blob: 481837b448c77a0a9e38a3c0807a6e1a07ad0aef [file] [log] [blame]
Daniel Veillard4255d502002-04-16 15:50:10 +00001/*
2 * regexp.c: generic and extensible Regular Expression engine
3 *
4 * Basically designed with the purpose of compiling regexps for
5 * the variety of validation/shemas mechanisms now available in
6 * XML related specifications thise includes:
7 * - XML-1.0 DTD validation
8 * - XML Schemas structure part 1
9 * - XML Schemas Datatypes part 2 especially Appendix F
10 * - RELAX-NG/TREX i.e. the counter proposal
11 *
12 * See Copyright for the status of this software.
13 *
14 * Daniel Veillard <veillard@redhat.com>
15 */
16
17#define IN_LIBXML
18#include "libxml.h"
19
20#ifdef LIBXML_REGEXP_ENABLED
21
22#include <stdio.h>
23#include <string.h>
24#include <libxml/tree.h>
25#include <libxml/parserInternals.h>
26#include <libxml/xmlregexp.h>
27#include <libxml/xmlautomata.h>
28#include <libxml/xmlunicode.h>
29
30/* #define DEBUG_REGEXP_GRAPH */
31/* #define DEBUG_REGEXP_EXEC */
32/* #define DEBUG_PUSH */
Daniel Veillard23e73572002-09-19 19:56:43 +000033/* #define DEBUG_COMPACTION */
Daniel Veillard4255d502002-04-16 15:50:10 +000034
35#define ERROR(str) ctxt->error = 1; \
36 xmlGenericError(xmlGenericErrorContext, "Regexp: %s: %s\n", str, ctxt->cur)
37#define NEXT ctxt->cur++
38#define CUR (*(ctxt->cur))
39#define NXT(index) (ctxt->cur[index])
40
41#define CUR_SCHAR(s, l) xmlStringCurrentChar(NULL, s, &l)
42#define NEXTL(l) ctxt->cur += l;
43
Daniel Veillarde19fc232002-04-22 16:01:24 +000044/**
45 * TODO:
46 *
47 * macro to flag unimplemented blocks
48 */
49#define TODO \
50 xmlGenericError(xmlGenericErrorContext, \
51 "Unimplemented block at %s:%d\n", \
52 __FILE__, __LINE__);
53
Daniel Veillard4255d502002-04-16 15:50:10 +000054
55/************************************************************************
56 * *
57 * Datatypes and structures *
58 * *
59 ************************************************************************/
60
61typedef enum {
62 XML_REGEXP_EPSILON = 1,
63 XML_REGEXP_CHARVAL,
64 XML_REGEXP_RANGES,
65 XML_REGEXP_SUBREG,
66 XML_REGEXP_STRING,
67 XML_REGEXP_ANYCHAR, /* . */
68 XML_REGEXP_ANYSPACE, /* \s */
69 XML_REGEXP_NOTSPACE, /* \S */
70 XML_REGEXP_INITNAME, /* \l */
71 XML_REGEXP_NOTINITNAME, /* \l */
72 XML_REGEXP_NAMECHAR, /* \c */
73 XML_REGEXP_NOTNAMECHAR, /* \C */
74 XML_REGEXP_DECIMAL, /* \d */
75 XML_REGEXP_NOTDECIMAL, /* \d */
76 XML_REGEXP_REALCHAR, /* \w */
77 XML_REGEXP_NOTREALCHAR, /* \w */
78 XML_REGEXP_LETTER,
79 XML_REGEXP_LETTER_UPPERCASE,
80 XML_REGEXP_LETTER_LOWERCASE,
81 XML_REGEXP_LETTER_TITLECASE,
82 XML_REGEXP_LETTER_MODIFIER,
83 XML_REGEXP_LETTER_OTHERS,
84 XML_REGEXP_MARK,
85 XML_REGEXP_MARK_NONSPACING,
86 XML_REGEXP_MARK_SPACECOMBINING,
87 XML_REGEXP_MARK_ENCLOSING,
88 XML_REGEXP_NUMBER,
89 XML_REGEXP_NUMBER_DECIMAL,
90 XML_REGEXP_NUMBER_LETTER,
91 XML_REGEXP_NUMBER_OTHERS,
92 XML_REGEXP_PUNCT,
93 XML_REGEXP_PUNCT_CONNECTOR,
94 XML_REGEXP_PUNCT_DASH,
95 XML_REGEXP_PUNCT_OPEN,
96 XML_REGEXP_PUNCT_CLOSE,
97 XML_REGEXP_PUNCT_INITQUOTE,
98 XML_REGEXP_PUNCT_FINQUOTE,
99 XML_REGEXP_PUNCT_OTHERS,
100 XML_REGEXP_SEPAR,
101 XML_REGEXP_SEPAR_SPACE,
102 XML_REGEXP_SEPAR_LINE,
103 XML_REGEXP_SEPAR_PARA,
104 XML_REGEXP_SYMBOL,
105 XML_REGEXP_SYMBOL_MATH,
106 XML_REGEXP_SYMBOL_CURRENCY,
107 XML_REGEXP_SYMBOL_MODIFIER,
108 XML_REGEXP_SYMBOL_OTHERS,
109 XML_REGEXP_OTHER,
110 XML_REGEXP_OTHER_CONTROL,
111 XML_REGEXP_OTHER_FORMAT,
112 XML_REGEXP_OTHER_PRIVATE,
113 XML_REGEXP_OTHER_NA,
114 XML_REGEXP_BLOCK_NAME
115} xmlRegAtomType;
116
117typedef enum {
118 XML_REGEXP_QUANT_EPSILON = 1,
119 XML_REGEXP_QUANT_ONCE,
120 XML_REGEXP_QUANT_OPT,
121 XML_REGEXP_QUANT_MULT,
122 XML_REGEXP_QUANT_PLUS,
Daniel Veillard7646b182002-04-20 06:41:40 +0000123 XML_REGEXP_QUANT_ONCEONLY,
124 XML_REGEXP_QUANT_ALL,
Daniel Veillard4255d502002-04-16 15:50:10 +0000125 XML_REGEXP_QUANT_RANGE
126} xmlRegQuantType;
127
128typedef enum {
129 XML_REGEXP_START_STATE = 1,
130 XML_REGEXP_FINAL_STATE,
131 XML_REGEXP_TRANS_STATE
132} xmlRegStateType;
133
134typedef enum {
135 XML_REGEXP_MARK_NORMAL = 0,
136 XML_REGEXP_MARK_START,
137 XML_REGEXP_MARK_VISITED
138} xmlRegMarkedType;
139
140typedef struct _xmlRegRange xmlRegRange;
141typedef xmlRegRange *xmlRegRangePtr;
142
143struct _xmlRegRange {
144 int neg;
145 xmlRegAtomType type;
146 int start;
147 int end;
148 xmlChar *blockName;
149};
150
151typedef struct _xmlRegAtom xmlRegAtom;
152typedef xmlRegAtom *xmlRegAtomPtr;
153
154typedef struct _xmlAutomataState xmlRegState;
155typedef xmlRegState *xmlRegStatePtr;
156
157struct _xmlRegAtom {
158 int no;
159 xmlRegAtomType type;
160 xmlRegQuantType quant;
161 int min;
162 int max;
163
164 void *valuep;
Daniel Veillarda646cfd2002-09-17 21:50:03 +0000165 void *valuep2;
Daniel Veillard4255d502002-04-16 15:50:10 +0000166 int neg;
167 int codepoint;
168 xmlRegStatePtr start;
169 xmlRegStatePtr stop;
170 int maxRanges;
171 int nbRanges;
172 xmlRegRangePtr *ranges;
173 void *data;
174};
175
176typedef struct _xmlRegCounter xmlRegCounter;
177typedef xmlRegCounter *xmlRegCounterPtr;
178
179struct _xmlRegCounter {
180 int min;
181 int max;
182};
183
184typedef struct _xmlRegTrans xmlRegTrans;
185typedef xmlRegTrans *xmlRegTransPtr;
186
187struct _xmlRegTrans {
188 xmlRegAtomPtr atom;
189 int to;
190 int counter;
191 int count;
192};
193
194struct _xmlAutomataState {
195 xmlRegStateType type;
196 xmlRegMarkedType mark;
Daniel Veillard23e73572002-09-19 19:56:43 +0000197 xmlRegMarkedType reached;
Daniel Veillard4255d502002-04-16 15:50:10 +0000198 int no;
199
200 int maxTrans;
201 int nbTrans;
202 xmlRegTrans *trans;
203};
204
205typedef struct _xmlAutomata xmlRegParserCtxt;
206typedef xmlRegParserCtxt *xmlRegParserCtxtPtr;
207
208struct _xmlAutomata {
209 xmlChar *string;
210 xmlChar *cur;
211
212 int error;
213 int neg;
214
215 xmlRegStatePtr start;
216 xmlRegStatePtr end;
217 xmlRegStatePtr state;
218
219 xmlRegAtomPtr atom;
220
221 int maxAtoms;
222 int nbAtoms;
223 xmlRegAtomPtr *atoms;
224
225 int maxStates;
226 int nbStates;
227 xmlRegStatePtr *states;
228
229 int maxCounters;
230 int nbCounters;
231 xmlRegCounter *counters;
Daniel Veillarde19fc232002-04-22 16:01:24 +0000232
233 int determinist;
Daniel Veillard4255d502002-04-16 15:50:10 +0000234};
235
236struct _xmlRegexp {
237 xmlChar *string;
238 int nbStates;
239 xmlRegStatePtr *states;
240 int nbAtoms;
241 xmlRegAtomPtr *atoms;
242 int nbCounters;
243 xmlRegCounter *counters;
Daniel Veillarde19fc232002-04-22 16:01:24 +0000244 int determinist;
Daniel Veillard23e73572002-09-19 19:56:43 +0000245 /*
246 * That's the compact form for determinists automatas
247 */
248 int nbstates;
249 int *compact;
Daniel Veillard118aed72002-09-24 14:13:13 +0000250 void **transdata;
Daniel Veillard23e73572002-09-19 19:56:43 +0000251 int nbstrings;
252 xmlChar **stringMap;
Daniel Veillard4255d502002-04-16 15:50:10 +0000253};
254
255typedef struct _xmlRegExecRollback xmlRegExecRollback;
256typedef xmlRegExecRollback *xmlRegExecRollbackPtr;
257
258struct _xmlRegExecRollback {
259 xmlRegStatePtr state;/* the current state */
260 int index; /* the index in the input stack */
261 int nextbranch; /* the next transition to explore in that state */
262 int *counts; /* save the automate state if it has some */
263};
264
265typedef struct _xmlRegInputToken xmlRegInputToken;
266typedef xmlRegInputToken *xmlRegInputTokenPtr;
267
268struct _xmlRegInputToken {
269 xmlChar *value;
270 void *data;
271};
272
273struct _xmlRegExecCtxt {
274 int status; /* execution status != 0 indicate an error */
275 int determinist; /* did we found an inderterministic behaviour */
276 xmlRegexpPtr comp; /* the compiled regexp */
277 xmlRegExecCallbacks callback;
278 void *data;
279
280 xmlRegStatePtr state;/* the current state */
281 int transno; /* the current transition on that state */
282 int transcount; /* the number of char in char counted transitions */
283
284 /*
285 * A stack of rollback states
286 */
287 int maxRollbacks;
288 int nbRollbacks;
289 xmlRegExecRollback *rollbacks;
290
291 /*
292 * The state of the automata if any
293 */
294 int *counts;
295
296 /*
297 * The input stack
298 */
299 int inputStackMax;
300 int inputStackNr;
301 int index;
302 int *charStack;
303 const xmlChar *inputString; /* when operating on characters */
304 xmlRegInputTokenPtr inputStack;/* when operating on strings */
305
306};
307
Daniel Veillard441bc322002-04-20 17:38:48 +0000308#define REGEXP_ALL_COUNTER 0x123456
309#define REGEXP_ALL_LAX_COUNTER 0x123457
Daniel Veillard7646b182002-04-20 06:41:40 +0000310
Daniel Veillard4255d502002-04-16 15:50:10 +0000311static void xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top);
Daniel Veillard23e73572002-09-19 19:56:43 +0000312static void xmlRegFreeState(xmlRegStatePtr state);
313static void xmlRegFreeAtom(xmlRegAtomPtr atom);
Daniel Veillard4255d502002-04-16 15:50:10 +0000314
315/************************************************************************
316 * *
317 * Allocation/Deallocation *
318 * *
319 ************************************************************************/
320
Daniel Veillard23e73572002-09-19 19:56:43 +0000321static int xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt);
Daniel Veillard4255d502002-04-16 15:50:10 +0000322/**
323 * xmlRegEpxFromParse:
324 * @ctxt: the parser context used to build it
325 *
326 * Allocate a new regexp and fill it with the reult from the parser
327 *
328 * Returns the new regexp or NULL in case of error
329 */
330static xmlRegexpPtr
331xmlRegEpxFromParse(xmlRegParserCtxtPtr ctxt) {
332 xmlRegexpPtr ret;
333
334 ret = (xmlRegexpPtr) xmlMalloc(sizeof(xmlRegexp));
335 if (ret == NULL)
336 return(NULL);
337 memset(ret, 0, sizeof(xmlRegexp));
338 ret->string = ctxt->string;
339 ctxt->string = NULL;
340 ret->nbStates = ctxt->nbStates;
341 ctxt->nbStates = 0;
342 ret->states = ctxt->states;
343 ctxt->states = NULL;
344 ret->nbAtoms = ctxt->nbAtoms;
345 ctxt->nbAtoms = 0;
346 ret->atoms = ctxt->atoms;
347 ctxt->atoms = NULL;
348 ret->nbCounters = ctxt->nbCounters;
349 ctxt->nbCounters = 0;
350 ret->counters = ctxt->counters;
351 ctxt->counters = NULL;
Daniel Veillarde19fc232002-04-22 16:01:24 +0000352 ret->determinist = ctxt->determinist;
Daniel Veillard23e73572002-09-19 19:56:43 +0000353
354 if ((ret->determinist != 0) &&
355 (ret->nbCounters == 0) &&
Daniel Veillard118aed72002-09-24 14:13:13 +0000356 (ret->atoms != NULL) &&
Daniel Veillard23e73572002-09-19 19:56:43 +0000357 (ret->atoms[0] != NULL) &&
358 (ret->atoms[0]->type == XML_REGEXP_STRING)) {
359 int i, j, nbstates = 0, nbatoms = 0;
360 int *stateRemap;
361 int *stringRemap;
362 int *transitions;
Daniel Veillard118aed72002-09-24 14:13:13 +0000363 void **transdata;
Daniel Veillard23e73572002-09-19 19:56:43 +0000364 xmlChar **stringMap;
365 xmlChar *value;
366
367 /*
368 * Switch to a compact representation
369 * 1/ counting the effective number of states left
370 * 2/ conting the unique number of atoms, and check that
371 * they are all of the string type
372 * 3/ build a table state x atom for the transitions
373 */
374
375 stateRemap = xmlMalloc(ret->nbStates * sizeof(int));
376 for (i = 0;i < ret->nbStates;i++) {
377 if (ret->states[i] != NULL) {
378 stateRemap[i] = nbstates;
379 nbstates++;
380 } else {
381 stateRemap[i] = -1;
382 }
383 }
384#ifdef DEBUG_COMPACTION
385 printf("Final: %d states\n", nbstates);
386#endif
387 stringMap = xmlMalloc(ret->nbAtoms * sizeof(char *));
388 stringRemap = xmlMalloc(ret->nbAtoms * sizeof(int));
389 for (i = 0;i < ret->nbAtoms;i++) {
390 if ((ret->atoms[i]->type == XML_REGEXP_STRING) &&
391 (ret->atoms[i]->quant == XML_REGEXP_QUANT_ONCE)) {
392 value = ret->atoms[i]->valuep;
393 for (j = 0;j < nbatoms;j++) {
394 if (xmlStrEqual(stringMap[j], value)) {
395 stringRemap[i] = j;
396 break;
397 }
398 }
399 if (j >= nbatoms) {
400 stringRemap[i] = nbatoms;
401 stringMap[nbatoms] = xmlStrdup(value);
402 nbatoms++;
403 }
404 } else {
405 xmlFree(stateRemap);
406 xmlFree(stringRemap);
407 for (i = 0;i < nbatoms;i++)
408 xmlFree(stringMap[i]);
409 xmlFree(stringMap);
410 goto fail_compact;
411 }
412 }
413#ifdef DEBUG_COMPACTION
414 printf("Final: %d atoms\n", nbatoms);
415#endif
416
417 /*
418 * Allocate the transition table. The first entry for each
419 * state correspond to the state type.
420 */
421 transitions = (int *) xmlMalloc(nbstates * (nbatoms + 1) * sizeof(int));
Daniel Veillard118aed72002-09-24 14:13:13 +0000422 transdata = NULL;
Daniel Veillard23e73572002-09-19 19:56:43 +0000423 memset(transitions, 0, nbstates * (nbatoms + 1) * sizeof(int));
424
425 for (i = 0;i < ret->nbStates;i++) {
426 int stateno, atomno, targetno, prev;
427 xmlRegStatePtr state;
428 xmlRegTransPtr trans;
429
430 stateno = stateRemap[i];
431 if (stateno == -1)
432 continue;
433 state = ret->states[i];
434
435 transitions[stateno * (nbatoms + 1)] = state->type;
436
437 for (j = 0;j < state->nbTrans;j++) {
438 trans = &(state->trans[j]);
439 if ((trans->to == -1) || (trans->atom == NULL))
440 continue;
441 atomno = stringRemap[trans->atom->no];
Daniel Veillard118aed72002-09-24 14:13:13 +0000442 if ((trans->atom->data != NULL) && (transdata == NULL)) {
443 transdata = (void **) xmlMalloc(nbstates * nbatoms *
444 sizeof(void *));
445 if (transdata != NULL)
446 memset(transdata, 0,
447 nbstates * nbatoms * sizeof(void *));
448 }
Daniel Veillard23e73572002-09-19 19:56:43 +0000449 targetno = stateRemap[trans->to];
450 /*
451 * if the same atome can generate transition to 2 different
452 * states then it means the automata is not determinist and
453 * the compact form can't be used !
454 */
455 prev = transitions[stateno * (nbatoms + 1) + atomno + 1];
456 if (prev != 0) {
457 if (prev != targetno + 1) {
458 printf("not determinist\n");
459 ret->determinist = 0;
460#ifdef DEBUG_COMPACTION
461 printf("Indet: state %d trans %d, atom %d to %d : %d to %d\n",
462 i, j, trans->atom->no, trans->to, atomno, targetno);
463 printf(" previous to is %d\n", prev);
464#endif
465 ret->determinist = 0;
Daniel Veillard118aed72002-09-24 14:13:13 +0000466 if (transdata != NULL)
467 xmlFree(transdata);
Daniel Veillard23e73572002-09-19 19:56:43 +0000468 xmlFree(transitions);
469 xmlFree(stateRemap);
470 xmlFree(stringRemap);
471 for (i = 0;i < nbatoms;i++)
472 xmlFree(stringMap[i]);
473 xmlFree(stringMap);
474 goto fail_compact;
475 }
476 } else {
477#if 0
478 printf("State %d trans %d: atom %d to %d : %d to %d\n",
479 i, j, trans->atom->no, trans->to, atomno, targetno);
480#endif
481 transitions[stateno * (nbatoms + 1) + atomno + 1] =
Daniel Veillard118aed72002-09-24 14:13:13 +0000482 targetno + 1; /* to avoid 0 */
483 if (transdata != NULL)
484 transdata[stateno * nbatoms + atomno] =
485 trans->atom->data;
Daniel Veillard23e73572002-09-19 19:56:43 +0000486 }
487 }
488 }
489 ret->determinist = 1;
490#ifdef DEBUG_COMPACTION
491 /*
492 * Debug
493 */
494 for (i = 0;i < nbstates;i++) {
495 for (j = 0;j < nbatoms + 1;j++) {
496 printf("%02d ", transitions[i * (nbatoms + 1) + j]);
497 }
498 printf("\n");
499 }
500 printf("\n");
501#endif
502 /*
503 * Cleanup of the old data
504 */
505 if (ret->states != NULL) {
506 for (i = 0;i < ret->nbStates;i++)
507 xmlRegFreeState(ret->states[i]);
508 xmlFree(ret->states);
509 }
510 ret->states = NULL;
511 ret->nbStates = 0;
512 if (ret->atoms != NULL) {
513 for (i = 0;i < ret->nbAtoms;i++)
514 xmlRegFreeAtom(ret->atoms[i]);
515 xmlFree(ret->atoms);
516 }
517 ret->atoms = NULL;
518 ret->nbAtoms = 0;
519
520 ret->compact = transitions;
Daniel Veillard118aed72002-09-24 14:13:13 +0000521 ret->transdata = transdata;
Daniel Veillard23e73572002-09-19 19:56:43 +0000522 ret->stringMap = stringMap;
523 ret->nbstrings = nbatoms;
524 ret->nbstates = nbstates;
525 xmlFree(stateRemap);
526 xmlFree(stringRemap);
527 }
528fail_compact:
Daniel Veillard4255d502002-04-16 15:50:10 +0000529 return(ret);
530}
531
532/**
533 * xmlRegNewParserCtxt:
534 * @string: the string to parse
535 *
536 * Allocate a new regexp parser context
537 *
538 * Returns the new context or NULL in case of error
539 */
540static xmlRegParserCtxtPtr
541xmlRegNewParserCtxt(const xmlChar *string) {
542 xmlRegParserCtxtPtr ret;
543
544 ret = (xmlRegParserCtxtPtr) xmlMalloc(sizeof(xmlRegParserCtxt));
545 if (ret == NULL)
546 return(NULL);
547 memset(ret, 0, sizeof(xmlRegParserCtxt));
548 if (string != NULL)
549 ret->string = xmlStrdup(string);
550 ret->cur = ret->string;
551 ret->neg = 0;
552 ret->error = 0;
Daniel Veillarde19fc232002-04-22 16:01:24 +0000553 ret->determinist = -1;
Daniel Veillard4255d502002-04-16 15:50:10 +0000554 return(ret);
555}
556
557/**
558 * xmlRegNewRange:
559 * @ctxt: the regexp parser context
560 * @neg: is that negative
561 * @type: the type of range
562 * @start: the start codepoint
563 * @end: the end codepoint
564 *
565 * Allocate a new regexp range
566 *
567 * Returns the new range or NULL in case of error
568 */
569static xmlRegRangePtr
570xmlRegNewRange(xmlRegParserCtxtPtr ctxt,
571 int neg, xmlRegAtomType type, int start, int end) {
572 xmlRegRangePtr ret;
573
574 ret = (xmlRegRangePtr) xmlMalloc(sizeof(xmlRegRange));
575 if (ret == NULL) {
576 ERROR("failed to allocate regexp range");
577 return(NULL);
578 }
579 ret->neg = neg;
580 ret->type = type;
581 ret->start = start;
582 ret->end = end;
583 return(ret);
584}
585
586/**
587 * xmlRegFreeRange:
588 * @range: the regexp range
589 *
590 * Free a regexp range
591 */
592static void
593xmlRegFreeRange(xmlRegRangePtr range) {
594 if (range == NULL)
595 return;
596
597 if (range->blockName != NULL)
598 xmlFree(range->blockName);
599 xmlFree(range);
600}
601
602/**
603 * xmlRegNewAtom:
604 * @ctxt: the regexp parser context
605 * @type: the type of atom
606 *
607 * Allocate a new regexp range
608 *
609 * Returns the new atom or NULL in case of error
610 */
611static xmlRegAtomPtr
612xmlRegNewAtom(xmlRegParserCtxtPtr ctxt, xmlRegAtomType type) {
613 xmlRegAtomPtr ret;
614
615 ret = (xmlRegAtomPtr) xmlMalloc(sizeof(xmlRegAtom));
616 if (ret == NULL) {
617 ERROR("failed to allocate regexp atom");
618 return(NULL);
619 }
620 memset(ret, 0, sizeof(xmlRegAtom));
621 ret->type = type;
622 ret->quant = XML_REGEXP_QUANT_ONCE;
623 ret->min = 0;
624 ret->max = 0;
625 return(ret);
626}
627
628/**
629 * xmlRegFreeAtom:
630 * @atom: the regexp atom
631 *
632 * Free a regexp atom
633 */
634static void
635xmlRegFreeAtom(xmlRegAtomPtr atom) {
636 int i;
637
638 if (atom == NULL)
639 return;
640
641 for (i = 0;i < atom->nbRanges;i++)
642 xmlRegFreeRange(atom->ranges[i]);
643 if (atom->ranges != NULL)
644 xmlFree(atom->ranges);
645 if (atom->type == XML_REGEXP_STRING)
646 xmlFree(atom->valuep);
647 xmlFree(atom);
648}
649
650static xmlRegStatePtr
651xmlRegNewState(xmlRegParserCtxtPtr ctxt) {
652 xmlRegStatePtr ret;
653
654 ret = (xmlRegStatePtr) xmlMalloc(sizeof(xmlRegState));
655 if (ret == NULL) {
656 ERROR("failed to allocate regexp state");
657 return(NULL);
658 }
659 memset(ret, 0, sizeof(xmlRegState));
660 ret->type = XML_REGEXP_TRANS_STATE;
661 ret->mark = XML_REGEXP_MARK_NORMAL;
662 return(ret);
663}
664
665/**
666 * xmlRegFreeState:
667 * @state: the regexp state
668 *
669 * Free a regexp state
670 */
671static void
672xmlRegFreeState(xmlRegStatePtr state) {
673 if (state == NULL)
674 return;
675
676 if (state->trans != NULL)
677 xmlFree(state->trans);
678 xmlFree(state);
679}
680
681/**
682 * xmlRegFreeParserCtxt:
683 * @ctxt: the regexp parser context
684 *
685 * Free a regexp parser context
686 */
687static void
688xmlRegFreeParserCtxt(xmlRegParserCtxtPtr ctxt) {
689 int i;
690 if (ctxt == NULL)
691 return;
692
693 if (ctxt->string != NULL)
694 xmlFree(ctxt->string);
695 if (ctxt->states != NULL) {
696 for (i = 0;i < ctxt->nbStates;i++)
697 xmlRegFreeState(ctxt->states[i]);
698 xmlFree(ctxt->states);
699 }
700 if (ctxt->atoms != NULL) {
701 for (i = 0;i < ctxt->nbAtoms;i++)
702 xmlRegFreeAtom(ctxt->atoms[i]);
703 xmlFree(ctxt->atoms);
704 }
705 if (ctxt->counters != NULL)
706 xmlFree(ctxt->counters);
707 xmlFree(ctxt);
708}
709
710/************************************************************************
711 * *
712 * Display of Data structures *
713 * *
714 ************************************************************************/
715
716static void
717xmlRegPrintAtomType(FILE *output, xmlRegAtomType type) {
718 switch (type) {
719 case XML_REGEXP_EPSILON:
720 fprintf(output, "epsilon "); break;
721 case XML_REGEXP_CHARVAL:
722 fprintf(output, "charval "); break;
723 case XML_REGEXP_RANGES:
724 fprintf(output, "ranges "); break;
725 case XML_REGEXP_SUBREG:
726 fprintf(output, "subexpr "); break;
727 case XML_REGEXP_STRING:
728 fprintf(output, "string "); break;
729 case XML_REGEXP_ANYCHAR:
730 fprintf(output, "anychar "); break;
731 case XML_REGEXP_ANYSPACE:
732 fprintf(output, "anyspace "); break;
733 case XML_REGEXP_NOTSPACE:
734 fprintf(output, "notspace "); break;
735 case XML_REGEXP_INITNAME:
736 fprintf(output, "initname "); break;
737 case XML_REGEXP_NOTINITNAME:
738 fprintf(output, "notinitname "); break;
739 case XML_REGEXP_NAMECHAR:
740 fprintf(output, "namechar "); break;
741 case XML_REGEXP_NOTNAMECHAR:
742 fprintf(output, "notnamechar "); break;
743 case XML_REGEXP_DECIMAL:
744 fprintf(output, "decimal "); break;
745 case XML_REGEXP_NOTDECIMAL:
746 fprintf(output, "notdecimal "); break;
747 case XML_REGEXP_REALCHAR:
748 fprintf(output, "realchar "); break;
749 case XML_REGEXP_NOTREALCHAR:
750 fprintf(output, "notrealchar "); break;
751 case XML_REGEXP_LETTER:
752 fprintf(output, "LETTER "); break;
753 case XML_REGEXP_LETTER_UPPERCASE:
754 fprintf(output, "LETTER_UPPERCASE "); break;
755 case XML_REGEXP_LETTER_LOWERCASE:
756 fprintf(output, "LETTER_LOWERCASE "); break;
757 case XML_REGEXP_LETTER_TITLECASE:
758 fprintf(output, "LETTER_TITLECASE "); break;
759 case XML_REGEXP_LETTER_MODIFIER:
760 fprintf(output, "LETTER_MODIFIER "); break;
761 case XML_REGEXP_LETTER_OTHERS:
762 fprintf(output, "LETTER_OTHERS "); break;
763 case XML_REGEXP_MARK:
764 fprintf(output, "MARK "); break;
765 case XML_REGEXP_MARK_NONSPACING:
766 fprintf(output, "MARK_NONSPACING "); break;
767 case XML_REGEXP_MARK_SPACECOMBINING:
768 fprintf(output, "MARK_SPACECOMBINING "); break;
769 case XML_REGEXP_MARK_ENCLOSING:
770 fprintf(output, "MARK_ENCLOSING "); break;
771 case XML_REGEXP_NUMBER:
772 fprintf(output, "NUMBER "); break;
773 case XML_REGEXP_NUMBER_DECIMAL:
774 fprintf(output, "NUMBER_DECIMAL "); break;
775 case XML_REGEXP_NUMBER_LETTER:
776 fprintf(output, "NUMBER_LETTER "); break;
777 case XML_REGEXP_NUMBER_OTHERS:
778 fprintf(output, "NUMBER_OTHERS "); break;
779 case XML_REGEXP_PUNCT:
780 fprintf(output, "PUNCT "); break;
781 case XML_REGEXP_PUNCT_CONNECTOR:
782 fprintf(output, "PUNCT_CONNECTOR "); break;
783 case XML_REGEXP_PUNCT_DASH:
784 fprintf(output, "PUNCT_DASH "); break;
785 case XML_REGEXP_PUNCT_OPEN:
786 fprintf(output, "PUNCT_OPEN "); break;
787 case XML_REGEXP_PUNCT_CLOSE:
788 fprintf(output, "PUNCT_CLOSE "); break;
789 case XML_REGEXP_PUNCT_INITQUOTE:
790 fprintf(output, "PUNCT_INITQUOTE "); break;
791 case XML_REGEXP_PUNCT_FINQUOTE:
792 fprintf(output, "PUNCT_FINQUOTE "); break;
793 case XML_REGEXP_PUNCT_OTHERS:
794 fprintf(output, "PUNCT_OTHERS "); break;
795 case XML_REGEXP_SEPAR:
796 fprintf(output, "SEPAR "); break;
797 case XML_REGEXP_SEPAR_SPACE:
798 fprintf(output, "SEPAR_SPACE "); break;
799 case XML_REGEXP_SEPAR_LINE:
800 fprintf(output, "SEPAR_LINE "); break;
801 case XML_REGEXP_SEPAR_PARA:
802 fprintf(output, "SEPAR_PARA "); break;
803 case XML_REGEXP_SYMBOL:
804 fprintf(output, "SYMBOL "); break;
805 case XML_REGEXP_SYMBOL_MATH:
806 fprintf(output, "SYMBOL_MATH "); break;
807 case XML_REGEXP_SYMBOL_CURRENCY:
808 fprintf(output, "SYMBOL_CURRENCY "); break;
809 case XML_REGEXP_SYMBOL_MODIFIER:
810 fprintf(output, "SYMBOL_MODIFIER "); break;
811 case XML_REGEXP_SYMBOL_OTHERS:
812 fprintf(output, "SYMBOL_OTHERS "); break;
813 case XML_REGEXP_OTHER:
814 fprintf(output, "OTHER "); break;
815 case XML_REGEXP_OTHER_CONTROL:
816 fprintf(output, "OTHER_CONTROL "); break;
817 case XML_REGEXP_OTHER_FORMAT:
818 fprintf(output, "OTHER_FORMAT "); break;
819 case XML_REGEXP_OTHER_PRIVATE:
820 fprintf(output, "OTHER_PRIVATE "); break;
821 case XML_REGEXP_OTHER_NA:
822 fprintf(output, "OTHER_NA "); break;
823 case XML_REGEXP_BLOCK_NAME:
824 fprintf(output, "BLOCK "); break;
825 }
826}
827
828static void
829xmlRegPrintQuantType(FILE *output, xmlRegQuantType type) {
830 switch (type) {
831 case XML_REGEXP_QUANT_EPSILON:
832 fprintf(output, "epsilon "); break;
833 case XML_REGEXP_QUANT_ONCE:
834 fprintf(output, "once "); break;
835 case XML_REGEXP_QUANT_OPT:
836 fprintf(output, "? "); break;
837 case XML_REGEXP_QUANT_MULT:
838 fprintf(output, "* "); break;
839 case XML_REGEXP_QUANT_PLUS:
840 fprintf(output, "+ "); break;
841 case XML_REGEXP_QUANT_RANGE:
842 fprintf(output, "range "); break;
Daniel Veillard7646b182002-04-20 06:41:40 +0000843 case XML_REGEXP_QUANT_ONCEONLY:
844 fprintf(output, "onceonly "); break;
845 case XML_REGEXP_QUANT_ALL:
846 fprintf(output, "all "); break;
Daniel Veillard4255d502002-04-16 15:50:10 +0000847 }
848}
849static void
850xmlRegPrintRange(FILE *output, xmlRegRangePtr range) {
851 fprintf(output, " range: ");
852 if (range->neg)
853 fprintf(output, "negative ");
854 xmlRegPrintAtomType(output, range->type);
855 fprintf(output, "%c - %c\n", range->start, range->end);
856}
857
858static void
859xmlRegPrintAtom(FILE *output, xmlRegAtomPtr atom) {
860 fprintf(output, " atom: ");
861 if (atom == NULL) {
862 fprintf(output, "NULL\n");
863 return;
864 }
865 xmlRegPrintAtomType(output, atom->type);
866 xmlRegPrintQuantType(output, atom->quant);
867 if (atom->quant == XML_REGEXP_QUANT_RANGE)
868 fprintf(output, "%d-%d ", atom->min, atom->max);
869 if (atom->type == XML_REGEXP_STRING)
870 fprintf(output, "'%s' ", (char *) atom->valuep);
871 if (atom->type == XML_REGEXP_CHARVAL)
872 fprintf(output, "char %c\n", atom->codepoint);
873 else if (atom->type == XML_REGEXP_RANGES) {
874 int i;
875 fprintf(output, "%d entries\n", atom->nbRanges);
876 for (i = 0; i < atom->nbRanges;i++)
877 xmlRegPrintRange(output, atom->ranges[i]);
878 } else if (atom->type == XML_REGEXP_SUBREG) {
879 fprintf(output, "start %d end %d\n", atom->start->no, atom->stop->no);
880 } else {
881 fprintf(output, "\n");
882 }
883}
884
885static void
886xmlRegPrintTrans(FILE *output, xmlRegTransPtr trans) {
887 fprintf(output, " trans: ");
888 if (trans == NULL) {
889 fprintf(output, "NULL\n");
890 return;
891 }
892 if (trans->to < 0) {
893 fprintf(output, "removed\n");
894 return;
895 }
896 if (trans->counter >= 0) {
897 fprintf(output, "counted %d, ", trans->counter);
898 }
Daniel Veillard8a001f62002-04-20 07:24:11 +0000899 if (trans->count == REGEXP_ALL_COUNTER) {
900 fprintf(output, "all transition, ");
901 } else if (trans->count >= 0) {
Daniel Veillard4255d502002-04-16 15:50:10 +0000902 fprintf(output, "count based %d, ", trans->count);
903 }
904 if (trans->atom == NULL) {
905 fprintf(output, "epsilon to %d\n", trans->to);
906 return;
907 }
908 if (trans->atom->type == XML_REGEXP_CHARVAL)
909 fprintf(output, "char %c ", trans->atom->codepoint);
910 fprintf(output, "atom %d, to %d\n", trans->atom->no, trans->to);
911}
912
913static void
914xmlRegPrintState(FILE *output, xmlRegStatePtr state) {
915 int i;
916
917 fprintf(output, " state: ");
918 if (state == NULL) {
919 fprintf(output, "NULL\n");
920 return;
921 }
922 if (state->type == XML_REGEXP_START_STATE)
923 fprintf(output, "START ");
924 if (state->type == XML_REGEXP_FINAL_STATE)
925 fprintf(output, "FINAL ");
926
927 fprintf(output, "%d, %d transitions:\n", state->no, state->nbTrans);
928 for (i = 0;i < state->nbTrans; i++) {
929 xmlRegPrintTrans(output, &(state->trans[i]));
930 }
931}
932
Daniel Veillard23e73572002-09-19 19:56:43 +0000933#ifdef DEBUG_REGEXP_GRAPH
Daniel Veillard4255d502002-04-16 15:50:10 +0000934static void
935xmlRegPrintCtxt(FILE *output, xmlRegParserCtxtPtr ctxt) {
936 int i;
937
938 fprintf(output, " ctxt: ");
939 if (ctxt == NULL) {
940 fprintf(output, "NULL\n");
941 return;
942 }
943 fprintf(output, "'%s' ", ctxt->string);
944 if (ctxt->error)
945 fprintf(output, "error ");
946 if (ctxt->neg)
947 fprintf(output, "neg ");
948 fprintf(output, "\n");
949 fprintf(output, "%d atoms:\n", ctxt->nbAtoms);
950 for (i = 0;i < ctxt->nbAtoms; i++) {
951 fprintf(output, " %02d ", i);
952 xmlRegPrintAtom(output, ctxt->atoms[i]);
953 }
954 if (ctxt->atom != NULL) {
955 fprintf(output, "current atom:\n");
956 xmlRegPrintAtom(output, ctxt->atom);
957 }
958 fprintf(output, "%d states:", ctxt->nbStates);
959 if (ctxt->start != NULL)
960 fprintf(output, " start: %d", ctxt->start->no);
961 if (ctxt->end != NULL)
962 fprintf(output, " end: %d", ctxt->end->no);
963 fprintf(output, "\n");
964 for (i = 0;i < ctxt->nbStates; i++) {
965 xmlRegPrintState(output, ctxt->states[i]);
966 }
967 fprintf(output, "%d counters:\n", ctxt->nbCounters);
968 for (i = 0;i < ctxt->nbCounters; i++) {
969 fprintf(output, " %d: min %d max %d\n", i, ctxt->counters[i].min,
970 ctxt->counters[i].max);
971 }
972}
Daniel Veillard23e73572002-09-19 19:56:43 +0000973#endif
Daniel Veillard4255d502002-04-16 15:50:10 +0000974
975/************************************************************************
976 * *
977 * Finite Automata structures manipulations *
978 * *
979 ************************************************************************/
980
981static void
982xmlRegAtomAddRange(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom,
983 int neg, xmlRegAtomType type, int start, int end,
984 xmlChar *blockName) {
985 xmlRegRangePtr range;
986
987 if (atom == NULL) {
988 ERROR("add range: atom is NULL");
989 return;
990 }
991 if (atom->type != XML_REGEXP_RANGES) {
992 ERROR("add range: atom is not ranges");
993 return;
994 }
995 if (atom->maxRanges == 0) {
996 atom->maxRanges = 4;
997 atom->ranges = (xmlRegRangePtr *) xmlMalloc(atom->maxRanges *
998 sizeof(xmlRegRangePtr));
999 if (atom->ranges == NULL) {
1000 ERROR("add range: allocation failed");
1001 atom->maxRanges = 0;
1002 return;
1003 }
1004 } else if (atom->nbRanges >= atom->maxRanges) {
1005 xmlRegRangePtr *tmp;
1006 atom->maxRanges *= 2;
1007 tmp = (xmlRegRangePtr *) xmlRealloc(atom->ranges, atom->maxRanges *
1008 sizeof(xmlRegRangePtr));
1009 if (tmp == NULL) {
1010 ERROR("add range: allocation failed");
1011 atom->maxRanges /= 2;
1012 return;
1013 }
1014 atom->ranges = tmp;
1015 }
1016 range = xmlRegNewRange(ctxt, neg, type, start, end);
1017 if (range == NULL)
1018 return;
1019 range->blockName = blockName;
1020 atom->ranges[atom->nbRanges++] = range;
1021
1022}
1023
1024static int
1025xmlRegGetCounter(xmlRegParserCtxtPtr ctxt) {
1026 if (ctxt->maxCounters == 0) {
1027 ctxt->maxCounters = 4;
1028 ctxt->counters = (xmlRegCounter *) xmlMalloc(ctxt->maxCounters *
1029 sizeof(xmlRegCounter));
1030 if (ctxt->counters == NULL) {
1031 ERROR("reg counter: allocation failed");
1032 ctxt->maxCounters = 0;
1033 return(-1);
1034 }
1035 } else if (ctxt->nbCounters >= ctxt->maxCounters) {
1036 xmlRegCounter *tmp;
1037 ctxt->maxCounters *= 2;
1038 tmp = (xmlRegCounter *) xmlRealloc(ctxt->counters, ctxt->maxCounters *
1039 sizeof(xmlRegCounter));
1040 if (tmp == NULL) {
1041 ERROR("reg counter: allocation failed");
1042 ctxt->maxCounters /= 2;
1043 return(-1);
1044 }
1045 ctxt->counters = tmp;
1046 }
1047 ctxt->counters[ctxt->nbCounters].min = -1;
1048 ctxt->counters[ctxt->nbCounters].max = -1;
1049 return(ctxt->nbCounters++);
1050}
1051
1052static void
1053xmlRegAtomPush(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom) {
1054 if (atom == NULL) {
1055 ERROR("atom push: atom is NULL");
1056 return;
1057 }
1058 if (ctxt->maxAtoms == 0) {
1059 ctxt->maxAtoms = 4;
1060 ctxt->atoms = (xmlRegAtomPtr *) xmlMalloc(ctxt->maxAtoms *
1061 sizeof(xmlRegAtomPtr));
1062 if (ctxt->atoms == NULL) {
1063 ERROR("atom push: allocation failed");
1064 ctxt->maxAtoms = 0;
1065 return;
1066 }
1067 } else if (ctxt->nbAtoms >= ctxt->maxAtoms) {
1068 xmlRegAtomPtr *tmp;
1069 ctxt->maxAtoms *= 2;
1070 tmp = (xmlRegAtomPtr *) xmlRealloc(ctxt->atoms, ctxt->maxAtoms *
1071 sizeof(xmlRegAtomPtr));
1072 if (tmp == NULL) {
1073 ERROR("atom push: allocation failed");
1074 ctxt->maxAtoms /= 2;
1075 return;
1076 }
1077 ctxt->atoms = tmp;
1078 }
1079 atom->no = ctxt->nbAtoms;
1080 ctxt->atoms[ctxt->nbAtoms++] = atom;
1081}
1082
1083static void
1084xmlRegStateAddTrans(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
1085 xmlRegAtomPtr atom, xmlRegStatePtr target,
1086 int counter, int count) {
1087 if (state == NULL) {
1088 ERROR("add state: state is NULL");
1089 return;
1090 }
1091 if (target == NULL) {
1092 ERROR("add state: target is NULL");
1093 return;
1094 }
1095 if (state->maxTrans == 0) {
1096 state->maxTrans = 4;
1097 state->trans = (xmlRegTrans *) xmlMalloc(state->maxTrans *
1098 sizeof(xmlRegTrans));
1099 if (state->trans == NULL) {
1100 ERROR("add range: allocation failed");
1101 state->maxTrans = 0;
1102 return;
1103 }
1104 } else if (state->nbTrans >= state->maxTrans) {
1105 xmlRegTrans *tmp;
1106 state->maxTrans *= 2;
1107 tmp = (xmlRegTrans *) xmlRealloc(state->trans, state->maxTrans *
1108 sizeof(xmlRegTrans));
1109 if (tmp == NULL) {
1110 ERROR("add range: allocation failed");
1111 state->maxTrans /= 2;
1112 return;
1113 }
1114 state->trans = tmp;
1115 }
1116#ifdef DEBUG_REGEXP_GRAPH
1117 printf("Add trans from %d to %d ", state->no, target->no);
Daniel Veillard8a001f62002-04-20 07:24:11 +00001118 if (count == REGEXP_ALL_COUNTER)
1119 printf("all transition");
Daniel Veillard4402ab42002-09-12 16:02:56 +00001120 else if (count >= 0)
Daniel Veillard4255d502002-04-16 15:50:10 +00001121 printf("count based %d", count);
1122 else if (counter >= 0)
1123 printf("counted %d", counter);
1124 else if (atom == NULL)
1125 printf("epsilon transition");
1126 printf("\n");
1127#endif
1128
1129 state->trans[state->nbTrans].atom = atom;
1130 state->trans[state->nbTrans].to = target->no;
1131 state->trans[state->nbTrans].counter = counter;
1132 state->trans[state->nbTrans].count = count;
1133 state->nbTrans++;
1134}
1135
1136static void
1137xmlRegStatePush(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state) {
1138 if (ctxt->maxStates == 0) {
1139 ctxt->maxStates = 4;
1140 ctxt->states = (xmlRegStatePtr *) xmlMalloc(ctxt->maxStates *
1141 sizeof(xmlRegStatePtr));
1142 if (ctxt->states == NULL) {
1143 ERROR("add range: allocation failed");
1144 ctxt->maxStates = 0;
1145 return;
1146 }
1147 } else if (ctxt->nbStates >= ctxt->maxStates) {
1148 xmlRegStatePtr *tmp;
1149 ctxt->maxStates *= 2;
1150 tmp = (xmlRegStatePtr *) xmlRealloc(ctxt->states, ctxt->maxStates *
1151 sizeof(xmlRegStatePtr));
1152 if (tmp == NULL) {
1153 ERROR("add range: allocation failed");
1154 ctxt->maxStates /= 2;
1155 return;
1156 }
1157 ctxt->states = tmp;
1158 }
1159 state->no = ctxt->nbStates;
1160 ctxt->states[ctxt->nbStates++] = state;
1161}
1162
1163/**
Daniel Veillard7646b182002-04-20 06:41:40 +00001164 * xmlFAGenerateAllTransition:
Daniel Veillard441bc322002-04-20 17:38:48 +00001165 * @ctxt: a regexp parser context
1166 * @from: the from state
1167 * @to: the target state or NULL for building a new one
1168 * @lax:
Daniel Veillard7646b182002-04-20 06:41:40 +00001169 *
1170 */
1171static void
1172xmlFAGenerateAllTransition(xmlRegParserCtxtPtr ctxt,
Daniel Veillard441bc322002-04-20 17:38:48 +00001173 xmlRegStatePtr from, xmlRegStatePtr to,
1174 int lax) {
Daniel Veillard7646b182002-04-20 06:41:40 +00001175 if (to == NULL) {
1176 to = xmlRegNewState(ctxt);
1177 xmlRegStatePush(ctxt, to);
1178 ctxt->state = to;
1179 }
Daniel Veillard441bc322002-04-20 17:38:48 +00001180 if (lax)
1181 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_LAX_COUNTER);
1182 else
1183 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_COUNTER);
Daniel Veillard7646b182002-04-20 06:41:40 +00001184}
1185
1186/**
Daniel Veillard4255d502002-04-16 15:50:10 +00001187 * xmlFAGenerateEpsilonTransition:
Daniel Veillard441bc322002-04-20 17:38:48 +00001188 * @ctxt: a regexp parser context
1189 * @from: the from state
1190 * @to: the target state or NULL for building a new one
Daniel Veillard4255d502002-04-16 15:50:10 +00001191 *
1192 */
1193static void
1194xmlFAGenerateEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1195 xmlRegStatePtr from, xmlRegStatePtr to) {
1196 if (to == NULL) {
1197 to = xmlRegNewState(ctxt);
1198 xmlRegStatePush(ctxt, to);
1199 ctxt->state = to;
1200 }
1201 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, -1);
1202}
1203
1204/**
1205 * xmlFAGenerateCountedEpsilonTransition:
Daniel Veillard441bc322002-04-20 17:38:48 +00001206 * @ctxt: a regexp parser context
1207 * @from: the from state
1208 * @to: the target state or NULL for building a new one
Daniel Veillard4255d502002-04-16 15:50:10 +00001209 * counter: the counter for that transition
1210 *
1211 */
1212static void
1213xmlFAGenerateCountedEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1214 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1215 if (to == NULL) {
1216 to = xmlRegNewState(ctxt);
1217 xmlRegStatePush(ctxt, to);
1218 ctxt->state = to;
1219 }
1220 xmlRegStateAddTrans(ctxt, from, NULL, to, counter, -1);
1221}
1222
1223/**
1224 * xmlFAGenerateCountedTransition:
Daniel Veillard441bc322002-04-20 17:38:48 +00001225 * @ctxt: a regexp parser context
1226 * @from: the from state
1227 * @to: the target state or NULL for building a new one
Daniel Veillard4255d502002-04-16 15:50:10 +00001228 * counter: the counter for that transition
1229 *
1230 */
1231static void
1232xmlFAGenerateCountedTransition(xmlRegParserCtxtPtr ctxt,
1233 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1234 if (to == NULL) {
1235 to = xmlRegNewState(ctxt);
1236 xmlRegStatePush(ctxt, to);
1237 ctxt->state = to;
1238 }
1239 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, counter);
1240}
1241
1242/**
1243 * xmlFAGenerateTransitions:
Daniel Veillard441bc322002-04-20 17:38:48 +00001244 * @ctxt: a regexp parser context
1245 * @from: the from state
1246 * @to: the target state or NULL for building a new one
1247 * @atom: the atom generating the transition
Daniel Veillard4255d502002-04-16 15:50:10 +00001248 *
1249 */
1250static void
1251xmlFAGenerateTransitions(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr from,
1252 xmlRegStatePtr to, xmlRegAtomPtr atom) {
1253 if (atom == NULL) {
1254 ERROR("genrate transition: atom == NULL");
1255 return;
1256 }
1257 if (atom->type == XML_REGEXP_SUBREG) {
1258 /*
1259 * this is a subexpression handling one should not need to
1260 * create a new node excep for XML_REGEXP_QUANT_RANGE.
1261 */
1262 xmlRegAtomPush(ctxt, atom);
1263 if ((to != NULL) && (atom->stop != to) &&
1264 (atom->quant != XML_REGEXP_QUANT_RANGE)) {
1265 /*
1266 * Generate an epsilon transition to link to the target
1267 */
1268 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, to);
1269 }
1270 switch (atom->quant) {
1271 case XML_REGEXP_QUANT_OPT:
1272 atom->quant = XML_REGEXP_QUANT_ONCE;
1273 xmlFAGenerateEpsilonTransition(ctxt, atom->start, atom->stop);
1274 break;
1275 case XML_REGEXP_QUANT_MULT:
1276 atom->quant = XML_REGEXP_QUANT_ONCE;
1277 xmlFAGenerateEpsilonTransition(ctxt, atom->start, atom->stop);
1278 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1279 break;
1280 case XML_REGEXP_QUANT_PLUS:
1281 atom->quant = XML_REGEXP_QUANT_ONCE;
1282 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1283 break;
1284 case XML_REGEXP_QUANT_RANGE: {
1285 int counter;
1286 xmlRegStatePtr newstate;
1287
1288 /*
1289 * This one is nasty:
1290 * 1/ register a new counter
1291 * 2/ register an epsilon transition associated to
1292 * this counter going from atom->stop to atom->start
1293 * 3/ create a new state
1294 * 4/ generate a counted transition from atom->stop to
1295 * that state
1296 */
1297 counter = xmlRegGetCounter(ctxt);
1298 ctxt->counters[counter].min = atom->min - 1;
1299 ctxt->counters[counter].max = atom->max - 1;
1300 atom->min = 0;
1301 atom->max = 0;
1302 atom->quant = XML_REGEXP_QUANT_ONCE;
1303 xmlFAGenerateCountedEpsilonTransition(ctxt, atom->stop,
1304 atom->start, counter);
1305 if (to != NULL) {
1306 newstate = to;
1307 } else {
1308 newstate = xmlRegNewState(ctxt);
1309 xmlRegStatePush(ctxt, newstate);
1310 ctxt->state = newstate;
1311 }
1312 xmlFAGenerateCountedTransition(ctxt, atom->stop,
1313 newstate, counter);
1314 }
1315 default:
1316 break;
1317 }
1318 return;
1319 } else {
1320 if (to == NULL) {
1321 to = xmlRegNewState(ctxt);
1322 xmlRegStatePush(ctxt, to);
1323 }
1324 xmlRegStateAddTrans(ctxt, from, atom, to, -1, -1);
1325 xmlRegAtomPush(ctxt, atom);
1326 ctxt->state = to;
1327 }
1328 switch (atom->quant) {
1329 case XML_REGEXP_QUANT_OPT:
1330 atom->quant = XML_REGEXP_QUANT_ONCE;
1331 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1332 break;
1333 case XML_REGEXP_QUANT_MULT:
1334 atom->quant = XML_REGEXP_QUANT_ONCE;
1335 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1336 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1337 break;
1338 case XML_REGEXP_QUANT_PLUS:
1339 atom->quant = XML_REGEXP_QUANT_ONCE;
1340 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1341 break;
1342 default:
1343 break;
1344 }
1345}
1346
1347/**
1348 * xmlFAReduceEpsilonTransitions:
Daniel Veillard441bc322002-04-20 17:38:48 +00001349 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00001350 * @fromnr: the from state
1351 * @tonr: the to state
1352 * @cpunter: should that transition be associted to a counted
1353 *
1354 */
1355static void
1356xmlFAReduceEpsilonTransitions(xmlRegParserCtxtPtr ctxt, int fromnr,
1357 int tonr, int counter) {
1358 int transnr;
1359 xmlRegStatePtr from;
1360 xmlRegStatePtr to;
1361
1362#ifdef DEBUG_REGEXP_GRAPH
1363 printf("xmlFAReduceEpsilonTransitions(%d, %d)\n", fromnr, tonr);
1364#endif
1365 from = ctxt->states[fromnr];
1366 if (from == NULL)
1367 return;
1368 to = ctxt->states[tonr];
1369 if (to == NULL)
1370 return;
1371 if ((to->mark == XML_REGEXP_MARK_START) ||
1372 (to->mark == XML_REGEXP_MARK_VISITED))
1373 return;
1374
1375 to->mark = XML_REGEXP_MARK_VISITED;
1376 if (to->type == XML_REGEXP_FINAL_STATE) {
1377#ifdef DEBUG_REGEXP_GRAPH
1378 printf("State %d is final, so %d becomes final\n", tonr, fromnr);
1379#endif
1380 from->type = XML_REGEXP_FINAL_STATE;
1381 }
1382 for (transnr = 0;transnr < to->nbTrans;transnr++) {
1383 if (to->trans[transnr].atom == NULL) {
1384 /*
1385 * Don't remove counted transitions
1386 * Don't loop either
1387 */
Daniel Veillardb509f152002-04-17 16:28:10 +00001388 if (to->trans[transnr].to != fromnr) {
1389 if (to->trans[transnr].count >= 0) {
1390 int newto = to->trans[transnr].to;
1391
1392 xmlRegStateAddTrans(ctxt, from, NULL,
1393 ctxt->states[newto],
1394 -1, to->trans[transnr].count);
1395 } else {
Daniel Veillard4255d502002-04-16 15:50:10 +00001396#ifdef DEBUG_REGEXP_GRAPH
Daniel Veillardb509f152002-04-17 16:28:10 +00001397 printf("Found epsilon trans %d from %d to %d\n",
1398 transnr, tonr, to->trans[transnr].to);
Daniel Veillard4255d502002-04-16 15:50:10 +00001399#endif
Daniel Veillardb509f152002-04-17 16:28:10 +00001400 if (to->trans[transnr].counter >= 0) {
1401 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1402 to->trans[transnr].to,
1403 to->trans[transnr].counter);
1404 } else {
1405 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1406 to->trans[transnr].to,
1407 counter);
1408 }
1409 }
Daniel Veillard4255d502002-04-16 15:50:10 +00001410 }
1411 } else {
1412 int newto = to->trans[transnr].to;
1413
Daniel Veillardb509f152002-04-17 16:28:10 +00001414 if (to->trans[transnr].counter >= 0) {
1415 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1416 ctxt->states[newto],
1417 to->trans[transnr].counter, -1);
1418 } else {
1419 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1420 ctxt->states[newto], counter, -1);
1421 }
Daniel Veillard4255d502002-04-16 15:50:10 +00001422 }
1423 }
1424 to->mark = XML_REGEXP_MARK_NORMAL;
1425}
1426
1427/**
1428 * xmlFAEliminateEpsilonTransitions:
Daniel Veillard441bc322002-04-20 17:38:48 +00001429 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00001430 *
1431 */
1432static void
1433xmlFAEliminateEpsilonTransitions(xmlRegParserCtxtPtr ctxt) {
1434 int statenr, transnr;
1435 xmlRegStatePtr state;
1436
1437 /*
1438 * build the completed transitions bypassing the epsilons
1439 * Use a marking algorithm to avoid loops
1440 */
1441 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1442 state = ctxt->states[statenr];
1443 if (state == NULL)
1444 continue;
1445 for (transnr = 0;transnr < state->nbTrans;transnr++) {
1446 if ((state->trans[transnr].atom == NULL) &&
1447 (state->trans[transnr].to >= 0)) {
1448 if (state->trans[transnr].to == statenr) {
1449 state->trans[transnr].to = -1;
1450#ifdef DEBUG_REGEXP_GRAPH
1451 printf("Removed loopback epsilon trans %d on %d\n",
1452 transnr, statenr);
1453#endif
1454 } else if (state->trans[transnr].count < 0) {
1455 int newto = state->trans[transnr].to;
1456
1457#ifdef DEBUG_REGEXP_GRAPH
1458 printf("Found epsilon trans %d from %d to %d\n",
1459 transnr, statenr, newto);
1460#endif
1461 state->mark = XML_REGEXP_MARK_START;
1462 xmlFAReduceEpsilonTransitions(ctxt, statenr,
1463 newto, state->trans[transnr].counter);
1464 state->mark = XML_REGEXP_MARK_NORMAL;
1465#ifdef DEBUG_REGEXP_GRAPH
1466 } else {
1467 printf("Found counted transition %d on %d\n",
1468 transnr, statenr);
1469#endif
1470 }
1471 }
1472 }
1473 }
1474 /*
1475 * Eliminate the epsilon transitions
1476 */
1477 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1478 state = ctxt->states[statenr];
1479 if (state == NULL)
1480 continue;
1481 for (transnr = 0;transnr < state->nbTrans;transnr++) {
1482 if ((state->trans[transnr].atom == NULL) &&
1483 (state->trans[transnr].count < 0) &&
1484 (state->trans[transnr].to >= 0)) {
1485 state->trans[transnr].to = -1;
1486 }
1487 }
1488 }
Daniel Veillard23e73572002-09-19 19:56:43 +00001489
1490 /*
1491 * Use this pass to detect unreachable states too
1492 */
1493 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1494 state = ctxt->states[statenr];
1495 if (state != NULL)
1496 state->reached = 0;
1497 }
1498 state = ctxt->states[0];
1499 if (state != NULL)
1500 state->reached = 1;
1501 while (state != NULL) {
1502 xmlRegStatePtr target = NULL;
1503 state->reached = 2;
1504 /*
1505 * Mark all state reachable from the current reachable state
1506 */
1507 for (transnr = 0;transnr < state->nbTrans;transnr++) {
1508 if ((state->trans[transnr].to >= 0) &&
1509 ((state->trans[transnr].atom != NULL) ||
1510 (state->trans[transnr].count >= 0))) {
1511 int newto = state->trans[transnr].to;
1512
1513 if (ctxt->states[newto] == NULL)
1514 continue;
1515 if (ctxt->states[newto]->reached == 0) {
1516 ctxt->states[newto]->reached = 1;
1517 target = ctxt->states[newto];
1518 }
1519 }
1520 }
1521 /*
1522 * find the next accessible state not explored
1523 */
1524 if (target == NULL) {
1525 for (statenr = 1;statenr < ctxt->nbStates;statenr++) {
1526 state = ctxt->states[statenr];
1527 if ((state != NULL) && (state->reached == 1)) {
1528 target = state;
1529 break;
1530 }
1531 }
1532 }
1533 state = target;
1534 }
1535 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1536 state = ctxt->states[statenr];
1537 if ((state != NULL) && (state->reached == 0)) {
1538#ifdef DEBUG_REGEXP_GRAPH
1539 printf("Removed unreachable state %d\n", statenr);
1540#endif
1541 xmlRegFreeState(state);
1542 ctxt->states[statenr] = NULL;
1543 }
1544 }
1545
Daniel Veillard4255d502002-04-16 15:50:10 +00001546}
1547
Daniel Veillarde19fc232002-04-22 16:01:24 +00001548/**
1549 * xmlFACompareAtoms:
1550 * @atom1: an atom
1551 * @atom2: an atom
1552 *
1553 * Compares two atoms to check whether they are equivatents
1554 *
1555 * Returns 1 if yes and 0 otherwise
1556 */
1557static int
1558xmlFACompareAtoms(xmlRegAtomPtr atom1, xmlRegAtomPtr atom2) {
1559 if (atom1 == atom2)
1560 return(1);
1561 if ((atom1 == NULL) || (atom2 == NULL))
1562 return(0);
1563
1564 if (atom1->type != atom2->type)
1565 return(0);
1566 switch (atom1->type) {
1567 case XML_REGEXP_STRING:
1568 return(xmlStrEqual((xmlChar *)atom1->valuep,
1569 (xmlChar *)atom2->valuep));
1570 case XML_REGEXP_EPSILON:
1571 return(1);
1572 case XML_REGEXP_CHARVAL:
1573 return(atom1->codepoint == atom2->codepoint);
1574 case XML_REGEXP_RANGES:
1575 TODO;
1576 return(0);
1577 default:
1578 break;
1579 }
1580 return(1);
1581}
1582
1583/**
1584 * xmlFARecurseDeterminism:
1585 * @ctxt: a regexp parser context
1586 *
1587 * Check whether the associated regexp is determinist,
1588 * should be called after xmlFAEliminateEpsilonTransitions()
1589 *
1590 */
1591static int
1592xmlFARecurseDeterminism(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
1593 int to, xmlRegAtomPtr atom) {
1594 int ret = 1;
1595 int transnr;
1596 xmlRegTransPtr t1;
1597
1598 if (state == NULL)
1599 return(ret);
1600 for (transnr = 0;transnr < state->nbTrans;transnr++) {
1601 t1 = &(state->trans[transnr]);
1602 /*
1603 * check transitions conflicting with the one looked at
1604 */
1605 if (t1->atom == NULL) {
1606 if (t1->to == -1)
1607 continue;
1608 ret = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
1609 to, atom);
1610 if (ret == 0)
1611 return(0);
1612 continue;
1613 }
1614 if (t1->to != to)
1615 continue;
1616 if (xmlFACompareAtoms(t1->atom, atom))
1617 return(0);
1618 }
1619 return(ret);
1620}
1621
1622/**
1623 * xmlFAComputesDeterminism:
1624 * @ctxt: a regexp parser context
1625 *
1626 * Check whether the associated regexp is determinist,
1627 * should be called after xmlFAEliminateEpsilonTransitions()
1628 *
1629 */
1630static int
1631xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt) {
1632 int statenr, transnr;
1633 xmlRegStatePtr state;
1634 xmlRegTransPtr t1, t2;
1635 int i;
1636 int ret = 1;
1637
Daniel Veillard4402ab42002-09-12 16:02:56 +00001638#ifdef DEBUG_REGEXP_GRAPH
1639 printf("xmlFAComputesDeterminism\n");
1640 xmlRegPrintCtxt(stdout, ctxt);
1641#endif
Daniel Veillarde19fc232002-04-22 16:01:24 +00001642 if (ctxt->determinist != -1)
1643 return(ctxt->determinist);
1644
1645 /*
1646 * Check for all states that there isn't 2 transitions
1647 * with the same atom and a different target.
1648 */
1649 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1650 state = ctxt->states[statenr];
1651 if (state == NULL)
1652 continue;
1653 for (transnr = 0;transnr < state->nbTrans;transnr++) {
1654 t1 = &(state->trans[transnr]);
1655 /*
1656 * Determinism checks in case of counted or all transitions
1657 * will have to be handled separately
1658 */
1659 if (t1->atom == NULL)
1660 continue;
1661 if (t1->to == -1) /* eliminated */
1662 continue;
1663 for (i = 0;i < transnr;i++) {
1664 t2 = &(state->trans[i]);
1665 if (t2->to == -1) /* eliminated */
1666 continue;
1667 if (t2->atom != NULL) {
1668 if (t1->to == t2->to) {
1669 if (xmlFACompareAtoms(t1->atom, t2->atom))
1670 t2->to = -1; /* eliminate */
1671 } else {
1672 /* not determinist ! */
1673 if (xmlFACompareAtoms(t1->atom, t2->atom))
1674 ret = 0;
1675 }
1676 } else if (t1->to != -1) {
1677 /*
1678 * do the closure in case of remaining specific
1679 * epsilon transitions like choices or all
1680 */
1681 ret = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
1682 t2->to, t2->atom);
1683 if (ret == 0)
1684 return(0);
1685 }
1686 }
1687 if (ret == 0)
1688 break;
1689 }
1690 if (ret == 0)
1691 break;
1692 }
1693 ctxt->determinist = ret;
1694 return(ret);
1695}
1696
Daniel Veillard4255d502002-04-16 15:50:10 +00001697/************************************************************************
1698 * *
1699 * Routines to check input against transition atoms *
1700 * *
1701 ************************************************************************/
1702
1703static int
1704xmlRegCheckCharacterRange(xmlRegAtomType type, int codepoint, int neg,
1705 int start, int end, const xmlChar *blockName) {
1706 int ret = 0;
1707
1708 switch (type) {
1709 case XML_REGEXP_STRING:
1710 case XML_REGEXP_SUBREG:
1711 case XML_REGEXP_RANGES:
1712 case XML_REGEXP_EPSILON:
1713 return(-1);
1714 case XML_REGEXP_ANYCHAR:
1715 ret = ((codepoint != '\n') && (codepoint != '\r'));
1716 break;
1717 case XML_REGEXP_CHARVAL:
1718 ret = ((codepoint >= start) && (codepoint <= end));
1719 break;
1720 case XML_REGEXP_NOTSPACE:
1721 neg = !neg;
1722 case XML_REGEXP_ANYSPACE:
1723 ret = ((codepoint == '\n') || (codepoint == '\r') ||
1724 (codepoint == '\t') || (codepoint == ' '));
1725 break;
1726 case XML_REGEXP_NOTINITNAME:
1727 neg = !neg;
1728 case XML_REGEXP_INITNAME:
1729 ret = (xmlIsLetter(codepoint) ||
1730 (codepoint == '_') || (codepoint == ':'));
1731 break;
1732 case XML_REGEXP_NOTNAMECHAR:
1733 neg = !neg;
1734 case XML_REGEXP_NAMECHAR:
1735 ret = (xmlIsLetter(codepoint) || xmlIsDigit(codepoint) ||
1736 (codepoint == '.') || (codepoint == '-') ||
1737 (codepoint == '_') || (codepoint == ':') ||
1738 xmlIsCombining(codepoint) || xmlIsExtender(codepoint));
1739 break;
1740 case XML_REGEXP_NOTDECIMAL:
1741 neg = !neg;
1742 case XML_REGEXP_DECIMAL:
1743 ret = xmlUCSIsCatNd(codepoint);
1744 break;
1745 case XML_REGEXP_REALCHAR:
1746 neg = !neg;
1747 case XML_REGEXP_NOTREALCHAR:
1748 ret = xmlUCSIsCatP(codepoint);
1749 if (ret == 0)
1750 ret = xmlUCSIsCatZ(codepoint);
1751 if (ret == 0)
1752 ret = xmlUCSIsCatC(codepoint);
1753 break;
1754 case XML_REGEXP_LETTER:
1755 ret = xmlUCSIsCatL(codepoint);
1756 break;
1757 case XML_REGEXP_LETTER_UPPERCASE:
1758 ret = xmlUCSIsCatLu(codepoint);
1759 break;
1760 case XML_REGEXP_LETTER_LOWERCASE:
1761 ret = xmlUCSIsCatLl(codepoint);
1762 break;
1763 case XML_REGEXP_LETTER_TITLECASE:
1764 ret = xmlUCSIsCatLt(codepoint);
1765 break;
1766 case XML_REGEXP_LETTER_MODIFIER:
1767 ret = xmlUCSIsCatLm(codepoint);
1768 break;
1769 case XML_REGEXP_LETTER_OTHERS:
1770 ret = xmlUCSIsCatLo(codepoint);
1771 break;
1772 case XML_REGEXP_MARK:
1773 ret = xmlUCSIsCatM(codepoint);
1774 break;
1775 case XML_REGEXP_MARK_NONSPACING:
1776 ret = xmlUCSIsCatMn(codepoint);
1777 break;
1778 case XML_REGEXP_MARK_SPACECOMBINING:
1779 ret = xmlUCSIsCatMc(codepoint);
1780 break;
1781 case XML_REGEXP_MARK_ENCLOSING:
1782 ret = xmlUCSIsCatMe(codepoint);
1783 break;
1784 case XML_REGEXP_NUMBER:
1785 ret = xmlUCSIsCatN(codepoint);
1786 break;
1787 case XML_REGEXP_NUMBER_DECIMAL:
1788 ret = xmlUCSIsCatNd(codepoint);
1789 break;
1790 case XML_REGEXP_NUMBER_LETTER:
1791 ret = xmlUCSIsCatNl(codepoint);
1792 break;
1793 case XML_REGEXP_NUMBER_OTHERS:
1794 ret = xmlUCSIsCatNo(codepoint);
1795 break;
1796 case XML_REGEXP_PUNCT:
1797 ret = xmlUCSIsCatP(codepoint);
1798 break;
1799 case XML_REGEXP_PUNCT_CONNECTOR:
1800 ret = xmlUCSIsCatPc(codepoint);
1801 break;
1802 case XML_REGEXP_PUNCT_DASH:
1803 ret = xmlUCSIsCatPd(codepoint);
1804 break;
1805 case XML_REGEXP_PUNCT_OPEN:
1806 ret = xmlUCSIsCatPs(codepoint);
1807 break;
1808 case XML_REGEXP_PUNCT_CLOSE:
1809 ret = xmlUCSIsCatPe(codepoint);
1810 break;
1811 case XML_REGEXP_PUNCT_INITQUOTE:
1812 ret = xmlUCSIsCatPi(codepoint);
1813 break;
1814 case XML_REGEXP_PUNCT_FINQUOTE:
1815 ret = xmlUCSIsCatPf(codepoint);
1816 break;
1817 case XML_REGEXP_PUNCT_OTHERS:
1818 ret = xmlUCSIsCatPo(codepoint);
1819 break;
1820 case XML_REGEXP_SEPAR:
1821 ret = xmlUCSIsCatZ(codepoint);
1822 break;
1823 case XML_REGEXP_SEPAR_SPACE:
1824 ret = xmlUCSIsCatZs(codepoint);
1825 break;
1826 case XML_REGEXP_SEPAR_LINE:
1827 ret = xmlUCSIsCatZl(codepoint);
1828 break;
1829 case XML_REGEXP_SEPAR_PARA:
1830 ret = xmlUCSIsCatZp(codepoint);
1831 break;
1832 case XML_REGEXP_SYMBOL:
1833 ret = xmlUCSIsCatS(codepoint);
1834 break;
1835 case XML_REGEXP_SYMBOL_MATH:
1836 ret = xmlUCSIsCatSm(codepoint);
1837 break;
1838 case XML_REGEXP_SYMBOL_CURRENCY:
1839 ret = xmlUCSIsCatSc(codepoint);
1840 break;
1841 case XML_REGEXP_SYMBOL_MODIFIER:
1842 ret = xmlUCSIsCatSk(codepoint);
1843 break;
1844 case XML_REGEXP_SYMBOL_OTHERS:
1845 ret = xmlUCSIsCatSo(codepoint);
1846 break;
1847 case XML_REGEXP_OTHER:
1848 ret = xmlUCSIsCatC(codepoint);
1849 break;
1850 case XML_REGEXP_OTHER_CONTROL:
1851 ret = xmlUCSIsCatCc(codepoint);
1852 break;
1853 case XML_REGEXP_OTHER_FORMAT:
1854 ret = xmlUCSIsCatCf(codepoint);
1855 break;
1856 case XML_REGEXP_OTHER_PRIVATE:
1857 ret = xmlUCSIsCatCo(codepoint);
1858 break;
1859 case XML_REGEXP_OTHER_NA:
1860 /* ret = xmlUCSIsCatCn(codepoint); */
1861 /* Seems it doesn't exist anymore in recent Unicode releases */
1862 ret = 0;
1863 break;
1864 case XML_REGEXP_BLOCK_NAME:
1865 ret = xmlUCSIsBlock(codepoint, (const char *) blockName);
1866 break;
1867 }
1868 if (neg)
1869 return(!ret);
1870 return(ret);
1871}
1872
1873static int
1874xmlRegCheckCharacter(xmlRegAtomPtr atom, int codepoint) {
1875 int i, ret = 0;
1876 xmlRegRangePtr range;
1877
1878 if ((atom == NULL) || (!xmlIsChar(codepoint)))
1879 return(-1);
1880
1881 switch (atom->type) {
1882 case XML_REGEXP_SUBREG:
1883 case XML_REGEXP_EPSILON:
1884 return(-1);
1885 case XML_REGEXP_CHARVAL:
1886 return(codepoint == atom->codepoint);
1887 case XML_REGEXP_RANGES: {
1888 int accept = 0;
1889 for (i = 0;i < atom->nbRanges;i++) {
1890 range = atom->ranges[i];
1891 if (range->neg) {
1892 ret = xmlRegCheckCharacterRange(range->type, codepoint,
1893 0, range->start, range->end,
1894 range->blockName);
1895 if (ret != 0)
1896 return(0); /* excluded char */
1897 } else {
1898 ret = xmlRegCheckCharacterRange(range->type, codepoint,
1899 0, range->start, range->end,
1900 range->blockName);
1901 if (ret != 0)
1902 accept = 1; /* might still be excluded */
1903 }
1904 }
1905 return(accept);
1906 }
1907 case XML_REGEXP_STRING:
1908 printf("TODO: XML_REGEXP_STRING\n");
1909 return(-1);
1910 case XML_REGEXP_ANYCHAR:
1911 case XML_REGEXP_ANYSPACE:
1912 case XML_REGEXP_NOTSPACE:
1913 case XML_REGEXP_INITNAME:
1914 case XML_REGEXP_NOTINITNAME:
1915 case XML_REGEXP_NAMECHAR:
1916 case XML_REGEXP_NOTNAMECHAR:
1917 case XML_REGEXP_DECIMAL:
1918 case XML_REGEXP_NOTDECIMAL:
1919 case XML_REGEXP_REALCHAR:
1920 case XML_REGEXP_NOTREALCHAR:
1921 case XML_REGEXP_LETTER:
1922 case XML_REGEXP_LETTER_UPPERCASE:
1923 case XML_REGEXP_LETTER_LOWERCASE:
1924 case XML_REGEXP_LETTER_TITLECASE:
1925 case XML_REGEXP_LETTER_MODIFIER:
1926 case XML_REGEXP_LETTER_OTHERS:
1927 case XML_REGEXP_MARK:
1928 case XML_REGEXP_MARK_NONSPACING:
1929 case XML_REGEXP_MARK_SPACECOMBINING:
1930 case XML_REGEXP_MARK_ENCLOSING:
1931 case XML_REGEXP_NUMBER:
1932 case XML_REGEXP_NUMBER_DECIMAL:
1933 case XML_REGEXP_NUMBER_LETTER:
1934 case XML_REGEXP_NUMBER_OTHERS:
1935 case XML_REGEXP_PUNCT:
1936 case XML_REGEXP_PUNCT_CONNECTOR:
1937 case XML_REGEXP_PUNCT_DASH:
1938 case XML_REGEXP_PUNCT_OPEN:
1939 case XML_REGEXP_PUNCT_CLOSE:
1940 case XML_REGEXP_PUNCT_INITQUOTE:
1941 case XML_REGEXP_PUNCT_FINQUOTE:
1942 case XML_REGEXP_PUNCT_OTHERS:
1943 case XML_REGEXP_SEPAR:
1944 case XML_REGEXP_SEPAR_SPACE:
1945 case XML_REGEXP_SEPAR_LINE:
1946 case XML_REGEXP_SEPAR_PARA:
1947 case XML_REGEXP_SYMBOL:
1948 case XML_REGEXP_SYMBOL_MATH:
1949 case XML_REGEXP_SYMBOL_CURRENCY:
1950 case XML_REGEXP_SYMBOL_MODIFIER:
1951 case XML_REGEXP_SYMBOL_OTHERS:
1952 case XML_REGEXP_OTHER:
1953 case XML_REGEXP_OTHER_CONTROL:
1954 case XML_REGEXP_OTHER_FORMAT:
1955 case XML_REGEXP_OTHER_PRIVATE:
1956 case XML_REGEXP_OTHER_NA:
1957 case XML_REGEXP_BLOCK_NAME:
1958 ret = xmlRegCheckCharacterRange(atom->type, codepoint, 0, 0, 0,
1959 (const xmlChar *)atom->valuep);
1960 if (atom->neg)
1961 ret = !ret;
1962 break;
1963 }
1964 return(ret);
1965}
1966
1967/************************************************************************
1968 * *
1969 * Saving an restoring state of an execution context *
1970 * *
1971 ************************************************************************/
1972
1973#ifdef DEBUG_REGEXP_EXEC
1974static void
1975xmlFARegDebugExec(xmlRegExecCtxtPtr exec) {
1976 printf("state: %d:%d:idx %d", exec->state->no, exec->transno, exec->index);
1977 if (exec->inputStack != NULL) {
1978 int i;
1979 printf(": ");
1980 for (i = 0;(i < 3) && (i < exec->inputStackNr);i++)
1981 printf("%s ", exec->inputStack[exec->inputStackNr - (i + 1)]);
1982 } else {
1983 printf(": %s", &(exec->inputString[exec->index]));
1984 }
1985 printf("\n");
1986}
1987#endif
1988
1989static void
1990xmlFARegExecSave(xmlRegExecCtxtPtr exec) {
1991#ifdef DEBUG_REGEXP_EXEC
1992 printf("saving ");
1993 exec->transno++;
1994 xmlFARegDebugExec(exec);
1995 exec->transno--;
1996#endif
1997
1998 if (exec->maxRollbacks == 0) {
1999 exec->maxRollbacks = 4;
2000 exec->rollbacks = (xmlRegExecRollback *) xmlMalloc(exec->maxRollbacks *
2001 sizeof(xmlRegExecRollback));
2002 if (exec->rollbacks == NULL) {
2003 fprintf(stderr, "exec save: allocation failed");
2004 exec->maxRollbacks = 0;
2005 return;
2006 }
2007 memset(exec->rollbacks, 0,
2008 exec->maxRollbacks * sizeof(xmlRegExecRollback));
2009 } else if (exec->nbRollbacks >= exec->maxRollbacks) {
2010 xmlRegExecRollback *tmp;
2011 int len = exec->maxRollbacks;
2012
2013 exec->maxRollbacks *= 2;
2014 tmp = (xmlRegExecRollback *) xmlRealloc(exec->rollbacks,
2015 exec->maxRollbacks * sizeof(xmlRegExecRollback));
2016 if (tmp == NULL) {
2017 fprintf(stderr, "exec save: allocation failed");
2018 exec->maxRollbacks /= 2;
2019 return;
2020 }
2021 exec->rollbacks = tmp;
2022 tmp = &exec->rollbacks[len];
2023 memset(tmp, 0, (exec->maxRollbacks - len) * sizeof(xmlRegExecRollback));
2024 }
2025 exec->rollbacks[exec->nbRollbacks].state = exec->state;
2026 exec->rollbacks[exec->nbRollbacks].index = exec->index;
2027 exec->rollbacks[exec->nbRollbacks].nextbranch = exec->transno + 1;
2028 if (exec->comp->nbCounters > 0) {
2029 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
2030 exec->rollbacks[exec->nbRollbacks].counts = (int *)
2031 xmlMalloc(exec->comp->nbCounters * sizeof(int));
2032 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
2033 fprintf(stderr, "exec save: allocation failed");
2034 exec->status = -5;
2035 return;
2036 }
2037 }
2038 memcpy(exec->rollbacks[exec->nbRollbacks].counts, exec->counts,
2039 exec->comp->nbCounters * sizeof(int));
2040 }
2041 exec->nbRollbacks++;
2042}
2043
2044static void
2045xmlFARegExecRollBack(xmlRegExecCtxtPtr exec) {
2046 if (exec->nbRollbacks <= 0) {
2047 exec->status = -1;
2048#ifdef DEBUG_REGEXP_EXEC
2049 printf("rollback failed on empty stack\n");
2050#endif
2051 return;
2052 }
2053 exec->nbRollbacks--;
2054 exec->state = exec->rollbacks[exec->nbRollbacks].state;
2055 exec->index = exec->rollbacks[exec->nbRollbacks].index;
2056 exec->transno = exec->rollbacks[exec->nbRollbacks].nextbranch;
2057 if (exec->comp->nbCounters > 0) {
2058 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
2059 fprintf(stderr, "exec save: allocation failed");
2060 exec->status = -6;
2061 return;
2062 }
2063 memcpy(exec->counts, exec->rollbacks[exec->nbRollbacks].counts,
2064 exec->comp->nbCounters * sizeof(int));
2065 }
2066
2067#ifdef DEBUG_REGEXP_EXEC
2068 printf("restored ");
2069 xmlFARegDebugExec(exec);
2070#endif
2071}
2072
2073/************************************************************************
2074 * *
2075 * Verifyer, running an input against a compiled regexp *
2076 * *
2077 ************************************************************************/
2078
2079static int
2080xmlFARegExec(xmlRegexpPtr comp, const xmlChar *content) {
2081 xmlRegExecCtxt execval;
2082 xmlRegExecCtxtPtr exec = &execval;
2083 int ret, codepoint, len;
2084
2085 exec->inputString = content;
2086 exec->index = 0;
2087 exec->determinist = 1;
2088 exec->maxRollbacks = 0;
2089 exec->nbRollbacks = 0;
2090 exec->rollbacks = NULL;
2091 exec->status = 0;
2092 exec->comp = comp;
2093 exec->state = comp->states[0];
2094 exec->transno = 0;
2095 exec->transcount = 0;
2096 if (comp->nbCounters > 0) {
2097 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int));
2098 if (exec->counts == NULL)
2099 return(-1);
2100 memset(exec->counts, 0, comp->nbCounters * sizeof(int));
2101 } else
2102 exec->counts = NULL;
2103 while ((exec->status == 0) &&
2104 ((exec->inputString[exec->index] != 0) ||
2105 (exec->state->type != XML_REGEXP_FINAL_STATE))) {
2106 xmlRegTransPtr trans;
2107 xmlRegAtomPtr atom;
2108
2109 /*
2110 * End of input on non-terminal state, rollback, however we may
2111 * still have epsilon like transition for counted transitions
2112 * on counters, in that case don't break too early.
2113 */
2114 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL))
2115 goto rollback;
2116
2117 exec->transcount = 0;
2118 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
2119 trans = &exec->state->trans[exec->transno];
2120 if (trans->to < 0)
2121 continue;
2122 atom = trans->atom;
2123 ret = 0;
2124 if (trans->count >= 0) {
2125 int count;
2126 xmlRegCounterPtr counter;
2127
2128 /*
2129 * A counted transition.
2130 */
2131
2132 count = exec->counts[trans->count];
2133 counter = &exec->comp->counters[trans->count];
2134#ifdef DEBUG_REGEXP_EXEC
2135 printf("testing count %d: val %d, min %d, max %d\n",
2136 trans->count, count, counter->min, counter->max);
2137#endif
2138 ret = ((count >= counter->min) && (count <= counter->max));
2139 } else if (atom == NULL) {
2140 fprintf(stderr, "epsilon transition left at runtime\n");
2141 exec->status = -2;
2142 break;
2143 } else if (exec->inputString[exec->index] != 0) {
2144 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
2145 ret = xmlRegCheckCharacter(atom, codepoint);
2146 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
2147 xmlRegStatePtr to = comp->states[trans->to];
2148
2149 /*
2150 * this is a multiple input sequence
2151 */
2152 if (exec->state->nbTrans > exec->transno + 1) {
2153 xmlFARegExecSave(exec);
2154 }
2155 exec->transcount = 1;
2156 do {
2157 /*
2158 * Try to progress as much as possible on the input
2159 */
2160 if (exec->transcount == atom->max) {
2161 break;
2162 }
2163 exec->index += len;
2164 /*
2165 * End of input: stop here
2166 */
2167 if (exec->inputString[exec->index] == 0) {
2168 exec->index -= len;
2169 break;
2170 }
2171 if (exec->transcount >= atom->min) {
2172 int transno = exec->transno;
2173 xmlRegStatePtr state = exec->state;
2174
2175 /*
2176 * The transition is acceptable save it
2177 */
2178 exec->transno = -1; /* trick */
2179 exec->state = to;
2180 xmlFARegExecSave(exec);
2181 exec->transno = transno;
2182 exec->state = state;
2183 }
2184 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
2185 len);
2186 ret = xmlRegCheckCharacter(atom, codepoint);
2187 exec->transcount++;
2188 } while (ret == 1);
2189 if (exec->transcount < atom->min)
2190 ret = 0;
2191
2192 /*
2193 * If the last check failed but one transition was found
2194 * possible, rollback
2195 */
2196 if (ret < 0)
2197 ret = 0;
2198 if (ret == 0) {
2199 goto rollback;
2200 }
2201 }
2202 }
2203 if (ret == 1) {
2204 if (exec->state->nbTrans > exec->transno + 1) {
2205 xmlFARegExecSave(exec);
2206 }
2207 if (trans->counter >= 0) {
2208#ifdef DEBUG_REGEXP_EXEC
2209 printf("Increasing count %d\n", trans->counter);
2210#endif
2211 exec->counts[trans->counter]++;
2212 }
2213#ifdef DEBUG_REGEXP_EXEC
2214 printf("entering state %d\n", trans->to);
2215#endif
2216 exec->state = comp->states[trans->to];
2217 exec->transno = 0;
2218 if (trans->atom != NULL) {
2219 exec->index += len;
2220 }
2221 goto progress;
2222 } else if (ret < 0) {
2223 exec->status = -4;
2224 break;
2225 }
2226 }
2227 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
2228rollback:
2229 /*
2230 * Failed to find a way out
2231 */
2232 exec->determinist = 0;
2233 xmlFARegExecRollBack(exec);
2234 }
2235progress:
2236 continue;
2237 }
2238 if (exec->rollbacks != NULL) {
2239 if (exec->counts != NULL) {
2240 int i;
2241
2242 for (i = 0;i < exec->maxRollbacks;i++)
2243 if (exec->rollbacks[i].counts != NULL)
2244 xmlFree(exec->rollbacks[i].counts);
2245 }
2246 xmlFree(exec->rollbacks);
2247 }
2248 if (exec->counts != NULL)
2249 xmlFree(exec->counts);
2250 if (exec->status == 0)
2251 return(1);
2252 if (exec->status == -1)
2253 return(0);
2254 return(exec->status);
2255}
2256
2257/************************************************************************
2258 * *
2259 * Progressive interface to the verifyer one atom at a time *
2260 * *
2261 ************************************************************************/
2262
2263/**
Daniel Veillard01c13b52002-12-10 15:19:08 +00002264 * xmlRegNewExecCtxt:
Daniel Veillard4255d502002-04-16 15:50:10 +00002265 * @comp: a precompiled regular expression
2266 * @callback: a callback function used for handling progresses in the
2267 * automata matching phase
2268 * @data: the context data associated to the callback in this context
2269 *
2270 * Build a context used for progressive evaluation of a regexp.
Daniel Veillard01c13b52002-12-10 15:19:08 +00002271 *
2272 * Returns the new context
Daniel Veillard4255d502002-04-16 15:50:10 +00002273 */
2274xmlRegExecCtxtPtr
2275xmlRegNewExecCtxt(xmlRegexpPtr comp, xmlRegExecCallbacks callback, void *data) {
2276 xmlRegExecCtxtPtr exec;
2277
2278 if (comp == NULL)
2279 return(NULL);
2280 exec = (xmlRegExecCtxtPtr) xmlMalloc(sizeof(xmlRegExecCtxt));
2281 if (exec == NULL) {
2282 return(NULL);
2283 }
2284 memset(exec, 0, sizeof(xmlRegExecCtxt));
2285 exec->inputString = NULL;
2286 exec->index = 0;
2287 exec->determinist = 1;
2288 exec->maxRollbacks = 0;
2289 exec->nbRollbacks = 0;
2290 exec->rollbacks = NULL;
2291 exec->status = 0;
2292 exec->comp = comp;
Daniel Veillard23e73572002-09-19 19:56:43 +00002293 if (comp->compact == NULL)
2294 exec->state = comp->states[0];
Daniel Veillard4255d502002-04-16 15:50:10 +00002295 exec->transno = 0;
2296 exec->transcount = 0;
2297 exec->callback = callback;
2298 exec->data = data;
2299 if (comp->nbCounters > 0) {
2300 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int));
2301 if (exec->counts == NULL) {
2302 xmlFree(exec);
2303 return(NULL);
2304 }
2305 memset(exec->counts, 0, comp->nbCounters * sizeof(int));
2306 } else
2307 exec->counts = NULL;
2308 exec->inputStackMax = 0;
2309 exec->inputStackNr = 0;
2310 exec->inputStack = NULL;
2311 return(exec);
2312}
2313
2314/**
2315 * xmlRegFreeExecCtxt:
2316 * @exec: a regular expression evaulation context
2317 *
2318 * Free the structures associated to a regular expression evaulation context.
2319 */
2320void
2321xmlRegFreeExecCtxt(xmlRegExecCtxtPtr exec) {
2322 if (exec == NULL)
2323 return;
2324
2325 if (exec->rollbacks != NULL) {
2326 if (exec->counts != NULL) {
2327 int i;
2328
2329 for (i = 0;i < exec->maxRollbacks;i++)
2330 if (exec->rollbacks[i].counts != NULL)
2331 xmlFree(exec->rollbacks[i].counts);
2332 }
2333 xmlFree(exec->rollbacks);
2334 }
2335 if (exec->counts != NULL)
2336 xmlFree(exec->counts);
2337 if (exec->inputStack != NULL) {
2338 int i;
2339
Daniel Veillard32370232002-10-16 14:08:14 +00002340 for (i = 0;i < exec->inputStackNr;i++) {
2341 if (exec->inputStack[i].value != NULL)
2342 xmlFree(exec->inputStack[i].value);
2343 }
Daniel Veillard4255d502002-04-16 15:50:10 +00002344 xmlFree(exec->inputStack);
2345 }
2346 xmlFree(exec);
2347}
2348
2349static void
2350xmlFARegExecSaveInputString(xmlRegExecCtxtPtr exec, const xmlChar *value,
2351 void *data) {
2352#ifdef DEBUG_PUSH
2353 printf("saving value: %d:%s\n", exec->inputStackNr, value);
2354#endif
2355 if (exec->inputStackMax == 0) {
2356 exec->inputStackMax = 4;
2357 exec->inputStack = (xmlRegInputTokenPtr)
2358 xmlMalloc(exec->inputStackMax * sizeof(xmlRegInputToken));
2359 if (exec->inputStack == NULL) {
2360 fprintf(stderr, "push input: allocation failed");
2361 exec->inputStackMax = 0;
2362 return;
2363 }
2364 } else if (exec->inputStackNr + 1 >= exec->inputStackMax) {
2365 xmlRegInputTokenPtr tmp;
2366
2367 exec->inputStackMax *= 2;
2368 tmp = (xmlRegInputTokenPtr) xmlRealloc(exec->inputStack,
2369 exec->inputStackMax * sizeof(xmlRegInputToken));
2370 if (tmp == NULL) {
2371 fprintf(stderr, "push input: allocation failed");
2372 exec->inputStackMax /= 2;
2373 return;
2374 }
2375 exec->inputStack = tmp;
2376 }
2377 exec->inputStack[exec->inputStackNr].value = xmlStrdup(value);
2378 exec->inputStack[exec->inputStackNr].data = data;
2379 exec->inputStackNr++;
2380 exec->inputStack[exec->inputStackNr].value = NULL;
2381 exec->inputStack[exec->inputStackNr].data = NULL;
2382}
2383
2384
2385/**
Daniel Veillard23e73572002-09-19 19:56:43 +00002386 * xmlRegCompactPushString:
2387 * @exec: a regexp execution context
2388 * @comp: the precompiled exec with a compact table
2389 * @value: a string token input
2390 * @data: data associated to the token to reuse in callbacks
2391 *
2392 * Push one input token in the execution context
2393 *
2394 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
2395 * a negative value in case of error.
2396 */
2397static int
2398xmlRegCompactPushString(xmlRegExecCtxtPtr exec,
2399 xmlRegexpPtr comp,
2400 const xmlChar *value,
2401 void *data) {
2402 int state = exec->index;
2403 int i, target;
2404
2405 if ((comp == NULL) || (comp->compact == NULL) || (comp->stringMap == NULL))
2406 return(-1);
2407
2408 if (value == NULL) {
2409 /*
2410 * are we at a final state ?
2411 */
2412 if (comp->compact[state * (comp->nbstrings + 1)] ==
2413 XML_REGEXP_FINAL_STATE)
2414 return(1);
2415 return(0);
2416 }
2417
2418#ifdef DEBUG_PUSH
2419 printf("value pushed: %s\n", value);
2420#endif
2421
2422 /*
2423 * Examine all outside transition from current state
2424 */
2425 for (i = 0;i < comp->nbstrings;i++) {
2426 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
2427 if ((target > 0) && (target <= comp->nbstates)) {
2428 target--; /* to avoid 0 */
2429 if (xmlStrEqual(comp->stringMap[i], value)) {
2430 exec->index = target;
Daniel Veillard118aed72002-09-24 14:13:13 +00002431 if ((exec->callback != NULL) && (comp->transdata != NULL)) {
2432 exec->callback(exec->data, value,
2433 comp->transdata[state * comp->nbstrings + i], data);
2434 }
Daniel Veillard23e73572002-09-19 19:56:43 +00002435#ifdef DEBUG_PUSH
2436 printf("entering state %d\n", target);
2437#endif
2438 if (comp->compact[target * (comp->nbstrings + 1)] ==
2439 XML_REGEXP_FINAL_STATE)
2440 return(1);
2441 return(0);
2442 }
2443 }
2444 }
2445 /*
2446 * Failed to find an exit transition out from current state for the
2447 * current token
2448 */
2449#ifdef DEBUG_PUSH
2450 printf("failed to find a transition for %s on state %d\n", value, state);
2451#endif
2452 exec->status = -1;
2453 return(-1);
2454}
2455
2456/**
Daniel Veillard4255d502002-04-16 15:50:10 +00002457 * xmlRegExecPushString:
Daniel Veillardea7751d2002-12-20 00:16:24 +00002458 * @exec: a regexp execution context or NULL to indicate the end
Daniel Veillard4255d502002-04-16 15:50:10 +00002459 * @value: a string token input
2460 * @data: data associated to the token to reuse in callbacks
2461 *
2462 * Push one input token in the execution context
2463 *
2464 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
2465 * a negative value in case of error.
2466 */
2467int
2468xmlRegExecPushString(xmlRegExecCtxtPtr exec, const xmlChar *value,
2469 void *data) {
2470 xmlRegTransPtr trans;
2471 xmlRegAtomPtr atom;
2472 int ret;
2473 int final = 0;
2474
2475 if (exec == NULL)
2476 return(-1);
Daniel Veillard23e73572002-09-19 19:56:43 +00002477 if (exec->comp == NULL)
2478 return(-1);
Daniel Veillard4255d502002-04-16 15:50:10 +00002479 if (exec->status != 0)
2480 return(exec->status);
2481
Daniel Veillard23e73572002-09-19 19:56:43 +00002482 if (exec->comp->compact != NULL)
2483 return(xmlRegCompactPushString(exec, exec->comp, value, data));
2484
Daniel Veillard4255d502002-04-16 15:50:10 +00002485 if (value == NULL) {
2486 if (exec->state->type == XML_REGEXP_FINAL_STATE)
2487 return(1);
2488 final = 1;
2489 }
2490
2491#ifdef DEBUG_PUSH
2492 printf("value pushed: %s\n", value);
2493#endif
2494 /*
2495 * If we have an active rollback stack push the new value there
2496 * and get back to where we were left
2497 */
2498 if ((value != NULL) && (exec->inputStackNr > 0)) {
2499 xmlFARegExecSaveInputString(exec, value, data);
2500 value = exec->inputStack[exec->index].value;
2501 data = exec->inputStack[exec->index].data;
2502#ifdef DEBUG_PUSH
2503 printf("value loaded: %s\n", value);
2504#endif
2505 }
2506
2507 while ((exec->status == 0) &&
2508 ((value != NULL) ||
2509 ((final == 1) &&
2510 (exec->state->type != XML_REGEXP_FINAL_STATE)))) {
2511
2512 /*
2513 * End of input on non-terminal state, rollback, however we may
2514 * still have epsilon like transition for counted transitions
2515 * on counters, in that case don't break too early.
2516 */
Daniel Veillardb509f152002-04-17 16:28:10 +00002517 if ((value == NULL) && (exec->counts == NULL))
Daniel Veillard4255d502002-04-16 15:50:10 +00002518 goto rollback;
2519
2520 exec->transcount = 0;
2521 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
2522 trans = &exec->state->trans[exec->transno];
2523 if (trans->to < 0)
2524 continue;
2525 atom = trans->atom;
2526 ret = 0;
Daniel Veillard441bc322002-04-20 17:38:48 +00002527 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
2528 int i;
2529 int count;
2530 xmlRegTransPtr t;
2531 xmlRegCounterPtr counter;
2532
2533 ret = 0;
2534
2535#ifdef DEBUG_PUSH
2536 printf("testing all lax %d\n", trans->count);
2537#endif
2538 /*
2539 * Check all counted transitions from the current state
2540 */
2541 if ((value == NULL) && (final)) {
2542 ret = 1;
2543 } else if (value != NULL) {
2544 for (i = 0;i < exec->state->nbTrans;i++) {
2545 t = &exec->state->trans[i];
2546 if ((t->counter < 0) || (t == trans))
2547 continue;
2548 counter = &exec->comp->counters[t->counter];
2549 count = exec->counts[t->counter];
2550 if ((count < counter->max) &&
2551 (t->atom != NULL) &&
2552 (xmlStrEqual(value, t->atom->valuep))) {
2553 ret = 0;
2554 break;
2555 }
2556 if ((count >= counter->min) &&
2557 (count < counter->max) &&
2558 (xmlStrEqual(value, t->atom->valuep))) {
2559 ret = 1;
2560 break;
2561 }
2562 }
2563 }
2564 } else if (trans->count == REGEXP_ALL_COUNTER) {
Daniel Veillard8a001f62002-04-20 07:24:11 +00002565 int i;
2566 int count;
2567 xmlRegTransPtr t;
2568 xmlRegCounterPtr counter;
2569
2570 ret = 1;
2571
2572#ifdef DEBUG_PUSH
2573 printf("testing all %d\n", trans->count);
2574#endif
2575 /*
2576 * Check all counted transitions from the current state
2577 */
2578 for (i = 0;i < exec->state->nbTrans;i++) {
2579 t = &exec->state->trans[i];
2580 if ((t->counter < 0) || (t == trans))
2581 continue;
2582 counter = &exec->comp->counters[t->counter];
2583 count = exec->counts[t->counter];
2584 if ((count < counter->min) || (count > counter->max)) {
2585 ret = 0;
2586 break;
2587 }
2588 }
2589 } else if (trans->count >= 0) {
Daniel Veillard4255d502002-04-16 15:50:10 +00002590 int count;
2591 xmlRegCounterPtr counter;
2592
2593 /*
2594 * A counted transition.
2595 */
2596
2597 count = exec->counts[trans->count];
2598 counter = &exec->comp->counters[trans->count];
2599#ifdef DEBUG_PUSH
2600 printf("testing count %d: val %d, min %d, max %d\n",
2601 trans->count, count, counter->min, counter->max);
2602#endif
2603 ret = ((count >= counter->min) && (count <= counter->max));
2604 } else if (atom == NULL) {
2605 fprintf(stderr, "epsilon transition left at runtime\n");
2606 exec->status = -2;
2607 break;
2608 } else if (value != NULL) {
2609 ret = xmlStrEqual(value, atom->valuep);
Daniel Veillard441bc322002-04-20 17:38:48 +00002610 if ((ret == 1) && (trans->counter >= 0)) {
2611 xmlRegCounterPtr counter;
2612 int count;
2613
2614 count = exec->counts[trans->counter];
2615 counter = &exec->comp->counters[trans->counter];
2616 if (count >= counter->max)
2617 ret = 0;
2618 }
2619
Daniel Veillard4255d502002-04-16 15:50:10 +00002620 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
2621 xmlRegStatePtr to = exec->comp->states[trans->to];
2622
2623 /*
2624 * this is a multiple input sequence
2625 */
2626 if (exec->state->nbTrans > exec->transno + 1) {
2627 if (exec->inputStackNr <= 0) {
2628 xmlFARegExecSaveInputString(exec, value, data);
2629 }
2630 xmlFARegExecSave(exec);
2631 }
2632 exec->transcount = 1;
2633 do {
2634 /*
2635 * Try to progress as much as possible on the input
2636 */
2637 if (exec->transcount == atom->max) {
2638 break;
2639 }
2640 exec->index++;
2641 value = exec->inputStack[exec->index].value;
2642 data = exec->inputStack[exec->index].data;
2643#ifdef DEBUG_PUSH
2644 printf("value loaded: %s\n", value);
2645#endif
2646
2647 /*
2648 * End of input: stop here
2649 */
2650 if (value == NULL) {
2651 exec->index --;
2652 break;
2653 }
2654 if (exec->transcount >= atom->min) {
2655 int transno = exec->transno;
2656 xmlRegStatePtr state = exec->state;
2657
2658 /*
2659 * The transition is acceptable save it
2660 */
2661 exec->transno = -1; /* trick */
2662 exec->state = to;
2663 if (exec->inputStackNr <= 0) {
2664 xmlFARegExecSaveInputString(exec, value, data);
2665 }
2666 xmlFARegExecSave(exec);
2667 exec->transno = transno;
2668 exec->state = state;
2669 }
2670 ret = xmlStrEqual(value, atom->valuep);
2671 exec->transcount++;
2672 } while (ret == 1);
2673 if (exec->transcount < atom->min)
2674 ret = 0;
2675
2676 /*
2677 * If the last check failed but one transition was found
2678 * possible, rollback
2679 */
2680 if (ret < 0)
2681 ret = 0;
2682 if (ret == 0) {
2683 goto rollback;
2684 }
2685 }
2686 }
2687 if (ret == 1) {
2688 if ((exec->callback != NULL) && (atom != NULL)) {
2689 exec->callback(exec->data, atom->valuep,
2690 atom->data, data);
2691 }
2692 if (exec->state->nbTrans > exec->transno + 1) {
2693 if (exec->inputStackNr <= 0) {
2694 xmlFARegExecSaveInputString(exec, value, data);
2695 }
2696 xmlFARegExecSave(exec);
2697 }
2698 if (trans->counter >= 0) {
2699#ifdef DEBUG_PUSH
2700 printf("Increasing count %d\n", trans->counter);
2701#endif
2702 exec->counts[trans->counter]++;
2703 }
2704#ifdef DEBUG_PUSH
2705 printf("entering state %d\n", trans->to);
2706#endif
2707 exec->state = exec->comp->states[trans->to];
2708 exec->transno = 0;
2709 if (trans->atom != NULL) {
2710 if (exec->inputStack != NULL) {
2711 exec->index++;
2712 if (exec->index < exec->inputStackNr) {
2713 value = exec->inputStack[exec->index].value;
2714 data = exec->inputStack[exec->index].data;
2715#ifdef DEBUG_PUSH
2716 printf("value loaded: %s\n", value);
2717#endif
2718 } else {
2719 value = NULL;
2720 data = NULL;
2721#ifdef DEBUG_PUSH
2722 printf("end of input\n");
2723#endif
2724 }
2725 } else {
2726 value = NULL;
2727 data = NULL;
2728#ifdef DEBUG_PUSH
2729 printf("end of input\n");
2730#endif
2731 }
2732 }
2733 goto progress;
2734 } else if (ret < 0) {
2735 exec->status = -4;
2736 break;
2737 }
2738 }
2739 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
2740rollback:
2741 /*
2742 * Failed to find a way out
2743 */
2744 exec->determinist = 0;
2745 xmlFARegExecRollBack(exec);
2746 if (exec->status == 0) {
2747 value = exec->inputStack[exec->index].value;
2748 data = exec->inputStack[exec->index].data;
2749#ifdef DEBUG_PUSH
2750 printf("value loaded: %s\n", value);
2751#endif
2752 }
2753 }
2754progress:
2755 continue;
2756 }
2757 if (exec->status == 0) {
2758 return(exec->state->type == XML_REGEXP_FINAL_STATE);
2759 }
2760 return(exec->status);
2761}
2762
2763#if 0
2764static int
2765xmlRegExecPushChar(xmlRegExecCtxtPtr exec, int UCS) {
2766 xmlRegTransPtr trans;
2767 xmlRegAtomPtr atom;
2768 int ret;
2769 int codepoint, len;
2770
2771 if (exec == NULL)
2772 return(-1);
2773 if (exec->status != 0)
2774 return(exec->status);
2775
2776 while ((exec->status == 0) &&
2777 ((exec->inputString[exec->index] != 0) ||
2778 (exec->state->type != XML_REGEXP_FINAL_STATE))) {
2779
2780 /*
2781 * End of input on non-terminal state, rollback, however we may
2782 * still have epsilon like transition for counted transitions
2783 * on counters, in that case don't break too early.
2784 */
2785 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL))
2786 goto rollback;
2787
2788 exec->transcount = 0;
2789 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
2790 trans = &exec->state->trans[exec->transno];
2791 if (trans->to < 0)
2792 continue;
2793 atom = trans->atom;
2794 ret = 0;
2795 if (trans->count >= 0) {
2796 int count;
2797 xmlRegCounterPtr counter;
2798
2799 /*
2800 * A counted transition.
2801 */
2802
2803 count = exec->counts[trans->count];
2804 counter = &exec->comp->counters[trans->count];
2805#ifdef DEBUG_REGEXP_EXEC
2806 printf("testing count %d: val %d, min %d, max %d\n",
2807 trans->count, count, counter->min, counter->max);
2808#endif
2809 ret = ((count >= counter->min) && (count <= counter->max));
2810 } else if (atom == NULL) {
2811 fprintf(stderr, "epsilon transition left at runtime\n");
2812 exec->status = -2;
2813 break;
2814 } else if (exec->inputString[exec->index] != 0) {
2815 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
2816 ret = xmlRegCheckCharacter(atom, codepoint);
2817 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
2818 xmlRegStatePtr to = exec->comp->states[trans->to];
2819
2820 /*
2821 * this is a multiple input sequence
2822 */
2823 if (exec->state->nbTrans > exec->transno + 1) {
2824 xmlFARegExecSave(exec);
2825 }
2826 exec->transcount = 1;
2827 do {
2828 /*
2829 * Try to progress as much as possible on the input
2830 */
2831 if (exec->transcount == atom->max) {
2832 break;
2833 }
2834 exec->index += len;
2835 /*
2836 * End of input: stop here
2837 */
2838 if (exec->inputString[exec->index] == 0) {
2839 exec->index -= len;
2840 break;
2841 }
2842 if (exec->transcount >= atom->min) {
2843 int transno = exec->transno;
2844 xmlRegStatePtr state = exec->state;
2845
2846 /*
2847 * The transition is acceptable save it
2848 */
2849 exec->transno = -1; /* trick */
2850 exec->state = to;
2851 xmlFARegExecSave(exec);
2852 exec->transno = transno;
2853 exec->state = state;
2854 }
2855 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
2856 len);
2857 ret = xmlRegCheckCharacter(atom, codepoint);
2858 exec->transcount++;
2859 } while (ret == 1);
2860 if (exec->transcount < atom->min)
2861 ret = 0;
2862
2863 /*
2864 * If the last check failed but one transition was found
2865 * possible, rollback
2866 */
2867 if (ret < 0)
2868 ret = 0;
2869 if (ret == 0) {
2870 goto rollback;
2871 }
2872 }
2873 }
2874 if (ret == 1) {
2875 if (exec->state->nbTrans > exec->transno + 1) {
2876 xmlFARegExecSave(exec);
2877 }
2878 if (trans->counter >= 0) {
2879#ifdef DEBUG_REGEXP_EXEC
2880 printf("Increasing count %d\n", trans->counter);
2881#endif
2882 exec->counts[trans->counter]++;
2883 }
2884#ifdef DEBUG_REGEXP_EXEC
2885 printf("entering state %d\n", trans->to);
2886#endif
2887 exec->state = exec->comp->states[trans->to];
2888 exec->transno = 0;
2889 if (trans->atom != NULL) {
2890 exec->index += len;
2891 }
2892 goto progress;
2893 } else if (ret < 0) {
2894 exec->status = -4;
2895 break;
2896 }
2897 }
2898 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
2899rollback:
2900 /*
2901 * Failed to find a way out
2902 */
2903 exec->determinist = 0;
2904 xmlFARegExecRollBack(exec);
2905 }
2906progress:
2907 continue;
2908 }
2909}
2910#endif
2911/************************************************************************
2912 * *
2913 * Parser for the Shemas Datatype Regular Expressions *
2914 * http://www.w3.org/TR/2001/REC-xmlschema-2-20010502/#regexs *
2915 * *
2916 ************************************************************************/
2917
2918/**
2919 * xmlFAIsChar:
Daniel Veillard441bc322002-04-20 17:38:48 +00002920 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00002921 *
2922 * [10] Char ::= [^.\?*+()|#x5B#x5D]
2923 */
2924static int
2925xmlFAIsChar(xmlRegParserCtxtPtr ctxt) {
2926 int cur;
2927 int len;
2928
2929 cur = CUR_SCHAR(ctxt->cur, len);
2930 if ((cur == '.') || (cur == '\\') || (cur == '?') ||
2931 (cur == '*') || (cur == '+') || (cur == '(') ||
2932 (cur == ')') || (cur == '|') || (cur == 0x5B) ||
2933 (cur == 0x5D) || (cur == 0))
2934 return(-1);
2935 return(cur);
2936}
2937
2938/**
2939 * xmlFAParseCharProp:
Daniel Veillard441bc322002-04-20 17:38:48 +00002940 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00002941 *
2942 * [27] charProp ::= IsCategory | IsBlock
2943 * [28] IsCategory ::= Letters | Marks | Numbers | Punctuation |
2944 * Separators | Symbols | Others
2945 * [29] Letters ::= 'L' [ultmo]?
2946 * [30] Marks ::= 'M' [nce]?
2947 * [31] Numbers ::= 'N' [dlo]?
2948 * [32] Punctuation ::= 'P' [cdseifo]?
2949 * [33] Separators ::= 'Z' [slp]?
2950 * [34] Symbols ::= 'S' [mcko]?
2951 * [35] Others ::= 'C' [cfon]?
2952 * [36] IsBlock ::= 'Is' [a-zA-Z0-9#x2D]+
2953 */
2954static void
2955xmlFAParseCharProp(xmlRegParserCtxtPtr ctxt) {
2956 int cur;
2957 xmlRegAtomType type = 0;
2958 xmlChar *blockName = NULL;
2959
2960 cur = CUR;
2961 if (cur == 'L') {
2962 NEXT;
2963 cur = CUR;
2964 if (cur == 'u') {
2965 NEXT;
2966 type = XML_REGEXP_LETTER_UPPERCASE;
2967 } else if (cur == 'l') {
2968 NEXT;
2969 type = XML_REGEXP_LETTER_LOWERCASE;
2970 } else if (cur == 't') {
2971 NEXT;
2972 type = XML_REGEXP_LETTER_TITLECASE;
2973 } else if (cur == 'm') {
2974 NEXT;
2975 type = XML_REGEXP_LETTER_MODIFIER;
2976 } else if (cur == 'o') {
2977 NEXT;
2978 type = XML_REGEXP_LETTER_OTHERS;
2979 } else {
2980 type = XML_REGEXP_LETTER;
2981 }
2982 } else if (cur == 'M') {
2983 NEXT;
2984 cur = CUR;
2985 if (cur == 'n') {
2986 NEXT;
2987 /* nonspacing */
2988 type = XML_REGEXP_MARK_NONSPACING;
2989 } else if (cur == 'c') {
2990 NEXT;
2991 /* spacing combining */
2992 type = XML_REGEXP_MARK_SPACECOMBINING;
2993 } else if (cur == 'e') {
2994 NEXT;
2995 /* enclosing */
2996 type = XML_REGEXP_MARK_ENCLOSING;
2997 } else {
2998 /* all marks */
2999 type = XML_REGEXP_MARK;
3000 }
3001 } else if (cur == 'N') {
3002 NEXT;
3003 cur = CUR;
3004 if (cur == 'd') {
3005 NEXT;
3006 /* digital */
3007 type = XML_REGEXP_NUMBER_DECIMAL;
3008 } else if (cur == 'l') {
3009 NEXT;
3010 /* letter */
3011 type = XML_REGEXP_NUMBER_LETTER;
3012 } else if (cur == 'o') {
3013 NEXT;
3014 /* other */
3015 type = XML_REGEXP_NUMBER_OTHERS;
3016 } else {
3017 /* all numbers */
3018 type = XML_REGEXP_NUMBER;
3019 }
3020 } else if (cur == 'P') {
3021 NEXT;
3022 cur = CUR;
3023 if (cur == 'c') {
3024 NEXT;
3025 /* connector */
3026 type = XML_REGEXP_PUNCT_CONNECTOR;
3027 } else if (cur == 'd') {
3028 NEXT;
3029 /* dash */
3030 type = XML_REGEXP_PUNCT_DASH;
3031 } else if (cur == 's') {
3032 NEXT;
3033 /* open */
3034 type = XML_REGEXP_PUNCT_OPEN;
3035 } else if (cur == 'e') {
3036 NEXT;
3037 /* close */
3038 type = XML_REGEXP_PUNCT_CLOSE;
3039 } else if (cur == 'i') {
3040 NEXT;
3041 /* initial quote */
3042 type = XML_REGEXP_PUNCT_INITQUOTE;
3043 } else if (cur == 'f') {
3044 NEXT;
3045 /* final quote */
3046 type = XML_REGEXP_PUNCT_FINQUOTE;
3047 } else if (cur == 'o') {
3048 NEXT;
3049 /* other */
3050 type = XML_REGEXP_PUNCT_OTHERS;
3051 } else {
3052 /* all punctuation */
3053 type = XML_REGEXP_PUNCT;
3054 }
3055 } else if (cur == 'Z') {
3056 NEXT;
3057 cur = CUR;
3058 if (cur == 's') {
3059 NEXT;
3060 /* space */
3061 type = XML_REGEXP_SEPAR_SPACE;
3062 } else if (cur == 'l') {
3063 NEXT;
3064 /* line */
3065 type = XML_REGEXP_SEPAR_LINE;
3066 } else if (cur == 'p') {
3067 NEXT;
3068 /* paragraph */
3069 type = XML_REGEXP_SEPAR_PARA;
3070 } else {
3071 /* all separators */
3072 type = XML_REGEXP_SEPAR;
3073 }
3074 } else if (cur == 'S') {
3075 NEXT;
3076 cur = CUR;
3077 if (cur == 'm') {
3078 NEXT;
3079 type = XML_REGEXP_SYMBOL_MATH;
3080 /* math */
3081 } else if (cur == 'c') {
3082 NEXT;
3083 type = XML_REGEXP_SYMBOL_CURRENCY;
3084 /* currency */
3085 } else if (cur == 'k') {
3086 NEXT;
3087 type = XML_REGEXP_SYMBOL_MODIFIER;
3088 /* modifiers */
3089 } else if (cur == 'o') {
3090 NEXT;
3091 type = XML_REGEXP_SYMBOL_OTHERS;
3092 /* other */
3093 } else {
3094 /* all symbols */
3095 type = XML_REGEXP_SYMBOL;
3096 }
3097 } else if (cur == 'C') {
3098 NEXT;
3099 cur = CUR;
3100 if (cur == 'c') {
3101 NEXT;
3102 /* control */
3103 type = XML_REGEXP_OTHER_CONTROL;
3104 } else if (cur == 'f') {
3105 NEXT;
3106 /* format */
3107 type = XML_REGEXP_OTHER_FORMAT;
3108 } else if (cur == 'o') {
3109 NEXT;
3110 /* private use */
3111 type = XML_REGEXP_OTHER_PRIVATE;
3112 } else if (cur == 'n') {
3113 NEXT;
3114 /* not assigned */
3115 type = XML_REGEXP_OTHER_NA;
3116 } else {
3117 /* all others */
3118 type = XML_REGEXP_OTHER;
3119 }
3120 } else if (cur == 'I') {
3121 const xmlChar *start;
3122 NEXT;
3123 cur = CUR;
3124 if (cur != 's') {
3125 ERROR("IsXXXX expected");
3126 return;
3127 }
3128 NEXT;
3129 start = ctxt->cur;
3130 cur = CUR;
3131 if (((cur >= 'a') && (cur <= 'z')) ||
3132 ((cur >= 'A') && (cur <= 'Z')) ||
3133 ((cur >= '0') && (cur <= '9')) ||
3134 (cur == 0x2D)) {
3135 NEXT;
3136 cur = CUR;
3137 while (((cur >= 'a') && (cur <= 'z')) ||
3138 ((cur >= 'A') && (cur <= 'Z')) ||
3139 ((cur >= '0') && (cur <= '9')) ||
3140 (cur == 0x2D)) {
3141 NEXT;
3142 cur = CUR;
3143 }
3144 }
3145 type = XML_REGEXP_BLOCK_NAME;
3146 blockName = xmlStrndup(start, ctxt->cur - start);
3147 } else {
3148 ERROR("Unknown char property");
3149 return;
3150 }
3151 if (ctxt->atom == NULL) {
3152 ctxt->atom = xmlRegNewAtom(ctxt, type);
3153 if (ctxt->atom != NULL)
3154 ctxt->atom->valuep = blockName;
3155 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
3156 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
3157 type, 0, 0, blockName);
3158 }
3159}
3160
3161/**
3162 * xmlFAParseCharClassEsc:
Daniel Veillard441bc322002-04-20 17:38:48 +00003163 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00003164 *
3165 * [23] charClassEsc ::= ( SingleCharEsc | MultiCharEsc | catEsc | complEsc )
3166 * [24] SingleCharEsc ::= '\' [nrt\|.?*+(){}#x2D#x5B#x5D#x5E]
3167 * [25] catEsc ::= '\p{' charProp '}'
3168 * [26] complEsc ::= '\P{' charProp '}'
3169 * [37] MultiCharEsc ::= '.' | ('\' [sSiIcCdDwW])
3170 */
3171static void
3172xmlFAParseCharClassEsc(xmlRegParserCtxtPtr ctxt) {
3173 int cur;
3174
3175 if (CUR == '.') {
3176 if (ctxt->atom == NULL) {
3177 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_ANYCHAR);
3178 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
3179 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
3180 XML_REGEXP_ANYCHAR, 0, 0, NULL);
3181 }
3182 NEXT;
3183 return;
3184 }
3185 if (CUR != '\\') {
3186 ERROR("Escaped sequence: expecting \\");
3187 return;
3188 }
3189 NEXT;
3190 cur = CUR;
3191 if (cur == 'p') {
3192 NEXT;
3193 if (CUR != '{') {
3194 ERROR("Expecting '{'");
3195 return;
3196 }
3197 NEXT;
3198 xmlFAParseCharProp(ctxt);
3199 if (CUR != '}') {
3200 ERROR("Expecting '}'");
3201 return;
3202 }
3203 NEXT;
3204 } else if (cur == 'P') {
3205 NEXT;
3206 if (CUR != '{') {
3207 ERROR("Expecting '{'");
3208 return;
3209 }
3210 NEXT;
3211 xmlFAParseCharProp(ctxt);
3212 ctxt->atom->neg = 1;
3213 if (CUR != '}') {
3214 ERROR("Expecting '}'");
3215 return;
3216 }
3217 NEXT;
3218 } else if ((cur == 'n') || (cur == 'r') || (cur == 't') || (cur == '\\') ||
3219 (cur == '|') || (cur == '.') || (cur == '?') || (cur == '*') ||
3220 (cur == '+') || (cur == '(') || (cur == ')') || (cur == '{') ||
3221 (cur == '}') || (cur == 0x2D) || (cur == 0x5B) || (cur == 0x5D) ||
3222 (cur == 0x5E)) {
3223 if (ctxt->atom == NULL) {
3224 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
3225 if (ctxt->atom != NULL)
3226 ctxt->atom->codepoint = cur;
3227 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
3228 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
3229 XML_REGEXP_CHARVAL, cur, cur, NULL);
3230 }
3231 NEXT;
3232 } else if ((cur == 's') || (cur == 'S') || (cur == 'i') || (cur == 'I') ||
3233 (cur == 'c') || (cur == 'C') || (cur == 'd') || (cur == 'D') ||
3234 (cur == 'w') || (cur == 'W')) {
Daniel Veillardb509f152002-04-17 16:28:10 +00003235 xmlRegAtomType type = XML_REGEXP_ANYSPACE;
Daniel Veillard4255d502002-04-16 15:50:10 +00003236
3237 switch (cur) {
3238 case 's':
3239 type = XML_REGEXP_ANYSPACE;
3240 break;
3241 case 'S':
3242 type = XML_REGEXP_NOTSPACE;
3243 break;
3244 case 'i':
3245 type = XML_REGEXP_INITNAME;
3246 break;
3247 case 'I':
3248 type = XML_REGEXP_NOTINITNAME;
3249 break;
3250 case 'c':
3251 type = XML_REGEXP_NAMECHAR;
3252 break;
3253 case 'C':
3254 type = XML_REGEXP_NOTNAMECHAR;
3255 break;
3256 case 'd':
3257 type = XML_REGEXP_DECIMAL;
3258 break;
3259 case 'D':
3260 type = XML_REGEXP_NOTDECIMAL;
3261 break;
3262 case 'w':
3263 type = XML_REGEXP_REALCHAR;
3264 break;
3265 case 'W':
3266 type = XML_REGEXP_NOTREALCHAR;
3267 break;
3268 }
3269 NEXT;
3270 if (ctxt->atom == NULL) {
3271 ctxt->atom = xmlRegNewAtom(ctxt, type);
3272 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
3273 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
3274 type, 0, 0, NULL);
3275 }
3276 }
3277}
3278
3279/**
3280 * xmlFAParseCharRef:
Daniel Veillard441bc322002-04-20 17:38:48 +00003281 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00003282 *
3283 * [19] XmlCharRef ::= ( '&#' [0-9]+ ';' ) | (' &#x' [0-9a-fA-F]+ ';' )
3284 */
3285static int
3286xmlFAParseCharRef(xmlRegParserCtxtPtr ctxt) {
3287 int ret = 0, cur;
3288
3289 if ((CUR != '&') || (NXT(1) != '#'))
3290 return(-1);
3291 NEXT;
3292 NEXT;
3293 cur = CUR;
3294 if (cur == 'x') {
3295 NEXT;
3296 cur = CUR;
3297 if (((cur >= '0') && (cur <= '9')) ||
3298 ((cur >= 'a') && (cur <= 'f')) ||
3299 ((cur >= 'A') && (cur <= 'F'))) {
3300 while (((cur >= '0') && (cur <= '9')) ||
3301 ((cur >= 'A') && (cur <= 'F'))) {
3302 if ((cur >= '0') && (cur <= '9'))
3303 ret = ret * 16 + cur - '0';
3304 else if ((cur >= 'a') && (cur <= 'f'))
3305 ret = ret * 16 + 10 + (cur - 'a');
3306 else
3307 ret = ret * 16 + 10 + (cur - 'A');
3308 NEXT;
3309 cur = CUR;
3310 }
3311 } else {
3312 ERROR("Char ref: expecting [0-9A-F]");
3313 return(-1);
3314 }
3315 } else {
3316 if ((cur >= '0') && (cur <= '9')) {
3317 while ((cur >= '0') && (cur <= '9')) {
3318 ret = ret * 10 + cur - '0';
3319 NEXT;
3320 cur = CUR;
3321 }
3322 } else {
3323 ERROR("Char ref: expecting [0-9]");
3324 return(-1);
3325 }
3326 }
3327 if (cur != ';') {
3328 ERROR("Char ref: expecting ';'");
3329 return(-1);
3330 } else {
3331 NEXT;
3332 }
3333 return(ret);
3334}
3335
3336/**
3337 * xmlFAParseCharRange:
Daniel Veillard441bc322002-04-20 17:38:48 +00003338 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00003339 *
3340 * [17] charRange ::= seRange | XmlCharRef | XmlCharIncDash
3341 * [18] seRange ::= charOrEsc '-' charOrEsc
3342 * [20] charOrEsc ::= XmlChar | SingleCharEsc
3343 * [21] XmlChar ::= [^\#x2D#x5B#x5D]
3344 * [22] XmlCharIncDash ::= [^\#x5B#x5D]
3345 */
3346static void
3347xmlFAParseCharRange(xmlRegParserCtxtPtr ctxt) {
3348 int cur;
3349 int start = -1;
3350 int end = -1;
3351
3352 if ((CUR == '&') && (NXT(1) == '#')) {
3353 end = start = xmlFAParseCharRef(ctxt);
3354 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
3355 XML_REGEXP_CHARVAL, start, end, NULL);
3356 return;
3357 }
3358 cur = CUR;
3359 if (cur == '\\') {
3360 NEXT;
3361 cur = CUR;
3362 switch (cur) {
3363 case 'n': start = 0xA; break;
3364 case 'r': start = 0xD; break;
3365 case 't': start = 0x9; break;
3366 case '\\': case '|': case '.': case '-': case '^': case '?':
3367 case '*': case '+': case '{': case '}': case '(': case ')':
3368 case '[': case ']':
3369 start = cur; break;
3370 default:
3371 ERROR("Invalid escape value");
3372 return;
3373 }
3374 end = start;
3375 } else if ((cur != 0x5B) && (cur != 0x5D)) {
3376 end = start = cur;
3377 } else {
3378 ERROR("Expecting a char range");
3379 return;
3380 }
3381 NEXT;
3382 if (start == '-') {
3383 return;
3384 }
3385 cur = CUR;
3386 if (cur != '-') {
3387 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
3388 XML_REGEXP_CHARVAL, start, end, NULL);
3389 return;
3390 }
3391 NEXT;
3392 cur = CUR;
3393 if (cur == '\\') {
3394 NEXT;
3395 cur = CUR;
3396 switch (cur) {
3397 case 'n': end = 0xA; break;
3398 case 'r': end = 0xD; break;
3399 case 't': end = 0x9; break;
3400 case '\\': case '|': case '.': case '-': case '^': case '?':
3401 case '*': case '+': case '{': case '}': case '(': case ')':
3402 case '[': case ']':
3403 end = cur; break;
3404 default:
3405 ERROR("Invalid escape value");
3406 return;
3407 }
3408 } else if ((cur != 0x5B) && (cur != 0x5D)) {
3409 end = cur;
3410 } else {
3411 ERROR("Expecting the end of a char range");
3412 return;
3413 }
3414 NEXT;
3415 /* TODO check that the values are acceptable character ranges for XML */
3416 if (end < start) {
3417 ERROR("End of range is before start of range");
3418 } else {
3419 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
3420 XML_REGEXP_CHARVAL, start, end, NULL);
3421 }
3422 return;
3423}
3424
3425/**
3426 * xmlFAParsePosCharGroup:
Daniel Veillard441bc322002-04-20 17:38:48 +00003427 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00003428 *
3429 * [14] posCharGroup ::= ( charRange | charClassEsc )+
3430 */
3431static void
3432xmlFAParsePosCharGroup(xmlRegParserCtxtPtr ctxt) {
3433 do {
3434 if ((CUR == '\\') || (CUR == '.')) {
3435 xmlFAParseCharClassEsc(ctxt);
3436 } else {
3437 xmlFAParseCharRange(ctxt);
3438 }
3439 } while ((CUR != ']') && (CUR != '^') && (CUR != '-') &&
3440 (ctxt->error == 0));
3441}
3442
3443/**
3444 * xmlFAParseCharGroup:
Daniel Veillard441bc322002-04-20 17:38:48 +00003445 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00003446 *
3447 * [13] charGroup ::= posCharGroup | negCharGroup | charClassSub
3448 * [15] negCharGroup ::= '^' posCharGroup
3449 * [16] charClassSub ::= ( posCharGroup | negCharGroup ) '-' charClassExpr
3450 * [12] charClassExpr ::= '[' charGroup ']'
3451 */
3452static void
3453xmlFAParseCharGroup(xmlRegParserCtxtPtr ctxt) {
3454 int n = ctxt->neg;
3455 while ((CUR != ']') && (ctxt->error == 0)) {
3456 if (CUR == '^') {
3457 int neg = ctxt->neg;
3458
3459 NEXT;
3460 ctxt->neg = !ctxt->neg;
3461 xmlFAParsePosCharGroup(ctxt);
3462 ctxt->neg = neg;
3463 } else if (CUR == '-') {
3464 NEXT;
3465 ctxt->neg = !ctxt->neg;
3466 if (CUR != '[') {
3467 ERROR("charClassExpr: '[' expected");
3468 break;
3469 }
3470 NEXT;
3471 xmlFAParseCharGroup(ctxt);
3472 if (CUR == ']') {
3473 NEXT;
3474 } else {
3475 ERROR("charClassExpr: ']' expected");
3476 break;
3477 }
3478 break;
3479 } else if (CUR != ']') {
3480 xmlFAParsePosCharGroup(ctxt);
3481 }
3482 }
3483 ctxt->neg = n;
3484}
3485
3486/**
3487 * xmlFAParseCharClass:
Daniel Veillard441bc322002-04-20 17:38:48 +00003488 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00003489 *
3490 * [11] charClass ::= charClassEsc | charClassExpr
3491 * [12] charClassExpr ::= '[' charGroup ']'
3492 */
3493static void
3494xmlFAParseCharClass(xmlRegParserCtxtPtr ctxt) {
3495 if (CUR == '[') {
3496 NEXT;
3497 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_RANGES);
3498 if (ctxt->atom == NULL)
3499 return;
3500 xmlFAParseCharGroup(ctxt);
3501 if (CUR == ']') {
3502 NEXT;
3503 } else {
3504 ERROR("xmlFAParseCharClass: ']' expected");
3505 }
3506 } else {
3507 xmlFAParseCharClassEsc(ctxt);
3508 }
3509}
3510
3511/**
3512 * xmlFAParseQuantExact:
Daniel Veillard441bc322002-04-20 17:38:48 +00003513 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00003514 *
3515 * [8] QuantExact ::= [0-9]+
3516 */
3517static int
3518xmlFAParseQuantExact(xmlRegParserCtxtPtr ctxt) {
3519 int ret = 0;
3520 int ok = 0;
3521
3522 while ((CUR >= '0') && (CUR <= '9')) {
3523 ret = ret * 10 + (CUR - '0');
3524 ok = 1;
3525 NEXT;
3526 }
3527 if (ok != 1) {
3528 return(-1);
3529 }
3530 return(ret);
3531}
3532
3533/**
3534 * xmlFAParseQuantifier:
Daniel Veillard441bc322002-04-20 17:38:48 +00003535 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00003536 *
3537 * [4] quantifier ::= [?*+] | ( '{' quantity '}' )
3538 * [5] quantity ::= quantRange | quantMin | QuantExact
3539 * [6] quantRange ::= QuantExact ',' QuantExact
3540 * [7] quantMin ::= QuantExact ','
3541 * [8] QuantExact ::= [0-9]+
3542 */
3543static int
3544xmlFAParseQuantifier(xmlRegParserCtxtPtr ctxt) {
3545 int cur;
3546
3547 cur = CUR;
3548 if ((cur == '?') || (cur == '*') || (cur == '+')) {
3549 if (ctxt->atom != NULL) {
3550 if (cur == '?')
3551 ctxt->atom->quant = XML_REGEXP_QUANT_OPT;
3552 else if (cur == '*')
3553 ctxt->atom->quant = XML_REGEXP_QUANT_MULT;
3554 else if (cur == '+')
3555 ctxt->atom->quant = XML_REGEXP_QUANT_PLUS;
3556 }
3557 NEXT;
3558 return(1);
3559 }
3560 if (cur == '{') {
3561 int min = 0, max = 0;
3562
3563 NEXT;
3564 cur = xmlFAParseQuantExact(ctxt);
3565 if (cur >= 0)
3566 min = cur;
3567 if (CUR == ',') {
3568 NEXT;
3569 cur = xmlFAParseQuantExact(ctxt);
3570 if (cur >= 0)
3571 max = cur;
3572 }
3573 if (CUR == '}') {
3574 NEXT;
3575 } else {
3576 ERROR("Unterminated quantifier");
3577 }
3578 if (max == 0)
3579 max = min;
3580 if (ctxt->atom != NULL) {
3581 ctxt->atom->quant = XML_REGEXP_QUANT_RANGE;
3582 ctxt->atom->min = min;
3583 ctxt->atom->max = max;
3584 }
3585 return(1);
3586 }
3587 return(0);
3588}
3589
3590/**
3591 * xmlFAParseAtom:
Daniel Veillard441bc322002-04-20 17:38:48 +00003592 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00003593 *
3594 * [9] atom ::= Char | charClass | ( '(' regExp ')' )
3595 */
3596static int
3597xmlFAParseAtom(xmlRegParserCtxtPtr ctxt) {
3598 int codepoint, len;
3599
3600 codepoint = xmlFAIsChar(ctxt);
3601 if (codepoint > 0) {
3602 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
3603 if (ctxt->atom == NULL)
3604 return(-1);
3605 codepoint = CUR_SCHAR(ctxt->cur, len);
3606 ctxt->atom->codepoint = codepoint;
3607 NEXTL(len);
3608 return(1);
3609 } else if (CUR == '|') {
3610 return(0);
3611 } else if (CUR == 0) {
3612 return(0);
3613 } else if (CUR == ')') {
3614 return(0);
3615 } else if (CUR == '(') {
3616 xmlRegStatePtr start, oldend;
3617
3618 NEXT;
3619 xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, NULL);
3620 start = ctxt->state;
3621 oldend = ctxt->end;
3622 ctxt->end = NULL;
3623 ctxt->atom = NULL;
3624 xmlFAParseRegExp(ctxt, 0);
3625 if (CUR == ')') {
3626 NEXT;
3627 } else {
3628 ERROR("xmlFAParseAtom: expecting ')'");
3629 }
3630 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_SUBREG);
3631 if (ctxt->atom == NULL)
3632 return(-1);
3633 ctxt->atom->start = start;
3634 ctxt->atom->stop = ctxt->state;
3635 ctxt->end = oldend;
3636 return(1);
3637 } else if ((CUR == '[') || (CUR == '\\') || (CUR == '.')) {
3638 xmlFAParseCharClass(ctxt);
3639 return(1);
3640 }
3641 return(0);
3642}
3643
3644/**
3645 * xmlFAParsePiece:
Daniel Veillard441bc322002-04-20 17:38:48 +00003646 * @ctxt: a regexp parser context
Daniel Veillard4255d502002-04-16 15:50:10 +00003647 *
3648 * [3] piece ::= atom quantifier?
3649 */
3650static int
3651xmlFAParsePiece(xmlRegParserCtxtPtr ctxt) {
3652 int ret;
3653
3654 ctxt->atom = NULL;
3655 ret = xmlFAParseAtom(ctxt);
3656 if (ret == 0)
3657 return(0);
3658 if (ctxt->atom == NULL) {
3659 ERROR("internal: no atom generated");
3660 }
3661 xmlFAParseQuantifier(ctxt);
3662 return(1);
3663}
3664
3665/**
3666 * xmlFAParseBranch:
Daniel Veillard441bc322002-04-20 17:38:48 +00003667 * @ctxt: a regexp parser context
3668 * @first: is taht the first
Daniel Veillard4255d502002-04-16 15:50:10 +00003669 *
3670 * [2] branch ::= piece*
3671 */
3672static void
3673xmlFAParseBranch(xmlRegParserCtxtPtr ctxt, int first) {
3674 xmlRegStatePtr previous;
3675 xmlRegAtomPtr prevatom = NULL;
3676 int ret;
3677
3678 previous = ctxt->state;
3679 ret = xmlFAParsePiece(ctxt);
3680 if (ret != 0) {
3681 if (first) {
3682 xmlFAGenerateTransitions(ctxt, previous, NULL, ctxt->atom);
3683 previous = ctxt->state;
3684 } else {
3685 prevatom = ctxt->atom;
3686 }
3687 ctxt->atom = NULL;
3688 }
3689 while ((ret != 0) && (ctxt->error == 0)) {
3690 ret = xmlFAParsePiece(ctxt);
3691 if (ret != 0) {
3692 if (first) {
3693 xmlFAGenerateTransitions(ctxt, previous, NULL, ctxt->atom);
3694 } else {
3695 xmlFAGenerateTransitions(ctxt, previous, NULL, prevatom);
3696 prevatom = ctxt->atom;
3697 }
3698 previous = ctxt->state;
3699 ctxt->atom = NULL;
3700 }
3701 }
3702 if (!first) {
3703 xmlFAGenerateTransitions(ctxt, previous, ctxt->end, prevatom);
3704 }
3705}
3706
3707/**
3708 * xmlFAParseRegExp:
Daniel Veillard441bc322002-04-20 17:38:48 +00003709 * @ctxt: a regexp parser context
3710 * @top: is that the top-level expressions ?
Daniel Veillard4255d502002-04-16 15:50:10 +00003711 *
3712 * [1] regExp ::= branch ( '|' branch )*
3713 */
3714static void
3715xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top) {
3716 xmlRegStatePtr start, end, oldend;
3717
3718 oldend = ctxt->end;
3719
3720 start = ctxt->state;
3721 xmlFAParseBranch(ctxt, (ctxt->end == NULL));
3722 if (CUR != '|') {
3723 ctxt->end = ctxt->state;
3724 return;
3725 }
3726 end = ctxt->state;
3727 while ((CUR == '|') && (ctxt->error == 0)) {
3728 NEXT;
3729 ctxt->state = start;
3730 ctxt->end = end;
3731 xmlFAParseBranch(ctxt, 0);
3732 }
3733 if (!top)
3734 ctxt->end = oldend;
3735}
3736
3737/************************************************************************
3738 * *
3739 * The basic API *
3740 * *
3741 ************************************************************************/
3742
3743/**
3744 * xmlRegexpPrint:
3745 * @output: the file for the output debug
3746 * @regexp: the compiled regexp
3747 *
3748 * Print the content of the compiled regular expression
3749 */
3750void
3751xmlRegexpPrint(FILE *output, xmlRegexpPtr regexp) {
3752 int i;
3753
3754 fprintf(output, " regexp: ");
3755 if (regexp == NULL) {
3756 fprintf(output, "NULL\n");
3757 return;
3758 }
3759 fprintf(output, "'%s' ", regexp->string);
3760 fprintf(output, "\n");
3761 fprintf(output, "%d atoms:\n", regexp->nbAtoms);
3762 for (i = 0;i < regexp->nbAtoms; i++) {
3763 fprintf(output, " %02d ", i);
3764 xmlRegPrintAtom(output, regexp->atoms[i]);
3765 }
3766 fprintf(output, "%d states:", regexp->nbStates);
3767 fprintf(output, "\n");
3768 for (i = 0;i < regexp->nbStates; i++) {
3769 xmlRegPrintState(output, regexp->states[i]);
3770 }
3771 fprintf(output, "%d counters:\n", regexp->nbCounters);
3772 for (i = 0;i < regexp->nbCounters; i++) {
3773 fprintf(output, " %d: min %d max %d\n", i, regexp->counters[i].min,
3774 regexp->counters[i].max);
3775 }
3776}
3777
3778/**
3779 * xmlRegexpCompile:
3780 * @regexp: a regular expression string
3781 *
3782 * Parses a regular expression conforming to XML Schemas Part 2 Datatype
3783 * Appendix F and build an automata suitable for testing strings against
3784 * that regular expression
3785 *
3786 * Returns the compiled expression or NULL in case of error
3787 */
3788xmlRegexpPtr
3789xmlRegexpCompile(const xmlChar *regexp) {
3790 xmlRegexpPtr ret;
3791 xmlRegParserCtxtPtr ctxt;
3792
3793 ctxt = xmlRegNewParserCtxt(regexp);
3794 if (ctxt == NULL)
3795 return(NULL);
3796
3797 /* initialize the parser */
3798 ctxt->end = NULL;
3799 ctxt->start = ctxt->state = xmlRegNewState(ctxt);
3800 xmlRegStatePush(ctxt, ctxt->start);
3801
3802 /* parse the expression building an automata */
3803 xmlFAParseRegExp(ctxt, 1);
3804 if (CUR != 0) {
3805 ERROR("xmlFAParseRegExp: extra characters");
3806 }
3807 ctxt->end = ctxt->state;
3808 ctxt->start->type = XML_REGEXP_START_STATE;
3809 ctxt->end->type = XML_REGEXP_FINAL_STATE;
3810
3811 /* remove the Epsilon except for counted transitions */
3812 xmlFAEliminateEpsilonTransitions(ctxt);
3813
3814
3815 if (ctxt->error != 0) {
3816 xmlRegFreeParserCtxt(ctxt);
3817 return(NULL);
3818 }
3819 ret = xmlRegEpxFromParse(ctxt);
3820 xmlRegFreeParserCtxt(ctxt);
3821 return(ret);
3822}
3823
3824/**
3825 * xmlRegexpExec:
3826 * @comp: the compiled regular expression
3827 * @content: the value to check against the regular expression
3828 *
3829 * Check if the regular expression generate the value
3830 *
3831 * Returns 1 if it matches, 0 if not and a negativa value in case of error
3832 */
3833int
3834xmlRegexpExec(xmlRegexpPtr comp, const xmlChar *content) {
3835 if ((comp == NULL) || (content == NULL))
3836 return(-1);
3837 return(xmlFARegExec(comp, content));
3838}
3839
3840/**
Daniel Veillard23e73572002-09-19 19:56:43 +00003841 * xmlRegexpIsDeterminist:
3842 * @comp: the compiled regular expression
3843 *
3844 * Check if the regular expression is determinist
3845 *
3846 * Returns 1 if it yes, 0 if not and a negativa value in case of error
3847 */
3848int
3849xmlRegexpIsDeterminist(xmlRegexpPtr comp) {
3850 xmlAutomataPtr am;
3851 int ret;
3852
3853 if (comp == NULL)
3854 return(-1);
3855 if (comp->determinist != -1)
3856 return(comp->determinist);
3857
3858 am = xmlNewAutomata();
Daniel Veillardbd9afb52002-09-25 22:25:35 +00003859 if (am->states != NULL) {
3860 int i;
3861
3862 for (i = 0;i < am->nbStates;i++)
3863 xmlRegFreeState(am->states[i]);
3864 xmlFree(am->states);
3865 }
Daniel Veillard23e73572002-09-19 19:56:43 +00003866 am->nbAtoms = comp->nbAtoms;
3867 am->atoms = comp->atoms;
3868 am->nbStates = comp->nbStates;
3869 am->states = comp->states;
3870 am->determinist = -1;
3871 ret = xmlFAComputesDeterminism(am);
3872 am->atoms = NULL;
3873 am->states = NULL;
3874 xmlFreeAutomata(am);
3875 return(ret);
3876}
3877
3878/**
Daniel Veillard4255d502002-04-16 15:50:10 +00003879 * xmlRegFreeRegexp:
3880 * @regexp: the regexp
3881 *
3882 * Free a regexp
3883 */
3884void
3885xmlRegFreeRegexp(xmlRegexpPtr regexp) {
3886 int i;
3887 if (regexp == NULL)
3888 return;
3889
3890 if (regexp->string != NULL)
3891 xmlFree(regexp->string);
3892 if (regexp->states != NULL) {
3893 for (i = 0;i < regexp->nbStates;i++)
3894 xmlRegFreeState(regexp->states[i]);
3895 xmlFree(regexp->states);
3896 }
3897 if (regexp->atoms != NULL) {
3898 for (i = 0;i < regexp->nbAtoms;i++)
3899 xmlRegFreeAtom(regexp->atoms[i]);
3900 xmlFree(regexp->atoms);
3901 }
3902 if (regexp->counters != NULL)
3903 xmlFree(regexp->counters);
Daniel Veillard23e73572002-09-19 19:56:43 +00003904 if (regexp->compact != NULL)
3905 xmlFree(regexp->compact);
Daniel Veillard118aed72002-09-24 14:13:13 +00003906 if (regexp->transdata != NULL)
3907 xmlFree(regexp->transdata);
Daniel Veillard23e73572002-09-19 19:56:43 +00003908 if (regexp->stringMap != NULL) {
3909 for (i = 0; i < regexp->nbstrings;i++)
3910 xmlFree(regexp->stringMap[i]);
3911 xmlFree(regexp->stringMap);
3912 }
3913
Daniel Veillard4255d502002-04-16 15:50:10 +00003914 xmlFree(regexp);
3915}
3916
3917#ifdef LIBXML_AUTOMATA_ENABLED
3918/************************************************************************
3919 * *
3920 * The Automata interface *
3921 * *
3922 ************************************************************************/
3923
3924/**
3925 * xmlNewAutomata:
3926 *
3927 * Create a new automata
3928 *
3929 * Returns the new object or NULL in case of failure
3930 */
3931xmlAutomataPtr
3932xmlNewAutomata(void) {
3933 xmlAutomataPtr ctxt;
3934
3935 ctxt = xmlRegNewParserCtxt(NULL);
3936 if (ctxt == NULL)
3937 return(NULL);
3938
3939 /* initialize the parser */
3940 ctxt->end = NULL;
3941 ctxt->start = ctxt->state = xmlRegNewState(ctxt);
3942 xmlRegStatePush(ctxt, ctxt->start);
3943
3944 return(ctxt);
3945}
3946
3947/**
3948 * xmlFreeAutomata:
3949 * @am: an automata
3950 *
3951 * Free an automata
3952 */
3953void
3954xmlFreeAutomata(xmlAutomataPtr am) {
3955 if (am == NULL)
3956 return;
3957 xmlRegFreeParserCtxt(am);
3958}
3959
3960/**
3961 * xmlAutomataGetInitState:
3962 * @am: an automata
3963 *
Daniel Veillarda9b66d02002-12-11 14:23:49 +00003964 * Initial state lookup
3965 *
Daniel Veillard4255d502002-04-16 15:50:10 +00003966 * Returns the initial state of the automata
3967 */
3968xmlAutomataStatePtr
3969xmlAutomataGetInitState(xmlAutomataPtr am) {
3970 if (am == NULL)
3971 return(NULL);
3972 return(am->start);
3973}
3974
3975/**
3976 * xmlAutomataSetFinalState:
3977 * @am: an automata
3978 * @state: a state in this automata
3979 *
3980 * Makes that state a final state
3981 *
3982 * Returns 0 or -1 in case of error
3983 */
3984int
3985xmlAutomataSetFinalState(xmlAutomataPtr am, xmlAutomataStatePtr state) {
3986 if ((am == NULL) || (state == NULL))
3987 return(-1);
3988 state->type = XML_REGEXP_FINAL_STATE;
3989 return(0);
3990}
3991
3992/**
3993 * xmlAutomataNewTransition:
3994 * @am: an automata
3995 * @from: the starting point of the transition
3996 * @to: the target point of the transition or NULL
3997 * @token: the input string associated to that transition
3998 * @data: data passed to the callback function if the transition is activated
3999 *
4000 * If @to is NULL, this create first a new target state in the automata
4001 * and then adds a transition from the @from state to the target state
4002 * activated by the value of @token
4003 *
4004 * Returns the target state or NULL in case of error
4005 */
4006xmlAutomataStatePtr
4007xmlAutomataNewTransition(xmlAutomataPtr am, xmlAutomataStatePtr from,
4008 xmlAutomataStatePtr to, const xmlChar *token,
4009 void *data) {
4010 xmlRegAtomPtr atom;
4011
4012 if ((am == NULL) || (from == NULL) || (token == NULL))
4013 return(NULL);
4014 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
4015 atom->data = data;
4016 if (atom == NULL)
4017 return(NULL);
4018 atom->valuep = xmlStrdup(token);
4019
4020 xmlFAGenerateTransitions(am, from, to, atom);
4021 if (to == NULL)
4022 return(am->state);
4023 return(to);
4024}
4025
4026/**
4027 * xmlAutomataNewCountTrans:
4028 * @am: an automata
4029 * @from: the starting point of the transition
4030 * @to: the target point of the transition or NULL
4031 * @token: the input string associated to that transition
4032 * @min: the minimum successive occurences of token
Daniel Veillarda9b66d02002-12-11 14:23:49 +00004033 * @max: the maximum successive occurences of token
4034 * @data: data associated to the transition
Daniel Veillard4255d502002-04-16 15:50:10 +00004035 *
4036 * If @to is NULL, this create first a new target state in the automata
4037 * and then adds a transition from the @from state to the target state
4038 * activated by a succession of input of value @token and whose number
4039 * is between @min and @max
4040 *
4041 * Returns the target state or NULL in case of error
4042 */
4043xmlAutomataStatePtr
4044xmlAutomataNewCountTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
4045 xmlAutomataStatePtr to, const xmlChar *token,
4046 int min, int max, void *data) {
4047 xmlRegAtomPtr atom;
4048
4049 if ((am == NULL) || (from == NULL) || (token == NULL))
4050 return(NULL);
4051 if (min < 0)
4052 return(NULL);
4053 if ((max < min) || (max < 1))
4054 return(NULL);
4055 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
4056 if (atom == NULL)
4057 return(NULL);
4058 atom->valuep = xmlStrdup(token);
4059 atom->data = data;
4060 if (min == 0)
4061 atom->min = 1;
4062 else
4063 atom->min = min;
4064 atom->max = max;
4065
4066 xmlFAGenerateTransitions(am, from, to, atom);
4067 if (to == NULL)
4068 to = am->state;
4069 if (to == NULL)
4070 return(NULL);
4071 if (min == 0)
4072 xmlFAGenerateEpsilonTransition(am, from, to);
4073 return(to);
4074}
4075
4076/**
Daniel Veillard7646b182002-04-20 06:41:40 +00004077 * xmlAutomataNewOnceTrans:
4078 * @am: an automata
4079 * @from: the starting point of the transition
4080 * @to: the target point of the transition or NULL
4081 * @token: the input string associated to that transition
4082 * @min: the minimum successive occurences of token
Daniel Veillarda9b66d02002-12-11 14:23:49 +00004083 * @max: the maximum successive occurences of token
4084 * @data: data associated to the transition
Daniel Veillard7646b182002-04-20 06:41:40 +00004085 *
4086 * If @to is NULL, this create first a new target state in the automata
4087 * and then adds a transition from the @from state to the target state
4088 * activated by a succession of input of value @token and whose number
4089 * is between @min and @max, moreover that transistion can only be crossed
4090 * once.
4091 *
4092 * Returns the target state or NULL in case of error
4093 */
4094xmlAutomataStatePtr
4095xmlAutomataNewOnceTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
4096 xmlAutomataStatePtr to, const xmlChar *token,
4097 int min, int max, void *data) {
4098 xmlRegAtomPtr atom;
4099 int counter;
4100
4101 if ((am == NULL) || (from == NULL) || (token == NULL))
4102 return(NULL);
4103 if (min < 1)
4104 return(NULL);
4105 if ((max < min) || (max < 1))
4106 return(NULL);
4107 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
4108 if (atom == NULL)
4109 return(NULL);
4110 atom->valuep = xmlStrdup(token);
4111 atom->data = data;
4112 atom->quant = XML_REGEXP_QUANT_ONCEONLY;
4113 if (min == 0)
4114 atom->min = 1;
4115 else
4116 atom->min = min;
4117 atom->max = max;
4118 /*
4119 * associate a counter to the transition.
4120 */
4121 counter = xmlRegGetCounter(am);
4122 am->counters[counter].min = 1;
4123 am->counters[counter].max = 1;
4124
4125 /* xmlFAGenerateTransitions(am, from, to, atom); */
4126 if (to == NULL) {
4127 to = xmlRegNewState(am);
4128 xmlRegStatePush(am, to);
4129 }
4130 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
4131 xmlRegAtomPush(am, atom);
4132 am->state = to;
4133 if (to == NULL)
4134 to = am->state;
4135 if (to == NULL)
4136 return(NULL);
4137 return(to);
4138}
4139
4140/**
Daniel Veillard4255d502002-04-16 15:50:10 +00004141 * xmlAutomataNewState:
4142 * @am: an automata
4143 *
4144 * Create a new disconnected state in the automata
4145 *
4146 * Returns the new state or NULL in case of error
4147 */
4148xmlAutomataStatePtr
4149xmlAutomataNewState(xmlAutomataPtr am) {
4150 xmlAutomataStatePtr to;
4151
4152 if (am == NULL)
4153 return(NULL);
4154 to = xmlRegNewState(am);
4155 xmlRegStatePush(am, to);
4156 return(to);
4157}
4158
4159/**
Daniel Veillarda9b66d02002-12-11 14:23:49 +00004160 * xmlAutomataNewEpsilon:
Daniel Veillard4255d502002-04-16 15:50:10 +00004161 * @am: an automata
4162 * @from: the starting point of the transition
4163 * @to: the target point of the transition or NULL
4164 *
4165 * If @to is NULL, this create first a new target state in the automata
4166 * and then adds a an epsilon transition from the @from state to the
4167 * target state
4168 *
4169 * Returns the target state or NULL in case of error
4170 */
4171xmlAutomataStatePtr
4172xmlAutomataNewEpsilon(xmlAutomataPtr am, xmlAutomataStatePtr from,
4173 xmlAutomataStatePtr to) {
4174 if ((am == NULL) || (from == NULL))
4175 return(NULL);
4176 xmlFAGenerateEpsilonTransition(am, from, to);
4177 if (to == NULL)
4178 return(am->state);
4179 return(to);
4180}
4181
Daniel Veillardb509f152002-04-17 16:28:10 +00004182/**
Daniel Veillard7646b182002-04-20 06:41:40 +00004183 * xmlAutomataNewAllTrans:
4184 * @am: an automata
4185 * @from: the starting point of the transition
4186 * @to: the target point of the transition or NULL
Daniel Veillarda9b66d02002-12-11 14:23:49 +00004187 * @lax: allow to transition if not all all transitions have been activated
Daniel Veillard7646b182002-04-20 06:41:40 +00004188 *
4189 * If @to is NULL, this create first a new target state in the automata
4190 * and then adds a an ALL transition from the @from state to the
4191 * target state. That transition is an epsilon transition allowed only when
4192 * all transitions from the @from node have been activated.
4193 *
4194 * Returns the target state or NULL in case of error
4195 */
4196xmlAutomataStatePtr
4197xmlAutomataNewAllTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
Daniel Veillard441bc322002-04-20 17:38:48 +00004198 xmlAutomataStatePtr to, int lax) {
Daniel Veillard7646b182002-04-20 06:41:40 +00004199 if ((am == NULL) || (from == NULL))
4200 return(NULL);
Daniel Veillard441bc322002-04-20 17:38:48 +00004201 xmlFAGenerateAllTransition(am, from, to, lax);
Daniel Veillard7646b182002-04-20 06:41:40 +00004202 if (to == NULL)
4203 return(am->state);
4204 return(to);
4205}
4206
4207/**
Daniel Veillardb509f152002-04-17 16:28:10 +00004208 * xmlAutomataNewCounter:
4209 * @am: an automata
4210 * @min: the minimal value on the counter
4211 * @max: the maximal value on the counter
4212 *
4213 * Create a new counter
4214 *
4215 * Returns the counter number or -1 in case of error
4216 */
4217int
4218xmlAutomataNewCounter(xmlAutomataPtr am, int min, int max) {
4219 int ret;
4220
4221 if (am == NULL)
4222 return(-1);
4223
4224 ret = xmlRegGetCounter(am);
4225 if (ret < 0)
4226 return(-1);
4227 am->counters[ret].min = min;
4228 am->counters[ret].max = max;
4229 return(ret);
4230}
4231
4232/**
4233 * xmlAutomataNewCountedTrans:
4234 * @am: an automata
4235 * @from: the starting point of the transition
4236 * @to: the target point of the transition or NULL
4237 * @counter: the counter associated to that transition
4238 *
4239 * If @to is NULL, this create first a new target state in the automata
4240 * and then adds an epsilon transition from the @from state to the target state
4241 * which will increment the counter provided
4242 *
4243 * Returns the target state or NULL in case of error
4244 */
4245xmlAutomataStatePtr
4246xmlAutomataNewCountedTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
4247 xmlAutomataStatePtr to, int counter) {
4248 if ((am == NULL) || (from == NULL) || (counter < 0))
4249 return(NULL);
4250 xmlFAGenerateCountedEpsilonTransition(am, from, to, counter);
4251 if (to == NULL)
4252 return(am->state);
4253 return(to);
4254}
4255
4256/**
4257 * xmlAutomataNewCounterTrans:
4258 * @am: an automata
4259 * @from: the starting point of the transition
4260 * @to: the target point of the transition or NULL
4261 * @counter: the counter associated to that transition
4262 *
4263 * If @to is NULL, this create first a new target state in the automata
4264 * and then adds an epsilon transition from the @from state to the target state
4265 * which will be allowed only if the counter is within the right range.
4266 *
4267 * Returns the target state or NULL in case of error
4268 */
4269xmlAutomataStatePtr
4270xmlAutomataNewCounterTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
4271 xmlAutomataStatePtr to, int counter) {
4272 if ((am == NULL) || (from == NULL) || (counter < 0))
4273 return(NULL);
4274 xmlFAGenerateCountedTransition(am, from, to, counter);
4275 if (to == NULL)
4276 return(am->state);
4277 return(to);
4278}
Daniel Veillard4255d502002-04-16 15:50:10 +00004279
4280/**
4281 * xmlAutomataCompile:
4282 * @am: an automata
4283 *
4284 * Compile the automata into a Reg Exp ready for being executed.
4285 * The automata should be free after this point.
4286 *
4287 * Returns the compiled regexp or NULL in case of error
4288 */
4289xmlRegexpPtr
4290xmlAutomataCompile(xmlAutomataPtr am) {
4291 xmlRegexpPtr ret;
4292
4293 xmlFAEliminateEpsilonTransitions(am);
Daniel Veillard23e73572002-09-19 19:56:43 +00004294 /* xmlFAComputesDeterminism(am); */
Daniel Veillard4255d502002-04-16 15:50:10 +00004295 ret = xmlRegEpxFromParse(am);
4296
4297 return(ret);
4298}
Daniel Veillarde19fc232002-04-22 16:01:24 +00004299
4300/**
4301 * xmlAutomataIsDeterminist:
4302 * @am: an automata
4303 *
4304 * Checks if an automata is determinist.
4305 *
4306 * Returns 1 if true, 0 if not, and -1 in case of error
4307 */
4308int
4309xmlAutomataIsDeterminist(xmlAutomataPtr am) {
4310 int ret;
4311
4312 if (am == NULL)
4313 return(-1);
4314
4315 ret = xmlFAComputesDeterminism(am);
4316 return(ret);
4317}
Daniel Veillard4255d502002-04-16 15:50:10 +00004318#endif /* LIBXML_AUTOMATA_ENABLED */
4319#endif /* LIBXML_REGEXP_ENABLED */