 | **** License or the Artistic License, as specified in the README file. |
| 50 | |
| 51 | * |
| 52 | * Beware that some of this code is subtly aware of the way operator |
| 53 | * precedence is structured in regular expressions. Serious changes in |
| 54 | * regular-expression syntax might require a total rethink. |
| 55 | */ |
| 56 | #include "EXTERN.h" |
| 57 | #include "perl.h" |
| 58 | #include "INTERN.h" |
| 59 | #include "regcomp.h" |
| 60 | |
| 61 | #ifdef MSDOS |
| 62 | # if defined(BUGGY_MSC6) |
| 63 | /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */ |
| 64 | # pragma optimize("a",off) |
| 65 | /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/ |
| 66 | # pragma optimize("w",on ) |
| 67 | # endif /* BUGGY_MSC6 */ |
| 68 | #endif /* MSDOS */ |
| 69 | |
| 70 | #ifndef STATIC |
| 71 | #define STATIC static |
| 72 | #endif |
| 73 | |
| 74 | #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?') |
| 75 | #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \ |
| 76 | ((*s) == '{' && regcurly(s))) |
| 77 | #ifdef atarist |
| 78 | #define PERL_META "^$.[()|?+*\\" |
| 79 | #else |
| 80 | #define META "^$.[()|?+*\\" |
| 81 | #endif |
| 82 | |
| 83 | #ifdef SPSTART |
| 84 | #undef SPSTART /* dratted cpp namespace... */ |
| 85 | #endif |
| 86 | /* |
| 87 | * Flags to be passed up and down. |
| 88 | */ |
| 89 | #define WORST 0 /* Worst case. */ |
| 90 | #define HASWIDTH 0x1 /* Known never to match null string. */ |
| 91 | #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ |
| 92 | #define SPSTART 0x4 /* Starts with * or +. */ |
| 93 | #define TRYAGAIN 0x8 /* Weeded out a declaration. */ |
| 94 | |
| 95 | /* |
| 96 | * Forward declarations for pregcomp()'s friends. |
| 97 | */ |
| 98 | |
| 99 | static char *reg _((I32, I32 *)); |
| 100 | static char *reganode _((char, unsigned short)); |
| 101 | static char *regatom _((I32 *)); |
| 102 | static char *regbranch _((I32 *)); |
| 103 | static void regc _((char)); |
| 104 | static char *regclass _((void)); |
| 105 | STATIC I32 regcurly _((char *)); |
| 106 | static char *regnode _((char)); |
| 107 | static char *regpiece _((I32 *)); |
| 108 | static void reginsert _((char, char *)); |
| 109 | static void regoptail _((char *, char *)); |
| 110 | static void regset _((char *, I32, I32)); |
| 111 | static void regtail _((char *, char *)); |
| 112 | static char* nextchar _((void)); |
| 113 | |
| 114 | /* |
| 115 | - pregcomp - compile a regular expression into internal code |
| 116 | * |
| 117 | * We can't allocate space until we know how big the compiled form will be, |
| 118 | * but we can't compile it (and thus know how big it is) until we've got a |
| 119 | * place to put the code. So we cheat: we compile it twice, once with code |
| 120 | * generation turned off and size counting turned on, and once "for real". |
| 121 | * This also means that we don't allocate space until we are sure that the |
| 122 | * thing really will compile successfully, and we never have to move the |
| 123 | * code and thus invalidate pointers into it. (Note that it has to be in |
| 124 | * one piece because free() must be able to free it all.) [NB: not true in perl] |
| 125 | * |
| 126 | * Beware that the optimization-preparation code in here knows about some |
| 127 | * of the structure of the compiled regexp. [I'll say.] |
| 128 | */ |
| 129 | regexp * |
| 130 | pregcomp(exp,xend,pm) |
| 131 | char* exp; |
| 132 | char* xend; |
| 133 | PMOP* pm; |
| 134 | { |
| 135 | I32 fold = pm->op_pmflags & PMf_FOLD; |
| 136 | register regexp *r; |
| 137 | register char *scan; |
| 138 | register SV *longish; |
| 139 | SV *longest; |
| 140 | register I32 len; |
| 141 | register char *first; |
| 142 | I32 flags; |
| 143 | I32 backish; |
| 144 | I32 backest; |
| 145 | I32 curback; |
| 146 | I32 minlen = 0; |
| 147 | I32 sawplus = 0; |
| 148 | I32 sawopen = 0; |
| 149 | |
| 150 | if (exp == NULL) |
| 151 | croak("NULL regexp argument"); |
| 152 | |
| 153 | /* First pass: determine size, legality. */ |
| 154 | regflags = pm->op_pmflags; |
| 155 | regparse = exp; |
| 156 | regxend = xend; |
| 157 | regprecomp = savepvn(exp,xend-exp); |
| 158 | regnaughty = 0; |
| 159 | regsawback = 0; |
| 160 | regnpar = 1; |
| 161 | regsize = 0L; |
| 162 | regcode = ®dummy; |
| 163 | regc((char)MAGIC); |
| 164 | if (reg(0, &flags) == NULL) { |
| 165 | Safefree(regprecomp); |
| 166 | regprecomp = Nullch; |
| 167 | return(NULL); |
| 168 | } |
| 169 | |
| 170 | /* Small enough for pointer-storage convention? */ |
| 171 | if (regsize >= 32767L) /* Probably could be 65535L. */ |
| 172 | FAIL("regexp too big"); |
| 173 | |
| 174 | /* Allocate space. */ |
| 175 | Newc(1001, r, sizeof(regexp) + (unsigned)regsize, char, regexp); |
| 176 | if (r == NULL) |
| 177 | FAIL("regexp out of space"); |
| 178 | |
| 179 | /* Second pass: emit code. */ |
| 180 | r->prelen = xend-exp; |
| 181 | r->precomp = regprecomp; |
| 182 | r->subbeg = r->subbase = NULL; |
| 183 | regnaughty = 0; |
| 184 | regparse = exp; |
| 185 | regnpar = 1; |
| 186 | regcode = r->program; |
| 187 | regc((char)MAGIC); |
| 188 | if (reg(0, &flags) == NULL) |
| 189 | return(NULL); |
| 190 | |
| 191 | /* Dig out information for optimizations. */ |
| 192 | pm->op_pmflags = regflags; |
| 193 | fold = pm->op_pmflags & PMf_FOLD; |
| 194 | r->regstart = Nullsv; /* Worst-case defaults. */ |
| 195 | r->reganch = 0; |
| 196 | r->regmust = Nullsv; |
| 197 | r->regback = -1; |
| 198 | r->regstclass = Nullch; |
| 199 | r->naughty = regnaughty >= 10; /* Probably an expensive pattern. */ |
| 200 | scan = r->program+1; /* First BRANCH. */ |
| 201 | if (OP(regnext(scan)) == END) {/* Only one top-level choice. */ |
| 202 | scan = NEXTOPER(scan); |
| 203 | |
| 204 | first = scan; |
| 205 | while ((OP(first) == OPEN && (sawopen = 1)) || |
| 206 | (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) || |
| 207 | (OP(first) == PLUS) || |
| 208 | (OP(first) == MINMOD) || |
| 209 | (regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) { |
| 210 | if (OP(first) == PLUS) |
| 211 | sawplus = 1; |
| 212 | else |
| 213 | first += regarglen[(U8)OP(first)]; |
| 214 | first = NEXTOPER(first); |
| 215 | } |
| 216 | |
| 217 | /* Starting-point info. */ |
| 218 | again: |
| 219 | if (OP(first) == EXACTLY) { |
| 220 | r->regstart = newSVpv(OPERAND(first)+1,*OPERAND(first)); |
| 221 | if (SvCUR(r->regstart) > !(sawstudy|fold)) |
| 222 | fbm_compile(r->regstart,fold); |
| 223 | else |
| 224 | sv_upgrade(r->regstart, SVt_PVBM); |
| 225 | } |
| 226 | else if (strchr(simple+2,OP(first))) |
| 227 | r->regstclass = first; |
| 228 | else if (OP(first) == BOUND || OP(first) == NBOUND) |
| 229 | r->regstclass = first; |
| 230 | else if (regkind[(U8)OP(first)] == BOL) { |
| 231 | r->reganch = ROPT_ANCH; |
| 232 | first = NEXTOPER(first); |
| 233 | goto again; |
| 234 | } |
| 235 | else if ((OP(first) == STAR && |
| 236 | regkind[(U8)OP(NEXTOPER(first))] == ANY) && |
| 237 | !(r->reganch & ROPT_ANCH) ) |
| 238 | { |
| 239 | /* turn .* into ^.* with an implied $*=1 */ |
| 240 | r->reganch = ROPT_ANCH | ROPT_IMPLICIT; |
| 241 | first = NEXTOPER(first); |
| 242 | goto again; |
| 243 | } |
| 244 | if (sawplus && (!sawopen || !regsawback)) |
| 245 | r->reganch |= ROPT_SKIP; /* x+ must match 1st of run */ |
| 246 | |
| 247 | DEBUG_r(PerlIO_printf(Perl_debug_log, "first %d next %d offset %d\n", |
| 248 | OP(first), OP(NEXTOPER(first)), first - scan)); |
| 249 | /* |
| 250 | * If there's something expensive in the r.e., find the |
| 251 | * longest literal string that must appear and make it the |
| 252 | * regmust. Resolve ties in favor of later strings, since |
| 253 | * the regstart check works with the beginning of the r.e. |
| 254 | * and avoiding duplication strengthens checking. Not a |
| 255 | * strong reason, but sufficient in the absence of others. |
| 256 | * [Now we resolve ties in favor of the earlier string if |
| 257 | * it happens that curback has been invalidated, since the |
| 258 | * earlier string may buy us something the later one won't.] |
| 259 | */ |
| 260 | longish = newSVpv("",0); |
| 261 | longest = newSVpv("",0); |
| 262 | len = 0; |
| 263 | minlen = 0; |
| 264 | curback = 0; |
| 265 | backish = 0; |
| 266 | backest = 0; |
| 267 | while (OP(scan) != END) { |
| 268 | if (OP(scan) == BRANCH) { |
| 269 | if (OP(regnext(scan)) == BRANCH) { |
| 270 | curback = -30000; |
| 271 | while (OP(scan) == BRANCH) |
| 272 | scan = regnext(scan); |
| 273 | } |
| 274 | else /* single branch is ok */ |
| 275 | scan = NEXTOPER(scan); |
| 276 | continue; |
| 277 | } |
| 278 | if (OP(scan) == UNLESSM) { |
| 279 | curback = -30000; |
| 280 | scan = regnext(scan); |
| 281 | continue; |
| 282 | } |
| 283 | if (OP(scan) == EXACTLY) { |
| 284 | char *t; |
| 285 | |
| 286 | first = scan; |
| 287 | while (OP(t = regnext(scan)) == CLOSE) |
| 288 | scan = t; |
| 289 | minlen += *OPERAND(first); |
| 290 | if (curback - backish == len) { |
| 291 | sv_catpvn(longish, OPERAND(first)+1, |
| 292 | *OPERAND(first)); |
| 293 | len += *OPERAND(first); |
| 294 | curback += *OPERAND(first); |
| 295 | first = regnext(scan); |
| 296 | } |
| 297 | else if (*OPERAND(first) >= len + (curback >= 0)) { |
| 298 | len = *OPERAND(first); |
| 299 | sv_setpvn(longish, OPERAND(first)+1,len); |
| 300 | backish = curback; |
| 301 | curback += len; |
| 302 | first = regnext(scan); |
| 303 | } |
| 304 | else |
| 305 | curback += *OPERAND(first); |
| 306 | } |
| 307 | else if (strchr(varies,OP(scan))) { |
| 308 | curback = -30000; |
| 309 | len = 0; |
| 310 | if (SvCUR(longish) > SvCUR(longest)) { |
| 311 | sv_setsv(longest,longish); |
| 312 | backest = backish; |
| 313 | } |
| 314 | sv_setpvn(longish,"",0); |
| 315 | if (OP(scan) == PLUS && strchr(simple,OP(NEXTOPER(scan)))) |
| 316 | minlen++; |
| 317 | else if (regkind[(U8)OP(scan)] == CURLY && |
| 318 | strchr(simple,OP(NEXTOPER(scan)+4))) |
| 319 | minlen += ARG1(scan); |
| 320 | } |
| 321 | else if (strchr(simple,OP(scan))) { |
| 322 | curback++; |
| 323 | minlen++; |
| 324 | len = 0; |
| 325 | if (SvCUR(longish) > SvCUR(longest)) { |
| 326 | sv_setsv(longest,longish); |
| 327 | backest = backish; |
| 328 | } |
| 329 | sv_setpvn(longish,"",0); |
| 330 | } |
| 331 | scan = regnext(scan); |
| 332 | } |
| 333 | |
| 334 | /* Prefer earlier on tie, unless we can tail match latter */ |
| 335 | |
| 336 | if (SvCUR(longish) + (regkind[(U8)OP(first)] == EOL) > |
| 337 | SvCUR(longest)) |
| 338 | { |
| 339 | sv_setsv(longest,longish); |
| 340 | backest = backish; |
| 341 | } |
| 342 | else |
| 343 | sv_setpvn(longish,"",0); |
| 344 | if (SvCUR(longest) |
| 345 | && |
| 346 | (!r->regstart |
| 347 | || |
| 348 | !fbm_instr((unsigned char*) SvPVX(r->regstart), |
| 349 | (unsigned char *) SvPVX(r->regstart) |
| 350 | + SvCUR(r->regstart), |
| 351 | longest) |
| 352 | ) |
| 353 | ) |
| 354 | { |
| 355 | r->regmust = longest; |
| 356 | if (backest < 0) |
| 357 | backest = -1; |
| 358 | r->regback = backest; |
| 359 | if (SvCUR(longest) > !(sawstudy || fold || |
| 360 | regkind[(U8)OP(first)]==EOL)) |
| 361 | fbm_compile(r->regmust,fold); |
| 362 | (void)SvUPGRADE(r->regmust, SVt_PVBM); |
| 363 | BmUSEFUL(r->regmust) = 100; |
| 364 | if (regkind[(U8)OP(first)] == EOL && SvCUR(longish)) |
| 365 | SvTAIL_on(r->regmust); |
| 366 | } |
| 367 | else { |
| 368 | SvREFCNT_dec(longest); |
| 369 | longest = Nullsv; |
| 370 | } |
| 371 | SvREFCNT_dec(longish); |
| 372 | } |
| 373 | |
| 374 | r->do_folding = fold; |
| 375 | r->nparens = regnpar - 1; |
| 376 | r->minlen = minlen; |
| 377 | Newz(1002, r->startp, regnpar, char*); |
| 378 | Newz(1002, r->endp, regnpar, char*); |
| 379 | DEBUG_r(regdump(r)); |
| 380 | return(r); |
| 381 | } |
| 382 | |
| 383 | /* |
| 384 | - reg - regular expression, i.e. main body or parenthesized thing |
| 385 | * |
| 386 | * Caller must absorb opening parenthesis. |
| 387 | * |
| 388 | * Combining parenthesis handling with the base level of regular expression |
| 389 | * is a trifle forced, but the need to tie the tails of the branches to what |
| 390 | * follows makes it hard to avoid. |
| 391 | */ |
| 392 | static char * |
| 393 | reg(paren, flagp) |
| 394 | I32 paren; /* Parenthesized? */ |
| 395 | I32 *flagp; |
| 396 | { |
| 397 | register char *ret; |
| 398 | register char *br; |
| 399 | register char *ender = 0; |
| 400 | register I32 parno = 0; |
| 401 | I32 flags; |
| 402 | |
| 403 | *flagp = HASWIDTH; /* Tentatively. */ |
| 404 | |
| 405 | /* Make an OPEN node, if parenthesized. */ |
| 406 | if (paren) { |
| 407 | if (*regparse == '?') { |
| 408 | regparse++; |
| 409 | paren = *regparse++; |
| 410 | ret = NULL; |
| 411 | switch (paren) { |
| 412 | case ':': |
| 413 | case '=': |
| 414 | case '!': |
| 415 | break; |
| 416 | case '$': |
| 417 | case '@': |
| 418 | croak("Sequence (?%c...) not implemented", paren); |
| 419 | break; |
| 420 | case '#': |
| 421 | while (*regparse && *regparse != ')') |
| 422 | regparse++; |
| 423 | if (*regparse != ')') |
| 424 | croak("Sequence (?#... not terminated"); |
| 425 | nextchar(); |
| 426 | *flagp = TRYAGAIN; |
| 427 | return NULL; |
| 428 | default: |
| 429 | --regparse; |
| 430 | while (*regparse && strchr("iogmsx", *regparse)) |
| 431 | pmflag(®flags, *regparse++); |
| 432 | if (*regparse != ')') |
| 433 | croak("Sequence (?%c...) not recognized", *regparse); |
| 434 | nextchar(); |
| 435 | *flagp = TRYAGAIN; |
| 436 | return NULL; |
| 437 | } |
| 438 | } |
| 439 | else { |
| 440 | parno = regnpar; |
| 441 | regnpar++; |
| 442 | ret = reganode(OPEN, parno); |
| 443 | } |
| 444 | } else |
| 445 | ret = NULL; |
| 446 | |
| 447 | /* Pick up the branches, linking them together. */ |
| 448 | br = regbranch(&flags); |
| 449 | if (br == NULL) |
| 450 | return(NULL); |
| 451 | if (ret != NULL) |
| 452 | regtail(ret, br); /* OPEN -> first. */ |
| 453 | else |
| 454 | ret = br; |
| 455 | if (!(flags&HASWIDTH)) |
| 456 | *flagp &= ~HASWIDTH; |
| 457 | *flagp |= flags&SPSTART; |
| 458 | while (*regparse == '|') { |
| 459 | nextchar(); |
| 460 | br = regbranch(&flags); |
| 461 | if (br == NULL) |
| 462 | return(NULL); |
| 463 | regtail(ret, br); /* BRANCH -> BRANCH. */ |
| 464 | if (!(flags&HASWIDTH)) |
| 465 | *flagp &= ~HASWIDTH; |
| 466 | *flagp |= flags&SPSTART; |
| 467 | } |
| 468 | |
| 469 | /* Make a closing node, and hook it on the end. */ |
| 470 | switch (paren) { |
| 471 | case ':': |
| 472 | ender = regnode(NOTHING); |
| 473 | break; |
| 474 | case 1: |
| 475 | ender = reganode(CLOSE, parno); |
| 476 | break; |
| 477 | case '=': |
| 478 | case '!': |
| 479 | ender = regnode(SUCCEED); |
| 480 | *flagp &= ~HASWIDTH; |
| 481 | break; |
| 482 | case 0: |
| 483 | ender = regnode(END); |
| 484 | break; |
| 485 | } |
| 486 | regtail(ret, ender); |
| 487 | |
| 488 | /* Hook the tails of the branches to the closing node. */ |
| 489 | for (br = ret; br != NULL; br = regnext(br)) |
| 490 | regoptail(br, ender); |
| 491 | |
| 492 | if (paren == '=') { |
| 493 | reginsert(IFMATCH,ret); |
| 494 | regtail(ret, regnode(NOTHING)); |
| 495 | } |
| 496 | else if (paren == '!') { |
| 497 | reginsert(UNLESSM,ret); |
| 498 | regtail(ret, regnode(NOTHING)); |
| 499 | } |
| 500 | |
| 501 | /* Check for proper termination. */ |
| 502 | if (paren && (regparse >= regxend || *nextchar() != ')')) { |
| 503 | FAIL("unmatched () in regexp"); |
| 504 | } else if (!paren && regparse < regxend) { |
| 505 | if (*regparse == ')') { |
| 506 | FAIL("unmatched () in regexp"); |
| 507 | } else |
| 508 | FAIL("junk on end of regexp"); /* "Can't happen". */ |
| 509 | /* NOTREACHED */ |
| 510 | } |
| 511 | |
| 512 | return(ret); |
| 513 | } |
| 514 | |
| 515 | /* |
| 516 | - regbranch - one alternative of an | operator |
| 517 | * |
| 518 | * Implements the concatenation operator. |
| 519 | */ |
| 520 | static char * |
| 521 | regbranch(flagp) |
| 522 | I32 *flagp; |
| 523 | { |
| 524 | register char *ret; |
| 525 | register char *chain; |
| 526 | register char *latest; |
| 527 | I32 flags = 0; |
| 528 | |
| 529 | *flagp = WORST; /* Tentatively. */ |
| 530 | |
| 531 | ret = regnode(BRANCH); |
| 532 | chain = NULL; |
| 533 | regparse--; |
| 534 | nextchar(); |
| 535 | while (regparse < regxend && *regparse != '|' && *regparse != ')') { |
| 536 | flags &= ~TRYAGAIN; |
| 537 | latest = regpiece(&flags); |
| 538 | if (latest == NULL) { |
| 539 | if (flags & TRYAGAIN) |
| 540 | continue; |
| 541 | return(NULL); |
| 542 | } |
| 543 | *flagp |= flags&HASWIDTH; |
| 544 | if (chain == NULL) /* First piece. */ |
| 545 | *flagp |= flags&SPSTART; |
| 546 | else { |
| 547 | regnaughty++; |
| 548 | regtail(chain, latest); |
| 549 | } |
| 550 | chain = latest; |
| 551 | } |
| 552 | if (chain == NULL) /* Loop ran zero times. */ |
| 553 | (void) regnode(NOTHING); |
| 554 | |
| 555 | return(ret); |
| 556 | } |
| 557 | |
| 558 | /* |
| 559 | - regpiece - something followed by possible [*+?] |
| 560 | * |
| 561 | * Note that the branching code sequences used for ? and the general cases |
| 562 | * of * and + are somewhat optimized: they use the same NOTHING node as |
| 563 | * both the endmarker for their branch list and the body of the last branch. |
| 564 | * It might seem that this node could be dispensed with entirely, but the |
| 565 | * endmarker role is not redundant. |
| 566 | */ |
| 567 | static char * |
| 568 | regpiece(flagp) |
| 569 | I32 *flagp; |
| 570 | { |
| 571 | register char *ret; |
| 572 | register char op; |
| 573 | register char *next; |
| 574 | I32 flags; |
| 575 | char *origparse = regparse; |
| 576 | char *maxpos; |
| 577 | I32 min; |
| 578 | I32 max = 32767; |
| 579 | |
| 580 | ret = regatom(&flags); |
| 581 | if (ret == NULL) { |
| 582 | if (flags & TRYAGAIN) |
| 583 | *flagp |= TRYAGAIN; |
| 584 | return(NULL); |
| 585 | } |
| 586 | |
| 587 | op = *regparse; |
| 588 | if (op == '(' && regparse[1] == '?' && regparse[2] == '#') { |
| 589 | while (op && op != ')') |
| 590 | op = *++regparse; |
| 591 | if (op) { |
| 592 | nextchar(); |
| 593 | op = *regparse; |
| 594 | } |
| 595 | } |
| 596 | |
| 597 | if (op == '{' && regcurly(regparse)) { |
| 598 | next = regparse + 1; |
| 599 | maxpos = Nullch; |
| 600 | while (isDIGIT(*next) || *next == ',') { |
| 601 | if (*next == ',') { |
| 602 | if (maxpos) |
| 603 | break; |
| 604 | else |
| 605 | maxpos = next; |
| 606 | } |
| 607 | next++; |
| 608 | } |
| 609 | if (*next == '}') { /* got one */ |
| 610 | if (!maxpos) |
| 611 | maxpos = next; |
| 612 | regparse++; |
| 613 | min = atoi(regparse); |
| 614 | if (*maxpos == ',') |
| 615 | maxpos++; |
| 616 | else |
| 617 | maxpos = regparse; |
| 618 | max = atoi(maxpos); |
| 619 | if (!max && *maxpos != '0') |
| 620 | max = 32767; /* meaning "infinity" */ |
| 621 | regparse = next; |
| 622 | nextchar(); |
| 623 | |
| 624 | do_curly: |
| 625 | if ((flags&SIMPLE)) { |
| 626 | regnaughty += 2 + regnaughty / 2; |
| 627 | reginsert(CURLY, ret); |
| 628 | } |
| 629 | else { |
| 630 | regnaughty += 4 + regnaughty; /* compound interest */ |
| 631 | regtail(ret, regnode(WHILEM)); |
| 632 | reginsert(CURLYX,ret); |
| 633 | regtail(ret, regnode(NOTHING)); |
| 634 | } |
| 635 | |
| 636 | if (min > 0) |
| 637 | *flagp = (WORST|HASWIDTH); |
| 638 | if (max && max < min) |
| 639 | croak("Can't do {n,m} with n > m"); |
| 640 | if (regcode != ®dummy) { |
| 641 | #ifdef REGALIGN |
| 642 | *(unsigned short *)(ret+3) = min; |
| 643 | *(unsigned short *)(ret+5) = max; |
| 644 | #else |
| 645 | ret[3] = min >> 8; ret[4] = min & 0377; |
| 646 | ret[5] = max >> 8; ret[6] = max & 0377; |
| 647 | #endif |
| 648 | } |
| 649 | |
| 650 | goto nest_check; |
| 651 | } |
| 652 | } |
| 653 | |
| 654 | if (!ISMULT1(op)) { |
| 655 | *flagp = flags; |
| 656 | return(ret); |
| 657 | } |
| 658 | nextchar(); |
| 659 | |
| 660 | *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH); |
| 661 | |
| 662 | if (op == '*' && (flags&SIMPLE)) { |
| 663 | reginsert(STAR, ret); |
| 664 | regnaughty += 4; |
| 665 | } |
| 666 | else if (op == '*') { |
| 667 | min = 0; |
| 668 | goto do_curly; |
| 669 | } else if (op == '+' && (flags&SIMPLE)) { |
| 670 | reginsert(PLUS, ret); |
| 671 | regnaughty += 3; |
| 672 | } |
| 673 | else if (op == '+') { |
| 674 | min = 1; |
| 675 | goto do_curly; |
| 676 | } else if (op == '?') { |
| 677 | min = 0; max = 1; |
| 678 | goto do_curly; |
| 679 | } |
| 680 | nest_check: |
| 681 | if (dowarn && regcode != ®dummy && !(flags&HASWIDTH) && max > 10000) { |
| 682 | warn("%.*s matches null string many times", |
| 683 | regparse - origparse, origparse); |
| 684 | } |
| 685 | |
| 686 | if (*regparse == '?') { |
| 687 | nextchar(); |
| 688 | reginsert(MINMOD, ret); |
| 689 | #ifdef REGALIGN |
| 690 | regtail(ret, ret + 4); |
| 691 | #else |
| 692 | regtail(ret, ret + 3); |
| 693 | #endif |
| 694 | } |
| 695 | if (ISMULT2(regparse)) |
| 696 | FAIL("nested *?+ in regexp"); |
| 697 | |
| 698 | return(ret); |
| 699 | } |
| 700 | |
| 701 | /* |
| 702 | - regatom - the lowest level |
| 703 | * |
| 704 | * Optimization: gobbles an entire sequence of ordinary characters so that |
| 705 | * it can turn them into a single node, which is smaller to store and |
| 706 | * faster to run. Backslashed characters are exceptions, each becoming a |
| 707 | * separate node; the code is simpler that way and it's not worth fixing. |
| 708 | * |
| 709 | * [Yes, it is worth fixing, some scripts can run twice the speed.] |
| 710 | */ |
| 711 | static char * |
| 712 | regatom(flagp) |
| 713 | I32 *flagp; |
| 714 | { |
| 715 | register char *ret = 0; |
| 716 | I32 flags; |
| 717 | |
| 718 | *flagp = WORST; /* Tentatively. */ |
| 719 | |
| 720 | tryagain: |
| 721 | switch (*regparse) { |
| 722 | case '^': |
| 723 | nextchar(); |
| 724 | if (regflags & PMf_MULTILINE) |
| 725 | ret = regnode(MBOL); |
| 726 | else if (regflags & PMf_SINGLELINE) |
| 727 | ret = regnode(SBOL); |
| 728 | else |
| 729 | ret = regnode(BOL); |
| 730 | break; |
| 731 | case '$': |
| 732 | nextchar(); |
| 733 | if (regflags & PMf_MULTILINE) |
| 734 | ret = regnode(MEOL); |
| 735 | else if (regflags & PMf_SINGLELINE) |
| 736 | ret = regnode(SEOL); |
| 737 | else |
| 738 | ret = regnode(EOL); |
| 739 | break; |
| 740 | case '.': |
| 741 | nextchar(); |
| 742 | if (regflags & PMf_SINGLELINE) |
| 743 | ret = regnode(SANY); |
| 744 | else |
| 745 | ret = regnode(ANY); |
| 746 | regnaughty++; |
| 747 | *flagp |= HASWIDTH|SIMPLE; |
| 748 | break; |
| 749 | case '[': |
| 750 | regparse++; |
| 751 | ret = regclass(); |
| 752 | *flagp |= HASWIDTH|SIMPLE; |
| 753 | break; |
| 754 | case '(': |
| 755 | nextchar(); |
| 756 | ret = reg(1, &flags); |
| 757 | if (ret == NULL) { |
| 758 | if (flags & TRYAGAIN) |
| 759 | goto tryagain; |
| 760 | return(NULL); |
| 761 | } |
| 762 | *flagp |= flags&(HASWIDTH|SPSTART); |
| 763 | break; |
| 764 | case '|': |
| 765 | case ')': |
| 766 | if (flags & TRYAGAIN) { |
| 767 | *flagp |= TRYAGAIN; |
| 768 | return NULL; |
| 769 | } |
| 770 | croak("internal urp in regexp at /%s/", regparse); |
| 771 | /* Supposed to be caught earlier. */ |
| 772 | break; |
| 773 | case '?': |
| 774 | case '+': |
| 775 | case '*': |
| 776 | FAIL("?+* follows nothing in regexp"); |
| 777 | break; |
| 778 | case '\\': |
| 779 | switch (*++regparse) { |
| 780 | case 'A': |
| 781 | ret = regnode(SBOL); |
| 782 | *flagp |= SIMPLE; |
| 783 | nextchar(); |
| 784 | break; |
| 785 | case 'G': |
| 786 | ret = regnode(GBOL); |
| 787 | *flagp |= SIMPLE; |
| 788 | nextchar(); |
| 789 | break; |
| 790 | case 'Z': |
| 791 | ret = regnode(SEOL); |
| 792 | *flagp |= SIMPLE; |
| 793 | nextchar(); |
| 794 | break; |
| 795 | case 'w': |
| 796 | ret = regnode(ALNUM); |
| 797 | *flagp |= HASWIDTH|SIMPLE; |
| 798 | nextchar(); |
| 799 | break; |
| 800 | case 'W': |
| 801 | ret = regnode(NALNUM); |
| 802 | *flagp |= HASWIDTH|SIMPLE; |
| 803 | nextchar(); |
| 804 | break; |
| 805 | case 'b': |
| 806 | ret = regnode(BOUND); |
| 807 | *flagp |= SIMPLE; |
| 808 | nextchar(); |
| 809 | break; |
| 810 | case 'B': |
| 811 | ret = regnode(NBOUND); |
| 812 | *flagp |= SIMPLE; |
| 813 | nextchar(); |
| 814 | break; |
| 815 | case 's': |
| 816 | ret = regnode(SPACE); |
| 817 | *flagp |= HASWIDTH|SIMPLE; |
| 818 | nextchar(); |
| 819 | break; |
| 820 | case 'S': |
| 821 | ret = regnode(NSPACE); |
| 822 | *flagp |= HASWIDTH|SIMPLE; |
| 823 | nextchar(); |
| 824 | break; |
| 825 | case 'd': |
| 826 | ret = regnode(DIGIT); |
| 827 | *flagp |= HASWIDTH|SIMPLE; |
| 828 | nextchar(); |
| 829 | break; |
| 830 | case 'D': |
| 831 | ret = regnode(NDIGIT); |
| 832 | *flagp |= HASWIDTH|SIMPLE; |
| 833 | nextchar(); |
| 834 | break; |
| 835 | case 'n': |
| 836 | case 'r': |
| 837 | case 't': |
| 838 | case 'f': |
| 839 | case 'e': |
| 840 | case 'a': |
| 841 | case 'x': |
| 842 | case 'c': |
| 843 | case '0': |
| 844 | goto defchar; |
| 845 | case '1': case '2': case '3': case '4': |
| 846 | case '5': case '6': case '7': case '8': case '9': |
| 847 | { |
| 848 | I32 num = atoi(regparse); |
| 849 | |
| 850 | if (num > 9 && num >= regnpar) |
| 851 | goto defchar; |
| 852 | else { |
| 853 | regsawback = 1; |
| 854 | ret = reganode(REF, num); |
| 855 | *flagp |= HASWIDTH; |
| 856 | while (isDIGIT(*regparse)) |
| 857 | regparse++; |
| 858 | regparse--; |
| 859 | nextchar(); |
| 860 | } |
| 861 | } |
| 862 | break; |
| 863 | case '\0': |
| 864 | if (regparse >= regxend) |
| 865 | FAIL("trailing \\ in regexp"); |
| 866 | /* FALL THROUGH */ |
| 867 | default: |
| 868 | goto defchar; |
| 869 | } |
| 870 | break; |
| 871 | |
| 872 | case '#': |
| 873 | if (regflags & PMf_EXTENDED) { |
| 874 | while (regparse < regxend && *regparse != '\n') regparse++; |
| 875 | if (regparse < regxend) |
| 876 | goto tryagain; |
| 877 | } |
| 878 | /* FALL THROUGH */ |
| 879 | |
| 880 | default: { |
| 881 | register I32 len; |
| 882 | register char ender; |
| 883 | register char *p; |
| 884 | char *oldp; |
| 885 | I32 numlen; |
| 886 | |
| 887 | regparse++; |
| 888 | |
| 889 | defchar: |
| 890 | ret = regnode(EXACTLY); |
| 891 | regc(0); /* save spot for len */ |
| 892 | for (len = 0, p = regparse - 1; |
| 893 | len < 127 && p < regxend; |
| 894 | len++) |
| 895 | { |
| 896 | oldp = p; |
| 897 | switch (*p) { |
| 898 | case '^': |
| 899 | case '$': |
| 900 | case '.': |
| 901 | case '[': |
| 902 | case '(': |
| 903 | case ')': |
| 904 | case '|': |
| 905 | goto loopdone; |
| 906 | case '\\': |
| 907 | switch (*++p) { |
| 908 | case 'A': |
| 909 | case 'G': |
| 910 | case 'Z': |
| 911 | case 'w': |
| 912 | case 'W': |
| 913 | case 'b': |
| 914 | case 'B': |
| 915 | case 's': |
| 916 | case 'S': |
| 917 | case 'd': |
| 918 | case 'D': |
| 919 | --p; |
| 920 | goto loopdone; |
| 921 | case 'n': |
| 922 | ender = '\n'; |
| 923 | p++; |
| 924 | break; |
| 925 | case 'r': |
| 926 | ender = '\r'; |
| 927 | p++; |
| 928 | break; |
| 929 | case 't': |
| 930 | ender = '\t'; |
| 931 | p++; |
| 932 | break; |
| 933 | case 'f': |
| 934 | ender = '\f'; |
| 935 | p++; |
| 936 | break; |
| 937 | case 'e': |
| 938 | ender = '\033'; |
| 939 | p++; |
| 940 | break; |
| 941 | case 'a': |
| 942 | ender = '\007'; |
| 943 | p++; |
| 944 | break; |
| 945 | case 'x': |
| 946 | ender = scan_hex(++p, 2, &numlen); |
| 947 | p += numlen; |
| 948 | break; |
| 949 | case 'c': |
| 950 | p++; |
| 951 | ender = *p++; |
| 952 | if (isLOWER(ender)) |
| 953 | ender = toUPPER(ender); |
| 954 | ender ^= 64; |
| 955 | break; |
| 956 | case '0': case '1': case '2': case '3':case '4': |
| 957 | case '5': case '6': case '7': case '8':case '9': |
| 958 | if (*p == '0' || |
| 959 | (isDIGIT(p[1]) && atoi(p) >= regnpar) ) { |
| 960 | ender = scan_oct(p, 3, &numlen); |
| 961 | p += numlen; |
| 962 | } |
| 963 | else { |
| 964 | --p; |
| 965 | goto loopdone; |
| 966 | } |
| 967 | break; |
| 968 | case '\0': |
| 969 | if (p >= regxend) |
| 970 | FAIL("trailing \\ in regexp"); |
| 971 | /* FALL THROUGH */ |
| 972 | default: |
| 973 | ender = *p++; |
| 974 | break; |
| 975 | } |
| 976 | break; |
| 977 | case '#': |
| 978 | if (regflags & PMf_EXTENDED) { |
| 979 | while (p < regxend && *p != '\n') p++; |
| 980 | } |
| 981 | /* FALL THROUGH */ |
| 982 | case ' ': case '\t': case '\n': case '\r': case '\f': case '\v': |
| 983 | if (regflags & PMf_EXTENDED) { |
| 984 | p++; |
| 985 | len--; |
| 986 | continue; |
| 987 | } |
| 988 | /* FALL THROUGH */ |
| 989 | default: |
| 990 | ender = *p++; |
| 991 | break; |
| 992 | } |
| 993 | if (regflags & PMf_FOLD && isUPPER(ender)) |
| 994 | ender = toLOWER(ender); |
| 995 | if (ISMULT2(p)) { /* Back off on ?+*. */ |
| 996 | if (len) |
| 997 | p = oldp; |
| 998 | else { |
| 999 | len++; |
| 1000 | regc(ender); |
| 1001 | } |
| 1002 | break; |
| 1003 | } |
| 1004 | regc(ender); |
| 1005 | } |
| 1006 | loopdone: |
| 1007 | regparse = p - 1; |
| 1008 | nextchar(); |
| 1009 | if (len < 0) |
| 1010 | FAIL("internal disaster in regexp"); |
| 1011 | if (len > 0) |
| 1012 | *flagp |= HASWIDTH; |
| 1013 | if (len == 1) |
| 1014 | *flagp |= SIMPLE; |
| 1015 | if (regcode != ®dummy) |
| 1016 | *OPERAND(ret) = len; |
| 1017 | regc('\0'); |
| 1018 | } |
| 1019 | break; |
| 1020 | } |
| 1021 | |
| 1022 | return(ret); |
| 1023 | } |
| 1024 | |
| 1025 | static void |
| 1026 | regset(bits,def,c) |
| 1027 | char *bits; |
| 1028 | I32 def; |
| 1029 | register I32 c; |
| 1030 | { |
| 1031 | if (regcode == ®dummy) |
| 1032 | return; |
| 1033 | c &= 255; |
| 1034 | if (def) |
| 1035 | bits[c >> 3] &= ~(1 << (c & 7)); |
| 1036 | else |
| 1037 | bits[c >> 3] |= (1 << (c & 7)); |
| 1038 | } |
| 1039 | |
| 1040 | static char * |
| 1041 | regclass() |
| 1042 | { |
| 1043 | register char *bits; |
| 1044 | register I32 class; |
| 1045 | register I32 lastclass = 1234; |
| 1046 | register I32 range = 0; |
| 1047 | register char *ret; |
| 1048 | register I32 def; |
| 1049 | I32 numlen; |
| 1050 | |
| 1051 | ret = regnode(ANYOF); |
| 1052 | if (*regparse == '^') { /* Complement of range. */ |
| 1053 | regnaughty++; |
| 1054 | regparse++; |
| 1055 | def = 0; |
| 1056 | } else { |
| 1057 | def = 255; |
| 1058 | } |
| 1059 | bits = regcode; |
| 1060 | for (class = 0; class < 32; class++) |
| 1061 | regc(def); |
| 1062 | if (*regparse == ']' || *regparse == '-') |
| 1063 | goto skipcond; /* allow 1st char to be ] or - */ |
| 1064 | while (regparse < regxend && *regparse != ']') { |
| 1065 | skipcond: |
| 1066 | class = UCHARAT(regparse++); |
| 1067 | if (class == '\\') { |
| 1068 | class = UCHARAT(regparse++); |
| 1069 | switch (class) { |
| 1070 | case 'w': |
| 1071 | for (class = 0; class < 256; class++) |
| 1072 | if (isALNUM(class)) |
| 1073 | regset(bits,def,class); |
| 1074 | lastclass = 1234; |
| 1075 | continue; |
| 1076 | case 'W': |
| 1077 | for (class = 0; class < 256; class++) |
| 1078 | if (!isALNUM(class)) |
| 1079 | regset(bits,def,class); |
| 1080 | lastclass = 1234; |
| 1081 | continue; |
| 1082 | case 's': |
| 1083 | for (class = 0; class < 256; class++) |
| 1084 | if (isSPACE(class)) |
| 1085 | regset(bits,def,class); |
| 1086 | lastclass = 1234; |
| 1087 | continue; |
| 1088 | case 'S': |
| 1089 | for (class = 0; class < 256; class++) |
| 1090 | if (!isSPACE(class)) |
| 1091 | regset(bits,def,class); |
| 1092 | lastclass = 1234; |
| 1093 | continue; |
| 1094 | case 'd': |
| 1095 | for (class = '0'; class <= '9'; class++) |
| 1096 | regset(bits,def,class); |
| 1097 | lastclass = 1234; |
| 1098 | continue; |
| 1099 | case 'D': |
| 1100 | for (class = 0; class < '0'; class++) |
| 1101 | regset(bits,def,class); |
| 1102 | for (class = '9' + 1; class < 256; class++) |
| 1103 | regset(bits,def,class); |
| 1104 | lastclass = 1234; |
| 1105 | continue; |
| 1106 | case 'n': |
| 1107 | class = '\n'; |
| 1108 | break; |
| 1109 | case 'r': |
| 1110 | class = '\r'; |
| 1111 | break; |
| 1112 | case 't': |
| 1113 | class = '\t'; |
| 1114 | break; |
| 1115 | case 'f': |
| 1116 | class = '\f'; |
| 1117 | break; |
| 1118 | case 'b': |
| 1119 | class = '\b'; |
| 1120 | break; |
| 1121 | case 'e': |
| 1122 | class = '\033'; |
| 1123 | break; |
| 1124 | case 'a': |
| 1125 | class = '\007'; |
| 1126 | break; |
| 1127 | case 'x': |
| 1128 | class = scan_hex(regparse, 2, &numlen); |
| 1129 | regparse += numlen; |
| 1130 | break; |
| 1131 | case 'c': |
| 1132 | class = *regparse++; |
| 1133 | if (isLOWER(class)) |
| 1134 | class = toUPPER(class); |
| 1135 | class ^= 64; |
| 1136 | break; |
| 1137 | case '0': case '1': case '2': case '3': case '4': |
| 1138 | case '5': case '6': case '7': case '8': case '9': |
| 1139 | class = scan_oct(--regparse, 3, &numlen); |
| 1140 | regparse += numlen; |
| 1141 | break; |
| 1142 | } |
| 1143 | } |
| 1144 | if (range) { |
| 1145 | if (lastclass > class) |
| 1146 | FAIL("invalid [] range in regexp"); |
| 1147 | range = 0; |
| 1148 | } |
| 1149 | else { |
| 1150 | lastclass = class; |
| 1151 | if (*regparse == '-' && regparse+1 < regxend && |
| 1152 | regparse[1] != ']') { |
| 1153 | regparse++; |
| 1154 | range = 1; |
| 1155 | continue; /* do it next time */ |
| 1156 | } |
| 1157 | } |
| 1158 | for ( ; lastclass <= class; lastclass++) { |
| 1159 | regset(bits,def,lastclass); |
| 1160 | if (regflags & PMf_FOLD && isUPPER(lastclass)) |
| 1161 | regset(bits,def,toLOWER(lastclass)); |
| 1162 | } |
| 1163 | lastclass = class; |
| 1164 | } |
| 1165 | if (*regparse != ']') |
| 1166 | FAIL("unmatched [] in regexp"); |
| 1167 | nextchar(); |
| 1168 | return ret; |
| 1169 | } |
| 1170 | |
| 1171 | static char* |
| 1172 | nextchar() |
| 1173 | { |
| 1174 | char* retval = regparse++; |
| 1175 | |
| 1176 | for (;;) { |
| 1177 | if (*regparse == '(' && regparse[1] == '?' && |
| 1178 | regparse[2] == '#') { |
| 1179 | while (*regparse && *regparse != ')') |
| 1180 | regparse++; |
| 1181 | regparse++; |
| 1182 | continue; |
| 1183 | } |
| 1184 | if (regflags & PMf_EXTENDED) { |
| 1185 | if (isSPACE(*regparse)) { |
| 1186 | regparse++; |
| 1187 | continue; |
| 1188 | } |
| 1189 | else if (*regparse == '#') { |
| 1190 | while (*regparse && *regparse != '\n') |
| 1191 | regparse++; |
| 1192 | regparse++; |
| 1193 | continue; |
| 1194 | } |
| 1195 | } |
| 1196 | return retval; |
| 1197 | } |
| 1198 | } |
| 1199 | |
| 1200 | /* |
| 1201 | - regnode - emit a node |
| 1202 | */ |
| 1203 | #ifdef CAN_PROTOTYPE |
| 1204 | static char * /* Location. */ |
| 1205 | regnode(char op) |
| 1206 | #else |
| 1207 | static char * /* Location. */ |
| 1208 | regnode(op) |
| 1209 | char op; |
| 1210 | #endif |
| 1211 | { |
| 1212 | register char *ret; |
| 1213 | register char *ptr; |
| 1214 | |
| 1215 | ret = regcode; |
| 1216 | if (ret == ®dummy) { |
| 1217 | #ifdef REGALIGN |
| 1218 | if (!(regsize & 1)) |
| 1219 | regsize++; |
| 1220 | #endif |
| 1221 | regsize += 3; |
| 1222 | return(ret); |
| 1223 | } |
| 1224 | |
| 1225 | #ifdef REGALIGN |
| 1226 | #ifndef lint |
| 1227 | if (!((long)ret & 1)) |
| 1228 | *ret++ = 127; |
| 1229 | #endif |
| 1230 | #endif |
| 1231 | ptr = ret; |
| 1232 | *ptr++ = op; |
| 1233 | *ptr++ = '\0'; /* Null "next" pointer. */ |
| 1234 | *ptr++ = '\0'; |
| 1235 | regcode = ptr; |
| 1236 | |
| 1237 | return(ret); |
| 1238 | } |
| 1239 | |
| 1240 | /* |
| 1241 | - reganode - emit a node with an argument |
| 1242 | */ |
| 1243 | #ifdef CAN_PROTOTYPE |
| 1244 | static char * /* Location. */ |
| 1245 | reganode(char op, unsigned short arg) |
| 1246 | #else |
| 1247 | static char * /* Location. */ |
| 1248 | reganode(op, arg) |
| 1249 | char op; |
| 1250 | unsigned short arg; |
| 1251 | #endif |
| 1252 | { |
| 1253 | register char *ret; |
| 1254 | register char *ptr; |
| 1255 | |
| 1256 | ret = regcode; |
| 1257 | if (ret == ®dummy) { |
| 1258 | #ifdef REGALIGN |
| 1259 | if (!(regsize & 1)) |
| 1260 | regsize++; |
| 1261 | #endif |
| 1262 | regsize += 5; |
| 1263 | return(ret); |
| 1264 | } |
| 1265 | |
| 1266 | #ifdef REGALIGN |
| 1267 | #ifndef lint |
| 1268 | if (!((long)ret & 1)) |
| 1269 | *ret++ = 127; |
| 1270 | #endif |
| 1271 | #endif |
| 1272 | ptr = ret; |
| 1273 | *ptr++ = op; |
| 1274 | *ptr++ = '\0'; /* Null "next" pointer. */ |
| 1275 | *ptr++ = '\0'; |
| 1276 | #ifdef REGALIGN |
| 1277 | *(unsigned short *)(ret+3) = arg; |
| 1278 | #else |
| 1279 | ret[3] = arg >> 8; ret[4] = arg & 0377; |
| 1280 | #endif |
| 1281 | ptr += 2; |
| 1282 | regcode = ptr; |
| 1283 | |
| 1284 | return(ret); |
| 1285 | } |
| 1286 | |
| 1287 | /* |
| 1288 | - regc - emit (if appropriate) a byte of code |
| 1289 | */ |
| 1290 | #ifdef CAN_PROTOTYPE |
| 1291 | static void |
| 1292 | regc(char b) |
| 1293 | #else |
| 1294 | static void |
| 1295 | regc(b) |
| 1296 | char b; |
| 1297 | #endif |
| 1298 | { |
| 1299 | if (regcode != ®dummy) |
| 1300 | *regcode++ = b; |
| 1301 | else |
| 1302 | regsize++; |
| 1303 | } |
| 1304 | |
| 1305 | /* |
| 1306 | - reginsert - insert an operator in front of already-emitted operand |
| 1307 | * |
| 1308 | * Means relocating the operand. |
| 1309 | */ |
| 1310 | #ifdef CAN_PROTOTYPE |
| 1311 | static void |
| 1312 | reginsert(char op, char *opnd) |
| 1313 | #else |
| 1314 | static void |
| 1315 | reginsert(op, opnd) |
| 1316 | char op; |
| 1317 | char *opnd; |
| 1318 | #endif |
| 1319 | { |
| 1320 | register char *src; |
| 1321 | register char *dst; |
| 1322 | register char *place; |
| 1323 | register int offset = (regkind[(U8)op] == CURLY ? 4 : 0); |
| 1324 | |
| 1325 | if (regcode == ®dummy) { |
| 1326 | #ifdef REGALIGN |
| 1327 | regsize += 4 + offset; |
| 1328 | #else |
| 1329 | regsize += 3 + offset; |
| 1330 | #endif |
| 1331 | return; |
| 1332 | } |
| 1333 | |
| 1334 | src = regcode; |
| 1335 | #ifdef REGALIGN |
| 1336 | regcode += 4 + offset; |
| 1337 | #else |
| 1338 | regcode += 3 + offset; |
| 1339 | #endif |
| 1340 | dst = regcode; |
| 1341 | while (src > opnd) |
| 1342 | *--dst = *--src; |
| 1343 | |
| 1344 | place = opnd; /* Op node, where operand used to be. */ |
| 1345 | *place++ = op; |
| 1346 | *place++ = '\0'; |
| 1347 | *place++ = '\0'; |
| 1348 | while (offset-- > 0) |
| 1349 | *place++ = '\0'; |
| 1350 | #ifdef REGALIGN |
| 1351 | *place++ = '\177'; |
| 1352 | #endif |
| 1353 | } |
| 1354 | |
| 1355 | /* |
| 1356 | - regtail - set the next-pointer at the end of a node chain |
| 1357 | */ |
| 1358 | static void |
| 1359 | regtail(p, val) |
| 1360 | char *p; |
| 1361 | char *val; |
| 1362 | { |
| 1363 | register char *scan; |
| 1364 | register char *temp; |
| 1365 | register I32 offset; |
| 1366 | |
| 1367 | if (p == ®dummy) |
| 1368 | return; |
| 1369 | |
| 1370 | /* Find last node. */ |
| 1371 | scan = p; |
| 1372 | for (;;) { |
| 1373 | temp = regnext(scan); |
| 1374 | if (temp == NULL) |
| 1375 | break; |
| 1376 | scan = temp; |
| 1377 | } |
| 1378 | |
| 1379 | #ifdef REGALIGN |
| 1380 | offset = val - scan; |
| 1381 | #ifndef lint |
| 1382 | *(short*)(scan+1) = offset; |
| 1383 | #else |
| 1384 | offset = offset; |
| 1385 | #endif |
| 1386 | #else |
| 1387 | if (OP(scan) == BACK) |
| 1388 | offset = scan - val; |
| 1389 | else |
| 1390 | offset = val - scan; |
| 1391 | *(scan+1) = (offset>>8)&0377; |
| 1392 | *(scan+2) = offset&0377; |
| 1393 | #endif |
| 1394 | } |
| 1395 | |
| 1396 | /* |
| 1397 | - regoptail - regtail on operand of first argument; nop if operandless |
| 1398 | */ |
| 1399 | static void |
| 1400 | regoptail(p, val) |
| 1401 | char *p; |
| 1402 | char *val; |
| 1403 | { |
| 1404 | /* "Operandless" and "op != BRANCH" are synonymous in practice. */ |
| 1405 | if (p == NULL || p == ®dummy || regkind[(U8)OP(p)] != BRANCH) |
| 1406 | return; |
| 1407 | regtail(NEXTOPER(p), val); |
| 1408 | } |
| 1409 | |
| 1410 | /* |
| 1411 | - regcurly - a little FSA that accepts {\d+,?\d*} |
| 1412 | */ |
| 1413 | STATIC I32 |
| 1414 | regcurly(s) |
| 1415 | register char *s; |
| 1416 | { |
| 1417 | if (*s++ != '{') |
| 1418 | return FALSE; |
| 1419 | if (!isDIGIT(*s)) |
| 1420 | return FALSE; |
| 1421 | while (isDIGIT(*s)) |
| 1422 | s++; |
| 1423 | if (*s == ',') |
| 1424 | s++; |
| 1425 | while (isDIGIT(*s)) |
| 1426 | s++; |
| 1427 | if (*s != '}') |
| 1428 | return FALSE; |
| 1429 | return TRUE; |
| 1430 | } |
| 1431 | |
| 1432 | #ifdef DEBUGGING |
| 1433 | |
| 1434 | /* |
| 1435 | - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form |
| 1436 | */ |
| 1437 | void |
| 1438 | regdump(r) |
| 1439 | regexp *r; |
| 1440 | { |
| 1441 | register char *s; |
| 1442 | register char op = EXACTLY; /* Arbitrary non-END op. */ |
| 1443 | register char *next; |
| 1444 | |
| 1445 | |
| 1446 | s = r->program + 1; |
| 1447 | while (op != END) { /* While that wasn't END last time... */ |
| 1448 | #ifdef REGALIGN |
| 1449 | if (!((long)s & 1)) |
| 1450 | s++; |
| 1451 | #endif |
| 1452 | op = OP(s); |
| 1453 | PerlIO_printf(Perl_debug_log, "%2d%s", s-r->program, regprop(s)); /* Where, what. */ |
| 1454 | next = regnext(s); |
| 1455 | s += regarglen[(U8)op]; |
| 1456 | if (next == NULL) /* Next ptr. */ |
| 1457 | PerlIO_printf(Perl_debug_log, "(0)"); |
| 1458 | else |
| 1459 | PerlIO_printf(Perl_debug_log, "(%d)", (s-r->program)+(next-s)); |
| 1460 | s += 3; |
| 1461 | if (op == ANYOF) { |
| 1462 | s += 32; |
| 1463 | } |
| 1464 | if (op == EXACTLY) { |
| 1465 | /* Literal string, where present. */ |
| 1466 | s++; |
| 1467 | (void)PerlIO_putc(Perl_debug_log, ' '); |
| 1468 | (void)PerlIO_putc(Perl_debug_log, '<'); |
| 1469 | while (*s != '\0') { |
| 1470 | (void)PerlIO_putc(Perl_debug_log,*s); |
| 1471 | s++; |
| 1472 | } |
| 1473 | (void)PerlIO_putc(Perl_debug_log, '>'); |
| 1474 | s++; |
| 1475 | } |
| 1476 | (void)PerlIO_putc(Perl_debug_log, '\n'); |
| 1477 | } |
| 1478 | |
| 1479 | /* Header fields of interest. */ |
| 1480 | if (r->regstart) |
| 1481 | PerlIO_printf(Perl_debug_log, "start `%s' ", SvPVX(r->regstart)); |
| 1482 | if (r->regstclass) |
| 1483 | PerlIO_printf(Perl_debug_log, "stclass `%s' ", regprop(r->regstclass)); |
| 1484 | if (r->reganch & ROPT_ANCH) |
| 1485 | PerlIO_printf(Perl_debug_log, "anchored "); |
| 1486 | if (r->reganch & ROPT_SKIP) |
| 1487 | PerlIO_printf(Perl_debug_log, "plus "); |
| 1488 | if (r->reganch & ROPT_IMPLICIT) |
| 1489 | PerlIO_printf(Perl_debug_log, "implicit "); |
| 1490 | if (r->regmust != NULL) |
| 1491 | PerlIO_printf(Perl_debug_log, "must have \"%s\" back %ld ", SvPVX(r->regmust), |
| 1492 | (long) r->regback); |
| 1493 | PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen); |
| 1494 | PerlIO_printf(Perl_debug_log, "\n"); |
| 1495 | } |
| 1496 | |
| 1497 | /* |
| 1498 | - regprop - printable representation of opcode |
| 1499 | */ |
| 1500 | char * |
| 1501 | regprop(op) |
| 1502 | char *op; |
| 1503 | { |
| 1504 | register char *p = 0; |
| 1505 | |
| 1506 | (void) strcpy(buf, ":"); |
| 1507 | |
| 1508 | switch (OP(op)) { |
| 1509 | case BOL: |
| 1510 | p = "BOL"; |
| 1511 | break; |
| 1512 | case MBOL: |
| 1513 | p = "MBOL"; |
| 1514 | break; |
| 1515 | case SBOL: |
| 1516 | p = "SBOL"; |
| 1517 | break; |
| 1518 | case EOL: |
| 1519 | p = "EOL"; |
| 1520 | break; |
| 1521 | case MEOL: |
| 1522 | p = "MEOL"; |
| 1523 | break; |
| 1524 | case SEOL: |
| 1525 | p = "SEOL"; |
| 1526 | break; |
| 1527 | case ANY: |
| 1528 | p = "ANY"; |
| 1529 | break; |
| 1530 | case SANY: |
| 1531 | p = "SANY"; |
| 1532 | break; |
| 1533 | case ANYOF: |
| 1534 | p = "ANYOF"; |
| 1535 | break; |
| 1536 | case BRANCH: |
| 1537 | p = "BRANCH"; |
| 1538 | break; |
| 1539 | case EXACTLY: |
| 1540 | p = "EXACTLY"; |
| 1541 | break; |
| 1542 | case NOTHING: |
| 1543 | p = "NOTHING"; |
| 1544 | break; |
| 1545 | case BACK: |
| 1546 | p = "BACK"; |
| 1547 | break; |
| 1548 | case END: |
| 1549 | p = "END"; |
| 1550 | break; |
| 1551 | case ALNUM: |
| 1552 | p = "ALNUM"; |
| 1553 | break; |
| 1554 | case NALNUM: |
| 1555 | p = "NALNUM"; |
| 1556 | break; |
| 1557 | case BOUND: |
| 1558 | p = "BOUND"; |
| 1559 | break; |
| 1560 | case NBOUND: |
| 1561 | p = "NBOUND"; |
| 1562 | break; |
| 1563 | case SPACE: |
| 1564 | p = "SPACE"; |
| 1565 | break; |
| 1566 | case NSPACE: |
| 1567 | p = "NSPACE"; |
| 1568 | break; |
| 1569 | case DIGIT: |
| 1570 | p = "DIGIT"; |
| 1571 | break; |
| 1572 | case NDIGIT: |
| 1573 | p = "NDIGIT"; |
| 1574 | break; |
| 1575 | case CURLY: |
| 1576 | (void)sprintf(buf+strlen(buf), "CURLY {%d,%d}", ARG1(op),ARG2(op)); |
| 1577 | p = NULL; |
| 1578 | break; |
| 1579 | case CURLYX: |
| 1580 | (void)sprintf(buf+strlen(buf), "CURLYX {%d,%d}", ARG1(op),ARG2(op)); |
| 1581 | p = NULL; |
| 1582 | break; |
| 1583 | case REF: |
| 1584 | (void)sprintf(buf+strlen(buf), "REF%d", ARG1(op)); |
| 1585 | p = NULL; |
| 1586 | break; |
| 1587 | case OPEN: |
| 1588 | (void)sprintf(buf+strlen(buf), "OPEN%d", ARG1(op)); |
| 1589 | p = NULL; |
| 1590 | break; |
| 1591 | case CLOSE: |
| 1592 | (void)sprintf(buf+strlen(buf), "CLOSE%d", ARG1(op)); |
| 1593 | p = NULL; |
| 1594 | break; |
| 1595 | case STAR: |
| 1596 | p = "STAR"; |
| 1597 | break; |
| 1598 | case PLUS: |
| 1599 | p = "PLUS"; |
| 1600 | break; |
| 1601 | case MINMOD: |
| 1602 | p = "MINMOD"; |
| 1603 | break; |
| 1604 | case GBOL: |
| 1605 | p = "GBOL"; |
| 1606 | break; |
| 1607 | case UNLESSM: |
| 1608 | p = "UNLESSM"; |
| 1609 | break; |
| 1610 | case IFMATCH: |
| 1611 | p = "IFMATCH"; |
| 1612 | break; |
| 1613 | case SUCCEED: |
| 1614 | p = "SUCCEED"; |
| 1615 | break; |
| 1616 | case WHILEM: |
| 1617 | p = "WHILEM"; |
| 1618 | break; |
| 1619 | default: |
| 1620 | FAIL("corrupted regexp opcode"); |
| 1621 | } |
| 1622 | if (p != NULL) |
| 1623 | (void) strcat(buf, p); |
| 1624 | return(buf); |
| 1625 | } |
| 1626 | #endif /* DEBUGGING */ |
| 1627 | |
| 1628 | void |
| 1629 | pregfree(r) |
| 1630 | struct regexp *r; |
| 1631 | { |
| 1632 | if (!r) |
| 1633 | return; |
| 1634 | if (r->precomp) { |
| 1635 | Safefree(r->precomp); |
| 1636 | r->precomp = Nullch; |
| 1637 | } |
| 1638 | if (r->subbase) { |
| 1639 | Safefree(r->subbase); |
| 1640 | r->subbase = Nullch; |
| 1641 | } |
| 1642 | if (r->regmust) { |
| 1643 | SvREFCNT_dec(r->regmust); |
| 1644 | r->regmust = Nullsv; |
| 1645 | } |
| 1646 | if (r->regstart) { |
| 1647 | SvREFCNT_dec(r->regstart); |
| 1648 | r->regstart = Nullsv; |
| 1649 | } |
| 1650 | Safefree(r->startp); |
| 1651 | Safefree(r->endp); |
| 1652 | Safefree(r); |
| 1653 | } |