| 1 |
/* |
| 2 |
* NFA utilities. |
| 3 |
* This file is #included by regcomp.c. |
| 4 |
* |
| 5 |
* Copyright (c) 1998, 1999 Henry Spencer. All rights reserved. |
| 6 |
* |
| 7 |
* Development of this software was funded, in part, by Cray Research Inc., |
| 8 |
* UUNET Communications Services Inc., Sun Microsystems Inc., and Scriptics |
| 9 |
* Corporation, none of whom are responsible for the results. The author |
| 10 |
* thanks all of them. |
| 11 |
* |
| 12 |
* Redistribution and use in source and binary forms -- with or without |
| 13 |
* modification -- are permitted for any purpose, provided that |
| 14 |
* redistributions in source form retain this entire copyright notice and |
| 15 |
* indicate the origin and nature of any modifications. |
| 16 |
* |
| 17 |
* I'd appreciate being given credit for this package in the documentation |
| 18 |
* of software which uses it, but that is not a requirement. |
| 19 |
* |
| 20 |
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, |
| 21 |
* INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY |
| 22 |
* AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL |
| 23 |
* HENRY SPENCER BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, |
| 24 |
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, |
| 25 |
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; |
| 26 |
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, |
| 27 |
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR |
| 28 |
* OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF |
| 29 |
* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 30 |
* |
| 31 |
* src/backend/regex/regc_nfa.c |
| 32 |
* |
| 33 |
* |
| 34 |
* One or two things that technically ought to be in here |
| 35 |
* are actually in color.c, thanks to some incestuous relationships in |
| 36 |
* the color chains. |
| 37 |
*/ |
| 38 |
|
| 39 |
#define NISERR() VISERR(nfa->v) |
| 40 |
#define NERR(e) VERR(nfa->v, (e)) |
| 41 |
|
| 42 |
|
| 43 |
/* |
| 44 |
* newnfa - set up an NFA |
| 45 |
*/ |
| 46 |
static struct nfa * /* the NFA, or NULL */ |
| 47 |
newnfa(struct vars *v, |
| 48 |
struct colormap *cm, |
| 49 |
struct nfa *parent) /* NULL if primary NFA */ |
| 50 |
{ |
| 51 |
struct nfa *nfa; |
| 52 |
|
| 53 |
nfa = (struct nfa *) MALLOC(sizeof(struct nfa)); |
| 54 |
if (nfa == NULL) |
| 55 |
{ |
| 56 |
ERR(REG_ESPACE); |
| 57 |
return NULL; |
| 58 |
} |
| 59 |
|
| 60 |
/* Make the NFA minimally valid, so freenfa() will behave sanely */ |
| 61 |
nfa->states = NULL; |
| 62 |
nfa->slast = NULL; |
| 63 |
nfa->freestates = NULL; |
| 64 |
nfa->freearcs = NULL; |
| 65 |
nfa->lastsb = NULL; |
| 66 |
nfa->lastab = NULL; |
| 67 |
nfa->lastsbused = 0; |
| 68 |
nfa->lastabused = 0; |
| 69 |
nfa->nstates = 0; |
| 70 |
nfa->cm = cm; |
| 71 |
nfa->v = v; |
| 72 |
nfa->bos[0] = nfa->bos[1] = COLORLESS; |
| 73 |
nfa->eos[0] = nfa->eos[1] = COLORLESS; |
| 74 |
nfa->flags = 0; |
| 75 |
nfa->minmatchall = nfa->maxmatchall = -1; |
| 76 |
nfa->parent = parent; /* Precedes newfstate so parent is valid. */ |
| 77 |
|
| 78 |
/* Create required infrastructure */ |
| 79 |
nfa->post = newfstate(nfa, '@'); /* number 0 */ |
| 80 |
nfa->pre = newfstate(nfa, '>'); /* number 1 */ |
| 81 |
nfa->init = newstate(nfa); /* may become invalid later */ |
| 82 |
nfa->final = newstate(nfa); |
| 83 |
if (ISERR()) |
| 84 |
{ |
| 85 |
freenfa(nfa); |
| 86 |
return NULL; |
| 87 |
} |
| 88 |
rainbow(nfa, nfa->cm, PLAIN, COLORLESS, nfa->pre, nfa->init); |
| 89 |
newarc(nfa, '^', 1, nfa->pre, nfa->init); |
| 90 |
newarc(nfa, '^', 0, nfa->pre, nfa->init); |
| 91 |
rainbow(nfa, nfa->cm, PLAIN, COLORLESS, nfa->final, nfa->post); |
| 92 |
newarc(nfa, '$', 1, nfa->final, nfa->post); |
| 93 |
newarc(nfa, '$', 0, nfa->final, nfa->post); |
| 94 |
|
| 95 |
if (ISERR()) |
| 96 |
{ |
| 97 |
freenfa(nfa); |
| 98 |
return NULL; |
| 99 |
} |
| 100 |
return nfa; |
| 101 |
} |
| 102 |
|
| 103 |
/* |
| 104 |
* freenfa - free an entire NFA |
| 105 |
*/ |
| 106 |
static void |
| 107 |
freenfa(struct nfa *nfa) |
| 108 |
{ |
| 109 |
struct statebatch *sb; |
| 110 |
struct statebatch *sbnext; |
| 111 |
struct arcbatch *ab; |
| 112 |
struct arcbatch *abnext; |
| 113 |
|
| 114 |
for (sb = nfa->lastsb; sb != NULL; sb = sbnext) |
| 115 |
{ |
| 116 |
sbnext = sb->next; |
| 117 |
nfa->v->spaceused -= STATEBATCHSIZE(sb->nstates); |
| 118 |
FREE(sb); |
| 119 |
} |
| 120 |
nfa->lastsb = NULL; |
| 121 |
for (ab = nfa->lastab; ab != NULL; ab = abnext) |
| 122 |
{ |
| 123 |
abnext = ab->next; |
| 124 |
nfa->v->spaceused -= ARCBATCHSIZE(ab->narcs); |
| 125 |
FREE(ab); |
| 126 |
} |
| 127 |
nfa->lastab = NULL; |
| 128 |
|
| 129 |
nfa->nstates = -1; |
| 130 |
FREE(nfa); |
| 131 |
} |
| 132 |
|
| 133 |
/* |
| 134 |
* newstate - allocate an NFA state, with zero flag value |
| 135 |
*/ |
| 136 |
static struct state * /* NULL on error */ |
| 137 |
newstate(struct nfa *nfa) |
| 138 |
{ |
| 139 |
struct state *s; |
| 140 |
|
| 141 |
/* |
| 142 |
* This is a handy place to check for operation cancel during regex |
| 143 |
* compilation, since no code path will go very long without making a new |
| 144 |
* state or arc. |
| 145 |
*/ |
| 146 |
INTERRUPT(nfa->v->re); |
| 147 |
|
| 148 |
/* first, recycle anything that's on the freelist */ |
| 149 |
if (nfa->freestates != NULL) |
| 150 |
{ |
| 151 |
s = nfa->freestates; |
| 152 |
nfa->freestates = s->next; |
| 153 |
} |
| 154 |
/* otherwise, is there anything left in the last statebatch? */ |
| 155 |
else if (nfa->lastsb != NULL && nfa->lastsbused < nfa->lastsb->nstates) |
| 156 |
{ |
| 157 |
s = &nfa->lastsb->s[nfa->lastsbused++]; |
| 158 |
} |
| 159 |
/* otherwise, need to allocate a new statebatch */ |
| 160 |
else |
| 161 |
{ |
| 162 |
struct statebatch *newSb; |
| 163 |
size_t nstates; |
| 164 |
|
| 165 |
if (nfa->v->spaceused >= REG_MAX_COMPILE_SPACE) |
| 166 |
{ |
| 167 |
NERR(REG_ETOOBIG); |
| 168 |
return NULL; |
| 169 |
} |
| 170 |
nstates = (nfa->lastsb != NULL) ? nfa->lastsb->nstates * 2 : FIRSTSBSIZE; |
| 171 |
if (nstates > MAXSBSIZE) |
| 172 |
nstates = MAXSBSIZE; |
| 173 |
newSb = (struct statebatch *) MALLOC(STATEBATCHSIZE(nstates)); |
| 174 |
if (newSb == NULL) |
| 175 |
{ |
| 176 |
NERR(REG_ESPACE); |
| 177 |
return NULL; |
| 178 |
} |
| 179 |
nfa->v->spaceused += STATEBATCHSIZE(nstates); |
| 180 |
newSb->nstates = nstates; |
| 181 |
newSb->next = nfa->lastsb; |
| 182 |
nfa->lastsb = newSb; |
| 183 |
nfa->lastsbused = 1; |
| 184 |
s = &newSb->s[0]; |
| 185 |
} |
| 186 |
|
| 187 |
assert(nfa->nstates >= 0); |
| 188 |
s->no = nfa->nstates++; |
| 189 |
s->flag = 0; |
| 190 |
if (nfa->states == NULL) |
| 191 |
nfa->states = s; |
| 192 |
s->nins = 0; |
| 193 |
s->ins = NULL; |
| 194 |
s->nouts = 0; |
| 195 |
s->outs = NULL; |
| 196 |
s->tmp = NULL; |
| 197 |
s->next = NULL; |
| 198 |
if (nfa->slast != NULL) |
| 199 |
{ |
| 200 |
assert(nfa->slast->next == NULL); |
| 201 |
nfa->slast->next = s; |
| 202 |
} |
| 203 |
s->prev = nfa->slast; |
| 204 |
nfa->slast = s; |
| 205 |
return s; |
| 206 |
} |
| 207 |
|
| 208 |
/* |
| 209 |
* newfstate - allocate an NFA state with a specified flag value |
| 210 |
*/ |
| 211 |
static struct state * /* NULL on error */ |
| 212 |
newfstate(struct nfa *nfa, int flag) |
| 213 |
{ |
| 214 |
struct state *s; |
| 215 |
|
| 216 |
s = newstate(nfa); |
| 217 |
if (s != NULL) |
| 218 |
s->flag = (char) flag; |
| 219 |
return s; |
| 220 |
} |
| 221 |
|
| 222 |
/* |
| 223 |
* dropstate - delete a state's inarcs and outarcs and free it |
| 224 |
*/ |
| 225 |
static void |
| 226 |
dropstate(struct nfa *nfa, |
| 227 |
struct state *s) |
| 228 |
{ |
| 229 |
struct arc *a; |
| 230 |
|
| 231 |
while ((a = s->ins) != NULL) |
| 232 |
freearc(nfa, a); |
| 233 |
while ((a = s->outs) != NULL) |
| 234 |
freearc(nfa, a); |
| 235 |
freestate(nfa, s); |
| 236 |
} |
| 237 |
|
| 238 |
/* |
| 239 |
* freestate - free a state, which has no in-arcs or out-arcs |
| 240 |
*/ |
| 241 |
static void |
| 242 |
freestate(struct nfa *nfa, |
| 243 |
struct state *s) |
| 244 |
{ |
| 245 |
assert(s != NULL); |
| 246 |
assert(s->nins == 0 && s->nouts == 0); |
| 247 |
|
| 248 |
s->no = FREESTATE; |
| 249 |
s->flag = 0; |
| 250 |
if (s->next != NULL) |
| 251 |
s->next->prev = s->prev; |
| 252 |
else |
| 253 |
{ |
| 254 |
assert(s == nfa->slast); |
| 255 |
nfa->slast = s->prev; |
| 256 |
} |
| 257 |
if (s->prev != NULL) |
| 258 |
s->prev->next = s->next; |
| 259 |
else |
| 260 |
{ |
| 261 |
assert(s == nfa->states); |
| 262 |
nfa->states = s->next; |
| 263 |
} |
| 264 |
s->prev = NULL; |
| 265 |
s->next = nfa->freestates; /* don't delete it, put it on the free list */ |
| 266 |
nfa->freestates = s; |
| 267 |
} |
| 268 |
|
| 269 |
/* |
| 270 |
* newarc - set up a new arc within an NFA |
| 271 |
* |
| 272 |
* This function checks to make sure that no duplicate arcs are created. |
| 273 |
* In general we never want duplicates. |
| 274 |
* |
| 275 |
* However: in principle, a RAINBOW arc is redundant with any plain arc |
| 276 |
* (unless that arc is for a pseudocolor). But we don't try to recognize |
| 277 |
* that redundancy, either here or in allied operations such as moveins(). |
| 278 |
* The pseudocolor consideration makes that more costly than it seems worth. |
| 279 |
*/ |
| 280 |
static void |
| 281 |
newarc(struct nfa *nfa, |
| 282 |
int t, |
| 283 |
color co, |
| 284 |
struct state *from, |
| 285 |
struct state *to) |
| 286 |
{ |
| 287 |
struct arc *a; |
| 288 |
|
| 289 |
assert(from != NULL && to != NULL); |
| 290 |
|
| 291 |
/* |
| 292 |
* This is a handy place to check for operation cancel during regex |
| 293 |
* compilation, since no code path will go very long without making a new |
| 294 |
* state or arc. |
| 295 |
*/ |
| 296 |
INTERRUPT(nfa->v->re); |
| 297 |
|
| 298 |
/* check for duplicate arc, using whichever chain is shorter */ |
| 299 |
if (from->nouts <= to->nins) |
| 300 |
{ |
| 301 |
for (a = from->outs; a != NULL; a = a->outchain) |
| 302 |
if (a->to == to && a->co == co && a->type == t) |
| 303 |
return; |
| 304 |
} |
| 305 |
else |
| 306 |
{ |
| 307 |
for (a = to->ins; a != NULL; a = a->inchain) |
| 308 |
if (a->from == from && a->co == co && a->type == t) |
| 309 |
return; |
| 310 |
} |
| 311 |
|
| 312 |
/* no dup, so create the arc */ |
| 313 |
createarc(nfa, t, co, from, to); |
| 314 |
} |
| 315 |
|
| 316 |
/* |
| 317 |
* createarc - create a new arc within an NFA |
| 318 |
* |
| 319 |
* This function must *only* be used after verifying that there is no existing |
| 320 |
* identical arc (same type/color/from/to). |
| 321 |
*/ |
| 322 |
static void |
| 323 |
createarc(struct nfa *nfa, |
| 324 |
int t, |
| 325 |
color co, |
| 326 |
struct state *from, |
| 327 |
struct state *to) |
| 328 |
{ |
| 329 |
struct arc *a; |
| 330 |
|
| 331 |
a = allocarc(nfa); |
| 332 |
if (NISERR()) |
| 333 |
return; |
| 334 |
assert(a != NULL); |
| 335 |
|
| 336 |
a->type = t; |
| 337 |
a->co = co; |
| 338 |
a->to = to; |
| 339 |
a->from = from; |
| 340 |
|
| 341 |
/* |
| 342 |
* Put the new arc on the beginning, not the end, of the chains; it's |
| 343 |
* simpler here, and freearc() is the same cost either way. See also the |
| 344 |
* logic in moveins() and its cohorts, as well as fixempties(). |
| 345 |
*/ |
| 346 |
a->inchain = to->ins; |
| 347 |
a->inchainRev = NULL; |
| 348 |
if (to->ins) |
| 349 |
to->ins->inchainRev = a; |
| 350 |
to->ins = a; |
| 351 |
a->outchain = from->outs; |
| 352 |
a->outchainRev = NULL; |
| 353 |
if (from->outs) |
| 354 |
from->outs->outchainRev = a; |
| 355 |
from->outs = a; |
| 356 |
|
| 357 |
from->nouts++; |
| 358 |
to->nins++; |
| 359 |
|
| 360 |
if (COLORED(a) && nfa->parent == NULL) |
| 361 |
colorchain(nfa->cm, a); |
| 362 |
} |
| 363 |
|
| 364 |
/* |
| 365 |
* allocarc - allocate a new arc within an NFA |
| 366 |
*/ |
| 367 |
static struct arc * /* NULL for failure */ |
| 368 |
allocarc(struct nfa *nfa) |
| 369 |
{ |
| 370 |
struct arc *a; |
| 371 |
|
| 372 |
/* first, recycle anything that's on the freelist */ |
| 373 |
if (nfa->freearcs != NULL) |
| 374 |
{ |
| 375 |
a = nfa->freearcs; |
| 376 |
nfa->freearcs = a->freechain; |
| 377 |
} |
| 378 |
/* otherwise, is there anything left in the last arcbatch? */ |
| 379 |
else if (nfa->lastab != NULL && nfa->lastabused < nfa->lastab->narcs) |
| 380 |
{ |
| 381 |
a = &nfa->lastab->a[nfa->lastabused++]; |
| 382 |
} |
| 383 |
/* otherwise, need to allocate a new arcbatch */ |
| 384 |
else |
| 385 |
{ |
| 386 |
struct arcbatch *newAb; |
| 387 |
size_t narcs; |
| 388 |
|
| 389 |
if (nfa->v->spaceused >= REG_MAX_COMPILE_SPACE) |
| 390 |
{ |
| 391 |
NERR(REG_ETOOBIG); |
| 392 |
return NULL; |
| 393 |
} |
| 394 |
narcs = (nfa->lastab != NULL) ? nfa->lastab->narcs * 2 : FIRSTABSIZE; |
| 395 |
if (narcs > MAXABSIZE) |
| 396 |
narcs = MAXABSIZE; |
| 397 |
newAb = (struct arcbatch *) MALLOC(ARCBATCHSIZE(narcs)); |
| 398 |
if (newAb == NULL) |
| 399 |
{ |
| 400 |
NERR(REG_ESPACE); |
| 401 |
return NULL; |
| 402 |
} |
| 403 |
nfa->v->spaceused += ARCBATCHSIZE(narcs); |
| 404 |
newAb->narcs = narcs; |
| 405 |
newAb->next = nfa->lastab; |
| 406 |
nfa->lastab = newAb; |
| 407 |
nfa->lastabused = 1; |
| 408 |
a = &newAb->a[0]; |
| 409 |
} |
| 410 |
|
| 411 |
return a; |
| 412 |
} |
| 413 |
|
| 414 |
/* |
| 415 |
* freearc - free an arc |
| 416 |
*/ |
| 417 |
static void |
| 418 |
freearc(struct nfa *nfa, |
| 419 |
struct arc *victim) |
| 420 |
{ |
| 421 |
struct state *from = victim->from; |
| 422 |
struct state *to = victim->to; |
| 423 |
struct arc *predecessor; |
| 424 |
|
| 425 |
assert(victim->type != 0); |
| 426 |
|
| 427 |
/* take it off color chain if necessary */ |
| 428 |
if (COLORED(victim) && nfa->parent == NULL) |
| 429 |
uncolorchain(nfa->cm, victim); |
| 430 |
|
| 431 |
/* take it off source's out-chain */ |
| 432 |
assert(from != NULL); |
| 433 |
predecessor = victim->outchainRev; |
| 434 |
if (predecessor == NULL) |
| 435 |
{ |
| 436 |
assert(from->outs == victim); |
| 437 |
from->outs = victim->outchain; |
| 438 |
} |
| 439 |
else |
| 440 |
{ |
| 441 |
assert(predecessor->outchain == victim); |
| 442 |
predecessor->outchain = victim->outchain; |
| 443 |
} |
| 444 |
if (victim->outchain != NULL) |
| 445 |
{ |
| 446 |
assert(victim->outchain->outchainRev == victim); |
| 447 |
victim->outchain->outchainRev = predecessor; |
| 448 |
} |
| 449 |
from->nouts--; |
| 450 |
|
| 451 |
/* take it off target's in-chain */ |
| 452 |
assert(to != NULL); |
| 453 |
predecessor = victim->inchainRev; |
| 454 |
if (predecessor == NULL) |
| 455 |
{ |
| 456 |
assert(to->ins == victim); |
| 457 |
to->ins = victim->inchain; |
| 458 |
} |
| 459 |
else |
| 460 |
{ |
| 461 |
assert(predecessor->inchain == victim); |
| 462 |
predecessor->inchain = victim->inchain; |
| 463 |
} |
| 464 |
if (victim->inchain != NULL) |
| 465 |
{ |
| 466 |
assert(victim->inchain->inchainRev == victim); |
| 467 |
victim->inchain->inchainRev = predecessor; |
| 468 |
} |
| 469 |
to->nins--; |
| 470 |
|
| 471 |
/* clean up and place on NFA's free list */ |
| 472 |
victim->type = 0; |
| 473 |
victim->from = NULL; /* precautions... */ |
| 474 |
victim->to = NULL; |
| 475 |
victim->inchain = NULL; |
| 476 |
victim->inchainRev = NULL; |
| 477 |
victim->outchain = NULL; |
| 478 |
victim->outchainRev = NULL; |
| 479 |
victim->freechain = nfa->freearcs; |
| 480 |
nfa->freearcs = victim; |
| 481 |
} |
| 482 |
|
| 483 |
/* |
| 484 |
* changearcsource - flip an arc to have a different from state |
| 485 |
* |
| 486 |
* Caller must have verified that there is no pre-existing duplicate arc. |
| 487 |
*/ |
| 488 |
static void |
| 489 |
changearcsource(struct arc *a, struct state *newfrom) |
| 490 |
{ |
| 491 |
struct state *oldfrom = a->from; |
| 492 |
struct arc *predecessor; |
| 493 |
|
| 494 |
assert(oldfrom != newfrom); |
| 495 |
|
| 496 |
/* take it off old source's out-chain */ |
| 497 |
assert(oldfrom != NULL); |
| 498 |
predecessor = a->outchainRev; |
| 499 |
if (predecessor == NULL) |
| 500 |
{ |
| 501 |
assert(oldfrom->outs == a); |
| 502 |
oldfrom->outs = a->outchain; |
| 503 |
} |
| 504 |
else |
| 505 |
{ |
| 506 |
assert(predecessor->outchain == a); |
| 507 |
predecessor->outchain = a->outchain; |
| 508 |
} |
| 509 |
if (a->outchain != NULL) |
| 510 |
{ |
| 511 |
assert(a->outchain->outchainRev == a); |
| 512 |
a->outchain->outchainRev = predecessor; |
| 513 |
} |
| 514 |
oldfrom->nouts--; |
| 515 |
|
| 516 |
a->from = newfrom; |
| 517 |
|
| 518 |
/* prepend it to new source's out-chain */ |
| 519 |
a->outchain = newfrom->outs; |
| 520 |
a->outchainRev = NULL; |
| 521 |
if (newfrom->outs) |
| 522 |
newfrom->outs->outchainRev = a; |
| 523 |
newfrom->outs = a; |
| 524 |
newfrom->nouts++; |
| 525 |
} |
| 526 |
|
| 527 |
/* |
| 528 |
* changearctarget - flip an arc to have a different to state |
| 529 |
* |
| 530 |
* Caller must have verified that there is no pre-existing duplicate arc. |
| 531 |
*/ |
| 532 |
static void |
| 533 |
changearctarget(struct arc *a, struct state *newto) |
| 534 |
{ |
| 535 |
struct state *oldto = a->to; |
| 536 |
struct arc *predecessor; |
| 537 |
|
| 538 |
assert(oldto != newto); |
| 539 |
|
| 540 |
/* take it off old target's in-chain */ |
| 541 |
assert(oldto != NULL); |
| 542 |
predecessor = a->inchainRev; |
| 543 |
if (predecessor == NULL) |
| 544 |
{ |
| 545 |
assert(oldto->ins == a); |
| 546 |
oldto->ins = a->inchain; |
| 547 |
} |
| 548 |
else |
| 549 |
{ |
| 550 |
assert(predecessor->inchain == a); |
| 551 |
predecessor->inchain = a->inchain; |
| 552 |
} |
| 553 |
if (a->inchain != NULL) |
| 554 |
{ |
| 555 |
assert(a->inchain->inchainRev == a); |
| 556 |
a->inchain->inchainRev = predecessor; |
| 557 |
} |
| 558 |
oldto->nins--; |
| 559 |
|
| 560 |
a->to = newto; |
| 561 |
|
| 562 |
/* prepend it to new target's in-chain */ |
| 563 |
a->inchain = newto->ins; |
| 564 |
a->inchainRev = NULL; |
| 565 |
if (newto->ins) |
| 566 |
newto->ins->inchainRev = a; |
| 567 |
newto->ins = a; |
| 568 |
newto->nins++; |
| 569 |
} |
| 570 |
|
| 571 |
/* |
| 572 |
* hasnonemptyout - Does state have a non-EMPTY out arc? |
| 573 |
*/ |
| 574 |
static int |
| 575 |
hasnonemptyout(struct state *s) |
| 576 |
{ |
| 577 |
struct arc *a; |
| 578 |
|
| 579 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 580 |
{ |
| 581 |
if (a->type != EMPTY) |
| 582 |
return 1; |
| 583 |
} |
| 584 |
return 0; |
| 585 |
} |
| 586 |
|
| 587 |
/* |
| 588 |
* findarc - find arc, if any, from given source with given type and color |
| 589 |
* If there is more than one such arc, the result is random. |
| 590 |
*/ |
| 591 |
static struct arc * |
| 592 |
findarc(struct state *s, |
| 593 |
int type, |
| 594 |
color co) |
| 595 |
{ |
| 596 |
struct arc *a; |
| 597 |
|
| 598 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 599 |
if (a->type == type && a->co == co) |
| 600 |
return a; |
| 601 |
return NULL; |
| 602 |
} |
| 603 |
|
| 604 |
/* |
| 605 |
* cparc - allocate a new arc within an NFA, copying details from old one |
| 606 |
*/ |
| 607 |
static void |
| 608 |
cparc(struct nfa *nfa, |
| 609 |
struct arc *oa, |
| 610 |
struct state *from, |
| 611 |
struct state *to) |
| 612 |
{ |
| 613 |
newarc(nfa, oa->type, oa->co, from, to); |
| 614 |
} |
| 615 |
|
| 616 |
/* |
| 617 |
* sortins - sort the in arcs of a state by from/color/type |
| 618 |
*/ |
| 619 |
static void |
| 620 |
sortins(struct nfa *nfa, |
| 621 |
struct state *s) |
| 622 |
{ |
| 623 |
struct arc **sortarray; |
| 624 |
struct arc *a; |
| 625 |
int n = s->nins; |
| 626 |
int i; |
| 627 |
|
| 628 |
if (n <= 1) |
| 629 |
return; /* nothing to do */ |
| 630 |
/* make an array of arc pointers ... */ |
| 631 |
sortarray = (struct arc **) MALLOC(n * sizeof(struct arc *)); |
| 632 |
if (sortarray == NULL) |
| 633 |
{ |
| 634 |
NERR(REG_ESPACE); |
| 635 |
return; |
| 636 |
} |
| 637 |
i = 0; |
| 638 |
for (a = s->ins; a != NULL; a = a->inchain) |
| 639 |
sortarray[i++] = a; |
| 640 |
assert(i == n); |
| 641 |
/* ... sort the array */ |
| 642 |
qsort(sortarray, n, sizeof(struct arc *), sortins_cmp); |
| 643 |
/* ... and rebuild arc list in order */ |
| 644 |
/* it seems worth special-casing first and last items to simplify loop */ |
| 645 |
a = sortarray[0]; |
| 646 |
s->ins = a; |
| 647 |
a->inchain = sortarray[1]; |
| 648 |
a->inchainRev = NULL; |
| 649 |
for (i = 1; i < n - 1; i++) |
| 650 |
{ |
| 651 |
a = sortarray[i]; |
| 652 |
a->inchain = sortarray[i + 1]; |
| 653 |
a->inchainRev = sortarray[i - 1]; |
| 654 |
} |
| 655 |
a = sortarray[i]; |
| 656 |
a->inchain = NULL; |
| 657 |
a->inchainRev = sortarray[i - 1]; |
| 658 |
FREE(sortarray); |
| 659 |
} |
| 660 |
|
| 661 |
static int |
| 662 |
sortins_cmp(const void *a, const void *b) |
| 663 |
{ |
| 664 |
const struct arc *aa = *((const struct arc *const *) a); |
| 665 |
const struct arc *bb = *((const struct arc *const *) b); |
| 666 |
|
| 667 |
/* we check the fields in the order they are most likely to be different */ |
| 668 |
if (aa->from->no < bb->from->no) |
| 669 |
return -1; |
| 670 |
if (aa->from->no > bb->from->no) |
| 671 |
return 1; |
| 672 |
if (aa->co < bb->co) |
| 673 |
return -1; |
| 674 |
if (aa->co > bb->co) |
| 675 |
return 1; |
| 676 |
if (aa->type < bb->type) |
| 677 |
return -1; |
| 678 |
if (aa->type > bb->type) |
| 679 |
return 1; |
| 680 |
return 0; |
| 681 |
} |
| 682 |
|
| 683 |
/* |
| 684 |
* sortouts - sort the out arcs of a state by to/color/type |
| 685 |
*/ |
| 686 |
static void |
| 687 |
sortouts(struct nfa *nfa, |
| 688 |
struct state *s) |
| 689 |
{ |
| 690 |
struct arc **sortarray; |
| 691 |
struct arc *a; |
| 692 |
int n = s->nouts; |
| 693 |
int i; |
| 694 |
|
| 695 |
if (n <= 1) |
| 696 |
return; /* nothing to do */ |
| 697 |
/* make an array of arc pointers ... */ |
| 698 |
sortarray = (struct arc **) MALLOC(n * sizeof(struct arc *)); |
| 699 |
if (sortarray == NULL) |
| 700 |
{ |
| 701 |
NERR(REG_ESPACE); |
| 702 |
return; |
| 703 |
} |
| 704 |
i = 0; |
| 705 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 706 |
sortarray[i++] = a; |
| 707 |
assert(i == n); |
| 708 |
/* ... sort the array */ |
| 709 |
qsort(sortarray, n, sizeof(struct arc *), sortouts_cmp); |
| 710 |
/* ... and rebuild arc list in order */ |
| 711 |
/* it seems worth special-casing first and last items to simplify loop */ |
| 712 |
a = sortarray[0]; |
| 713 |
s->outs = a; |
| 714 |
a->outchain = sortarray[1]; |
| 715 |
a->outchainRev = NULL; |
| 716 |
for (i = 1; i < n - 1; i++) |
| 717 |
{ |
| 718 |
a = sortarray[i]; |
| 719 |
a->outchain = sortarray[i + 1]; |
| 720 |
a->outchainRev = sortarray[i - 1]; |
| 721 |
} |
| 722 |
a = sortarray[i]; |
| 723 |
a->outchain = NULL; |
| 724 |
a->outchainRev = sortarray[i - 1]; |
| 725 |
FREE(sortarray); |
| 726 |
} |
| 727 |
|
| 728 |
static int |
| 729 |
sortouts_cmp(const void *a, const void *b) |
| 730 |
{ |
| 731 |
const struct arc *aa = *((const struct arc *const *) a); |
| 732 |
const struct arc *bb = *((const struct arc *const *) b); |
| 733 |
|
| 734 |
/* we check the fields in the order they are most likely to be different */ |
| 735 |
if (aa->to->no < bb->to->no) |
| 736 |
return -1; |
| 737 |
if (aa->to->no > bb->to->no) |
| 738 |
return 1; |
| 739 |
if (aa->co < bb->co) |
| 740 |
return -1; |
| 741 |
if (aa->co > bb->co) |
| 742 |
return 1; |
| 743 |
if (aa->type < bb->type) |
| 744 |
return -1; |
| 745 |
if (aa->type > bb->type) |
| 746 |
return 1; |
| 747 |
return 0; |
| 748 |
} |
| 749 |
|
| 750 |
/* |
| 751 |
* Common decision logic about whether to use arc-by-arc operations or |
| 752 |
* sort/merge. If there's just a few source arcs we cannot recoup the |
| 753 |
* cost of sorting the destination arc list, no matter how large it is. |
| 754 |
* Otherwise, limit the number of arc-by-arc comparisons to about 1000 |
| 755 |
* (a somewhat arbitrary choice, but the breakeven point would probably |
| 756 |
* be machine dependent anyway). |
| 757 |
*/ |
| 758 |
#define BULK_ARC_OP_USE_SORT(nsrcarcs, ndestarcs) \ |
| 759 |
((nsrcarcs) < 4 ? 0 : ((nsrcarcs) > 32 || (ndestarcs) > 32)) |
| 760 |
|
| 761 |
/* |
| 762 |
* moveins - move all in arcs of a state to another state |
| 763 |
* |
| 764 |
* You might think this could be done better by just updating the |
| 765 |
* existing arcs, and you would be right if it weren't for the need |
| 766 |
* for duplicate suppression, which makes it easier to just make new |
| 767 |
* ones to exploit the suppression built into newarc. |
| 768 |
* |
| 769 |
* However, if we have a whole lot of arcs to deal with, retail duplicate |
| 770 |
* checks become too slow. In that case we proceed by sorting and merging |
| 771 |
* the arc lists, and then we can indeed just update the arcs in-place. |
| 772 |
* |
| 773 |
* On the other hand, it's also true that this is frequently called with |
| 774 |
* a brand-new newState that has no existing in-arcs. In that case, |
| 775 |
* de-duplication is unnecessary, so we can just blindly move all the arcs. |
| 776 |
*/ |
| 777 |
static void |
| 778 |
moveins(struct nfa *nfa, |
| 779 |
struct state *oldState, |
| 780 |
struct state *newState) |
| 781 |
{ |
| 782 |
assert(oldState != newState); |
| 783 |
|
| 784 |
if (newState->nins == 0) |
| 785 |
{ |
| 786 |
/* No need for de-duplication */ |
| 787 |
struct arc *a; |
| 788 |
|
| 789 |
while ((a = oldState->ins) != NULL) |
| 790 |
{ |
| 791 |
createarc(nfa, a->type, a->co, a->from, newState); |
| 792 |
freearc(nfa, a); |
| 793 |
} |
| 794 |
} |
| 795 |
else if (!BULK_ARC_OP_USE_SORT(oldState->nins, newState->nins)) |
| 796 |
{ |
| 797 |
/* With not too many arcs, just do them one at a time */ |
| 798 |
struct arc *a; |
| 799 |
|
| 800 |
while ((a = oldState->ins) != NULL) |
| 801 |
{ |
| 802 |
cparc(nfa, a, a->from, newState); |
| 803 |
freearc(nfa, a); |
| 804 |
} |
| 805 |
} |
| 806 |
else |
| 807 |
{ |
| 808 |
/* |
| 809 |
* With many arcs, use a sort-merge approach. Note changearctarget() |
| 810 |
* will put the arc onto the front of newState's chain, so it does not |
| 811 |
* break our walk through the sorted part of the chain. |
| 812 |
*/ |
| 813 |
struct arc *oa; |
| 814 |
struct arc *na; |
| 815 |
|
| 816 |
/* |
| 817 |
* Because we bypass newarc() in this code path, we'd better include a |
| 818 |
* cancel check. |
| 819 |
*/ |
| 820 |
INTERRUPT(nfa->v->re); |
| 821 |
|
| 822 |
sortins(nfa, oldState); |
| 823 |
sortins(nfa, newState); |
| 824 |
if (NISERR()) |
| 825 |
return; /* might have failed to sort */ |
| 826 |
oa = oldState->ins; |
| 827 |
na = newState->ins; |
| 828 |
while (oa != NULL && na != NULL) |
| 829 |
{ |
| 830 |
struct arc *a = oa; |
| 831 |
|
| 832 |
switch (sortins_cmp(&oa, &na)) |
| 833 |
{ |
| 834 |
case -1: |
| 835 |
/* newState does not have anything matching oa */ |
| 836 |
oa = oa->inchain; |
| 837 |
|
| 838 |
/* |
| 839 |
* Rather than doing createarc+freearc, we can just unlink |
| 840 |
* and relink the existing arc struct. |
| 841 |
*/ |
| 842 |
changearctarget(a, newState); |
| 843 |
break; |
| 844 |
case 0: |
| 845 |
/* match, advance in both lists */ |
| 846 |
oa = oa->inchain; |
| 847 |
na = na->inchain; |
| 848 |
/* ... and drop duplicate arc from oldState */ |
| 849 |
freearc(nfa, a); |
| 850 |
break; |
| 851 |
case +1: |
| 852 |
/* advance only na; oa might have a match later */ |
| 853 |
na = na->inchain; |
| 854 |
break; |
| 855 |
default: |
| 856 |
assert(NOTREACHED); |
| 857 |
} |
| 858 |
} |
| 859 |
while (oa != NULL) |
| 860 |
{ |
| 861 |
/* newState does not have anything matching oa */ |
| 862 |
struct arc *a = oa; |
| 863 |
|
| 864 |
oa = oa->inchain; |
| 865 |
changearctarget(a, newState); |
| 866 |
} |
| 867 |
} |
| 868 |
|
| 869 |
assert(oldState->nins == 0); |
| 870 |
assert(oldState->ins == NULL); |
| 871 |
} |
| 872 |
|
| 873 |
/* |
| 874 |
* copyins - copy in arcs of a state to another state |
| 875 |
* |
| 876 |
* The comments for moveins() apply here as well. However, in current |
| 877 |
* usage, this is *only* called with brand-new target states, so that |
| 878 |
* only the "no need for de-duplication" code path is ever reached. |
| 879 |
* We keep the rest #ifdef'd out in case it's needed in the future. |
| 880 |
*/ |
| 881 |
static void |
| 882 |
copyins(struct nfa *nfa, |
| 883 |
struct state *oldState, |
| 884 |
struct state *newState) |
| 885 |
{ |
| 886 |
assert(oldState != newState); |
| 887 |
assert(newState->nins == 0); /* see comment above */ |
| 888 |
|
| 889 |
if (newState->nins == 0) |
| 890 |
{ |
| 891 |
/* No need for de-duplication */ |
| 892 |
struct arc *a; |
| 893 |
|
| 894 |
for (a = oldState->ins; a != NULL; a = a->inchain) |
| 895 |
createarc(nfa, a->type, a->co, a->from, newState); |
| 896 |
} |
| 897 |
#ifdef NOT_USED /* see comment above */ |
| 898 |
else if (!BULK_ARC_OP_USE_SORT(oldState->nins, newState->nins)) |
| 899 |
{ |
| 900 |
/* With not too many arcs, just do them one at a time */ |
| 901 |
struct arc *a; |
| 902 |
|
| 903 |
for (a = oldState->ins; a != NULL; a = a->inchain) |
| 904 |
cparc(nfa, a, a->from, newState); |
| 905 |
} |
| 906 |
else |
| 907 |
{ |
| 908 |
/* |
| 909 |
* With many arcs, use a sort-merge approach. Note that createarc() |
| 910 |
* will put new arcs onto the front of newState's chain, so it does |
| 911 |
* not break our walk through the sorted part of the chain. |
| 912 |
*/ |
| 913 |
struct arc *oa; |
| 914 |
struct arc *na; |
| 915 |
|
| 916 |
/* |
| 917 |
* Because we bypass newarc() in this code path, we'd better include a |
| 918 |
* cancel check. |
| 919 |
*/ |
| 920 |
INTERRUPT(nfa->v->re); |
| 921 |
|
| 922 |
sortins(nfa, oldState); |
| 923 |
sortins(nfa, newState); |
| 924 |
if (NISERR()) |
| 925 |
return; /* might have failed to sort */ |
| 926 |
oa = oldState->ins; |
| 927 |
na = newState->ins; |
| 928 |
while (oa != NULL && na != NULL) |
| 929 |
{ |
| 930 |
struct arc *a = oa; |
| 931 |
|
| 932 |
switch (sortins_cmp(&oa, &na)) |
| 933 |
{ |
| 934 |
case -1: |
| 935 |
/* newState does not have anything matching oa */ |
| 936 |
oa = oa->inchain; |
| 937 |
createarc(nfa, a->type, a->co, a->from, newState); |
| 938 |
break; |
| 939 |
case 0: |
| 940 |
/* match, advance in both lists */ |
| 941 |
oa = oa->inchain; |
| 942 |
na = na->inchain; |
| 943 |
break; |
| 944 |
case +1: |
| 945 |
/* advance only na; oa might have a match later */ |
| 946 |
na = na->inchain; |
| 947 |
break; |
| 948 |
default: |
| 949 |
assert(NOTREACHED); |
| 950 |
} |
| 951 |
} |
| 952 |
while (oa != NULL) |
| 953 |
{ |
| 954 |
/* newState does not have anything matching oa */ |
| 955 |
struct arc *a = oa; |
| 956 |
|
| 957 |
oa = oa->inchain; |
| 958 |
createarc(nfa, a->type, a->co, a->from, newState); |
| 959 |
} |
| 960 |
} |
| 961 |
#endif /* NOT_USED */ |
| 962 |
} |
| 963 |
|
| 964 |
/* |
| 965 |
* mergeins - merge a list of inarcs into a state |
| 966 |
* |
| 967 |
* This is much like copyins, but the source arcs are listed in an array, |
| 968 |
* and are not guaranteed unique. It's okay to clobber the array contents. |
| 969 |
*/ |
| 970 |
static void |
| 971 |
mergeins(struct nfa *nfa, |
| 972 |
struct state *s, |
| 973 |
struct arc **arcarray, |
| 974 |
int arccount) |
| 975 |
{ |
| 976 |
struct arc *na; |
| 977 |
int i; |
| 978 |
int j; |
| 979 |
|
| 980 |
if (arccount <= 0) |
| 981 |
return; |
| 982 |
|
| 983 |
/* |
| 984 |
* Because we bypass newarc() in this code path, we'd better include a |
| 985 |
* cancel check. |
| 986 |
*/ |
| 987 |
INTERRUPT(nfa->v->re); |
| 988 |
|
| 989 |
/* Sort existing inarcs as well as proposed new ones */ |
| 990 |
sortins(nfa, s); |
| 991 |
if (NISERR()) |
| 992 |
return; /* might have failed to sort */ |
| 993 |
|
| 994 |
qsort(arcarray, arccount, sizeof(struct arc *), sortins_cmp); |
| 995 |
|
| 996 |
/* |
| 997 |
* arcarray very likely includes dups, so we must eliminate them. (This |
| 998 |
* could be folded into the next loop, but it's not worth the trouble.) |
| 999 |
*/ |
| 1000 |
j = 0; |
| 1001 |
for (i = 1; i < arccount; i++) |
| 1002 |
{ |
| 1003 |
switch (sortins_cmp(&arcarray[j], &arcarray[i])) |
| 1004 |
{ |
| 1005 |
case -1: |
| 1006 |
/* non-dup */ |
| 1007 |
arcarray[++j] = arcarray[i]; |
| 1008 |
break; |
| 1009 |
case 0: |
| 1010 |
/* dup */ |
| 1011 |
break; |
| 1012 |
default: |
| 1013 |
/* trouble */ |
| 1014 |
assert(NOTREACHED); |
| 1015 |
} |
| 1016 |
} |
| 1017 |
arccount = j + 1; |
| 1018 |
|
| 1019 |
/* |
| 1020 |
* Now merge into s' inchain. Note that createarc() will put new arcs |
| 1021 |
* onto the front of s's chain, so it does not break our walk through the |
| 1022 |
* sorted part of the chain. |
| 1023 |
*/ |
| 1024 |
i = 0; |
| 1025 |
na = s->ins; |
| 1026 |
while (i < arccount && na != NULL) |
| 1027 |
{ |
| 1028 |
struct arc *a = arcarray[i]; |
| 1029 |
|
| 1030 |
switch (sortins_cmp(&a, &na)) |
| 1031 |
{ |
| 1032 |
case -1: |
| 1033 |
/* s does not have anything matching a */ |
| 1034 |
createarc(nfa, a->type, a->co, a->from, s); |
| 1035 |
i++; |
| 1036 |
break; |
| 1037 |
case 0: |
| 1038 |
/* match, advance in both lists */ |
| 1039 |
i++; |
| 1040 |
na = na->inchain; |
| 1041 |
break; |
| 1042 |
case +1: |
| 1043 |
/* advance only na; array might have a match later */ |
| 1044 |
na = na->inchain; |
| 1045 |
break; |
| 1046 |
default: |
| 1047 |
assert(NOTREACHED); |
| 1048 |
} |
| 1049 |
} |
| 1050 |
while (i < arccount) |
| 1051 |
{ |
| 1052 |
/* s does not have anything matching a */ |
| 1053 |
struct arc *a = arcarray[i]; |
| 1054 |
|
| 1055 |
createarc(nfa, a->type, a->co, a->from, s); |
| 1056 |
i++; |
| 1057 |
} |
| 1058 |
} |
| 1059 |
|
| 1060 |
/* |
| 1061 |
* moveouts - move all out arcs of a state to another state |
| 1062 |
* |
| 1063 |
* See comments for moveins() |
| 1064 |
*/ |
| 1065 |
static void |
| 1066 |
moveouts(struct nfa *nfa, |
| 1067 |
struct state *oldState, |
| 1068 |
struct state *newState) |
| 1069 |
{ |
| 1070 |
assert(oldState != newState); |
| 1071 |
|
| 1072 |
if (newState->nouts == 0) |
| 1073 |
{ |
| 1074 |
/* No need for de-duplication */ |
| 1075 |
struct arc *a; |
| 1076 |
|
| 1077 |
while ((a = oldState->outs) != NULL) |
| 1078 |
{ |
| 1079 |
createarc(nfa, a->type, a->co, newState, a->to); |
| 1080 |
freearc(nfa, a); |
| 1081 |
} |
| 1082 |
} |
| 1083 |
else if (!BULK_ARC_OP_USE_SORT(oldState->nouts, newState->nouts)) |
| 1084 |
{ |
| 1085 |
/* With not too many arcs, just do them one at a time */ |
| 1086 |
struct arc *a; |
| 1087 |
|
| 1088 |
while ((a = oldState->outs) != NULL) |
| 1089 |
{ |
| 1090 |
cparc(nfa, a, newState, a->to); |
| 1091 |
freearc(nfa, a); |
| 1092 |
} |
| 1093 |
} |
| 1094 |
else |
| 1095 |
{ |
| 1096 |
/* |
| 1097 |
* With many arcs, use a sort-merge approach. Note changearcsource() |
| 1098 |
* will put the arc onto the front of newState's chain, so it does not |
| 1099 |
* break our walk through the sorted part of the chain. |
| 1100 |
*/ |
| 1101 |
struct arc *oa; |
| 1102 |
struct arc *na; |
| 1103 |
|
| 1104 |
/* |
| 1105 |
* Because we bypass newarc() in this code path, we'd better include a |
| 1106 |
* cancel check. |
| 1107 |
*/ |
| 1108 |
INTERRUPT(nfa->v->re); |
| 1109 |
|
| 1110 |
sortouts(nfa, oldState); |
| 1111 |
sortouts(nfa, newState); |
| 1112 |
if (NISERR()) |
| 1113 |
return; /* might have failed to sort */ |
| 1114 |
oa = oldState->outs; |
| 1115 |
na = newState->outs; |
| 1116 |
while (oa != NULL && na != NULL) |
| 1117 |
{ |
| 1118 |
struct arc *a = oa; |
| 1119 |
|
| 1120 |
switch (sortouts_cmp(&oa, &na)) |
| 1121 |
{ |
| 1122 |
case -1: |
| 1123 |
/* newState does not have anything matching oa */ |
| 1124 |
oa = oa->outchain; |
| 1125 |
|
| 1126 |
/* |
| 1127 |
* Rather than doing createarc+freearc, we can just unlink |
| 1128 |
* and relink the existing arc struct. |
| 1129 |
*/ |
| 1130 |
changearcsource(a, newState); |
| 1131 |
break; |
| 1132 |
case 0: |
| 1133 |
/* match, advance in both lists */ |
| 1134 |
oa = oa->outchain; |
| 1135 |
na = na->outchain; |
| 1136 |
/* ... and drop duplicate arc from oldState */ |
| 1137 |
freearc(nfa, a); |
| 1138 |
break; |
| 1139 |
case +1: |
| 1140 |
/* advance only na; oa might have a match later */ |
| 1141 |
na = na->outchain; |
| 1142 |
break; |
| 1143 |
default: |
| 1144 |
assert(NOTREACHED); |
| 1145 |
} |
| 1146 |
} |
| 1147 |
while (oa != NULL) |
| 1148 |
{ |
| 1149 |
/* newState does not have anything matching oa */ |
| 1150 |
struct arc *a = oa; |
| 1151 |
|
| 1152 |
oa = oa->outchain; |
| 1153 |
changearcsource(a, newState); |
| 1154 |
} |
| 1155 |
} |
| 1156 |
|
| 1157 |
assert(oldState->nouts == 0); |
| 1158 |
assert(oldState->outs == NULL); |
| 1159 |
} |
| 1160 |
|
| 1161 |
/* |
| 1162 |
* copyouts - copy out arcs of a state to another state |
| 1163 |
* |
| 1164 |
* See comments for copyins() |
| 1165 |
*/ |
| 1166 |
static void |
| 1167 |
copyouts(struct nfa *nfa, |
| 1168 |
struct state *oldState, |
| 1169 |
struct state *newState) |
| 1170 |
{ |
| 1171 |
assert(oldState != newState); |
| 1172 |
assert(newState->nouts == 0); /* see comment above */ |
| 1173 |
|
| 1174 |
if (newState->nouts == 0) |
| 1175 |
{ |
| 1176 |
/* No need for de-duplication */ |
| 1177 |
struct arc *a; |
| 1178 |
|
| 1179 |
for (a = oldState->outs; a != NULL; a = a->outchain) |
| 1180 |
createarc(nfa, a->type, a->co, newState, a->to); |
| 1181 |
} |
| 1182 |
#ifdef NOT_USED /* see comment above */ |
| 1183 |
else if (!BULK_ARC_OP_USE_SORT(oldState->nouts, newState->nouts)) |
| 1184 |
{ |
| 1185 |
/* With not too many arcs, just do them one at a time */ |
| 1186 |
struct arc *a; |
| 1187 |
|
| 1188 |
for (a = oldState->outs; a != NULL; a = a->outchain) |
| 1189 |
cparc(nfa, a, newState, a->to); |
| 1190 |
} |
| 1191 |
else |
| 1192 |
{ |
| 1193 |
/* |
| 1194 |
* With many arcs, use a sort-merge approach. Note that createarc() |
| 1195 |
* will put new arcs onto the front of newState's chain, so it does |
| 1196 |
* not break our walk through the sorted part of the chain. |
| 1197 |
*/ |
| 1198 |
struct arc *oa; |
| 1199 |
struct arc *na; |
| 1200 |
|
| 1201 |
/* |
| 1202 |
* Because we bypass newarc() in this code path, we'd better include a |
| 1203 |
* cancel check. |
| 1204 |
*/ |
| 1205 |
INTERRUPT(nfa->v->re); |
| 1206 |
|
| 1207 |
sortouts(nfa, oldState); |
| 1208 |
sortouts(nfa, newState); |
| 1209 |
if (NISERR()) |
| 1210 |
return; /* might have failed to sort */ |
| 1211 |
oa = oldState->outs; |
| 1212 |
na = newState->outs; |
| 1213 |
while (oa != NULL && na != NULL) |
| 1214 |
{ |
| 1215 |
struct arc *a = oa; |
| 1216 |
|
| 1217 |
switch (sortouts_cmp(&oa, &na)) |
| 1218 |
{ |
| 1219 |
case -1: |
| 1220 |
/* newState does not have anything matching oa */ |
| 1221 |
oa = oa->outchain; |
| 1222 |
createarc(nfa, a->type, a->co, newState, a->to); |
| 1223 |
break; |
| 1224 |
case 0: |
| 1225 |
/* match, advance in both lists */ |
| 1226 |
oa = oa->outchain; |
| 1227 |
na = na->outchain; |
| 1228 |
break; |
| 1229 |
case +1: |
| 1230 |
/* advance only na; oa might have a match later */ |
| 1231 |
na = na->outchain; |
| 1232 |
break; |
| 1233 |
default: |
| 1234 |
assert(NOTREACHED); |
| 1235 |
} |
| 1236 |
} |
| 1237 |
while (oa != NULL) |
| 1238 |
{ |
| 1239 |
/* newState does not have anything matching oa */ |
| 1240 |
struct arc *a = oa; |
| 1241 |
|
| 1242 |
oa = oa->outchain; |
| 1243 |
createarc(nfa, a->type, a->co, newState, a->to); |
| 1244 |
} |
| 1245 |
} |
| 1246 |
#endif /* NOT_USED */ |
| 1247 |
} |
| 1248 |
|
| 1249 |
/* |
| 1250 |
* cloneouts - copy out arcs of a state to another state pair, modifying type |
| 1251 |
* |
| 1252 |
* This is only used to convert PLAIN arcs to AHEAD/BEHIND arcs, which share |
| 1253 |
* the same interpretation of "co". It wouldn't be sensible with LACONs. |
| 1254 |
*/ |
| 1255 |
static void |
| 1256 |
cloneouts(struct nfa *nfa, |
| 1257 |
struct state *old, |
| 1258 |
struct state *from, |
| 1259 |
struct state *to, |
| 1260 |
int type) |
| 1261 |
{ |
| 1262 |
struct arc *a; |
| 1263 |
|
| 1264 |
assert(old != from); |
| 1265 |
assert(type == AHEAD || type == BEHIND); |
| 1266 |
|
| 1267 |
for (a = old->outs; a != NULL; a = a->outchain) |
| 1268 |
{ |
| 1269 |
assert(a->type == PLAIN); |
| 1270 |
newarc(nfa, type, a->co, from, to); |
| 1271 |
} |
| 1272 |
} |
| 1273 |
|
| 1274 |
/* |
| 1275 |
* delsub - delete a sub-NFA, updating subre pointers if necessary |
| 1276 |
* |
| 1277 |
* This uses a recursive traversal of the sub-NFA, marking already-seen |
| 1278 |
* states using their tmp pointer. |
| 1279 |
*/ |
| 1280 |
static void |
| 1281 |
delsub(struct nfa *nfa, |
| 1282 |
struct state *lp, /* the sub-NFA goes from here... */ |
| 1283 |
struct state *rp) /* ...to here, *not* inclusive */ |
| 1284 |
{ |
| 1285 |
assert(lp != rp); |
| 1286 |
|
| 1287 |
rp->tmp = rp; /* mark end */ |
| 1288 |
|
| 1289 |
deltraverse(nfa, lp, lp); |
| 1290 |
if (NISERR()) |
| 1291 |
return; /* asserts might not hold after failure */ |
| 1292 |
assert(lp->nouts == 0 && rp->nins == 0); /* did the job */ |
| 1293 |
assert(lp->no != FREESTATE && rp->no != FREESTATE); /* no more */ |
| 1294 |
|
| 1295 |
rp->tmp = NULL; /* unmark end */ |
| 1296 |
lp->tmp = NULL; /* and begin, marked by deltraverse */ |
| 1297 |
} |
| 1298 |
|
| 1299 |
/* |
| 1300 |
* deltraverse - the recursive heart of delsub |
| 1301 |
* This routine's basic job is to destroy all out-arcs of the state. |
| 1302 |
*/ |
| 1303 |
static void |
| 1304 |
deltraverse(struct nfa *nfa, |
| 1305 |
struct state *leftend, |
| 1306 |
struct state *s) |
| 1307 |
{ |
| 1308 |
struct arc *a; |
| 1309 |
struct state *to; |
| 1310 |
|
| 1311 |
/* Since this is recursive, it could be driven to stack overflow */ |
| 1312 |
if (STACK_TOO_DEEP(nfa->v->re)) |
| 1313 |
{ |
| 1314 |
NERR(REG_ETOOBIG); |
| 1315 |
return; |
| 1316 |
} |
| 1317 |
|
| 1318 |
if (s->nouts == 0) |
| 1319 |
return; /* nothing to do */ |
| 1320 |
if (s->tmp != NULL) |
| 1321 |
return; /* already in progress */ |
| 1322 |
|
| 1323 |
s->tmp = s; /* mark as in progress */ |
| 1324 |
|
| 1325 |
while ((a = s->outs) != NULL) |
| 1326 |
{ |
| 1327 |
to = a->to; |
| 1328 |
deltraverse(nfa, leftend, to); |
| 1329 |
if (NISERR()) |
| 1330 |
return; /* asserts might not hold after failure */ |
| 1331 |
assert(to->nouts == 0 || to->tmp != NULL); |
| 1332 |
freearc(nfa, a); |
| 1333 |
if (to->nins == 0 && to->tmp == NULL) |
| 1334 |
{ |
| 1335 |
assert(to->nouts == 0); |
| 1336 |
freestate(nfa, to); |
| 1337 |
} |
| 1338 |
} |
| 1339 |
|
| 1340 |
assert(s->no != FREESTATE); /* we're still here */ |
| 1341 |
assert(s == leftend || s->nins != 0); /* and still reachable */ |
| 1342 |
assert(s->nouts == 0); /* but have no outarcs */ |
| 1343 |
|
| 1344 |
s->tmp = NULL; /* we're done here */ |
| 1345 |
} |
| 1346 |
|
| 1347 |
/* |
| 1348 |
* dupnfa - duplicate sub-NFA |
| 1349 |
* |
| 1350 |
* Another recursive traversal, this time using tmp to point to duplicates |
| 1351 |
* as well as mark already-seen states. (You knew there was a reason why |
| 1352 |
* it's a state pointer, didn't you? :-)) |
| 1353 |
*/ |
| 1354 |
static void |
| 1355 |
dupnfa(struct nfa *nfa, |
| 1356 |
struct state *start, /* duplicate of subNFA starting here */ |
| 1357 |
struct state *stop, /* and stopping here */ |
| 1358 |
struct state *from, /* stringing duplicate from here */ |
| 1359 |
struct state *to) /* to here */ |
| 1360 |
{ |
| 1361 |
if (start == stop) |
| 1362 |
{ |
| 1363 |
newarc(nfa, EMPTY, 0, from, to); |
| 1364 |
return; |
| 1365 |
} |
| 1366 |
|
| 1367 |
stop->tmp = to; |
| 1368 |
duptraverse(nfa, start, from); |
| 1369 |
/* done, except for clearing out the tmp pointers */ |
| 1370 |
|
| 1371 |
stop->tmp = NULL; |
| 1372 |
cleartraverse(nfa, start); |
| 1373 |
} |
| 1374 |
|
| 1375 |
/* |
| 1376 |
* duptraverse - recursive heart of dupnfa |
| 1377 |
*/ |
| 1378 |
static void |
| 1379 |
duptraverse(struct nfa *nfa, |
| 1380 |
struct state *s, |
| 1381 |
struct state *stmp) /* s's duplicate, or NULL */ |
| 1382 |
{ |
| 1383 |
struct arc *a; |
| 1384 |
|
| 1385 |
/* Since this is recursive, it could be driven to stack overflow */ |
| 1386 |
if (STACK_TOO_DEEP(nfa->v->re)) |
| 1387 |
{ |
| 1388 |
NERR(REG_ETOOBIG); |
| 1389 |
return; |
| 1390 |
} |
| 1391 |
|
| 1392 |
if (s->tmp != NULL) |
| 1393 |
return; /* already done */ |
| 1394 |
|
| 1395 |
s->tmp = (stmp == NULL) ? newstate(nfa) : stmp; |
| 1396 |
if (s->tmp == NULL) |
| 1397 |
{ |
| 1398 |
assert(NISERR()); |
| 1399 |
return; |
| 1400 |
} |
| 1401 |
|
| 1402 |
for (a = s->outs; a != NULL && !NISERR(); a = a->outchain) |
| 1403 |
{ |
| 1404 |
duptraverse(nfa, a->to, (struct state *) NULL); |
| 1405 |
if (NISERR()) |
| 1406 |
break; |
| 1407 |
assert(a->to->tmp != NULL); |
| 1408 |
cparc(nfa, a, s->tmp, a->to->tmp); |
| 1409 |
} |
| 1410 |
} |
| 1411 |
|
| 1412 |
/* |
| 1413 |
* removeconstraints - remove any constraints in an NFA |
| 1414 |
* |
| 1415 |
* Constraint arcs are replaced by empty arcs, essentially treating all |
| 1416 |
* constraints as automatically satisfied. |
| 1417 |
*/ |
| 1418 |
static void |
| 1419 |
removeconstraints(struct nfa *nfa, |
| 1420 |
struct state *start, /* process subNFA starting here */ |
| 1421 |
struct state *stop) /* and stopping here */ |
| 1422 |
{ |
| 1423 |
if (start == stop) |
| 1424 |
return; |
| 1425 |
|
| 1426 |
stop->tmp = stop; |
| 1427 |
removetraverse(nfa, start); |
| 1428 |
/* done, except for clearing out the tmp pointers */ |
| 1429 |
|
| 1430 |
stop->tmp = NULL; |
| 1431 |
cleartraverse(nfa, start); |
| 1432 |
} |
| 1433 |
|
| 1434 |
/* |
| 1435 |
* removetraverse - recursive heart of removeconstraints |
| 1436 |
*/ |
| 1437 |
static void |
| 1438 |
removetraverse(struct nfa *nfa, |
| 1439 |
struct state *s) |
| 1440 |
{ |
| 1441 |
struct arc *a; |
| 1442 |
struct arc *oa; |
| 1443 |
|
| 1444 |
/* Since this is recursive, it could be driven to stack overflow */ |
| 1445 |
if (STACK_TOO_DEEP(nfa->v->re)) |
| 1446 |
{ |
| 1447 |
NERR(REG_ETOOBIG); |
| 1448 |
return; |
| 1449 |
} |
| 1450 |
|
| 1451 |
if (s->tmp != NULL) |
| 1452 |
return; /* already done */ |
| 1453 |
|
| 1454 |
s->tmp = s; |
| 1455 |
for (a = s->outs; a != NULL && !NISERR(); a = oa) |
| 1456 |
{ |
| 1457 |
removetraverse(nfa, a->to); |
| 1458 |
if (NISERR()) |
| 1459 |
break; |
| 1460 |
oa = a->outchain; |
| 1461 |
switch (a->type) |
| 1462 |
{ |
| 1463 |
case PLAIN: |
| 1464 |
case EMPTY: |
| 1465 |
case CANTMATCH: |
| 1466 |
/* nothing to do */ |
| 1467 |
break; |
| 1468 |
case AHEAD: |
| 1469 |
case BEHIND: |
| 1470 |
case '^': |
| 1471 |
case '$': |
| 1472 |
case LACON: |
| 1473 |
/* replace it */ |
| 1474 |
newarc(nfa, EMPTY, 0, s, a->to); |
| 1475 |
freearc(nfa, a); |
| 1476 |
break; |
| 1477 |
default: |
| 1478 |
NERR(REG_ASSERT); |
| 1479 |
break; |
| 1480 |
} |
| 1481 |
} |
| 1482 |
} |
| 1483 |
|
| 1484 |
/* |
| 1485 |
* cleartraverse - recursive cleanup for algorithms that leave tmp ptrs set |
| 1486 |
*/ |
| 1487 |
static void |
| 1488 |
cleartraverse(struct nfa *nfa, |
| 1489 |
struct state *s) |
| 1490 |
{ |
| 1491 |
struct arc *a; |
| 1492 |
|
| 1493 |
/* Since this is recursive, it could be driven to stack overflow */ |
| 1494 |
if (STACK_TOO_DEEP(nfa->v->re)) |
| 1495 |
{ |
| 1496 |
NERR(REG_ETOOBIG); |
| 1497 |
return; |
| 1498 |
} |
| 1499 |
|
| 1500 |
if (s->tmp == NULL) |
| 1501 |
return; |
| 1502 |
s->tmp = NULL; |
| 1503 |
|
| 1504 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 1505 |
cleartraverse(nfa, a->to); |
| 1506 |
} |
| 1507 |
|
| 1508 |
/* |
| 1509 |
* single_color_transition - does getting from s1 to s2 cross one PLAIN arc? |
| 1510 |
* |
| 1511 |
* If traversing from s1 to s2 requires a single PLAIN match (possibly of any |
| 1512 |
* of a set of colors), return a state whose outarc list contains only PLAIN |
| 1513 |
* arcs of those color(s). Otherwise return NULL. |
| 1514 |
* |
| 1515 |
* This is used before optimizing the NFA, so there may be EMPTY arcs, which |
| 1516 |
* we should ignore; the possibility of an EMPTY is why the result state could |
| 1517 |
* be different from s1. |
| 1518 |
* |
| 1519 |
* It's worth troubling to handle multiple parallel PLAIN arcs here because a |
| 1520 |
* bracket construct such as [abc] might yield either one or several parallel |
| 1521 |
* PLAIN arcs depending on earlier atoms in the expression. We'd rather that |
| 1522 |
* that implementation detail not create user-visible performance differences. |
| 1523 |
*/ |
| 1524 |
static struct state * |
| 1525 |
single_color_transition(struct state *s1, struct state *s2) |
| 1526 |
{ |
| 1527 |
struct arc *a; |
| 1528 |
|
| 1529 |
/* Ignore leading EMPTY arc, if any */ |
| 1530 |
if (s1->nouts == 1 && s1->outs->type == EMPTY) |
| 1531 |
s1 = s1->outs->to; |
| 1532 |
/* Likewise for any trailing EMPTY arc */ |
| 1533 |
if (s2->nins == 1 && s2->ins->type == EMPTY) |
| 1534 |
s2 = s2->ins->from; |
| 1535 |
/* Perhaps we could have a single-state loop in between, if so reject */ |
| 1536 |
if (s1 == s2) |
| 1537 |
return NULL; |
| 1538 |
/* s1 must have at least one outarc... */ |
| 1539 |
if (s1->outs == NULL) |
| 1540 |
return NULL; |
| 1541 |
/* ... and they must all be PLAIN arcs to s2 */ |
| 1542 |
for (a = s1->outs; a != NULL; a = a->outchain) |
| 1543 |
{ |
| 1544 |
if (a->type != PLAIN || a->to != s2) |
| 1545 |
return NULL; |
| 1546 |
} |
| 1547 |
/* OK, return s1 as the possessor of the relevant outarcs */ |
| 1548 |
return s1; |
| 1549 |
} |
| 1550 |
|
| 1551 |
/* |
| 1552 |
* specialcolors - fill in special colors for an NFA |
| 1553 |
*/ |
| 1554 |
static void |
| 1555 |
specialcolors(struct nfa *nfa) |
| 1556 |
{ |
| 1557 |
/* false colors for BOS, BOL, EOS, EOL */ |
| 1558 |
if (nfa->parent == NULL) |
| 1559 |
{ |
| 1560 |
nfa->bos[0] = pseudocolor(nfa->cm); |
| 1561 |
nfa->bos[1] = pseudocolor(nfa->cm); |
| 1562 |
nfa->eos[0] = pseudocolor(nfa->cm); |
| 1563 |
nfa->eos[1] = pseudocolor(nfa->cm); |
| 1564 |
} |
| 1565 |
else |
| 1566 |
{ |
| 1567 |
assert(nfa->parent->bos[0] != COLORLESS); |
| 1568 |
nfa->bos[0] = nfa->parent->bos[0]; |
| 1569 |
assert(nfa->parent->bos[1] != COLORLESS); |
| 1570 |
nfa->bos[1] = nfa->parent->bos[1]; |
| 1571 |
assert(nfa->parent->eos[0] != COLORLESS); |
| 1572 |
nfa->eos[0] = nfa->parent->eos[0]; |
| 1573 |
assert(nfa->parent->eos[1] != COLORLESS); |
| 1574 |
nfa->eos[1] = nfa->parent->eos[1]; |
| 1575 |
} |
| 1576 |
} |
| 1577 |
|
| 1578 |
/* |
| 1579 |
* optimize - optimize an NFA |
| 1580 |
* |
| 1581 |
* The main goal of this function is not so much "optimization" (though it |
| 1582 |
* does try to get rid of useless NFA states) as reducing the NFA to a form |
| 1583 |
* the regex executor can handle. The executor, and indeed the cNFA format |
| 1584 |
* that is its input, can only handle PLAIN and LACON arcs. The output of |
| 1585 |
* the regex parser also includes EMPTY (do-nothing) arcs, as well as |
| 1586 |
* ^, $, AHEAD, and BEHIND constraint arcs, which we must get rid of here. |
| 1587 |
* We first get rid of EMPTY arcs and then deal with the constraint arcs. |
| 1588 |
* The hardest part of either job is to get rid of circular loops of the |
| 1589 |
* target arc type. We would have to do that in any case, though, as such a |
| 1590 |
* loop would otherwise allow the executor to cycle through the loop endlessly |
| 1591 |
* without making any progress in the input string. |
| 1592 |
*/ |
| 1593 |
static long /* re_info bits */ |
| 1594 |
optimize(struct nfa *nfa, |
| 1595 |
FILE *f) /* for debug output; NULL none */ |
| 1596 |
{ |
| 1597 |
#ifdef REG_DEBUG |
| 1598 |
int verbose = (f != NULL) ? 1 : 0; |
| 1599 |
|
| 1600 |
if (verbose) |
| 1601 |
fprintf(f, "\ninitial cleanup:\n"); |
| 1602 |
#endif |
| 1603 |
/* If we have any CANTMATCH arcs, drop them; but this is uncommon */ |
| 1604 |
if (nfa->flags & HASCANTMATCH) |
| 1605 |
{ |
| 1606 |
removecantmatch(nfa); |
| 1607 |
nfa->flags &= ~HASCANTMATCH; |
| 1608 |
} |
| 1609 |
cleanup(nfa); /* may simplify situation */ |
| 1610 |
#ifdef REG_DEBUG |
| 1611 |
if (verbose) |
| 1612 |
dumpnfa(nfa, f); |
| 1613 |
if (verbose) |
| 1614 |
fprintf(f, "\nempties:\n"); |
| 1615 |
#endif |
| 1616 |
fixempties(nfa, f); /* get rid of EMPTY arcs */ |
| 1617 |
#ifdef REG_DEBUG |
| 1618 |
if (verbose) |
| 1619 |
fprintf(f, "\nconstraints:\n"); |
| 1620 |
#endif |
| 1621 |
fixconstraintloops(nfa, f); /* get rid of constraint loops */ |
| 1622 |
pullback(nfa, f); /* pull back constraints backward */ |
| 1623 |
pushfwd(nfa, f); /* push fwd constraints forward */ |
| 1624 |
#ifdef REG_DEBUG |
| 1625 |
if (verbose) |
| 1626 |
fprintf(f, "\nfinal cleanup:\n"); |
| 1627 |
#endif |
| 1628 |
cleanup(nfa); /* final tidying */ |
| 1629 |
#ifdef REG_DEBUG |
| 1630 |
if (verbose) |
| 1631 |
dumpnfa(nfa, f); |
| 1632 |
#endif |
| 1633 |
return analyze(nfa); /* and analysis */ |
| 1634 |
} |
| 1635 |
|
| 1636 |
/* |
| 1637 |
* pullback - pull back constraints backward to eliminate them |
| 1638 |
*/ |
| 1639 |
static void |
| 1640 |
pullback(struct nfa *nfa, |
| 1641 |
FILE *f) /* for debug output; NULL none */ |
| 1642 |
{ |
| 1643 |
struct state *s; |
| 1644 |
struct state *nexts; |
| 1645 |
struct arc *a; |
| 1646 |
struct arc *nexta; |
| 1647 |
struct state *intermediates; |
| 1648 |
int progress; |
| 1649 |
|
| 1650 |
/* find and pull until there are no more */ |
| 1651 |
do |
| 1652 |
{ |
| 1653 |
progress = 0; |
| 1654 |
for (s = nfa->states; s != NULL && !NISERR(); s = nexts) |
| 1655 |
{ |
| 1656 |
nexts = s->next; |
| 1657 |
intermediates = NULL; |
| 1658 |
for (a = s->outs; a != NULL && !NISERR(); a = nexta) |
| 1659 |
{ |
| 1660 |
nexta = a->outchain; |
| 1661 |
if (a->type == '^' || a->type == BEHIND) |
| 1662 |
if (pull(nfa, a, &intermediates)) |
| 1663 |
progress = 1; |
| 1664 |
} |
| 1665 |
/* clear tmp fields of intermediate states created here */ |
| 1666 |
while (intermediates != NULL) |
| 1667 |
{ |
| 1668 |
struct state *ns = intermediates->tmp; |
| 1669 |
|
| 1670 |
intermediates->tmp = NULL; |
| 1671 |
intermediates = ns; |
| 1672 |
} |
| 1673 |
/* if s is now useless, get rid of it */ |
| 1674 |
if ((s->nins == 0 || s->nouts == 0) && !s->flag) |
| 1675 |
dropstate(nfa, s); |
| 1676 |
} |
| 1677 |
if (progress && f != NULL) |
| 1678 |
dumpnfa(nfa, f); |
| 1679 |
} while (progress && !NISERR()); |
| 1680 |
if (NISERR()) |
| 1681 |
return; |
| 1682 |
|
| 1683 |
/* |
| 1684 |
* Any ^ constraints we were able to pull to the start state can now be |
| 1685 |
* replaced by PLAIN arcs referencing the BOS or BOL colors. There should |
| 1686 |
* be no other ^ or BEHIND arcs left in the NFA, though we do not check |
| 1687 |
* that here (compact() will fail if so). |
| 1688 |
*/ |
| 1689 |
for (a = nfa->pre->outs; a != NULL; a = nexta) |
| 1690 |
{ |
| 1691 |
nexta = a->outchain; |
| 1692 |
if (a->type == '^') |
| 1693 |
{ |
| 1694 |
assert(a->co == 0 || a->co == 1); |
| 1695 |
newarc(nfa, PLAIN, nfa->bos[a->co], a->from, a->to); |
| 1696 |
freearc(nfa, a); |
| 1697 |
} |
| 1698 |
} |
| 1699 |
} |
| 1700 |
|
| 1701 |
/* |
| 1702 |
* pull - pull a back constraint backward past its source state |
| 1703 |
* |
| 1704 |
* Returns 1 if successful (which it always is unless the source is the |
| 1705 |
* start state or we have an internal error), 0 if nothing happened. |
| 1706 |
* |
| 1707 |
* A significant property of this function is that it deletes no pre-existing |
| 1708 |
* states, and no outarcs of the constraint's from state other than the given |
| 1709 |
* constraint arc. This makes the loops in pullback() safe, at the cost that |
| 1710 |
* we may leave useless states behind. Therefore, we leave it to pullback() |
| 1711 |
* to delete such states. |
| 1712 |
* |
| 1713 |
* If the from state has multiple back-constraint outarcs, and/or multiple |
| 1714 |
* compatible constraint inarcs, we only need to create one new intermediate |
| 1715 |
* state per combination of predecessor and successor states. *intermediates |
| 1716 |
* points to a list of such intermediate states for this from state (chained |
| 1717 |
* through their tmp fields). |
| 1718 |
*/ |
| 1719 |
static int |
| 1720 |
pull(struct nfa *nfa, |
| 1721 |
struct arc *con, |
| 1722 |
struct state **intermediates) |
| 1723 |
{ |
| 1724 |
struct state *from = con->from; |
| 1725 |
struct state *to = con->to; |
| 1726 |
struct arc *a; |
| 1727 |
struct arc *nexta; |
| 1728 |
struct state *s; |
| 1729 |
|
| 1730 |
assert(from != to); /* should have gotten rid of this earlier */ |
| 1731 |
if (from->flag) /* can't pull back beyond start */ |
| 1732 |
return 0; |
| 1733 |
if (from->nins == 0) |
| 1734 |
{ /* unreachable */ |
| 1735 |
freearc(nfa, con); |
| 1736 |
return 1; |
| 1737 |
} |
| 1738 |
|
| 1739 |
/* |
| 1740 |
* First, clone from state if necessary to avoid other outarcs. This may |
| 1741 |
* seem wasteful, but it simplifies the logic, and we'll get rid of the |
| 1742 |
* clone state again at the bottom. |
| 1743 |
*/ |
| 1744 |
if (from->nouts > 1) |
| 1745 |
{ |
| 1746 |
s = newstate(nfa); |
| 1747 |
if (NISERR()) |
| 1748 |
return 0; |
| 1749 |
copyins(nfa, from, s); /* duplicate inarcs */ |
| 1750 |
cparc(nfa, con, s, to); /* move constraint arc */ |
| 1751 |
freearc(nfa, con); |
| 1752 |
if (NISERR()) |
| 1753 |
return 0; |
| 1754 |
from = s; |
| 1755 |
con = from->outs; |
| 1756 |
} |
| 1757 |
assert(from->nouts == 1); |
| 1758 |
|
| 1759 |
/* propagate the constraint into the from state's inarcs */ |
| 1760 |
for (a = from->ins; a != NULL && !NISERR(); a = nexta) |
| 1761 |
{ |
| 1762 |
nexta = a->inchain; |
| 1763 |
switch (combine(nfa, con, a)) |
| 1764 |
{ |
| 1765 |
case INCOMPATIBLE: /* destroy the arc */ |
| 1766 |
freearc(nfa, a); |
| 1767 |
break; |
| 1768 |
case SATISFIED: /* no action needed */ |
| 1769 |
break; |
| 1770 |
case COMPATIBLE: /* swap the two arcs, more or less */ |
| 1771 |
/* need an intermediate state, but might have one already */ |
| 1772 |
for (s = *intermediates; s != NULL; s = s->tmp) |
| 1773 |
{ |
| 1774 |
assert(s->nins > 0 && s->nouts > 0); |
| 1775 |
if (s->ins->from == a->from && s->outs->to == to) |
| 1776 |
break; |
| 1777 |
} |
| 1778 |
if (s == NULL) |
| 1779 |
{ |
| 1780 |
s = newstate(nfa); |
| 1781 |
if (NISERR()) |
| 1782 |
return 0; |
| 1783 |
s->tmp = *intermediates; |
| 1784 |
*intermediates = s; |
| 1785 |
} |
| 1786 |
cparc(nfa, con, a->from, s); |
| 1787 |
cparc(nfa, a, s, to); |
| 1788 |
freearc(nfa, a); |
| 1789 |
break; |
| 1790 |
case REPLACEARC: /* replace arc's color */ |
| 1791 |
newarc(nfa, a->type, con->co, a->from, to); |
| 1792 |
freearc(nfa, a); |
| 1793 |
break; |
| 1794 |
default: |
| 1795 |
assert(NOTREACHED); |
| 1796 |
break; |
| 1797 |
} |
| 1798 |
} |
| 1799 |
|
| 1800 |
/* remaining inarcs, if any, incorporate the constraint */ |
| 1801 |
moveins(nfa, from, to); |
| 1802 |
freearc(nfa, con); |
| 1803 |
/* from state is now useless, but we leave it to pullback() to clean up */ |
| 1804 |
return 1; |
| 1805 |
} |
| 1806 |
|
| 1807 |
/* |
| 1808 |
* pushfwd - push forward constraints forward to eliminate them |
| 1809 |
*/ |
| 1810 |
static void |
| 1811 |
pushfwd(struct nfa *nfa, |
| 1812 |
FILE *f) /* for debug output; NULL none */ |
| 1813 |
{ |
| 1814 |
struct state *s; |
| 1815 |
struct state *nexts; |
| 1816 |
struct arc *a; |
| 1817 |
struct arc *nexta; |
| 1818 |
struct state *intermediates; |
| 1819 |
int progress; |
| 1820 |
|
| 1821 |
/* find and push until there are no more */ |
| 1822 |
do |
| 1823 |
{ |
| 1824 |
progress = 0; |
| 1825 |
for (s = nfa->states; s != NULL && !NISERR(); s = nexts) |
| 1826 |
{ |
| 1827 |
nexts = s->next; |
| 1828 |
intermediates = NULL; |
| 1829 |
for (a = s->ins; a != NULL && !NISERR(); a = nexta) |
| 1830 |
{ |
| 1831 |
nexta = a->inchain; |
| 1832 |
if (a->type == '$' || a->type == AHEAD) |
| 1833 |
if (push(nfa, a, &intermediates)) |
| 1834 |
progress = 1; |
| 1835 |
} |
| 1836 |
/* clear tmp fields of intermediate states created here */ |
| 1837 |
while (intermediates != NULL) |
| 1838 |
{ |
| 1839 |
struct state *ns = intermediates->tmp; |
| 1840 |
|
| 1841 |
intermediates->tmp = NULL; |
| 1842 |
intermediates = ns; |
| 1843 |
} |
| 1844 |
/* if s is now useless, get rid of it */ |
| 1845 |
if ((s->nins == 0 || s->nouts == 0) && !s->flag) |
| 1846 |
dropstate(nfa, s); |
| 1847 |
} |
| 1848 |
if (progress && f != NULL) |
| 1849 |
dumpnfa(nfa, f); |
| 1850 |
} while (progress && !NISERR()); |
| 1851 |
if (NISERR()) |
| 1852 |
return; |
| 1853 |
|
| 1854 |
/* |
| 1855 |
* Any $ constraints we were able to push to the post state can now be |
| 1856 |
* replaced by PLAIN arcs referencing the EOS or EOL colors. There should |
| 1857 |
* be no other $ or AHEAD arcs left in the NFA, though we do not check |
| 1858 |
* that here (compact() will fail if so). |
| 1859 |
*/ |
| 1860 |
for (a = nfa->post->ins; a != NULL; a = nexta) |
| 1861 |
{ |
| 1862 |
nexta = a->inchain; |
| 1863 |
if (a->type == '$') |
| 1864 |
{ |
| 1865 |
assert(a->co == 0 || a->co == 1); |
| 1866 |
newarc(nfa, PLAIN, nfa->eos[a->co], a->from, a->to); |
| 1867 |
freearc(nfa, a); |
| 1868 |
} |
| 1869 |
} |
| 1870 |
} |
| 1871 |
|
| 1872 |
/* |
| 1873 |
* push - push a forward constraint forward past its destination state |
| 1874 |
* |
| 1875 |
* Returns 1 if successful (which it always is unless the destination is the |
| 1876 |
* post state or we have an internal error), 0 if nothing happened. |
| 1877 |
* |
| 1878 |
* A significant property of this function is that it deletes no pre-existing |
| 1879 |
* states, and no inarcs of the constraint's to state other than the given |
| 1880 |
* constraint arc. This makes the loops in pushfwd() safe, at the cost that |
| 1881 |
* we may leave useless states behind. Therefore, we leave it to pushfwd() |
| 1882 |
* to delete such states. |
| 1883 |
* |
| 1884 |
* If the to state has multiple forward-constraint inarcs, and/or multiple |
| 1885 |
* compatible constraint outarcs, we only need to create one new intermediate |
| 1886 |
* state per combination of predecessor and successor states. *intermediates |
| 1887 |
* points to a list of such intermediate states for this to state (chained |
| 1888 |
* through their tmp fields). |
| 1889 |
*/ |
| 1890 |
static int |
| 1891 |
push(struct nfa *nfa, |
| 1892 |
struct arc *con, |
| 1893 |
struct state **intermediates) |
| 1894 |
{ |
| 1895 |
struct state *from = con->from; |
| 1896 |
struct state *to = con->to; |
| 1897 |
struct arc *a; |
| 1898 |
struct arc *nexta; |
| 1899 |
struct state *s; |
| 1900 |
|
| 1901 |
assert(to != from); /* should have gotten rid of this earlier */ |
| 1902 |
if (to->flag) /* can't push forward beyond end */ |
| 1903 |
return 0; |
| 1904 |
if (to->nouts == 0) |
| 1905 |
{ /* dead end */ |
| 1906 |
freearc(nfa, con); |
| 1907 |
return 1; |
| 1908 |
} |
| 1909 |
|
| 1910 |
/* |
| 1911 |
* First, clone to state if necessary to avoid other inarcs. This may |
| 1912 |
* seem wasteful, but it simplifies the logic, and we'll get rid of the |
| 1913 |
* clone state again at the bottom. |
| 1914 |
*/ |
| 1915 |
if (to->nins > 1) |
| 1916 |
{ |
| 1917 |
s = newstate(nfa); |
| 1918 |
if (NISERR()) |
| 1919 |
return 0; |
| 1920 |
copyouts(nfa, to, s); /* duplicate outarcs */ |
| 1921 |
cparc(nfa, con, from, s); /* move constraint arc */ |
| 1922 |
freearc(nfa, con); |
| 1923 |
if (NISERR()) |
| 1924 |
return 0; |
| 1925 |
to = s; |
| 1926 |
con = to->ins; |
| 1927 |
} |
| 1928 |
assert(to->nins == 1); |
| 1929 |
|
| 1930 |
/* propagate the constraint into the to state's outarcs */ |
| 1931 |
for (a = to->outs; a != NULL && !NISERR(); a = nexta) |
| 1932 |
{ |
| 1933 |
nexta = a->outchain; |
| 1934 |
switch (combine(nfa, con, a)) |
| 1935 |
{ |
| 1936 |
case INCOMPATIBLE: /* destroy the arc */ |
| 1937 |
freearc(nfa, a); |
| 1938 |
break; |
| 1939 |
case SATISFIED: /* no action needed */ |
| 1940 |
break; |
| 1941 |
case COMPATIBLE: /* swap the two arcs, more or less */ |
| 1942 |
/* need an intermediate state, but might have one already */ |
| 1943 |
for (s = *intermediates; s != NULL; s = s->tmp) |
| 1944 |
{ |
| 1945 |
assert(s->nins > 0 && s->nouts > 0); |
| 1946 |
if (s->ins->from == from && s->outs->to == a->to) |
| 1947 |
break; |
| 1948 |
} |
| 1949 |
if (s == NULL) |
| 1950 |
{ |
| 1951 |
s = newstate(nfa); |
| 1952 |
if (NISERR()) |
| 1953 |
return 0; |
| 1954 |
s->tmp = *intermediates; |
| 1955 |
*intermediates = s; |
| 1956 |
} |
| 1957 |
cparc(nfa, con, s, a->to); |
| 1958 |
cparc(nfa, a, from, s); |
| 1959 |
freearc(nfa, a); |
| 1960 |
break; |
| 1961 |
case REPLACEARC: /* replace arc's color */ |
| 1962 |
newarc(nfa, a->type, con->co, from, a->to); |
| 1963 |
freearc(nfa, a); |
| 1964 |
break; |
| 1965 |
default: |
| 1966 |
assert(NOTREACHED); |
| 1967 |
break; |
| 1968 |
} |
| 1969 |
} |
| 1970 |
|
| 1971 |
/* remaining outarcs, if any, incorporate the constraint */ |
| 1972 |
moveouts(nfa, to, from); |
| 1973 |
freearc(nfa, con); |
| 1974 |
/* to state is now useless, but we leave it to pushfwd() to clean up */ |
| 1975 |
return 1; |
| 1976 |
} |
| 1977 |
|
| 1978 |
/* |
| 1979 |
* combine - constraint lands on an arc, what happens? |
| 1980 |
* |
| 1981 |
* #def INCOMPATIBLE 1 // destroys arc |
| 1982 |
* #def SATISFIED 2 // constraint satisfied |
| 1983 |
* #def COMPATIBLE 3 // compatible but not satisfied yet |
| 1984 |
* #def REPLACEARC 4 // replace arc's color with constraint color |
| 1985 |
*/ |
| 1986 |
static int |
| 1987 |
combine(struct nfa *nfa, |
| 1988 |
struct arc *con, |
| 1989 |
struct arc *a) |
| 1990 |
{ |
| 1991 |
#define CA(ct,at) (((ct)<<CHAR_BIT) | (at)) |
| 1992 |
|
| 1993 |
switch (CA(con->type, a->type)) |
| 1994 |
{ |
| 1995 |
case CA('^', PLAIN): /* newlines are handled separately */ |
| 1996 |
case CA('$', PLAIN): |
| 1997 |
return INCOMPATIBLE; |
| 1998 |
break; |
| 1999 |
case CA(AHEAD, PLAIN): /* color constraints meet colors */ |
| 2000 |
case CA(BEHIND, PLAIN): |
| 2001 |
if (con->co == a->co) |
| 2002 |
return SATISFIED; |
| 2003 |
if (con->co == RAINBOW) |
| 2004 |
{ |
| 2005 |
/* con is satisfied unless arc's color is a pseudocolor */ |
| 2006 |
if (!(nfa->cm->cd[a->co].flags & PSEUDO)) |
| 2007 |
return SATISFIED; |
| 2008 |
} |
| 2009 |
else if (a->co == RAINBOW) |
| 2010 |
{ |
| 2011 |
/* con is incompatible if it's for a pseudocolor */ |
| 2012 |
/* (this is hypothetical; we make no such constraints today) */ |
| 2013 |
if (nfa->cm->cd[con->co].flags & PSEUDO) |
| 2014 |
return INCOMPATIBLE; |
| 2015 |
/* otherwise, constraint constrains arc to be only its color */ |
| 2016 |
return REPLACEARC; |
| 2017 |
} |
| 2018 |
return INCOMPATIBLE; |
| 2019 |
break; |
| 2020 |
case CA('^', '^'): /* collision, similar constraints */ |
| 2021 |
case CA('$', '$'): |
| 2022 |
if (con->co == a->co) /* true duplication */ |
| 2023 |
return SATISFIED; |
| 2024 |
return INCOMPATIBLE; |
| 2025 |
break; |
| 2026 |
case CA(AHEAD, AHEAD): /* collision, similar constraints */ |
| 2027 |
case CA(BEHIND, BEHIND): |
| 2028 |
if (con->co == a->co) /* true duplication */ |
| 2029 |
return SATISFIED; |
| 2030 |
if (con->co == RAINBOW) |
| 2031 |
{ |
| 2032 |
/* con is satisfied unless arc's color is a pseudocolor */ |
| 2033 |
if (!(nfa->cm->cd[a->co].flags & PSEUDO)) |
| 2034 |
return SATISFIED; |
| 2035 |
} |
| 2036 |
else if (a->co == RAINBOW) |
| 2037 |
{ |
| 2038 |
/* con is incompatible if it's for a pseudocolor */ |
| 2039 |
/* (this is hypothetical; we make no such constraints today) */ |
| 2040 |
if (nfa->cm->cd[con->co].flags & PSEUDO) |
| 2041 |
return INCOMPATIBLE; |
| 2042 |
/* otherwise, constraint constrains arc to be only its color */ |
| 2043 |
return REPLACEARC; |
| 2044 |
} |
| 2045 |
return INCOMPATIBLE; |
| 2046 |
break; |
| 2047 |
case CA('^', BEHIND): /* collision, dissimilar constraints */ |
| 2048 |
case CA(BEHIND, '^'): |
| 2049 |
case CA('$', AHEAD): |
| 2050 |
case CA(AHEAD, '$'): |
| 2051 |
return INCOMPATIBLE; |
| 2052 |
break; |
| 2053 |
case CA('^', '$'): /* constraints passing each other */ |
| 2054 |
case CA('^', AHEAD): |
| 2055 |
case CA(BEHIND, '$'): |
| 2056 |
case CA(BEHIND, AHEAD): |
| 2057 |
case CA('$', '^'): |
| 2058 |
case CA('$', BEHIND): |
| 2059 |
case CA(AHEAD, '^'): |
| 2060 |
case CA(AHEAD, BEHIND): |
| 2061 |
case CA('^', LACON): |
| 2062 |
case CA(BEHIND, LACON): |
| 2063 |
case CA('$', LACON): |
| 2064 |
case CA(AHEAD, LACON): |
| 2065 |
return COMPATIBLE; |
| 2066 |
break; |
| 2067 |
} |
| 2068 |
assert(NOTREACHED); |
| 2069 |
return INCOMPATIBLE; /* for benefit of blind compilers */ |
| 2070 |
} |
| 2071 |
|
| 2072 |
/* |
| 2073 |
* fixempties - get rid of EMPTY arcs |
| 2074 |
*/ |
| 2075 |
static void |
| 2076 |
fixempties(struct nfa *nfa, |
| 2077 |
FILE *f) /* for debug output; NULL none */ |
| 2078 |
{ |
| 2079 |
struct state *s; |
| 2080 |
struct state *s2; |
| 2081 |
struct state *nexts; |
| 2082 |
struct arc *a; |
| 2083 |
struct arc *nexta; |
| 2084 |
int totalinarcs; |
| 2085 |
struct arc **inarcsorig; |
| 2086 |
struct arc **arcarray; |
| 2087 |
int arccount; |
| 2088 |
int prevnins; |
| 2089 |
int nskip; |
| 2090 |
|
| 2091 |
/* |
| 2092 |
* First, get rid of any states whose sole out-arc is an EMPTY, since |
| 2093 |
* they're basically just aliases for their successor. The parsing |
| 2094 |
* algorithm creates enough of these that it's worth special-casing this. |
| 2095 |
*/ |
| 2096 |
for (s = nfa->states; s != NULL && !NISERR(); s = nexts) |
| 2097 |
{ |
| 2098 |
nexts = s->next; |
| 2099 |
if (s->flag || s->nouts != 1) |
| 2100 |
continue; |
| 2101 |
a = s->outs; |
| 2102 |
assert(a != NULL && a->outchain == NULL); |
| 2103 |
if (a->type != EMPTY) |
| 2104 |
continue; |
| 2105 |
if (s != a->to) |
| 2106 |
moveins(nfa, s, a->to); |
| 2107 |
dropstate(nfa, s); |
| 2108 |
} |
| 2109 |
|
| 2110 |
/* |
| 2111 |
* Similarly, get rid of any state with a single EMPTY in-arc, by folding |
| 2112 |
* it into its predecessor. |
| 2113 |
*/ |
| 2114 |
for (s = nfa->states; s != NULL && !NISERR(); s = nexts) |
| 2115 |
{ |
| 2116 |
nexts = s->next; |
| 2117 |
/* while we're at it, ensure tmp fields are clear for next step */ |
| 2118 |
assert(s->tmp == NULL); |
| 2119 |
if (s->flag || s->nins != 1) |
| 2120 |
continue; |
| 2121 |
a = s->ins; |
| 2122 |
assert(a != NULL && a->inchain == NULL); |
| 2123 |
if (a->type != EMPTY) |
| 2124 |
continue; |
| 2125 |
if (s != a->from) |
| 2126 |
moveouts(nfa, s, a->from); |
| 2127 |
dropstate(nfa, s); |
| 2128 |
} |
| 2129 |
|
| 2130 |
if (NISERR()) |
| 2131 |
return; |
| 2132 |
|
| 2133 |
/* |
| 2134 |
* For each remaining NFA state, find all other states from which it is |
| 2135 |
* reachable by a chain of one or more EMPTY arcs. Then generate new arcs |
| 2136 |
* that eliminate the need for each such chain. |
| 2137 |
* |
| 2138 |
* We could replace a chain of EMPTY arcs that leads from a "from" state |
| 2139 |
* to a "to" state either by pushing non-EMPTY arcs forward (linking |
| 2140 |
* directly from "from"'s predecessors to "to") or by pulling them back |
| 2141 |
* (linking directly from "from" to "to"'s successors). We choose to |
| 2142 |
* always do the former; this choice is somewhat arbitrary, but the |
| 2143 |
* approach below requires that we uniformly do one or the other. |
| 2144 |
* |
| 2145 |
* Suppose we have a chain of N successive EMPTY arcs (where N can easily |
| 2146 |
* approach the size of the NFA). All of the intermediate states must |
| 2147 |
* have additional inarcs and outarcs, else they'd have been removed by |
| 2148 |
* the steps above. Assuming their inarcs are mostly not empties, we will |
| 2149 |
* add O(N^2) arcs to the NFA, since a non-EMPTY inarc leading to any one |
| 2150 |
* state in the chain must be duplicated to lead to all its successor |
| 2151 |
* states as well. So there is no hope of doing less than O(N^2) work; |
| 2152 |
* however, we should endeavor to keep the big-O cost from being even |
| 2153 |
* worse than that, which it can easily become without care. In |
| 2154 |
* particular, suppose we were to copy all S1's inarcs forward to S2, and |
| 2155 |
* then also to S3, and then later we consider pushing S2's inarcs forward |
| 2156 |
* to S3. If we include the arcs already copied from S1 in that, we'd be |
| 2157 |
* doing O(N^3) work. (The duplicate-arc elimination built into newarc() |
| 2158 |
* and its cohorts would get rid of the extra arcs, but not without cost.) |
| 2159 |
* |
| 2160 |
* We can avoid this cost by treating only arcs that existed at the start |
| 2161 |
* of this phase as candidates to be pushed forward. To identify those, |
| 2162 |
* we remember the first inarc each state had to start with. We rely on |
| 2163 |
* the fact that newarc() and friends put new arcs on the front of their |
| 2164 |
* to-states' inchains, and that this phase never deletes arcs, so that |
| 2165 |
* the original arcs must be the last arcs in their to-states' inchains. |
| 2166 |
* |
| 2167 |
* So the process here is that, for each state in the NFA, we gather up |
| 2168 |
* all non-EMPTY inarcs of states that can reach the target state via |
| 2169 |
* EMPTY arcs. We then sort, de-duplicate, and merge these arcs into the |
| 2170 |
* target state's inchain. (We can safely use sort-merge for this as long |
| 2171 |
* as we update each state's original-arcs pointer after we add arcs to |
| 2172 |
* it; the sort step of mergeins probably changed the order of the old |
| 2173 |
* arcs.) |
| 2174 |
* |
| 2175 |
* Another refinement worth making is that, because we only add non-EMPTY |
| 2176 |
* arcs during this phase, and all added arcs have the same from-state as |
| 2177 |
* the non-EMPTY arc they were cloned from, we know ahead of time that any |
| 2178 |
* states having only EMPTY outarcs will be useless for lack of outarcs |
| 2179 |
* after we drop the EMPTY arcs. (They cannot gain non-EMPTY outarcs if |
| 2180 |
* they had none to start with.) So we need not bother to update the |
| 2181 |
* inchains of such states at all. |
| 2182 |
*/ |
| 2183 |
|
| 2184 |
/* Remember the states' first original inarcs */ |
| 2185 |
/* ... and while at it, count how many old inarcs there are altogether */ |
| 2186 |
inarcsorig = (struct arc **) MALLOC(nfa->nstates * sizeof(struct arc *)); |
| 2187 |
if (inarcsorig == NULL) |
| 2188 |
{ |
| 2189 |
NERR(REG_ESPACE); |
| 2190 |
return; |
| 2191 |
} |
| 2192 |
totalinarcs = 0; |
| 2193 |
for (s = nfa->states; s != NULL; s = s->next) |
| 2194 |
{ |
| 2195 |
inarcsorig[s->no] = s->ins; |
| 2196 |
totalinarcs += s->nins; |
| 2197 |
} |
| 2198 |
|
| 2199 |
/* |
| 2200 |
* Create a workspace for accumulating the inarcs to be added to the |
| 2201 |
* current target state. totalinarcs is probably a considerable |
| 2202 |
* overestimate of the space needed, but the NFA is unlikely to be large |
| 2203 |
* enough at this point to make it worth being smarter. |
| 2204 |
*/ |
| 2205 |
arcarray = (struct arc **) MALLOC(totalinarcs * sizeof(struct arc *)); |
| 2206 |
if (arcarray == NULL) |
| 2207 |
{ |
| 2208 |
NERR(REG_ESPACE); |
| 2209 |
FREE(inarcsorig); |
| 2210 |
return; |
| 2211 |
} |
| 2212 |
|
| 2213 |
/* And iterate over the target states */ |
| 2214 |
for (s = nfa->states; s != NULL && !NISERR(); s = s->next) |
| 2215 |
{ |
| 2216 |
/* Ignore target states without non-EMPTY outarcs, per note above */ |
| 2217 |
if (!s->flag && !hasnonemptyout(s)) |
| 2218 |
continue; |
| 2219 |
|
| 2220 |
/* Find predecessor states and accumulate their original inarcs */ |
| 2221 |
arccount = 0; |
| 2222 |
for (s2 = emptyreachable(nfa, s, s, inarcsorig); s2 != s; s2 = nexts) |
| 2223 |
{ |
| 2224 |
/* Add s2's original inarcs to arcarray[], but ignore empties */ |
| 2225 |
for (a = inarcsorig[s2->no]; a != NULL; a = a->inchain) |
| 2226 |
{ |
| 2227 |
if (a->type != EMPTY) |
| 2228 |
arcarray[arccount++] = a; |
| 2229 |
} |
| 2230 |
|
| 2231 |
/* Reset the tmp fields as we walk back */ |
| 2232 |
nexts = s2->tmp; |
| 2233 |
s2->tmp = NULL; |
| 2234 |
} |
| 2235 |
s->tmp = NULL; |
| 2236 |
assert(arccount <= totalinarcs); |
| 2237 |
|
| 2238 |
/* Remember how many original inarcs this state has */ |
| 2239 |
prevnins = s->nins; |
| 2240 |
|
| 2241 |
/* Add non-duplicate inarcs to target state */ |
| 2242 |
mergeins(nfa, s, arcarray, arccount); |
| 2243 |
|
| 2244 |
/* Now we must update the state's inarcsorig pointer */ |
| 2245 |
nskip = s->nins - prevnins; |
| 2246 |
a = s->ins; |
| 2247 |
while (nskip-- > 0) |
| 2248 |
a = a->inchain; |
| 2249 |
inarcsorig[s->no] = a; |
| 2250 |
} |
| 2251 |
|
| 2252 |
FREE(arcarray); |
| 2253 |
FREE(inarcsorig); |
| 2254 |
|
| 2255 |
if (NISERR()) |
| 2256 |
return; |
| 2257 |
|
| 2258 |
/* |
| 2259 |
* Now remove all the EMPTY arcs, since we don't need them anymore. |
| 2260 |
*/ |
| 2261 |
for (s = nfa->states; s != NULL; s = s->next) |
| 2262 |
{ |
| 2263 |
for (a = s->outs; a != NULL; a = nexta) |
| 2264 |
{ |
| 2265 |
nexta = a->outchain; |
| 2266 |
if (a->type == EMPTY) |
| 2267 |
freearc(nfa, a); |
| 2268 |
} |
| 2269 |
} |
| 2270 |
|
| 2271 |
/* |
| 2272 |
* And remove any states that have become useless. (This cleanup is not |
| 2273 |
* very thorough, and would be even less so if we tried to combine it with |
| 2274 |
* the previous step; but cleanup() will take care of anything we miss.) |
| 2275 |
*/ |
| 2276 |
for (s = nfa->states; s != NULL; s = nexts) |
| 2277 |
{ |
| 2278 |
nexts = s->next; |
| 2279 |
if ((s->nins == 0 || s->nouts == 0) && !s->flag) |
| 2280 |
dropstate(nfa, s); |
| 2281 |
} |
| 2282 |
|
| 2283 |
if (f != NULL) |
| 2284 |
dumpnfa(nfa, f); |
| 2285 |
} |
| 2286 |
|
| 2287 |
/* |
| 2288 |
* emptyreachable - recursively find all states that can reach s by EMPTY arcs |
| 2289 |
* |
| 2290 |
* The return value is the last such state found. Its tmp field links back |
| 2291 |
* to the next-to-last such state, and so on back to s, so that all these |
| 2292 |
* states can be located without searching the whole NFA. |
| 2293 |
* |
| 2294 |
* Since this is only used in fixempties(), we pass in the inarcsorig[] array |
| 2295 |
* maintained by that function. This lets us skip over all new inarcs, which |
| 2296 |
* are certainly not EMPTY arcs. |
| 2297 |
* |
| 2298 |
* The maximum recursion depth here is equal to the length of the longest |
| 2299 |
* loop-free chain of EMPTY arcs, which is surely no more than the size of |
| 2300 |
* the NFA ... but that could still be enough to cause trouble. |
| 2301 |
*/ |
| 2302 |
static struct state * |
| 2303 |
emptyreachable(struct nfa *nfa, |
| 2304 |
struct state *s, |
| 2305 |
struct state *lastfound, |
| 2306 |
struct arc **inarcsorig) |
| 2307 |
{ |
| 2308 |
struct arc *a; |
| 2309 |
|
| 2310 |
/* Since this is recursive, it could be driven to stack overflow */ |
| 2311 |
if (STACK_TOO_DEEP(nfa->v->re)) |
| 2312 |
{ |
| 2313 |
NERR(REG_ETOOBIG); |
| 2314 |
return lastfound; |
| 2315 |
} |
| 2316 |
|
| 2317 |
s->tmp = lastfound; |
| 2318 |
lastfound = s; |
| 2319 |
for (a = inarcsorig[s->no]; a != NULL; a = a->inchain) |
| 2320 |
{ |
| 2321 |
if (a->type == EMPTY && a->from->tmp == NULL) |
| 2322 |
lastfound = emptyreachable(nfa, a->from, lastfound, inarcsorig); |
| 2323 |
} |
| 2324 |
return lastfound; |
| 2325 |
} |
| 2326 |
|
| 2327 |
/* |
| 2328 |
* isconstraintarc - detect whether an arc is of a constraint type |
| 2329 |
*/ |
| 2330 |
static inline int |
| 2331 |
isconstraintarc(struct arc *a) |
| 2332 |
{ |
| 2333 |
switch (a->type) |
| 2334 |
{ |
| 2335 |
case '^': |
| 2336 |
case '$': |
| 2337 |
case BEHIND: |
| 2338 |
case AHEAD: |
| 2339 |
case LACON: |
| 2340 |
return 1; |
| 2341 |
} |
| 2342 |
return 0; |
| 2343 |
} |
| 2344 |
|
| 2345 |
/* |
| 2346 |
* hasconstraintout - does state have a constraint out arc? |
| 2347 |
*/ |
| 2348 |
static int |
| 2349 |
hasconstraintout(struct state *s) |
| 2350 |
{ |
| 2351 |
struct arc *a; |
| 2352 |
|
| 2353 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 2354 |
{ |
| 2355 |
if (isconstraintarc(a)) |
| 2356 |
return 1; |
| 2357 |
} |
| 2358 |
return 0; |
| 2359 |
} |
| 2360 |
|
| 2361 |
/* |
| 2362 |
* fixconstraintloops - get rid of loops containing only constraint arcs |
| 2363 |
* |
| 2364 |
* A loop of states that contains only constraint arcs is useless, since |
| 2365 |
* passing around the loop represents no forward progress. Moreover, it |
| 2366 |
* would cause infinite looping in pullback/pushfwd, so we need to get rid |
| 2367 |
* of such loops before doing that. |
| 2368 |
*/ |
| 2369 |
static void |
| 2370 |
fixconstraintloops(struct nfa *nfa, |
| 2371 |
FILE *f) /* for debug output; NULL none */ |
| 2372 |
{ |
| 2373 |
struct state *s; |
| 2374 |
struct state *nexts; |
| 2375 |
struct arc *a; |
| 2376 |
struct arc *nexta; |
| 2377 |
int hasconstraints; |
| 2378 |
|
| 2379 |
/* |
| 2380 |
* In the trivial case of a state that loops to itself, we can just drop |
| 2381 |
* the constraint arc altogether. This is worth special-casing because |
| 2382 |
* such loops are far more common than loops containing multiple states. |
| 2383 |
* While we're at it, note whether any constraint arcs survive. |
| 2384 |
*/ |
| 2385 |
hasconstraints = 0; |
| 2386 |
for (s = nfa->states; s != NULL && !NISERR(); s = nexts) |
| 2387 |
{ |
| 2388 |
nexts = s->next; |
| 2389 |
/* while we're at it, ensure tmp fields are clear for next step */ |
| 2390 |
assert(s->tmp == NULL); |
| 2391 |
for (a = s->outs; a != NULL && !NISERR(); a = nexta) |
| 2392 |
{ |
| 2393 |
nexta = a->outchain; |
| 2394 |
if (isconstraintarc(a)) |
| 2395 |
{ |
| 2396 |
if (a->to == s) |
| 2397 |
freearc(nfa, a); |
| 2398 |
else |
| 2399 |
hasconstraints = 1; |
| 2400 |
} |
| 2401 |
} |
| 2402 |
/* If we removed all the outarcs, the state is useless. */ |
| 2403 |
if (s->nouts == 0 && !s->flag) |
| 2404 |
dropstate(nfa, s); |
| 2405 |
} |
| 2406 |
|
| 2407 |
/* Nothing to do if no remaining constraint arcs */ |
| 2408 |
if (NISERR() || !hasconstraints) |
| 2409 |
return; |
| 2410 |
|
| 2411 |
/* |
| 2412 |
* Starting from each remaining NFA state, search outwards for a |
| 2413 |
* constraint loop. If we find a loop, break the loop, then start the |
| 2414 |
* search over. (We could possibly retain some state from the first scan, |
| 2415 |
* but it would complicate things greatly, and multi-state constraint |
| 2416 |
* loops are rare enough that it's not worth optimizing the case.) |
| 2417 |
*/ |
| 2418 |
restart: |
| 2419 |
for (s = nfa->states; s != NULL && !NISERR(); s = s->next) |
| 2420 |
{ |
| 2421 |
if (findconstraintloop(nfa, s)) |
| 2422 |
goto restart; |
| 2423 |
} |
| 2424 |
|
| 2425 |
if (NISERR()) |
| 2426 |
return; |
| 2427 |
|
| 2428 |
/* |
| 2429 |
* Now remove any states that have become useless. (This cleanup is not |
| 2430 |
* very thorough, and would be even less so if we tried to combine it with |
| 2431 |
* the previous step; but cleanup() will take care of anything we miss.) |
| 2432 |
* |
| 2433 |
* Because findconstraintloop intentionally doesn't reset all tmp fields, |
| 2434 |
* we have to clear them after it's done. This is a convenient place to |
| 2435 |
* do that, too. |
| 2436 |
*/ |
| 2437 |
for (s = nfa->states; s != NULL; s = nexts) |
| 2438 |
{ |
| 2439 |
nexts = s->next; |
| 2440 |
s->tmp = NULL; |
| 2441 |
if ((s->nins == 0 || s->nouts == 0) && !s->flag) |
| 2442 |
dropstate(nfa, s); |
| 2443 |
} |
| 2444 |
|
| 2445 |
if (f != NULL) |
| 2446 |
dumpnfa(nfa, f); |
| 2447 |
} |
| 2448 |
|
| 2449 |
/* |
| 2450 |
* findconstraintloop - recursively find a loop of constraint arcs |
| 2451 |
* |
| 2452 |
* If we find a loop, break it by calling breakconstraintloop(), then |
| 2453 |
* return 1; otherwise return 0. |
| 2454 |
* |
| 2455 |
* State tmp fields are guaranteed all NULL on a success return, because |
| 2456 |
* breakconstraintloop does that. After a failure return, any state that |
| 2457 |
* is known not to be part of a loop is marked with s->tmp == s; this allows |
| 2458 |
* us not to have to re-prove that fact on later calls. (This convention is |
| 2459 |
* workable because we already eliminated single-state loops.) |
| 2460 |
* |
| 2461 |
* Note that the found loop doesn't necessarily include the first state we |
| 2462 |
* are called on. Any loop reachable from that state will do. |
| 2463 |
* |
| 2464 |
* The maximum recursion depth here is one more than the length of the longest |
| 2465 |
* loop-free chain of constraint arcs, which is surely no more than the size |
| 2466 |
* of the NFA ... but that could still be enough to cause trouble. |
| 2467 |
*/ |
| 2468 |
static int |
| 2469 |
findconstraintloop(struct nfa *nfa, struct state *s) |
| 2470 |
{ |
| 2471 |
struct arc *a; |
| 2472 |
|
| 2473 |
/* Since this is recursive, it could be driven to stack overflow */ |
| 2474 |
if (STACK_TOO_DEEP(nfa->v->re)) |
| 2475 |
{ |
| 2476 |
NERR(REG_ETOOBIG); |
| 2477 |
return 1; /* to exit as quickly as possible */ |
| 2478 |
} |
| 2479 |
|
| 2480 |
if (s->tmp != NULL) |
| 2481 |
{ |
| 2482 |
/* Already proven uninteresting? */ |
| 2483 |
if (s->tmp == s) |
| 2484 |
return 0; |
| 2485 |
/* Found a loop involving s */ |
| 2486 |
breakconstraintloop(nfa, s); |
| 2487 |
/* The tmp fields have been cleaned up by breakconstraintloop */ |
| 2488 |
return 1; |
| 2489 |
} |
| 2490 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 2491 |
{ |
| 2492 |
if (isconstraintarc(a)) |
| 2493 |
{ |
| 2494 |
struct state *sto = a->to; |
| 2495 |
|
| 2496 |
assert(sto != s); |
| 2497 |
s->tmp = sto; |
| 2498 |
if (findconstraintloop(nfa, sto)) |
| 2499 |
return 1; |
| 2500 |
} |
| 2501 |
} |
| 2502 |
|
| 2503 |
/* |
| 2504 |
* If we get here, no constraint loop exists leading out from s. Mark it |
| 2505 |
* with s->tmp == s so we need not rediscover that fact again later. |
| 2506 |
*/ |
| 2507 |
s->tmp = s; |
| 2508 |
return 0; |
| 2509 |
} |
| 2510 |
|
| 2511 |
/* |
| 2512 |
* breakconstraintloop - break a loop of constraint arcs |
| 2513 |
* |
| 2514 |
* sinitial is any one member state of the loop. Each loop member's tmp |
| 2515 |
* field links to its successor within the loop. (Note that this function |
| 2516 |
* will reset all the tmp fields to NULL.) |
| 2517 |
* |
| 2518 |
* We can break the loop by, for any one state S1 in the loop, cloning its |
| 2519 |
* loop successor state S2 (and possibly following states), and then moving |
| 2520 |
* all S1->S2 constraint arcs to point to the cloned S2. The cloned S2 should |
| 2521 |
* copy any non-constraint outarcs of S2. Constraint outarcs should be |
| 2522 |
* dropped if they point back to S1, else they need to be copied as arcs to |
| 2523 |
* similarly cloned states S3, S4, etc. In general, each cloned state copies |
| 2524 |
* non-constraint outarcs, drops constraint outarcs that would lead to itself |
| 2525 |
* or any earlier cloned state, and sends other constraint outarcs to newly |
| 2526 |
* cloned states. No cloned state will have any inarcs that aren't constraint |
| 2527 |
* arcs or do not lead from S1 or earlier-cloned states. It's okay to drop |
| 2528 |
* constraint back-arcs since they would not take us to any state we've not |
| 2529 |
* already been in; therefore, no new constraint loop is created. In this way |
| 2530 |
* we generate a modified NFA that can still represent every useful state |
| 2531 |
* sequence, but not sequences that represent state loops with no consumption |
| 2532 |
* of input data. Note that the set of cloned states will certainly include |
| 2533 |
* all of the loop member states other than S1, and it may also include |
| 2534 |
* non-loop states that are reachable from S2 via constraint arcs. This is |
| 2535 |
* important because there is no guarantee that findconstraintloop found a |
| 2536 |
* maximal loop (and searching for one would be NP-hard, so don't try). |
| 2537 |
* Frequently the "non-loop states" are actually part of a larger loop that |
| 2538 |
* we didn't notice, and indeed there may be several overlapping loops. |
| 2539 |
* This technique ensures convergence in such cases, while considering only |
| 2540 |
* the originally-found loop does not. |
| 2541 |
* |
| 2542 |
* If there is only one S1->S2 constraint arc, then that constraint is |
| 2543 |
* certainly satisfied when we enter any of the clone states. This means that |
| 2544 |
* in the common case where many of the constraint arcs are identically |
| 2545 |
* labeled, we can merge together clone states linked by a similarly-labeled |
| 2546 |
* constraint: if we can get to the first one we can certainly get to the |
| 2547 |
* second, so there's no need to distinguish. This greatly reduces the number |
| 2548 |
* of new states needed, so we preferentially break the given loop at a state |
| 2549 |
* pair where this is true. |
| 2550 |
* |
| 2551 |
* Furthermore, it's fairly common to find that a cloned successor state has |
| 2552 |
* no outarcs, especially if we're a bit aggressive about removing unnecessary |
| 2553 |
* outarcs. If that happens, then there is simply not any interesting state |
| 2554 |
* that can be reached through the predecessor's loop arcs, which means we can |
| 2555 |
* break the loop just by removing those loop arcs, with no new states added. |
| 2556 |
*/ |
| 2557 |
static void |
| 2558 |
breakconstraintloop(struct nfa *nfa, struct state *sinitial) |
| 2559 |
{ |
| 2560 |
struct state *s; |
| 2561 |
struct state *shead; |
| 2562 |
struct state *stail; |
| 2563 |
struct state *sclone; |
| 2564 |
struct state *nexts; |
| 2565 |
struct arc *refarc; |
| 2566 |
struct arc *a; |
| 2567 |
struct arc *nexta; |
| 2568 |
|
| 2569 |
/* |
| 2570 |
* Start by identifying which loop step we want to break at. |
| 2571 |
* Preferentially this is one with only one constraint arc. (XXX are |
| 2572 |
* there any other secondary heuristics we want to use here?) Set refarc |
| 2573 |
* to point to the selected lone constraint arc, if there is one. |
| 2574 |
*/ |
| 2575 |
refarc = NULL; |
| 2576 |
s = sinitial; |
| 2577 |
do |
| 2578 |
{ |
| 2579 |
nexts = s->tmp; |
| 2580 |
assert(nexts != s); /* should not see any one-element loops */ |
| 2581 |
if (refarc == NULL) |
| 2582 |
{ |
| 2583 |
int narcs = 0; |
| 2584 |
|
| 2585 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 2586 |
{ |
| 2587 |
if (a->to == nexts && isconstraintarc(a)) |
| 2588 |
{ |
| 2589 |
refarc = a; |
| 2590 |
narcs++; |
| 2591 |
} |
| 2592 |
} |
| 2593 |
assert(narcs > 0); |
| 2594 |
if (narcs > 1) |
| 2595 |
refarc = NULL; /* multiple constraint arcs here, no good */ |
| 2596 |
} |
| 2597 |
s = nexts; |
| 2598 |
} while (s != sinitial); |
| 2599 |
|
| 2600 |
if (refarc) |
| 2601 |
{ |
| 2602 |
/* break at the refarc */ |
| 2603 |
shead = refarc->from; |
| 2604 |
stail = refarc->to; |
| 2605 |
assert(stail == shead->tmp); |
| 2606 |
} |
| 2607 |
else |
| 2608 |
{ |
| 2609 |
/* for lack of a better idea, break after sinitial */ |
| 2610 |
shead = sinitial; |
| 2611 |
stail = sinitial->tmp; |
| 2612 |
} |
| 2613 |
|
| 2614 |
/* |
| 2615 |
* Reset the tmp fields so that we can use them for local storage in |
| 2616 |
* clonesuccessorstates. (findconstraintloop won't mind, since it's just |
| 2617 |
* going to abandon its search anyway.) |
| 2618 |
*/ |
| 2619 |
for (s = nfa->states; s != NULL; s = s->next) |
| 2620 |
s->tmp = NULL; |
| 2621 |
|
| 2622 |
/* |
| 2623 |
* Recursively build clone state(s) as needed. |
| 2624 |
*/ |
| 2625 |
sclone = newstate(nfa); |
| 2626 |
if (sclone == NULL) |
| 2627 |
{ |
| 2628 |
assert(NISERR()); |
| 2629 |
return; |
| 2630 |
} |
| 2631 |
|
| 2632 |
clonesuccessorstates(nfa, stail, sclone, shead, refarc, |
| 2633 |
NULL, NULL, nfa->nstates); |
| 2634 |
|
| 2635 |
if (NISERR()) |
| 2636 |
return; |
| 2637 |
|
| 2638 |
/* |
| 2639 |
* It's possible that sclone has no outarcs at all, in which case it's |
| 2640 |
* useless. (We don't try extremely hard to get rid of useless states |
| 2641 |
* here, but this is an easy and fairly common case.) |
| 2642 |
*/ |
| 2643 |
if (sclone->nouts == 0) |
| 2644 |
{ |
| 2645 |
freestate(nfa, sclone); |
| 2646 |
sclone = NULL; |
| 2647 |
} |
| 2648 |
|
| 2649 |
/* |
| 2650 |
* Move shead's constraint-loop arcs to point to sclone, or just drop them |
| 2651 |
* if we discovered we don't need sclone. |
| 2652 |
*/ |
| 2653 |
for (a = shead->outs; a != NULL; a = nexta) |
| 2654 |
{ |
| 2655 |
nexta = a->outchain; |
| 2656 |
if (a->to == stail && isconstraintarc(a)) |
| 2657 |
{ |
| 2658 |
if (sclone) |
| 2659 |
cparc(nfa, a, shead, sclone); |
| 2660 |
freearc(nfa, a); |
| 2661 |
if (NISERR()) |
| 2662 |
break; |
| 2663 |
} |
| 2664 |
} |
| 2665 |
} |
| 2666 |
|
| 2667 |
/* |
| 2668 |
* clonesuccessorstates - create a tree of constraint-arc successor states |
| 2669 |
* |
| 2670 |
* ssource is the state to be cloned, and sclone is the state to copy its |
| 2671 |
* outarcs into. sclone's inarcs, if any, should already be set up. |
| 2672 |
* |
| 2673 |
* spredecessor is the original predecessor state that we are trying to build |
| 2674 |
* successors for (it may not be the immediate predecessor of ssource). |
| 2675 |
* refarc, if not NULL, is the original constraint arc that is known to have |
| 2676 |
* been traversed out of spredecessor to reach the successor(s). |
| 2677 |
* |
| 2678 |
* For each cloned successor state, we transiently create a "donemap" that is |
| 2679 |
* a boolean array showing which source states we've already visited for this |
| 2680 |
* clone state. This prevents infinite recursion as well as useless repeat |
| 2681 |
* visits to the same state subtree (which can add up fast, since typical NFAs |
| 2682 |
* have multiple redundant arc pathways). Each donemap is a char array |
| 2683 |
* indexed by state number. The donemaps are all of the same size "nstates", |
| 2684 |
* which is nfa->nstates as of the start of the recursion. This is enough to |
| 2685 |
* have entries for all pre-existing states, but *not* entries for clone |
| 2686 |
* states created during the recursion. That's okay since we have no need to |
| 2687 |
* mark those. |
| 2688 |
* |
| 2689 |
* curdonemap is NULL when recursing to a new sclone state, or sclone's |
| 2690 |
* donemap when we are recursing without having created a new state (which we |
| 2691 |
* do when we decide we can merge a successor state into the current clone |
| 2692 |
* state). outerdonemap is NULL at the top level and otherwise the parent |
| 2693 |
* clone state's donemap. |
| 2694 |
* |
| 2695 |
* The successor states we create and fill here form a strict tree structure, |
| 2696 |
* with each state having exactly one predecessor, except that the toplevel |
| 2697 |
* state has no inarcs as yet (breakconstraintloop will add its inarcs from |
| 2698 |
* spredecessor after we're done). Thus, we can examine sclone's inarcs back |
| 2699 |
* to the root, plus refarc if any, to identify the set of constraints already |
| 2700 |
* known valid at the current point. This allows us to avoid generating extra |
| 2701 |
* successor states. |
| 2702 |
*/ |
| 2703 |
static void |
| 2704 |
clonesuccessorstates(struct nfa *nfa, |
| 2705 |
struct state *ssource, |
| 2706 |
struct state *sclone, |
| 2707 |
struct state *spredecessor, |
| 2708 |
struct arc *refarc, |
| 2709 |
char *curdonemap, |
| 2710 |
char *outerdonemap, |
| 2711 |
int nstates) |
| 2712 |
{ |
| 2713 |
char *donemap; |
| 2714 |
struct arc *a; |
| 2715 |
|
| 2716 |
/* Since this is recursive, it could be driven to stack overflow */ |
| 2717 |
if (STACK_TOO_DEEP(nfa->v->re)) |
| 2718 |
{ |
| 2719 |
NERR(REG_ETOOBIG); |
| 2720 |
return; |
| 2721 |
} |
| 2722 |
|
| 2723 |
/* If this state hasn't already got a donemap, create one */ |
| 2724 |
donemap = curdonemap; |
| 2725 |
if (donemap == NULL) |
| 2726 |
{ |
| 2727 |
donemap = (char *) MALLOC(nstates * sizeof(char)); |
| 2728 |
if (donemap == NULL) |
| 2729 |
{ |
| 2730 |
NERR(REG_ESPACE); |
| 2731 |
return; |
| 2732 |
} |
| 2733 |
|
| 2734 |
if (outerdonemap != NULL) |
| 2735 |
{ |
| 2736 |
/* |
| 2737 |
* Not at outermost recursion level, so copy the outer level's |
| 2738 |
* donemap; this ensures that we see states in process of being |
| 2739 |
* visited at outer levels, or already merged into predecessor |
| 2740 |
* states, as ones we shouldn't traverse back to. |
| 2741 |
*/ |
| 2742 |
memcpy(donemap, outerdonemap, nstates * sizeof(char)); |
| 2743 |
} |
| 2744 |
else |
| 2745 |
{ |
| 2746 |
/* At outermost level, only spredecessor is off-limits */ |
| 2747 |
memset(donemap, 0, nstates * sizeof(char)); |
| 2748 |
assert(spredecessor->no < nstates); |
| 2749 |
donemap[spredecessor->no] = 1; |
| 2750 |
} |
| 2751 |
} |
| 2752 |
|
| 2753 |
/* Mark ssource as visited in the donemap */ |
| 2754 |
assert(ssource->no < nstates); |
| 2755 |
assert(donemap[ssource->no] == 0); |
| 2756 |
donemap[ssource->no] = 1; |
| 2757 |
|
| 2758 |
/* |
| 2759 |
* We proceed by first cloning all of ssource's outarcs, creating new |
| 2760 |
* clone states as needed but not doing more with them than that. Then in |
| 2761 |
* a second pass, recurse to process the child clone states. This allows |
| 2762 |
* us to have only one child clone state per reachable source state, even |
| 2763 |
* when there are multiple outarcs leading to the same state. Also, when |
| 2764 |
* we do visit a child state, its set of inarcs is known exactly, which |
| 2765 |
* makes it safe to apply the constraint-is-already-checked optimization. |
| 2766 |
* Also, this ensures that we've merged all the states we can into the |
| 2767 |
* current clone before we recurse to any children, thus possibly saving |
| 2768 |
* them from making extra images of those states. |
| 2769 |
* |
| 2770 |
* While this function runs, child clone states of the current state are |
| 2771 |
* marked by setting their tmp fields to point to the original state they |
| 2772 |
* were cloned from. This makes it possible to detect multiple outarcs |
| 2773 |
* leading to the same state, and also makes it easy to distinguish clone |
| 2774 |
* states from original states (which will have tmp == NULL). |
| 2775 |
*/ |
| 2776 |
for (a = ssource->outs; a != NULL && !NISERR(); a = a->outchain) |
| 2777 |
{ |
| 2778 |
struct state *sto = a->to; |
| 2779 |
|
| 2780 |
/* |
| 2781 |
* We do not consider cloning successor states that have no constraint |
| 2782 |
* outarcs; just link to them as-is. They cannot be part of a |
| 2783 |
* constraint loop so there is no need to make copies. In particular, |
| 2784 |
* this rule keeps us from trying to clone the post state, which would |
| 2785 |
* be a bad idea. |
| 2786 |
*/ |
| 2787 |
if (isconstraintarc(a) && hasconstraintout(sto)) |
| 2788 |
{ |
| 2789 |
struct state *prevclone; |
| 2790 |
int canmerge; |
| 2791 |
struct arc *a2; |
| 2792 |
|
| 2793 |
/* |
| 2794 |
* Back-link constraint arcs must not be followed. Nor is there a |
| 2795 |
* need to revisit states previously merged into this clone. |
| 2796 |
*/ |
| 2797 |
assert(sto->no < nstates); |
| 2798 |
if (donemap[sto->no] != 0) |
| 2799 |
continue; |
| 2800 |
|
| 2801 |
/* |
| 2802 |
* Check whether we already have a child clone state for this |
| 2803 |
* source state. |
| 2804 |
*/ |
| 2805 |
prevclone = NULL; |
| 2806 |
for (a2 = sclone->outs; a2 != NULL; a2 = a2->outchain) |
| 2807 |
{ |
| 2808 |
if (a2->to->tmp == sto) |
| 2809 |
{ |
| 2810 |
prevclone = a2->to; |
| 2811 |
break; |
| 2812 |
} |
| 2813 |
} |
| 2814 |
|
| 2815 |
/* |
| 2816 |
* If this arc is labeled the same as refarc, or the same as any |
| 2817 |
* arc we must have traversed to get to sclone, then no additional |
| 2818 |
* constraints need to be met to get to sto, so we should just |
| 2819 |
* merge its outarcs into sclone. |
| 2820 |
*/ |
| 2821 |
if (refarc && a->type == refarc->type && a->co == refarc->co) |
| 2822 |
canmerge = 1; |
| 2823 |
else |
| 2824 |
{ |
| 2825 |
struct state *s; |
| 2826 |
|
| 2827 |
canmerge = 0; |
| 2828 |
for (s = sclone; s->ins; s = s->ins->from) |
| 2829 |
{ |
| 2830 |
if (s->nins == 1 && |
| 2831 |
a->type == s->ins->type && a->co == s->ins->co) |
| 2832 |
{ |
| 2833 |
canmerge = 1; |
| 2834 |
break; |
| 2835 |
} |
| 2836 |
} |
| 2837 |
} |
| 2838 |
|
| 2839 |
if (canmerge) |
| 2840 |
{ |
| 2841 |
/* |
| 2842 |
* We can merge into sclone. If we previously made a child |
| 2843 |
* clone state, drop it; there's no need to visit it. (This |
| 2844 |
* can happen if ssource has multiple pathways to sto, and we |
| 2845 |
* only just now found one that is provably a no-op.) |
| 2846 |
*/ |
| 2847 |
if (prevclone) |
| 2848 |
dropstate(nfa, prevclone); /* kills our outarc, too */ |
| 2849 |
|
| 2850 |
/* Recurse to merge sto's outarcs into sclone */ |
| 2851 |
clonesuccessorstates(nfa, |
| 2852 |
sto, |
| 2853 |
sclone, |
| 2854 |
spredecessor, |
| 2855 |
refarc, |
| 2856 |
donemap, |
| 2857 |
outerdonemap, |
| 2858 |
nstates); |
| 2859 |
/* sto should now be marked as previously visited */ |
| 2860 |
assert(NISERR() || donemap[sto->no] == 1); |
| 2861 |
} |
| 2862 |
else if (prevclone) |
| 2863 |
{ |
| 2864 |
/* |
| 2865 |
* We already have a clone state for this successor, so just |
| 2866 |
* make another arc to it. |
| 2867 |
*/ |
| 2868 |
cparc(nfa, a, sclone, prevclone); |
| 2869 |
} |
| 2870 |
else |
| 2871 |
{ |
| 2872 |
/* |
| 2873 |
* We need to create a new successor clone state. |
| 2874 |
*/ |
| 2875 |
struct state *stoclone; |
| 2876 |
|
| 2877 |
stoclone = newstate(nfa); |
| 2878 |
if (stoclone == NULL) |
| 2879 |
{ |
| 2880 |
assert(NISERR()); |
| 2881 |
break; |
| 2882 |
} |
| 2883 |
/* Mark it as to what it's a clone of */ |
| 2884 |
stoclone->tmp = sto; |
| 2885 |
/* ... and add the outarc leading to it */ |
| 2886 |
cparc(nfa, a, sclone, stoclone); |
| 2887 |
} |
| 2888 |
} |
| 2889 |
else |
| 2890 |
{ |
| 2891 |
/* |
| 2892 |
* Non-constraint outarcs just get copied to sclone, as do outarcs |
| 2893 |
* leading to states with no constraint outarc. |
| 2894 |
*/ |
| 2895 |
cparc(nfa, a, sclone, sto); |
| 2896 |
} |
| 2897 |
} |
| 2898 |
|
| 2899 |
/* |
| 2900 |
* If we are at outer level for this clone state, recurse to all its child |
| 2901 |
* clone states, clearing their tmp fields as we go. (If we're not |
| 2902 |
* outermost for sclone, leave this to be done by the outer call level.) |
| 2903 |
* Note that if we have multiple outarcs leading to the same clone state, |
| 2904 |
* it will only be recursed-to once. |
| 2905 |
*/ |
| 2906 |
if (curdonemap == NULL) |
| 2907 |
{ |
| 2908 |
for (a = sclone->outs; a != NULL && !NISERR(); a = a->outchain) |
| 2909 |
{ |
| 2910 |
struct state *stoclone = a->to; |
| 2911 |
struct state *sto = stoclone->tmp; |
| 2912 |
|
| 2913 |
if (sto != NULL) |
| 2914 |
{ |
| 2915 |
stoclone->tmp = NULL; |
| 2916 |
clonesuccessorstates(nfa, |
| 2917 |
sto, |
| 2918 |
stoclone, |
| 2919 |
spredecessor, |
| 2920 |
refarc, |
| 2921 |
NULL, |
| 2922 |
donemap, |
| 2923 |
nstates); |
| 2924 |
} |
| 2925 |
} |
| 2926 |
|
| 2927 |
/* Don't forget to free sclone's donemap when done with it */ |
| 2928 |
FREE(donemap); |
| 2929 |
} |
| 2930 |
} |
| 2931 |
|
| 2932 |
/* |
| 2933 |
* removecantmatch - remove CANTMATCH arcs, which are no longer useful |
| 2934 |
* once we are done with the parsing phase. (We need them only to |
| 2935 |
* preserve connectedness of NFA subgraphs during parsing.) |
| 2936 |
*/ |
| 2937 |
static void |
| 2938 |
removecantmatch(struct nfa *nfa) |
| 2939 |
{ |
| 2940 |
struct state *s; |
| 2941 |
|
| 2942 |
for (s = nfa->states; s != NULL; s = s->next) |
| 2943 |
{ |
| 2944 |
struct arc *a; |
| 2945 |
struct arc *nexta; |
| 2946 |
|
| 2947 |
for (a = s->outs; a != NULL; a = nexta) |
| 2948 |
{ |
| 2949 |
nexta = a->outchain; |
| 2950 |
if (a->type == CANTMATCH) |
| 2951 |
{ |
| 2952 |
freearc(nfa, a); |
| 2953 |
if (NISERR()) |
| 2954 |
return; |
| 2955 |
} |
| 2956 |
} |
| 2957 |
} |
| 2958 |
} |
| 2959 |
|
| 2960 |
/* |
| 2961 |
* cleanup - clean up NFA after optimizations |
| 2962 |
*/ |
| 2963 |
static void |
| 2964 |
cleanup(struct nfa *nfa) |
| 2965 |
{ |
| 2966 |
struct state *s; |
| 2967 |
struct state *nexts; |
| 2968 |
int n; |
| 2969 |
|
| 2970 |
if (NISERR()) |
| 2971 |
return; |
| 2972 |
|
| 2973 |
/* clear out unreachable or dead-end states */ |
| 2974 |
/* use pre to mark reachable, then post to mark can-reach-post */ |
| 2975 |
markreachable(nfa, nfa->pre, (struct state *) NULL, nfa->pre); |
| 2976 |
markcanreach(nfa, nfa->post, nfa->pre, nfa->post); |
| 2977 |
for (s = nfa->states; s != NULL && !NISERR(); s = nexts) |
| 2978 |
{ |
| 2979 |
nexts = s->next; |
| 2980 |
if (s->tmp != nfa->post && !s->flag) |
| 2981 |
dropstate(nfa, s); |
| 2982 |
} |
| 2983 |
assert(NISERR() || nfa->post->nins == 0 || nfa->post->tmp == nfa->post); |
| 2984 |
cleartraverse(nfa, nfa->pre); |
| 2985 |
assert(NISERR() || nfa->post->nins == 0 || nfa->post->tmp == NULL); |
| 2986 |
/* the nins==0 (final unreachable) case will be caught later */ |
| 2987 |
|
| 2988 |
/* renumber surviving states */ |
| 2989 |
n = 0; |
| 2990 |
for (s = nfa->states; s != NULL; s = s->next) |
| 2991 |
s->no = n++; |
| 2992 |
nfa->nstates = n; |
| 2993 |
} |
| 2994 |
|
| 2995 |
/* |
| 2996 |
* markreachable - recursive marking of reachable states |
| 2997 |
*/ |
| 2998 |
static void |
| 2999 |
markreachable(struct nfa *nfa, |
| 3000 |
struct state *s, |
| 3001 |
struct state *okay, /* consider only states with this mark */ |
| 3002 |
struct state *mark) /* the value to mark with */ |
| 3003 |
{ |
| 3004 |
struct arc *a; |
| 3005 |
|
| 3006 |
/* Since this is recursive, it could be driven to stack overflow */ |
| 3007 |
if (STACK_TOO_DEEP(nfa->v->re)) |
| 3008 |
{ |
| 3009 |
NERR(REG_ETOOBIG); |
| 3010 |
return; |
| 3011 |
} |
| 3012 |
|
| 3013 |
if (s->tmp != okay) |
| 3014 |
return; |
| 3015 |
s->tmp = mark; |
| 3016 |
|
| 3017 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 3018 |
markreachable(nfa, a->to, okay, mark); |
| 3019 |
} |
| 3020 |
|
| 3021 |
/* |
| 3022 |
* markcanreach - recursive marking of states which can reach here |
| 3023 |
*/ |
| 3024 |
static void |
| 3025 |
markcanreach(struct nfa *nfa, |
| 3026 |
struct state *s, |
| 3027 |
struct state *okay, /* consider only states with this mark */ |
| 3028 |
struct state *mark) /* the value to mark with */ |
| 3029 |
{ |
| 3030 |
struct arc *a; |
| 3031 |
|
| 3032 |
/* Since this is recursive, it could be driven to stack overflow */ |
| 3033 |
if (STACK_TOO_DEEP(nfa->v->re)) |
| 3034 |
{ |
| 3035 |
NERR(REG_ETOOBIG); |
| 3036 |
return; |
| 3037 |
} |
| 3038 |
|
| 3039 |
if (s->tmp != okay) |
| 3040 |
return; |
| 3041 |
s->tmp = mark; |
| 3042 |
|
| 3043 |
for (a = s->ins; a != NULL; a = a->inchain) |
| 3044 |
markcanreach(nfa, a->from, okay, mark); |
| 3045 |
} |
| 3046 |
|
| 3047 |
/* |
| 3048 |
* analyze - ascertain potentially-useful facts about an optimized NFA |
| 3049 |
*/ |
| 3050 |
static long /* re_info bits to be ORed in */ |
| 3051 |
analyze(struct nfa *nfa) |
| 3052 |
{ |
| 3053 |
struct arc *a; |
| 3054 |
struct arc *aa; |
| 3055 |
|
| 3056 |
if (NISERR()) |
| 3057 |
return 0; |
| 3058 |
|
| 3059 |
/* Detect whether NFA can't match anything */ |
| 3060 |
if (nfa->pre->outs == NULL) |
| 3061 |
return REG_UIMPOSSIBLE; |
| 3062 |
|
| 3063 |
/* Detect whether NFA matches all strings (possibly with length bounds) */ |
| 3064 |
checkmatchall(nfa); |
| 3065 |
|
| 3066 |
/* Detect whether NFA can possibly match a zero-length string */ |
| 3067 |
for (a = nfa->pre->outs; a != NULL; a = a->outchain) |
| 3068 |
for (aa = a->to->outs; aa != NULL; aa = aa->outchain) |
| 3069 |
if (aa->to == nfa->post) |
| 3070 |
return REG_UEMPTYMATCH; |
| 3071 |
return 0; |
| 3072 |
} |
| 3073 |
|
| 3074 |
/* |
| 3075 |
* checkmatchall - does the NFA represent no more than a string length test? |
| 3076 |
* |
| 3077 |
* If so, set nfa->minmatchall and nfa->maxmatchall correctly (they are -1 |
| 3078 |
* to begin with) and set the MATCHALL bit in nfa->flags. |
| 3079 |
* |
| 3080 |
* To succeed, we require all arcs to be PLAIN RAINBOW arcs, except for those |
| 3081 |
* for pseudocolors (i.e., BOS/BOL/EOS/EOL). We must be able to reach the |
| 3082 |
* post state via RAINBOW arcs, and if there are any loops in the graph, they |
| 3083 |
* must be loop-to-self arcs, ensuring that each loop iteration consumes |
| 3084 |
* exactly one character. (Longer loops are problematic because they create |
| 3085 |
* non-consecutive possible match lengths; we have no good way to represent |
| 3086 |
* that situation for lengths beyond the DUPINF limit.) |
| 3087 |
* |
| 3088 |
* Pseudocolor arcs complicate things a little. We know that they can only |
| 3089 |
* appear as pre-state outarcs (for BOS/BOL) or post-state inarcs (for |
| 3090 |
* EOS/EOL). There, they must exactly replicate the parallel RAINBOW arcs, |
| 3091 |
* e.g. if the pre state has one RAINBOW outarc to state 2, it must have BOS |
| 3092 |
* and BOL outarcs to state 2, and no others. Missing or extra pseudocolor |
| 3093 |
* arcs can occur, meaning that the NFA involves some constraint on the |
| 3094 |
* adjacent characters, which makes it not a matchall NFA. |
| 3095 |
*/ |
| 3096 |
static void |
| 3097 |
checkmatchall(struct nfa *nfa) |
| 3098 |
{ |
| 3099 |
bool **haspaths; |
| 3100 |
struct state *s; |
| 3101 |
int i; |
| 3102 |
|
| 3103 |
/* |
| 3104 |
* If there are too many states, don't bother trying to detect matchall. |
| 3105 |
* This limit serves to bound the time and memory we could consume below. |
| 3106 |
* Note that even if the graph is all-RAINBOW, if there are significantly |
| 3107 |
* more than DUPINF states then it's likely that there are paths of length |
| 3108 |
* more than DUPINF, which would force us to fail anyhow. In practice, |
| 3109 |
* plausible ways of writing a matchall regex with maximum finite path |
| 3110 |
* length K tend not to have very many more than K states. |
| 3111 |
*/ |
| 3112 |
if (nfa->nstates > DUPINF * 2) |
| 3113 |
return; |
| 3114 |
|
| 3115 |
/* |
| 3116 |
* First, scan all the states to verify that only RAINBOW arcs appear, |
| 3117 |
* plus pseudocolor arcs adjacent to the pre and post states. This lets |
| 3118 |
* us quickly eliminate most cases that aren't matchall NFAs. |
| 3119 |
*/ |
| 3120 |
for (s = nfa->states; s != NULL; s = s->next) |
| 3121 |
{ |
| 3122 |
struct arc *a; |
| 3123 |
|
| 3124 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 3125 |
{ |
| 3126 |
if (a->type != PLAIN) |
| 3127 |
return; /* any LACONs make it non-matchall */ |
| 3128 |
if (a->co != RAINBOW) |
| 3129 |
{ |
| 3130 |
if (nfa->cm->cd[a->co].flags & PSEUDO) |
| 3131 |
{ |
| 3132 |
/* |
| 3133 |
* Pseudocolor arc: verify it's in a valid place (this |
| 3134 |
* seems quite unlikely to fail, but let's be sure). |
| 3135 |
*/ |
| 3136 |
if (s == nfa->pre && |
| 3137 |
(a->co == nfa->bos[0] || a->co == nfa->bos[1])) |
| 3138 |
/* okay BOS/BOL arc */ ; |
| 3139 |
else if (a->to == nfa->post && |
| 3140 |
(a->co == nfa->eos[0] || a->co == nfa->eos[1])) |
| 3141 |
/* okay EOS/EOL arc */ ; |
| 3142 |
else |
| 3143 |
return; /* unexpected pseudocolor arc */ |
| 3144 |
/* We'll check these arcs some more below. */ |
| 3145 |
} |
| 3146 |
else |
| 3147 |
return; /* any other color makes it non-matchall */ |
| 3148 |
} |
| 3149 |
} |
| 3150 |
/* Also, assert that the tmp fields are available for use. */ |
| 3151 |
assert(s->tmp == NULL); |
| 3152 |
} |
| 3153 |
|
| 3154 |
/* |
| 3155 |
* The next cheapest check we can make is to verify that the BOS/BOL |
| 3156 |
* outarcs of the pre state reach the same states as its RAINBOW outarcs. |
| 3157 |
* If they don't, the NFA expresses some constraints on the character |
| 3158 |
* before the matched string, making it non-matchall. Likewise, the |
| 3159 |
* EOS/EOL inarcs of the post state must match its RAINBOW inarcs. |
| 3160 |
*/ |
| 3161 |
if (!check_out_colors_match(nfa->pre, RAINBOW, nfa->bos[0]) || |
| 3162 |
!check_out_colors_match(nfa->pre, RAINBOW, nfa->bos[1]) || |
| 3163 |
!check_in_colors_match(nfa->post, RAINBOW, nfa->eos[0]) || |
| 3164 |
!check_in_colors_match(nfa->post, RAINBOW, nfa->eos[1])) |
| 3165 |
return; |
| 3166 |
|
| 3167 |
/* |
| 3168 |
* Initialize an array of path-length arrays, in which |
| 3169 |
* checkmatchall_recurse will return per-state results. This lets us |
| 3170 |
* memo-ize the recursive search and avoid exponential time consumption. |
| 3171 |
*/ |
| 3172 |
haspaths = (bool **) MALLOC(nfa->nstates * sizeof(bool *)); |
| 3173 |
if (haspaths == NULL) |
| 3174 |
return; /* fail quietly */ |
| 3175 |
memset(haspaths, 0, nfa->nstates * sizeof(bool *)); |
| 3176 |
|
| 3177 |
/* |
| 3178 |
* Recursively search the graph for all-RAINBOW paths to the "post" state, |
| 3179 |
* starting at the "pre" state, and computing the lengths of the paths. |
| 3180 |
* (Given the preceding checks, there should be at least one such path. |
| 3181 |
* However we could get back a false result anyway, in case there are |
| 3182 |
* multi-state loops, paths exceeding DUPINF+1 length, or non-algorithmic |
| 3183 |
* failures such as ENOMEM.) |
| 3184 |
*/ |
| 3185 |
if (checkmatchall_recurse(nfa, nfa->pre, haspaths)) |
| 3186 |
{ |
| 3187 |
/* The useful result is the path length array for the pre state */ |
| 3188 |
bool *haspath = haspaths[nfa->pre->no]; |
| 3189 |
int minmatch, |
| 3190 |
maxmatch, |
| 3191 |
morematch; |
| 3192 |
|
| 3193 |
assert(haspath != NULL); |
| 3194 |
|
| 3195 |
/* |
| 3196 |
* haspath[] now represents the set of possible path lengths; but we |
| 3197 |
* want to reduce that to a min and max value, because it doesn't seem |
| 3198 |
* worth complicating regexec.c to deal with nonconsecutive possible |
| 3199 |
* match lengths. Find min and max of first run of lengths, then |
| 3200 |
* verify there are no nonconsecutive lengths. |
| 3201 |
*/ |
| 3202 |
for (minmatch = 0; minmatch <= DUPINF + 1; minmatch++) |
| 3203 |
{ |
| 3204 |
if (haspath[minmatch]) |
| 3205 |
break; |
| 3206 |
} |
| 3207 |
assert(minmatch <= DUPINF + 1); /* else checkmatchall_recurse lied */ |
| 3208 |
for (maxmatch = minmatch; maxmatch < DUPINF + 1; maxmatch++) |
| 3209 |
{ |
| 3210 |
if (!haspath[maxmatch + 1]) |
| 3211 |
break; |
| 3212 |
} |
| 3213 |
for (morematch = maxmatch + 1; morematch <= DUPINF + 1; morematch++) |
| 3214 |
{ |
| 3215 |
if (haspath[morematch]) |
| 3216 |
{ |
| 3217 |
haspath = NULL; /* fail, there are nonconsecutive lengths */ |
| 3218 |
break; |
| 3219 |
} |
| 3220 |
} |
| 3221 |
|
| 3222 |
if (haspath != NULL) |
| 3223 |
{ |
| 3224 |
/* |
| 3225 |
* Success, so record the info. Here we have a fine point: the |
| 3226 |
* path length from the pre state includes the pre-to-initial |
| 3227 |
* transition, so it's one more than the actually matched string |
| 3228 |
* length. (We avoided counting the final-to-post transition |
| 3229 |
* within checkmatchall_recurse, but not this one.) This is why |
| 3230 |
* checkmatchall_recurse allows one more level of path length than |
| 3231 |
* might seem necessary. This decrement also takes care of |
| 3232 |
* converting checkmatchall_recurse's definition of "infinity" as |
| 3233 |
* "DUPINF+1" to our normal representation as "DUPINF". |
| 3234 |
*/ |
| 3235 |
assert(minmatch > 0); /* else pre and post states were adjacent */ |
| 3236 |
nfa->minmatchall = minmatch - 1; |
| 3237 |
nfa->maxmatchall = maxmatch - 1; |
| 3238 |
nfa->flags |= MATCHALL; |
| 3239 |
} |
| 3240 |
} |
| 3241 |
|
| 3242 |
/* Clean up */ |
| 3243 |
for (i = 0; i < nfa->nstates; i++) |
| 3244 |
{ |
| 3245 |
if (haspaths[i] != NULL) |
| 3246 |
FREE(haspaths[i]); |
| 3247 |
} |
| 3248 |
FREE(haspaths); |
| 3249 |
} |
| 3250 |
|
| 3251 |
/* |
| 3252 |
* checkmatchall_recurse - recursive search for checkmatchall |
| 3253 |
* |
| 3254 |
* s is the state to be examined in this recursion level. |
| 3255 |
* haspaths[] is an array of per-state exit path length arrays. |
| 3256 |
* |
| 3257 |
* We return true if the search was performed successfully, false if |
| 3258 |
* we had to fail because of multi-state loops or other internal reasons. |
| 3259 |
* (Because "dead" states that can't reach the post state have been |
| 3260 |
* eliminated, and we already verified that only RAINBOW and matching |
| 3261 |
* pseudocolor arcs exist, every state should have RAINBOW path(s) to |
| 3262 |
* the post state. Hence we take a false result from recursive calls |
| 3263 |
* as meaning that we'd better fail altogether, not just that that |
| 3264 |
* particular state can't reach the post state.) |
| 3265 |
* |
| 3266 |
* On success, we store a malloc'd result array in haspaths[s->no], |
| 3267 |
* showing the possible path lengths from s to the post state. |
| 3268 |
* Each state's haspath[] array is of length DUPINF+2. The entries from |
| 3269 |
* k = 0 to DUPINF are true if there is an all-RAINBOW path of length k |
| 3270 |
* from this state to the string end. haspath[DUPINF+1] is true if all |
| 3271 |
* path lengths >= DUPINF+1 are possible. (Situations that cannot be |
| 3272 |
* represented under these rules cause failure.) |
| 3273 |
* |
| 3274 |
* checkmatchall is responsible for eventually freeing the haspath[] arrays. |
| 3275 |
*/ |
| 3276 |
static bool |
| 3277 |
checkmatchall_recurse(struct nfa *nfa, struct state *s, bool **haspaths) |
| 3278 |
{ |
| 3279 |
bool result = false; |
| 3280 |
bool foundloop = false; |
| 3281 |
bool *haspath; |
| 3282 |
struct arc *a; |
| 3283 |
|
| 3284 |
/* |
| 3285 |
* Since this is recursive, it could be driven to stack overflow. But we |
| 3286 |
* need not treat that as a hard failure; just deem the NFA non-matchall. |
| 3287 |
*/ |
| 3288 |
if (STACK_TOO_DEEP(nfa->v->re)) |
| 3289 |
return false; |
| 3290 |
|
| 3291 |
/* In case the search takes a long time, check for cancel */ |
| 3292 |
INTERRUPT(nfa->v->re); |
| 3293 |
|
| 3294 |
/* Create a haspath array for this state */ |
| 3295 |
haspath = (bool *) MALLOC((DUPINF + 2) * sizeof(bool)); |
| 3296 |
if (haspath == NULL) |
| 3297 |
return false; /* again, treat as non-matchall */ |
| 3298 |
memset(haspath, 0, (DUPINF + 2) * sizeof(bool)); |
| 3299 |
|
| 3300 |
/* Mark this state as being visited */ |
| 3301 |
assert(s->tmp == NULL); |
| 3302 |
s->tmp = s; |
| 3303 |
|
| 3304 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 3305 |
{ |
| 3306 |
if (a->co != RAINBOW) |
| 3307 |
continue; /* ignore pseudocolor arcs */ |
| 3308 |
if (a->to == nfa->post) |
| 3309 |
{ |
| 3310 |
/* We found an all-RAINBOW path to the post state */ |
| 3311 |
result = true; |
| 3312 |
|
| 3313 |
/* |
| 3314 |
* Mark this state as being zero steps away from the string end |
| 3315 |
* (the transition to the post state isn't counted). |
| 3316 |
*/ |
| 3317 |
haspath[0] = true; |
| 3318 |
} |
| 3319 |
else if (a->to == s) |
| 3320 |
{ |
| 3321 |
/* We found a cycle of length 1, which we'll deal with below. */ |
| 3322 |
foundloop = true; |
| 3323 |
} |
| 3324 |
else if (a->to->tmp != NULL) |
| 3325 |
{ |
| 3326 |
/* It's busy, so we found a cycle of length > 1, so fail. */ |
| 3327 |
result = false; |
| 3328 |
break; |
| 3329 |
} |
| 3330 |
else |
| 3331 |
{ |
| 3332 |
/* Consider paths forward through this to-state. */ |
| 3333 |
bool *nexthaspath; |
| 3334 |
int i; |
| 3335 |
|
| 3336 |
/* If to-state was not already visited, recurse */ |
| 3337 |
if (haspaths[a->to->no] == NULL) |
| 3338 |
{ |
| 3339 |
result = checkmatchall_recurse(nfa, a->to, haspaths); |
| 3340 |
/* Fail if any recursive path fails */ |
| 3341 |
if (!result) |
| 3342 |
break; |
| 3343 |
} |
| 3344 |
else |
| 3345 |
{ |
| 3346 |
/* The previous visit must have found path(s) to the end */ |
| 3347 |
result = true; |
| 3348 |
} |
| 3349 |
assert(a->to->tmp == NULL); |
| 3350 |
nexthaspath = haspaths[a->to->no]; |
| 3351 |
assert(nexthaspath != NULL); |
| 3352 |
|
| 3353 |
/* |
| 3354 |
* Now, for every path of length i from a->to to the string end, |
| 3355 |
* there is a path of length i + 1 from s to the string end. |
| 3356 |
*/ |
| 3357 |
if (nexthaspath[DUPINF] != nexthaspath[DUPINF + 1]) |
| 3358 |
{ |
| 3359 |
/* |
| 3360 |
* a->to has a path of length exactly DUPINF, but not longer; |
| 3361 |
* or it has paths of all lengths > DUPINF but not one of |
| 3362 |
* exactly that length. In either case, we cannot represent |
| 3363 |
* the possible path lengths from s correctly, so fail. |
| 3364 |
*/ |
| 3365 |
result = false; |
| 3366 |
break; |
| 3367 |
} |
| 3368 |
/* Merge knowledge of these path lengths into what we have */ |
| 3369 |
for (i = 0; i < DUPINF; i++) |
| 3370 |
haspath[i + 1] |= nexthaspath[i]; |
| 3371 |
/* Infinity + 1 is still infinity */ |
| 3372 |
haspath[DUPINF + 1] |= nexthaspath[DUPINF + 1]; |
| 3373 |
} |
| 3374 |
} |
| 3375 |
|
| 3376 |
if (result && foundloop) |
| 3377 |
{ |
| 3378 |
/* |
| 3379 |
* If there is a length-1 loop at this state, then find the shortest |
| 3380 |
* known path length to the end. The loop means that every larger |
| 3381 |
* path length is possible, too. (It doesn't matter whether any of |
| 3382 |
* the longer lengths were already known possible.) |
| 3383 |
*/ |
| 3384 |
int i; |
| 3385 |
|
| 3386 |
for (i = 0; i <= DUPINF; i++) |
| 3387 |
{ |
| 3388 |
if (haspath[i]) |
| 3389 |
break; |
| 3390 |
} |
| 3391 |
for (i++; i <= DUPINF + 1; i++) |
| 3392 |
haspath[i] = true; |
| 3393 |
} |
| 3394 |
|
| 3395 |
/* Report out the completed path length map */ |
| 3396 |
assert(s->no < nfa->nstates); |
| 3397 |
assert(haspaths[s->no] == NULL); |
| 3398 |
haspaths[s->no] = haspath; |
| 3399 |
|
| 3400 |
/* Mark state no longer busy */ |
| 3401 |
s->tmp = NULL; |
| 3402 |
|
| 3403 |
return result; |
| 3404 |
} |
| 3405 |
|
| 3406 |
/* |
| 3407 |
* check_out_colors_match - subroutine for checkmatchall |
| 3408 |
* |
| 3409 |
* Check whether the set of states reachable from s by arcs of color co1 |
| 3410 |
* is equivalent to the set reachable by arcs of color co2. |
| 3411 |
* checkmatchall already verified that all of the NFA's arcs are PLAIN, |
| 3412 |
* so we need not examine arc types here. |
| 3413 |
*/ |
| 3414 |
static bool |
| 3415 |
check_out_colors_match(struct state *s, color co1, color co2) |
| 3416 |
{ |
| 3417 |
bool result = true; |
| 3418 |
struct arc *a; |
| 3419 |
|
| 3420 |
/* |
| 3421 |
* To do this in linear time, we assume that the NFA contains no duplicate |
| 3422 |
* arcs. Run through the out-arcs, marking states reachable by arcs of |
| 3423 |
* color co1. Run through again, un-marking states reachable by arcs of |
| 3424 |
* color co2; if we see a not-marked state, we know this co2 arc is |
| 3425 |
* unmatched. Then run through again, checking for still-marked states, |
| 3426 |
* and in any case leaving all the tmp fields reset to NULL. |
| 3427 |
*/ |
| 3428 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 3429 |
{ |
| 3430 |
if (a->co == co1) |
| 3431 |
{ |
| 3432 |
assert(a->to->tmp == NULL); |
| 3433 |
a->to->tmp = a->to; |
| 3434 |
} |
| 3435 |
} |
| 3436 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 3437 |
{ |
| 3438 |
if (a->co == co2) |
| 3439 |
{ |
| 3440 |
if (a->to->tmp != NULL) |
| 3441 |
a->to->tmp = NULL; |
| 3442 |
else |
| 3443 |
result = false; /* unmatched co2 arc */ |
| 3444 |
} |
| 3445 |
} |
| 3446 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 3447 |
{ |
| 3448 |
if (a->co == co1) |
| 3449 |
{ |
| 3450 |
if (a->to->tmp != NULL) |
| 3451 |
{ |
| 3452 |
result = false; /* unmatched co1 arc */ |
| 3453 |
a->to->tmp = NULL; |
| 3454 |
} |
| 3455 |
} |
| 3456 |
} |
| 3457 |
return result; |
| 3458 |
} |
| 3459 |
|
| 3460 |
/* |
| 3461 |
* check_in_colors_match - subroutine for checkmatchall |
| 3462 |
* |
| 3463 |
* Check whether the set of states that can reach s by arcs of color co1 |
| 3464 |
* is equivalent to the set that can reach s by arcs of color co2. |
| 3465 |
* checkmatchall already verified that all of the NFA's arcs are PLAIN, |
| 3466 |
* so we need not examine arc types here. |
| 3467 |
*/ |
| 3468 |
static bool |
| 3469 |
check_in_colors_match(struct state *s, color co1, color co2) |
| 3470 |
{ |
| 3471 |
bool result = true; |
| 3472 |
struct arc *a; |
| 3473 |
|
| 3474 |
/* |
| 3475 |
* Identical algorithm to check_out_colors_match, except examine the |
| 3476 |
* from-states of s' inarcs. |
| 3477 |
*/ |
| 3478 |
for (a = s->ins; a != NULL; a = a->inchain) |
| 3479 |
{ |
| 3480 |
if (a->co == co1) |
| 3481 |
{ |
| 3482 |
assert(a->from->tmp == NULL); |
| 3483 |
a->from->tmp = a->from; |
| 3484 |
} |
| 3485 |
} |
| 3486 |
for (a = s->ins; a != NULL; a = a->inchain) |
| 3487 |
{ |
| 3488 |
if (a->co == co2) |
| 3489 |
{ |
| 3490 |
if (a->from->tmp != NULL) |
| 3491 |
a->from->tmp = NULL; |
| 3492 |
else |
| 3493 |
result = false; /* unmatched co2 arc */ |
| 3494 |
} |
| 3495 |
} |
| 3496 |
for (a = s->ins; a != NULL; a = a->inchain) |
| 3497 |
{ |
| 3498 |
if (a->co == co1) |
| 3499 |
{ |
| 3500 |
if (a->from->tmp != NULL) |
| 3501 |
{ |
| 3502 |
result = false; /* unmatched co1 arc */ |
| 3503 |
a->from->tmp = NULL; |
| 3504 |
} |
| 3505 |
} |
| 3506 |
} |
| 3507 |
return result; |
| 3508 |
} |
| 3509 |
|
| 3510 |
/* |
| 3511 |
* compact - construct the compact representation of an NFA |
| 3512 |
*/ |
| 3513 |
static void |
| 3514 |
compact(struct nfa *nfa, |
| 3515 |
struct cnfa *cnfa) |
| 3516 |
{ |
| 3517 |
struct state *s; |
| 3518 |
struct arc *a; |
| 3519 |
size_t nstates; |
| 3520 |
size_t narcs; |
| 3521 |
struct carc *ca; |
| 3522 |
struct carc *first; |
| 3523 |
|
| 3524 |
assert(!NISERR()); |
| 3525 |
|
| 3526 |
/* |
| 3527 |
* The REG_MAX_COMPILE_SPACE restriction ensures that integer overflow |
| 3528 |
* can't occur in this loop nor in the allocation requests below. |
| 3529 |
*/ |
| 3530 |
nstates = 0; |
| 3531 |
narcs = 0; |
| 3532 |
for (s = nfa->states; s != NULL; s = s->next) |
| 3533 |
{ |
| 3534 |
nstates++; |
| 3535 |
narcs += s->nouts + 1; /* need one extra for endmarker */ |
| 3536 |
} |
| 3537 |
|
| 3538 |
cnfa->stflags = (char *) MALLOC(nstates * sizeof(char)); |
| 3539 |
cnfa->states = (struct carc **) MALLOC(nstates * sizeof(struct carc *)); |
| 3540 |
cnfa->arcs = (struct carc *) MALLOC(narcs * sizeof(struct carc)); |
| 3541 |
if (cnfa->stflags == NULL || cnfa->states == NULL || cnfa->arcs == NULL) |
| 3542 |
{ |
| 3543 |
if (cnfa->stflags != NULL) |
| 3544 |
FREE(cnfa->stflags); |
| 3545 |
if (cnfa->states != NULL) |
| 3546 |
FREE(cnfa->states); |
| 3547 |
if (cnfa->arcs != NULL) |
| 3548 |
FREE(cnfa->arcs); |
| 3549 |
NERR(REG_ESPACE); |
| 3550 |
return; |
| 3551 |
} |
| 3552 |
cnfa->nstates = nstates; |
| 3553 |
cnfa->pre = nfa->pre->no; |
| 3554 |
cnfa->post = nfa->post->no; |
| 3555 |
cnfa->bos[0] = nfa->bos[0]; |
| 3556 |
cnfa->bos[1] = nfa->bos[1]; |
| 3557 |
cnfa->eos[0] = nfa->eos[0]; |
| 3558 |
cnfa->eos[1] = nfa->eos[1]; |
| 3559 |
cnfa->ncolors = maxcolor(nfa->cm) + 1; |
| 3560 |
cnfa->flags = nfa->flags; |
| 3561 |
cnfa->minmatchall = nfa->minmatchall; |
| 3562 |
cnfa->maxmatchall = nfa->maxmatchall; |
| 3563 |
|
| 3564 |
ca = cnfa->arcs; |
| 3565 |
for (s = nfa->states; s != NULL; s = s->next) |
| 3566 |
{ |
| 3567 |
assert((size_t) s->no < nstates); |
| 3568 |
cnfa->stflags[s->no] = 0; |
| 3569 |
cnfa->states[s->no] = ca; |
| 3570 |
first = ca; |
| 3571 |
for (a = s->outs; a != NULL; a = a->outchain) |
| 3572 |
switch (a->type) |
| 3573 |
{ |
| 3574 |
case PLAIN: |
| 3575 |
ca->co = a->co; |
| 3576 |
ca->to = a->to->no; |
| 3577 |
ca++; |
| 3578 |
break; |
| 3579 |
case LACON: |
| 3580 |
assert(s->no != cnfa->pre); |
| 3581 |
assert(a->co >= 0); |
| 3582 |
/* make sure the modified color number will fit */ |
| 3583 |
if (a->co > MAX_COLOR - cnfa->ncolors) |
| 3584 |
{ |
| 3585 |
NERR(REG_ECOLORS); |
| 3586 |
return; |
| 3587 |
} |
| 3588 |
ca->co = (color) (cnfa->ncolors + a->co); |
| 3589 |
ca->to = a->to->no; |
| 3590 |
ca++; |
| 3591 |
cnfa->flags |= HASLACONS; |
| 3592 |
break; |
| 3593 |
default: |
| 3594 |
NERR(REG_ASSERT); |
| 3595 |
return; |
| 3596 |
} |
| 3597 |
carcsort(first, ca - first); |
| 3598 |
ca->co = COLORLESS; |
| 3599 |
ca->to = 0; |
| 3600 |
ca++; |
| 3601 |
} |
| 3602 |
assert(ca == &cnfa->arcs[narcs]); |
| 3603 |
assert(cnfa->nstates != 0); |
| 3604 |
|
| 3605 |
/* mark no-progress states */ |
| 3606 |
for (a = nfa->pre->outs; a != NULL; a = a->outchain) |
| 3607 |
cnfa->stflags[a->to->no] = CNFA_NOPROGRESS; |
| 3608 |
cnfa->stflags[nfa->pre->no] = CNFA_NOPROGRESS; |
| 3609 |
} |
| 3610 |
|
| 3611 |
/* |
| 3612 |
* carcsort - sort compacted-NFA arcs by color |
| 3613 |
*/ |
| 3614 |
static void |
| 3615 |
carcsort(struct carc *first, size_t n) |
| 3616 |
{ |
| 3617 |
if (n > 1) |
| 3618 |
qsort(first, n, sizeof(struct carc), carc_cmp); |
| 3619 |
} |
| 3620 |
|
| 3621 |
static int |
| 3622 |
carc_cmp(const void *a, const void *b) |
| 3623 |
{ |
| 3624 |
const struct carc *aa = (const struct carc *) a; |
| 3625 |
const struct carc *bb = (const struct carc *) b; |
| 3626 |
|
| 3627 |
if (aa->co < bb->co) |
| 3628 |
return -1; |
| 3629 |
if (aa->co > bb->co) |
| 3630 |
return +1; |
| 3631 |
if (aa->to < bb->to) |
| 3632 |
return -1; |
| 3633 |
if (aa->to > bb->to) |
| 3634 |
return +1; |
| 3635 |
/* This is unreached, since there should be no duplicate arcs now: */ |
| 3636 |
return 0; |
| 3637 |
} |
| 3638 |
|
| 3639 |
/* |
| 3640 |
* freecnfa - free a compacted NFA |
| 3641 |
*/ |
| 3642 |
static void |
| 3643 |
freecnfa(struct cnfa *cnfa) |
| 3644 |
{ |
| 3645 |
assert(!NULLCNFA(*cnfa)); /* not empty already */ |
| 3646 |
FREE(cnfa->stflags); |
| 3647 |
FREE(cnfa->states); |
| 3648 |
FREE(cnfa->arcs); |
| 3649 |
ZAPCNFA(*cnfa); |
| 3650 |
} |
| 3651 |
|
| 3652 |
/* |
| 3653 |
* dumpnfa - dump an NFA in human-readable form |
| 3654 |
*/ |
| 3655 |
static void |
| 3656 |
dumpnfa(struct nfa *nfa, |
| 3657 |
FILE *f) |
| 3658 |
{ |
| 3659 |
#ifdef REG_DEBUG |
| 3660 |
struct state *s; |
| 3661 |
int nstates = 0; |
| 3662 |
int narcs = 0; |
| 3663 |
|
| 3664 |
fprintf(f, "pre %d, post %d", nfa->pre->no, nfa->post->no); |
| 3665 |
if (nfa->bos[0] != COLORLESS) |
| 3666 |
fprintf(f, ", bos [%ld]", (long) nfa->bos[0]); |
| 3667 |
if (nfa->bos[1] != COLORLESS) |
| 3668 |
fprintf(f, ", bol [%ld]", (long) nfa->bos[1]); |
| 3669 |
if (nfa->eos[0] != COLORLESS) |
| 3670 |
fprintf(f, ", eos [%ld]", (long) nfa->eos[0]); |
| 3671 |
if (nfa->eos[1] != COLORLESS) |
| 3672 |
fprintf(f, ", eol [%ld]", (long) nfa->eos[1]); |
| 3673 |
if (nfa->flags & HASLACONS) |
| 3674 |
fprintf(f, ", haslacons"); |
| 3675 |
if (nfa->flags & HASCANTMATCH) |
| 3676 |
fprintf(f, ", hascantmatch"); |
| 3677 |
if (nfa->flags & MATCHALL) |
| 3678 |
{ |
| 3679 |
fprintf(f, ", minmatchall %d", nfa->minmatchall); |
| 3680 |
if (nfa->maxmatchall == DUPINF) |
| 3681 |
fprintf(f, ", maxmatchall inf"); |
| 3682 |
else |
| 3683 |
fprintf(f, ", maxmatchall %d", nfa->maxmatchall); |
| 3684 |
} |
| 3685 |
fprintf(f, "\n"); |
| 3686 |
for (s = nfa->states; s != NULL; s = s->next) |
| 3687 |
{ |
| 3688 |
dumpstate(s, f); |
| 3689 |
nstates++; |
| 3690 |
narcs += s->nouts; |
| 3691 |
} |
| 3692 |
fprintf(f, "total of %d states, %d arcs\n", nstates, narcs); |
| 3693 |
if (nfa->parent == NULL) |
| 3694 |
dumpcolors(nfa->cm, f); |
| 3695 |
fflush(f); |
| 3696 |
#endif |
| 3697 |
} |
| 3698 |
|
| 3699 |
#ifdef REG_DEBUG /* subordinates of dumpnfa */ |
| 3700 |
|
| 3701 |
/* |
| 3702 |
* dumpstate - dump an NFA state in human-readable form |
| 3703 |
*/ |
| 3704 |
static void |
| 3705 |
dumpstate(struct state *s, |
| 3706 |
FILE *f) |
| 3707 |
{ |
| 3708 |
struct arc *a; |
| 3709 |
|
| 3710 |
fprintf(f, "%d%s%c", s->no, (s->tmp != NULL) ? "T" : "", |
| 3711 |
(s->flag) ? s->flag : '.'); |
| 3712 |
if (s->prev != NULL && s->prev->next != s) |
| 3713 |
fprintf(f, "\tstate chain bad\n"); |
| 3714 |
if (s->nouts == 0) |
| 3715 |
fprintf(f, "\tno out arcs\n"); |
| 3716 |
else |
| 3717 |
dumparcs(s, f); |
| 3718 |
for (a = s->ins; a != NULL; a = a->inchain) |
| 3719 |
{ |
| 3720 |
if (a->to != s) |
| 3721 |
fprintf(f, "\tlink from %d to %d on %d's in-chain\n", |
| 3722 |
a->from->no, a->to->no, s->no); |
| 3723 |
} |
| 3724 |
fflush(f); |
| 3725 |
} |
| 3726 |
|
| 3727 |
/* |
| 3728 |
* dumparcs - dump out-arcs in human-readable form |
| 3729 |
*/ |
| 3730 |
static void |
| 3731 |
dumparcs(struct state *s, |
| 3732 |
FILE *f) |
| 3733 |
{ |
| 3734 |
int pos; |
| 3735 |
struct arc *a; |
| 3736 |
|
| 3737 |
/* printing oldest arcs first is usually clearer */ |
| 3738 |
a = s->outs; |
| 3739 |
assert(a != NULL); |
| 3740 |
while (a->outchain != NULL) |
| 3741 |
a = a->outchain; |
| 3742 |
pos = 1; |
| 3743 |
do |
| 3744 |
{ |
| 3745 |
dumparc(a, s, f); |
| 3746 |
if (pos == 5) |
| 3747 |
{ |
| 3748 |
fprintf(f, "\n"); |
| 3749 |
pos = 1; |
| 3750 |
} |
| 3751 |
else |
| 3752 |
pos++; |
| 3753 |
a = a->outchainRev; |
| 3754 |
} while (a != NULL); |
| 3755 |
if (pos != 1) |
| 3756 |
fprintf(f, "\n"); |
| 3757 |
} |
| 3758 |
|
| 3759 |
/* |
| 3760 |
* dumparc - dump one outarc in readable form, including prefixing tab |
| 3761 |
*/ |
| 3762 |
static void |
| 3763 |
dumparc(struct arc *a, |
| 3764 |
struct state *s, |
| 3765 |
FILE *f) |
| 3766 |
{ |
| 3767 |
struct arc *aa; |
| 3768 |
|
| 3769 |
fprintf(f, "\t"); |
| 3770 |
switch (a->type) |
| 3771 |
{ |
| 3772 |
case PLAIN: |
| 3773 |
if (a->co == RAINBOW) |
| 3774 |
fprintf(f, "[*]"); |
| 3775 |
else |
| 3776 |
fprintf(f, "[%ld]", (long) a->co); |
| 3777 |
break; |
| 3778 |
case AHEAD: |
| 3779 |
if (a->co == RAINBOW) |
| 3780 |
fprintf(f, ">*>"); |
| 3781 |
else |
| 3782 |
fprintf(f, ">%ld>", (long) a->co); |
| 3783 |
break; |
| 3784 |
case BEHIND: |
| 3785 |
if (a->co == RAINBOW) |
| 3786 |
fprintf(f, "<*<"); |
| 3787 |
else |
| 3788 |
fprintf(f, "<%ld<", (long) a->co); |
| 3789 |
break; |
| 3790 |
case LACON: |
| 3791 |
fprintf(f, ":%ld:", (long) a->co); |
| 3792 |
break; |
| 3793 |
case '^': |
| 3794 |
case '$': |
| 3795 |
fprintf(f, "%c%d", a->type, (int) a->co); |
| 3796 |
break; |
| 3797 |
case EMPTY: |
| 3798 |
break; |
| 3799 |
case CANTMATCH: |
| 3800 |
fprintf(f, "X"); |
| 3801 |
break; |
| 3802 |
default: |
| 3803 |
fprintf(f, "0x%x/0%lo", a->type, (long) a->co); |
| 3804 |
break; |
| 3805 |
} |
| 3806 |
if (a->from != s) |
| 3807 |
fprintf(f, "?%d?", a->from->no); |
| 3808 |
for (aa = a->from->outs; aa != NULL; aa = aa->outchain) |
| 3809 |
if (aa == a) |
| 3810 |
break; /* NOTE BREAK OUT */ |
| 3811 |
if (aa == NULL) |
| 3812 |
fprintf(f, "?!?"); /* missing from out-chain */ |
| 3813 |
fprintf(f, "->"); |
| 3814 |
if (a->to == NULL) |
| 3815 |
{ |
| 3816 |
fprintf(f, "NULL"); |
| 3817 |
return; |
| 3818 |
} |
| 3819 |
fprintf(f, "%d", a->to->no); |
| 3820 |
for (aa = a->to->ins; aa != NULL; aa = aa->inchain) |
| 3821 |
if (aa == a) |
| 3822 |
break; /* NOTE BREAK OUT */ |
| 3823 |
if (aa == NULL) |
| 3824 |
fprintf(f, "?!?"); /* missing from in-chain */ |
| 3825 |
} |
| 3826 |
#endif /* REG_DEBUG */ |
| 3827 |
|
| 3828 |
/* |
| 3829 |
* dumpcnfa - dump a compacted NFA in human-readable form |
| 3830 |
*/ |
| 3831 |
#ifdef REG_DEBUG |
| 3832 |
static void |
| 3833 |
dumpcnfa(struct cnfa *cnfa, |
| 3834 |
FILE *f) |
| 3835 |
{ |
| 3836 |
int st; |
| 3837 |
|
| 3838 |
fprintf(f, "pre %d, post %d", cnfa->pre, cnfa->post); |
| 3839 |
if (cnfa->bos[0] != COLORLESS) |
| 3840 |
fprintf(f, ", bos [%ld]", (long) cnfa->bos[0]); |
| 3841 |
if (cnfa->bos[1] != COLORLESS) |
| 3842 |
fprintf(f, ", bol [%ld]", (long) cnfa->bos[1]); |
| 3843 |
if (cnfa->eos[0] != COLORLESS) |
| 3844 |
fprintf(f, ", eos [%ld]", (long) cnfa->eos[0]); |
| 3845 |
if (cnfa->eos[1] != COLORLESS) |
| 3846 |
fprintf(f, ", eol [%ld]", (long) cnfa->eos[1]); |
| 3847 |
if (cnfa->flags & HASLACONS) |
| 3848 |
fprintf(f, ", haslacons"); |
| 3849 |
if (cnfa->flags & MATCHALL) |
| 3850 |
{ |
| 3851 |
fprintf(f, ", minmatchall %d", cnfa->minmatchall); |
| 3852 |
if (cnfa->maxmatchall == DUPINF) |
| 3853 |
fprintf(f, ", maxmatchall inf"); |
| 3854 |
else |
| 3855 |
fprintf(f, ", maxmatchall %d", cnfa->maxmatchall); |
| 3856 |
} |
| 3857 |
fprintf(f, "\n"); |
| 3858 |
for (st = 0; st < cnfa->nstates; st++) |
| 3859 |
dumpcstate(st, cnfa, f); |
| 3860 |
fflush(f); |
| 3861 |
} |
| 3862 |
#endif |
| 3863 |
|
| 3864 |
#ifdef REG_DEBUG /* subordinates of dumpcnfa */ |
| 3865 |
|
| 3866 |
/* |
| 3867 |
* dumpcstate - dump a compacted-NFA state in human-readable form |
| 3868 |
*/ |
| 3869 |
static void |
| 3870 |
dumpcstate(int st, |
| 3871 |
struct cnfa *cnfa, |
| 3872 |
FILE *f) |
| 3873 |
{ |
| 3874 |
struct carc *ca; |
| 3875 |
int pos; |
| 3876 |
|
| 3877 |
fprintf(f, "%d%s", st, (cnfa->stflags[st] & CNFA_NOPROGRESS) ? ":" : "."); |
| 3878 |
pos = 1; |
| 3879 |
for (ca = cnfa->states[st]; ca->co != COLORLESS; ca++) |
| 3880 |
{ |
| 3881 |
if (ca->co == RAINBOW) |
| 3882 |
fprintf(f, "\t[*]->%d", ca->to); |
| 3883 |
else if (ca->co < cnfa->ncolors) |
| 3884 |
fprintf(f, "\t[%ld]->%d", (long) ca->co, ca->to); |
| 3885 |
else |
| 3886 |
fprintf(f, "\t:%ld:->%d", (long) (ca->co - cnfa->ncolors), ca->to); |
| 3887 |
if (pos == 5) |
| 3888 |
{ |
| 3889 |
fprintf(f, "\n"); |
| 3890 |
pos = 1; |
| 3891 |
} |
| 3892 |
else |
| 3893 |
pos++; |
| 3894 |
} |
| 3895 |
if (ca == cnfa->states[st] || pos != 1) |
| 3896 |
fprintf(f, "\n"); |
| 3897 |
fflush(f); |
| 3898 |
} |
| 3899 |
|
| 3900 |
#endif /* REG_DEBUG */ |