VirtualBox

source: vbox/trunk/src/libs/libxml2-2.6.31/xmlregexp.c@ 53428

Last change on this file since 53428 was 39915, checked in by vboxsync, 13 years ago

libxml-2.6.31 unmodified
  • Property svn:eol-style set to native
File size: 213.5 KB
Line 
1/*
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 these include:
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 <[email protected]>
15 */
16
17#define IN_LIBXML
18#include "libxml.h"
19
20#ifdef LIBXML_REGEXP_ENABLED
21
22/* #define DEBUG_ERR */
23
24#include <stdio.h>
25#include <string.h>
26#ifdef HAVE_LIMITS_H
27#include <limits.h>
28#endif
29
30#include <libxml/tree.h>
31#include <libxml/parserInternals.h>
32#include <libxml/xmlregexp.h>
33#include <libxml/xmlautomata.h>
34#include <libxml/xmlunicode.h>
35
36#ifndef INT_MAX
37#define INT_MAX 123456789 /* easy to flag and big enough for our needs */
38#endif
39
40/* #define DEBUG_REGEXP_GRAPH */
41/* #define DEBUG_REGEXP_EXEC */
42/* #define DEBUG_PUSH */
43/* #define DEBUG_COMPACTION */
44
45#define MAX_PUSH 10000000
46
47#define ERROR(str) \
48 ctxt->error = XML_REGEXP_COMPILE_ERROR; \
49 xmlRegexpErrCompile(ctxt, str);
50#define NEXT ctxt->cur++
51#define CUR (*(ctxt->cur))
52#define NXT(index) (ctxt->cur[index])
53
54#define CUR_SCHAR(s, l) xmlStringCurrentChar(NULL, s, &l)
55#define NEXTL(l) ctxt->cur += l;
56#define XML_REG_STRING_SEPARATOR '|'
57/*
58 * Need PREV to check on a '-' within a Character Group. May only be used
59 * when it's guaranteed that cur is not at the beginning of ctxt->string!
60 */
61#define PREV (ctxt->cur[-1])
62
63/**
64 * TODO:
65 *
66 * macro to flag unimplemented blocks
67 */
68#define TODO \
69 xmlGenericError(xmlGenericErrorContext, \
70 "Unimplemented block at %s:%d\n", \
71 __FILE__, __LINE__);
72
73/************************************************************************
74 * *
75 * Datatypes and structures *
76 * *
77 ************************************************************************/
78
79/*
80 * Note: the order of the enums below is significant, do not shuffle
81 */
82typedef enum {
83 XML_REGEXP_EPSILON = 1,
84 XML_REGEXP_CHARVAL,
85 XML_REGEXP_RANGES,
86 XML_REGEXP_SUBREG, /* used for () sub regexps */
87 XML_REGEXP_STRING,
88 XML_REGEXP_ANYCHAR, /* . */
89 XML_REGEXP_ANYSPACE, /* \s */
90 XML_REGEXP_NOTSPACE, /* \S */
91 XML_REGEXP_INITNAME, /* \l */
92 XML_REGEXP_NOTINITNAME, /* \L */
93 XML_REGEXP_NAMECHAR, /* \c */
94 XML_REGEXP_NOTNAMECHAR, /* \C */
95 XML_REGEXP_DECIMAL, /* \d */
96 XML_REGEXP_NOTDECIMAL, /* \D */
97 XML_REGEXP_REALCHAR, /* \w */
98 XML_REGEXP_NOTREALCHAR, /* \W */
99 XML_REGEXP_LETTER = 100,
100 XML_REGEXP_LETTER_UPPERCASE,
101 XML_REGEXP_LETTER_LOWERCASE,
102 XML_REGEXP_LETTER_TITLECASE,
103 XML_REGEXP_LETTER_MODIFIER,
104 XML_REGEXP_LETTER_OTHERS,
105 XML_REGEXP_MARK,
106 XML_REGEXP_MARK_NONSPACING,
107 XML_REGEXP_MARK_SPACECOMBINING,
108 XML_REGEXP_MARK_ENCLOSING,
109 XML_REGEXP_NUMBER,
110 XML_REGEXP_NUMBER_DECIMAL,
111 XML_REGEXP_NUMBER_LETTER,
112 XML_REGEXP_NUMBER_OTHERS,
113 XML_REGEXP_PUNCT,
114 XML_REGEXP_PUNCT_CONNECTOR,
115 XML_REGEXP_PUNCT_DASH,
116 XML_REGEXP_PUNCT_OPEN,
117 XML_REGEXP_PUNCT_CLOSE,
118 XML_REGEXP_PUNCT_INITQUOTE,
119 XML_REGEXP_PUNCT_FINQUOTE,
120 XML_REGEXP_PUNCT_OTHERS,
121 XML_REGEXP_SEPAR,
122 XML_REGEXP_SEPAR_SPACE,
123 XML_REGEXP_SEPAR_LINE,
124 XML_REGEXP_SEPAR_PARA,
125 XML_REGEXP_SYMBOL,
126 XML_REGEXP_SYMBOL_MATH,
127 XML_REGEXP_SYMBOL_CURRENCY,
128 XML_REGEXP_SYMBOL_MODIFIER,
129 XML_REGEXP_SYMBOL_OTHERS,
130 XML_REGEXP_OTHER,
131 XML_REGEXP_OTHER_CONTROL,
132 XML_REGEXP_OTHER_FORMAT,
133 XML_REGEXP_OTHER_PRIVATE,
134 XML_REGEXP_OTHER_NA,
135 XML_REGEXP_BLOCK_NAME
136} xmlRegAtomType;
137
138typedef enum {
139 XML_REGEXP_QUANT_EPSILON = 1,
140 XML_REGEXP_QUANT_ONCE,
141 XML_REGEXP_QUANT_OPT,
142 XML_REGEXP_QUANT_MULT,
143 XML_REGEXP_QUANT_PLUS,
144 XML_REGEXP_QUANT_ONCEONLY,
145 XML_REGEXP_QUANT_ALL,
146 XML_REGEXP_QUANT_RANGE
147} xmlRegQuantType;
148
149typedef enum {
150 XML_REGEXP_START_STATE = 1,
151 XML_REGEXP_FINAL_STATE,
152 XML_REGEXP_TRANS_STATE,
153 XML_REGEXP_SINK_STATE,
154 XML_REGEXP_UNREACH_STATE
155} xmlRegStateType;
156
157typedef enum {
158 XML_REGEXP_MARK_NORMAL = 0,
159 XML_REGEXP_MARK_START,
160 XML_REGEXP_MARK_VISITED
161} xmlRegMarkedType;
162
163typedef struct _xmlRegRange xmlRegRange;
164typedef xmlRegRange *xmlRegRangePtr;
165
166struct _xmlRegRange {
167 int neg; /* 0 normal, 1 not, 2 exclude */
168 xmlRegAtomType type;
169 int start;
170 int end;
171 xmlChar *blockName;
172};
173
174typedef struct _xmlRegAtom xmlRegAtom;
175typedef xmlRegAtom *xmlRegAtomPtr;
176
177typedef struct _xmlAutomataState xmlRegState;
178typedef xmlRegState *xmlRegStatePtr;
179
180struct _xmlRegAtom {
181 int no;
182 xmlRegAtomType type;
183 xmlRegQuantType quant;
184 int min;
185 int max;
186
187 void *valuep;
188 void *valuep2;
189 int neg;
190 int codepoint;
191 xmlRegStatePtr start;
192 xmlRegStatePtr start0;
193 xmlRegStatePtr stop;
194 int maxRanges;
195 int nbRanges;
196 xmlRegRangePtr *ranges;
197 void *data;
198};
199
200typedef struct _xmlRegCounter xmlRegCounter;
201typedef xmlRegCounter *xmlRegCounterPtr;
202
203struct _xmlRegCounter {
204 int min;
205 int max;
206};
207
208typedef struct _xmlRegTrans xmlRegTrans;
209typedef xmlRegTrans *xmlRegTransPtr;
210
211struct _xmlRegTrans {
212 xmlRegAtomPtr atom;
213 int to;
214 int counter;
215 int count;
216 int nd;
217};
218
219struct _xmlAutomataState {
220 xmlRegStateType type;
221 xmlRegMarkedType mark;
222 xmlRegMarkedType reached;
223 int no;
224 int maxTrans;
225 int nbTrans;
226 xmlRegTrans *trans;
227 /* knowing states ponting to us can speed things up */
228 int maxTransTo;
229 int nbTransTo;
230 int *transTo;
231};
232
233typedef struct _xmlAutomata xmlRegParserCtxt;
234typedef xmlRegParserCtxt *xmlRegParserCtxtPtr;
235
236struct _xmlAutomata {
237 xmlChar *string;
238 xmlChar *cur;
239
240 int error;
241 int neg;
242
243 xmlRegStatePtr start;
244 xmlRegStatePtr end;
245 xmlRegStatePtr state;
246
247 xmlRegAtomPtr atom;
248
249 int maxAtoms;
250 int nbAtoms;
251 xmlRegAtomPtr *atoms;
252
253 int maxStates;
254 int nbStates;
255 xmlRegStatePtr *states;
256
257 int maxCounters;
258 int nbCounters;
259 xmlRegCounter *counters;
260
261 int determinist;
262 int negs;
263};
264
265struct _xmlRegexp {
266 xmlChar *string;
267 int nbStates;
268 xmlRegStatePtr *states;
269 int nbAtoms;
270 xmlRegAtomPtr *atoms;
271 int nbCounters;
272 xmlRegCounter *counters;
273 int determinist;
274 /*
275 * That's the compact form for determinists automatas
276 */
277 int nbstates;
278 int *compact;
279 void **transdata;
280 int nbstrings;
281 xmlChar **stringMap;
282};
283
284typedef struct _xmlRegExecRollback xmlRegExecRollback;
285typedef xmlRegExecRollback *xmlRegExecRollbackPtr;
286
287struct _xmlRegExecRollback {
288 xmlRegStatePtr state;/* the current state */
289 int index; /* the index in the input stack */
290 int nextbranch; /* the next transition to explore in that state */
291 int *counts; /* save the automata state if it has some */
292};
293
294typedef struct _xmlRegInputToken xmlRegInputToken;
295typedef xmlRegInputToken *xmlRegInputTokenPtr;
296
297struct _xmlRegInputToken {
298 xmlChar *value;
299 void *data;
300};
301
302struct _xmlRegExecCtxt {
303 int status; /* execution status != 0 indicate an error */
304 int determinist; /* did we find an indeterministic behaviour */
305 xmlRegexpPtr comp; /* the compiled regexp */
306 xmlRegExecCallbacks callback;
307 void *data;
308
309 xmlRegStatePtr state;/* the current state */
310 int transno; /* the current transition on that state */
311 int transcount; /* the number of chars in char counted transitions */
312
313 /*
314 * A stack of rollback states
315 */
316 int maxRollbacks;
317 int nbRollbacks;
318 xmlRegExecRollback *rollbacks;
319
320 /*
321 * The state of the automata if any
322 */
323 int *counts;
324
325 /*
326 * The input stack
327 */
328 int inputStackMax;
329 int inputStackNr;
330 int index;
331 int *charStack;
332 const xmlChar *inputString; /* when operating on characters */
333 xmlRegInputTokenPtr inputStack;/* when operating on strings */
334
335 /*
336 * error handling
337 */
338 int errStateNo; /* the error state number */
339 xmlRegStatePtr errState; /* the error state */
340 xmlChar *errString; /* the string raising the error */
341 int *errCounts; /* counters at the error state */
342 int nbPush;
343};
344
345#define REGEXP_ALL_COUNTER 0x123456
346#define REGEXP_ALL_LAX_COUNTER 0x123457
347
348static void xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top);
349static void xmlRegFreeState(xmlRegStatePtr state);
350static void xmlRegFreeAtom(xmlRegAtomPtr atom);
351static int xmlRegStrEqualWildcard(const xmlChar *expStr, const xmlChar *valStr);
352static int xmlRegCheckCharacter(xmlRegAtomPtr atom, int codepoint);
353static int xmlRegCheckCharacterRange(xmlRegAtomType type, int codepoint,
354 int neg, int start, int end, const xmlChar *blockName);
355
356/************************************************************************
357 * *
358 * Regexp memory error handler *
359 * *
360 ************************************************************************/
361/**
362 * xmlRegexpErrMemory:
363 * @extra: extra information
364 *
365 * Handle an out of memory condition
366 */
367static void
368xmlRegexpErrMemory(xmlRegParserCtxtPtr ctxt, const char *extra)
369{
370 const char *regexp = NULL;
371 if (ctxt != NULL) {
372 regexp = (const char *) ctxt->string;
373 ctxt->error = XML_ERR_NO_MEMORY;
374 }
375 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_REGEXP,
376 XML_ERR_NO_MEMORY, XML_ERR_FATAL, NULL, 0, extra,
377 regexp, NULL, 0, 0,
378 "Memory allocation failed : %s\n", extra);
379}
380
381/**
382 * xmlRegexpErrCompile:
383 * @extra: extra information
384 *
385 * Handle a compilation failure
386 */
387static void
388xmlRegexpErrCompile(xmlRegParserCtxtPtr ctxt, const char *extra)
389{
390 const char *regexp = NULL;
391 int idx = 0;
392
393 if (ctxt != NULL) {
394 regexp = (const char *) ctxt->string;
395 idx = ctxt->cur - ctxt->string;
396 ctxt->error = XML_REGEXP_COMPILE_ERROR;
397 }
398 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_REGEXP,
399 XML_REGEXP_COMPILE_ERROR, XML_ERR_FATAL, NULL, 0, extra,
400 regexp, NULL, idx, 0,
401 "failed to compile: %s\n", extra);
402}
403
404/************************************************************************
405 * *
406 * Allocation/Deallocation *
407 * *
408 ************************************************************************/
409
410static int xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt);
411/**
412 * xmlRegEpxFromParse:
413 * @ctxt: the parser context used to build it
414 *
415 * Allocate a new regexp and fill it with the result from the parser
416 *
417 * Returns the new regexp or NULL in case of error
418 */
419static xmlRegexpPtr
420xmlRegEpxFromParse(xmlRegParserCtxtPtr ctxt) {
421 xmlRegexpPtr ret;
422
423 ret = (xmlRegexpPtr) xmlMalloc(sizeof(xmlRegexp));
424 if (ret == NULL) {
425 xmlRegexpErrMemory(ctxt, "compiling regexp");
426 return(NULL);
427 }
428 memset(ret, 0, sizeof(xmlRegexp));
429 ret->string = ctxt->string;
430 ret->nbStates = ctxt->nbStates;
431 ret->states = ctxt->states;
432 ret->nbAtoms = ctxt->nbAtoms;
433 ret->atoms = ctxt->atoms;
434 ret->nbCounters = ctxt->nbCounters;
435 ret->counters = ctxt->counters;
436 ret->determinist = ctxt->determinist;
437 if (ret->determinist == -1) {
438 xmlRegexpIsDeterminist(ret);
439 }
440
441 if ((ret->determinist != 0) &&
442 (ret->nbCounters == 0) &&
443 (ctxt->negs == 0) &&
444 (ret->atoms != NULL) &&
445 (ret->atoms[0] != NULL) &&
446 (ret->atoms[0]->type == XML_REGEXP_STRING)) {
447 int i, j, nbstates = 0, nbatoms = 0;
448 int *stateRemap;
449 int *stringRemap;
450 int *transitions;
451 void **transdata;
452 xmlChar **stringMap;
453 xmlChar *value;
454
455 /*
456 * Switch to a compact representation
457 * 1/ counting the effective number of states left
458 * 2/ counting the unique number of atoms, and check that
459 * they are all of the string type
460 * 3/ build a table state x atom for the transitions
461 */
462
463 stateRemap = xmlMalloc(ret->nbStates * sizeof(int));
464 if (stateRemap == NULL) {
465 xmlRegexpErrMemory(ctxt, "compiling regexp");
466 xmlFree(ret);
467 return(NULL);
468 }
469 for (i = 0;i < ret->nbStates;i++) {
470 if (ret->states[i] != NULL) {
471 stateRemap[i] = nbstates;
472 nbstates++;
473 } else {
474 stateRemap[i] = -1;
475 }
476 }
477#ifdef DEBUG_COMPACTION
478 printf("Final: %d states\n", nbstates);
479#endif
480 stringMap = xmlMalloc(ret->nbAtoms * sizeof(char *));
481 if (stringMap == NULL) {
482 xmlRegexpErrMemory(ctxt, "compiling regexp");
483 xmlFree(stateRemap);
484 xmlFree(ret);
485 return(NULL);
486 }
487 stringRemap = xmlMalloc(ret->nbAtoms * sizeof(int));
488 if (stringRemap == NULL) {
489 xmlRegexpErrMemory(ctxt, "compiling regexp");
490 xmlFree(stringMap);
491 xmlFree(stateRemap);
492 xmlFree(ret);
493 return(NULL);
494 }
495 for (i = 0;i < ret->nbAtoms;i++) {
496 if ((ret->atoms[i]->type == XML_REGEXP_STRING) &&
497 (ret->atoms[i]->quant == XML_REGEXP_QUANT_ONCE)) {
498 value = ret->atoms[i]->valuep;
499 for (j = 0;j < nbatoms;j++) {
500 if (xmlStrEqual(stringMap[j], value)) {
501 stringRemap[i] = j;
502 break;
503 }
504 }
505 if (j >= nbatoms) {
506 stringRemap[i] = nbatoms;
507 stringMap[nbatoms] = xmlStrdup(value);
508 if (stringMap[nbatoms] == NULL) {
509 for (i = 0;i < nbatoms;i++)
510 xmlFree(stringMap[i]);
511 xmlFree(stringRemap);
512 xmlFree(stringMap);
513 xmlFree(stateRemap);
514 xmlFree(ret);
515 return(NULL);
516 }
517 nbatoms++;
518 }
519 } else {
520 xmlFree(stateRemap);
521 xmlFree(stringRemap);
522 for (i = 0;i < nbatoms;i++)
523 xmlFree(stringMap[i]);
524 xmlFree(stringMap);
525 xmlFree(ret);
526 return(NULL);
527 }
528 }
529#ifdef DEBUG_COMPACTION
530 printf("Final: %d atoms\n", nbatoms);
531#endif
532 transitions = (int *) xmlMalloc((nbstates + 1) *
533 (nbatoms + 1) * sizeof(int));
534 if (transitions == NULL) {
535 xmlFree(stateRemap);
536 xmlFree(stringRemap);
537 xmlFree(stringMap);
538 xmlFree(ret);
539 return(NULL);
540 }
541 memset(transitions, 0, (nbstates + 1) * (nbatoms + 1) * sizeof(int));
542
543 /*
544 * Allocate the transition table. The first entry for each
545 * state corresponds to the state type.
546 */
547 transdata = NULL;
548
549 for (i = 0;i < ret->nbStates;i++) {
550 int stateno, atomno, targetno, prev;
551 xmlRegStatePtr state;
552 xmlRegTransPtr trans;
553
554 stateno = stateRemap[i];
555 if (stateno == -1)
556 continue;
557 state = ret->states[i];
558
559 transitions[stateno * (nbatoms + 1)] = state->type;
560
561 for (j = 0;j < state->nbTrans;j++) {
562 trans = &(state->trans[j]);
563 if ((trans->to == -1) || (trans->atom == NULL))
564 continue;
565 atomno = stringRemap[trans->atom->no];
566 if ((trans->atom->data != NULL) && (transdata == NULL)) {
567 transdata = (void **) xmlMalloc(nbstates * nbatoms *
568 sizeof(void *));
569 if (transdata != NULL)
570 memset(transdata, 0,
571 nbstates * nbatoms * sizeof(void *));
572 else {
573 xmlRegexpErrMemory(ctxt, "compiling regexp");
574 break;
575 }
576 }
577 targetno = stateRemap[trans->to];
578 /*
579 * if the same atom can generate transitions to 2 different
580 * states then it means the automata is not determinist and
581 * the compact form can't be used !
582 */
583 prev = transitions[stateno * (nbatoms + 1) + atomno + 1];
584 if (prev != 0) {
585 if (prev != targetno + 1) {
586 ret->determinist = 0;
587#ifdef DEBUG_COMPACTION
588 printf("Indet: state %d trans %d, atom %d to %d : %d to %d\n",
589 i, j, trans->atom->no, trans->to, atomno, targetno);
590 printf(" previous to is %d\n", prev);
591#endif
592 if (transdata != NULL)
593 xmlFree(transdata);
594 xmlFree(transitions);
595 xmlFree(stateRemap);
596 xmlFree(stringRemap);
597 for (i = 0;i < nbatoms;i++)
598 xmlFree(stringMap[i]);
599 xmlFree(stringMap);
600 goto not_determ;
601 }
602 } else {
603#if 0
604 printf("State %d trans %d: atom %d to %d : %d to %d\n",
605 i, j, trans->atom->no, trans->to, atomno, targetno);
606#endif
607 transitions[stateno * (nbatoms + 1) + atomno + 1] =
608 targetno + 1; /* to avoid 0 */
609 if (transdata != NULL)
610 transdata[stateno * nbatoms + atomno] =
611 trans->atom->data;
612 }
613 }
614 }
615 ret->determinist = 1;
616#ifdef DEBUG_COMPACTION
617 /*
618 * Debug
619 */
620 for (i = 0;i < nbstates;i++) {
621 for (j = 0;j < nbatoms + 1;j++) {
622 printf("%02d ", transitions[i * (nbatoms + 1) + j]);
623 }
624 printf("\n");
625 }
626 printf("\n");
627#endif
628 /*
629 * Cleanup of the old data
630 */
631 if (ret->states != NULL) {
632 for (i = 0;i < ret->nbStates;i++)
633 xmlRegFreeState(ret->states[i]);
634 xmlFree(ret->states);
635 }
636 ret->states = NULL;
637 ret->nbStates = 0;
638 if (ret->atoms != NULL) {
639 for (i = 0;i < ret->nbAtoms;i++)
640 xmlRegFreeAtom(ret->atoms[i]);
641 xmlFree(ret->atoms);
642 }
643 ret->atoms = NULL;
644 ret->nbAtoms = 0;
645
646 ret->compact = transitions;
647 ret->transdata = transdata;
648 ret->stringMap = stringMap;
649 ret->nbstrings = nbatoms;
650 ret->nbstates = nbstates;
651 xmlFree(stateRemap);
652 xmlFree(stringRemap);
653 }
654not_determ:
655 ctxt->string = NULL;
656 ctxt->nbStates = 0;
657 ctxt->states = NULL;
658 ctxt->nbAtoms = 0;
659 ctxt->atoms = NULL;
660 ctxt->nbCounters = 0;
661 ctxt->counters = NULL;
662 return(ret);
663}
664
665/**
666 * xmlRegNewParserCtxt:
667 * @string: the string to parse
668 *
669 * Allocate a new regexp parser context
670 *
671 * Returns the new context or NULL in case of error
672 */
673static xmlRegParserCtxtPtr
674xmlRegNewParserCtxt(const xmlChar *string) {
675 xmlRegParserCtxtPtr ret;
676
677 ret = (xmlRegParserCtxtPtr) xmlMalloc(sizeof(xmlRegParserCtxt));
678 if (ret == NULL)
679 return(NULL);
680 memset(ret, 0, sizeof(xmlRegParserCtxt));
681 if (string != NULL)
682 ret->string = xmlStrdup(string);
683 ret->cur = ret->string;
684 ret->neg = 0;
685 ret->negs = 0;
686 ret->error = 0;
687 ret->determinist = -1;
688 return(ret);
689}
690
691/**
692 * xmlRegNewRange:
693 * @ctxt: the regexp parser context
694 * @neg: is that negative
695 * @type: the type of range
696 * @start: the start codepoint
697 * @end: the end codepoint
698 *
699 * Allocate a new regexp range
700 *
701 * Returns the new range or NULL in case of error
702 */
703static xmlRegRangePtr
704xmlRegNewRange(xmlRegParserCtxtPtr ctxt,
705 int neg, xmlRegAtomType type, int start, int end) {
706 xmlRegRangePtr ret;
707
708 ret = (xmlRegRangePtr) xmlMalloc(sizeof(xmlRegRange));
709 if (ret == NULL) {
710 xmlRegexpErrMemory(ctxt, "allocating range");
711 return(NULL);
712 }
713 ret->neg = neg;
714 ret->type = type;
715 ret->start = start;
716 ret->end = end;
717 return(ret);
718}
719
720/**
721 * xmlRegFreeRange:
722 * @range: the regexp range
723 *
724 * Free a regexp range
725 */
726static void
727xmlRegFreeRange(xmlRegRangePtr range) {
728 if (range == NULL)
729 return;
730
731 if (range->blockName != NULL)
732 xmlFree(range->blockName);
733 xmlFree(range);
734}
735
736/**
737 * xmlRegCopyRange:
738 * @range: the regexp range
739 *
740 * Copy a regexp range
741 *
742 * Returns the new copy or NULL in case of error.
743 */
744static xmlRegRangePtr
745xmlRegCopyRange(xmlRegParserCtxtPtr ctxt, xmlRegRangePtr range) {
746 xmlRegRangePtr ret;
747
748 if (range == NULL)
749 return(NULL);
750
751 ret = xmlRegNewRange(ctxt, range->neg, range->type, range->start,
752 range->end);
753 if (ret == NULL)
754 return(NULL);
755 if (range->blockName != NULL) {
756 ret->blockName = xmlStrdup(range->blockName);
757 if (ret->blockName == NULL) {
758 xmlRegexpErrMemory(ctxt, "allocating range");
759 xmlRegFreeRange(ret);
760 return(NULL);
761 }
762 }
763 return(ret);
764}
765
766/**
767 * xmlRegNewAtom:
768 * @ctxt: the regexp parser context
769 * @type: the type of atom
770 *
771 * Allocate a new atom
772 *
773 * Returns the new atom or NULL in case of error
774 */
775static xmlRegAtomPtr
776xmlRegNewAtom(xmlRegParserCtxtPtr ctxt, xmlRegAtomType type) {
777 xmlRegAtomPtr ret;
778
779 ret = (xmlRegAtomPtr) xmlMalloc(sizeof(xmlRegAtom));
780 if (ret == NULL) {
781 xmlRegexpErrMemory(ctxt, "allocating atom");
782 return(NULL);
783 }
784 memset(ret, 0, sizeof(xmlRegAtom));
785 ret->type = type;
786 ret->quant = XML_REGEXP_QUANT_ONCE;
787 ret->min = 0;
788 ret->max = 0;
789 return(ret);
790}
791
792/**
793 * xmlRegFreeAtom:
794 * @atom: the regexp atom
795 *
796 * Free a regexp atom
797 */
798static void
799xmlRegFreeAtom(xmlRegAtomPtr atom) {
800 int i;
801
802 if (atom == NULL)
803 return;
804
805 for (i = 0;i < atom->nbRanges;i++)
806 xmlRegFreeRange(atom->ranges[i]);
807 if (atom->ranges != NULL)
808 xmlFree(atom->ranges);
809 if ((atom->type == XML_REGEXP_STRING) && (atom->valuep != NULL))
810 xmlFree(atom->valuep);
811 if ((atom->type == XML_REGEXP_STRING) && (atom->valuep2 != NULL))
812 xmlFree(atom->valuep2);
813 if ((atom->type == XML_REGEXP_BLOCK_NAME) && (atom->valuep != NULL))
814 xmlFree(atom->valuep);
815 xmlFree(atom);
816}
817
818/**
819 * xmlRegCopyAtom:
820 * @ctxt: the regexp parser context
821 * @atom: the oiginal atom
822 *
823 * Allocate a new regexp range
824 *
825 * Returns the new atom or NULL in case of error
826 */
827static xmlRegAtomPtr
828xmlRegCopyAtom(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom) {
829 xmlRegAtomPtr ret;
830
831 ret = (xmlRegAtomPtr) xmlMalloc(sizeof(xmlRegAtom));
832 if (ret == NULL) {
833 xmlRegexpErrMemory(ctxt, "copying atom");
834 return(NULL);
835 }
836 memset(ret, 0, sizeof(xmlRegAtom));
837 ret->type = atom->type;
838 ret->quant = atom->quant;
839 ret->min = atom->min;
840 ret->max = atom->max;
841 if (atom->nbRanges > 0) {
842 int i;
843
844 ret->ranges = (xmlRegRangePtr *) xmlMalloc(sizeof(xmlRegRangePtr) *
845 atom->nbRanges);
846 if (ret->ranges == NULL) {
847 xmlRegexpErrMemory(ctxt, "copying atom");
848 goto error;
849 }
850 for (i = 0;i < atom->nbRanges;i++) {
851 ret->ranges[i] = xmlRegCopyRange(ctxt, atom->ranges[i]);
852 if (ret->ranges[i] == NULL)
853 goto error;
854 ret->nbRanges = i + 1;
855 }
856 }
857 return(ret);
858
859error:
860 xmlRegFreeAtom(ret);
861 return(NULL);
862}
863
864static xmlRegStatePtr
865xmlRegNewState(xmlRegParserCtxtPtr ctxt) {
866 xmlRegStatePtr ret;
867
868 ret = (xmlRegStatePtr) xmlMalloc(sizeof(xmlRegState));
869 if (ret == NULL) {
870 xmlRegexpErrMemory(ctxt, "allocating state");
871 return(NULL);
872 }
873 memset(ret, 0, sizeof(xmlRegState));
874 ret->type = XML_REGEXP_TRANS_STATE;
875 ret->mark = XML_REGEXP_MARK_NORMAL;
876 return(ret);
877}
878
879/**
880 * xmlRegFreeState:
881 * @state: the regexp state
882 *
883 * Free a regexp state
884 */
885static void
886xmlRegFreeState(xmlRegStatePtr state) {
887 if (state == NULL)
888 return;
889
890 if (state->trans != NULL)
891 xmlFree(state->trans);
892 if (state->transTo != NULL)
893 xmlFree(state->transTo);
894 xmlFree(state);
895}
896
897/**
898 * xmlRegFreeParserCtxt:
899 * @ctxt: the regexp parser context
900 *
901 * Free a regexp parser context
902 */
903static void
904xmlRegFreeParserCtxt(xmlRegParserCtxtPtr ctxt) {
905 int i;
906 if (ctxt == NULL)
907 return;
908
909 if (ctxt->string != NULL)
910 xmlFree(ctxt->string);
911 if (ctxt->states != NULL) {
912 for (i = 0;i < ctxt->nbStates;i++)
913 xmlRegFreeState(ctxt->states[i]);
914 xmlFree(ctxt->states);
915 }
916 if (ctxt->atoms != NULL) {
917 for (i = 0;i < ctxt->nbAtoms;i++)
918 xmlRegFreeAtom(ctxt->atoms[i]);
919 xmlFree(ctxt->atoms);
920 }
921 if (ctxt->counters != NULL)
922 xmlFree(ctxt->counters);
923 xmlFree(ctxt);
924}
925
926/************************************************************************
927 * *
928 * Display of Data structures *
929 * *
930 ************************************************************************/
931
932static void
933xmlRegPrintAtomType(FILE *output, xmlRegAtomType type) {
934 switch (type) {
935 case XML_REGEXP_EPSILON:
936 fprintf(output, "epsilon "); break;
937 case XML_REGEXP_CHARVAL:
938 fprintf(output, "charval "); break;
939 case XML_REGEXP_RANGES:
940 fprintf(output, "ranges "); break;
941 case XML_REGEXP_SUBREG:
942 fprintf(output, "subexpr "); break;
943 case XML_REGEXP_STRING:
944 fprintf(output, "string "); break;
945 case XML_REGEXP_ANYCHAR:
946 fprintf(output, "anychar "); break;
947 case XML_REGEXP_ANYSPACE:
948 fprintf(output, "anyspace "); break;
949 case XML_REGEXP_NOTSPACE:
950 fprintf(output, "notspace "); break;
951 case XML_REGEXP_INITNAME:
952 fprintf(output, "initname "); break;
953 case XML_REGEXP_NOTINITNAME:
954 fprintf(output, "notinitname "); break;
955 case XML_REGEXP_NAMECHAR:
956 fprintf(output, "namechar "); break;
957 case XML_REGEXP_NOTNAMECHAR:
958 fprintf(output, "notnamechar "); break;
959 case XML_REGEXP_DECIMAL:
960 fprintf(output, "decimal "); break;
961 case XML_REGEXP_NOTDECIMAL:
962 fprintf(output, "notdecimal "); break;
963 case XML_REGEXP_REALCHAR:
964 fprintf(output, "realchar "); break;
965 case XML_REGEXP_NOTREALCHAR:
966 fprintf(output, "notrealchar "); break;
967 case XML_REGEXP_LETTER:
968 fprintf(output, "LETTER "); break;
969 case XML_REGEXP_LETTER_UPPERCASE:
970 fprintf(output, "LETTER_UPPERCASE "); break;
971 case XML_REGEXP_LETTER_LOWERCASE:
972 fprintf(output, "LETTER_LOWERCASE "); break;
973 case XML_REGEXP_LETTER_TITLECASE:
974 fprintf(output, "LETTER_TITLECASE "); break;
975 case XML_REGEXP_LETTER_MODIFIER:
976 fprintf(output, "LETTER_MODIFIER "); break;
977 case XML_REGEXP_LETTER_OTHERS:
978 fprintf(output, "LETTER_OTHERS "); break;
979 case XML_REGEXP_MARK:
980 fprintf(output, "MARK "); break;
981 case XML_REGEXP_MARK_NONSPACING:
982 fprintf(output, "MARK_NONSPACING "); break;
983 case XML_REGEXP_MARK_SPACECOMBINING:
984 fprintf(output, "MARK_SPACECOMBINING "); break;
985 case XML_REGEXP_MARK_ENCLOSING:
986 fprintf(output, "MARK_ENCLOSING "); break;
987 case XML_REGEXP_NUMBER:
988 fprintf(output, "NUMBER "); break;
989 case XML_REGEXP_NUMBER_DECIMAL:
990 fprintf(output, "NUMBER_DECIMAL "); break;
991 case XML_REGEXP_NUMBER_LETTER:
992 fprintf(output, "NUMBER_LETTER "); break;
993 case XML_REGEXP_NUMBER_OTHERS:
994 fprintf(output, "NUMBER_OTHERS "); break;
995 case XML_REGEXP_PUNCT:
996 fprintf(output, "PUNCT "); break;
997 case XML_REGEXP_PUNCT_CONNECTOR:
998 fprintf(output, "PUNCT_CONNECTOR "); break;
999 case XML_REGEXP_PUNCT_DASH:
1000 fprintf(output, "PUNCT_DASH "); break;
1001 case XML_REGEXP_PUNCT_OPEN:
1002 fprintf(output, "PUNCT_OPEN "); break;
1003 case XML_REGEXP_PUNCT_CLOSE:
1004 fprintf(output, "PUNCT_CLOSE "); break;
1005 case XML_REGEXP_PUNCT_INITQUOTE:
1006 fprintf(output, "PUNCT_INITQUOTE "); break;
1007 case XML_REGEXP_PUNCT_FINQUOTE:
1008 fprintf(output, "PUNCT_FINQUOTE "); break;
1009 case XML_REGEXP_PUNCT_OTHERS:
1010 fprintf(output, "PUNCT_OTHERS "); break;
1011 case XML_REGEXP_SEPAR:
1012 fprintf(output, "SEPAR "); break;
1013 case XML_REGEXP_SEPAR_SPACE:
1014 fprintf(output, "SEPAR_SPACE "); break;
1015 case XML_REGEXP_SEPAR_LINE:
1016 fprintf(output, "SEPAR_LINE "); break;
1017 case XML_REGEXP_SEPAR_PARA:
1018 fprintf(output, "SEPAR_PARA "); break;
1019 case XML_REGEXP_SYMBOL:
1020 fprintf(output, "SYMBOL "); break;
1021 case XML_REGEXP_SYMBOL_MATH:
1022 fprintf(output, "SYMBOL_MATH "); break;
1023 case XML_REGEXP_SYMBOL_CURRENCY:
1024 fprintf(output, "SYMBOL_CURRENCY "); break;
1025 case XML_REGEXP_SYMBOL_MODIFIER:
1026 fprintf(output, "SYMBOL_MODIFIER "); break;
1027 case XML_REGEXP_SYMBOL_OTHERS:
1028 fprintf(output, "SYMBOL_OTHERS "); break;
1029 case XML_REGEXP_OTHER:
1030 fprintf(output, "OTHER "); break;
1031 case XML_REGEXP_OTHER_CONTROL:
1032 fprintf(output, "OTHER_CONTROL "); break;
1033 case XML_REGEXP_OTHER_FORMAT:
1034 fprintf(output, "OTHER_FORMAT "); break;
1035 case XML_REGEXP_OTHER_PRIVATE:
1036 fprintf(output, "OTHER_PRIVATE "); break;
1037 case XML_REGEXP_OTHER_NA:
1038 fprintf(output, "OTHER_NA "); break;
1039 case XML_REGEXP_BLOCK_NAME:
1040 fprintf(output, "BLOCK "); break;
1041 }
1042}
1043
1044static void
1045xmlRegPrintQuantType(FILE *output, xmlRegQuantType type) {
1046 switch (type) {
1047 case XML_REGEXP_QUANT_EPSILON:
1048 fprintf(output, "epsilon "); break;
1049 case XML_REGEXP_QUANT_ONCE:
1050 fprintf(output, "once "); break;
1051 case XML_REGEXP_QUANT_OPT:
1052 fprintf(output, "? "); break;
1053 case XML_REGEXP_QUANT_MULT:
1054 fprintf(output, "* "); break;
1055 case XML_REGEXP_QUANT_PLUS:
1056 fprintf(output, "+ "); break;
1057 case XML_REGEXP_QUANT_RANGE:
1058 fprintf(output, "range "); break;
1059 case XML_REGEXP_QUANT_ONCEONLY:
1060 fprintf(output, "onceonly "); break;
1061 case XML_REGEXP_QUANT_ALL:
1062 fprintf(output, "all "); break;
1063 }
1064}
1065static void
1066xmlRegPrintRange(FILE *output, xmlRegRangePtr range) {
1067 fprintf(output, " range: ");
1068 if (range->neg)
1069 fprintf(output, "negative ");
1070 xmlRegPrintAtomType(output, range->type);
1071 fprintf(output, "%c - %c\n", range->start, range->end);
1072}
1073
1074static void
1075xmlRegPrintAtom(FILE *output, xmlRegAtomPtr atom) {
1076 fprintf(output, " atom: ");
1077 if (atom == NULL) {
1078 fprintf(output, "NULL\n");
1079 return;
1080 }
1081 if (atom->neg)
1082 fprintf(output, "not ");
1083 xmlRegPrintAtomType(output, atom->type);
1084 xmlRegPrintQuantType(output, atom->quant);
1085 if (atom->quant == XML_REGEXP_QUANT_RANGE)
1086 fprintf(output, "%d-%d ", atom->min, atom->max);
1087 if (atom->type == XML_REGEXP_STRING)
1088 fprintf(output, "'%s' ", (char *) atom->valuep);
1089 if (atom->type == XML_REGEXP_CHARVAL)
1090 fprintf(output, "char %c\n", atom->codepoint);
1091 else if (atom->type == XML_REGEXP_RANGES) {
1092 int i;
1093 fprintf(output, "%d entries\n", atom->nbRanges);
1094 for (i = 0; i < atom->nbRanges;i++)
1095 xmlRegPrintRange(output, atom->ranges[i]);
1096 } else if (atom->type == XML_REGEXP_SUBREG) {
1097 fprintf(output, "start %d end %d\n", atom->start->no, atom->stop->no);
1098 } else {
1099 fprintf(output, "\n");
1100 }
1101}
1102
1103static void
1104xmlRegPrintTrans(FILE *output, xmlRegTransPtr trans) {
1105 fprintf(output, " trans: ");
1106 if (trans == NULL) {
1107 fprintf(output, "NULL\n");
1108 return;
1109 }
1110 if (trans->to < 0) {
1111 fprintf(output, "removed\n");
1112 return;
1113 }
1114 if (trans->nd != 0) {
1115 if (trans->nd == 2)
1116 fprintf(output, "last not determinist, ");
1117 else
1118 fprintf(output, "not determinist, ");
1119 }
1120 if (trans->counter >= 0) {
1121 fprintf(output, "counted %d, ", trans->counter);
1122 }
1123 if (trans->count == REGEXP_ALL_COUNTER) {
1124 fprintf(output, "all transition, ");
1125 } else if (trans->count >= 0) {
1126 fprintf(output, "count based %d, ", trans->count);
1127 }
1128 if (trans->atom == NULL) {
1129 fprintf(output, "epsilon to %d\n", trans->to);
1130 return;
1131 }
1132 if (trans->atom->type == XML_REGEXP_CHARVAL)
1133 fprintf(output, "char %c ", trans->atom->codepoint);
1134 fprintf(output, "atom %d, to %d\n", trans->atom->no, trans->to);
1135}
1136
1137static void
1138xmlRegPrintState(FILE *output, xmlRegStatePtr state) {
1139 int i;
1140
1141 fprintf(output, " state: ");
1142 if (state == NULL) {
1143 fprintf(output, "NULL\n");
1144 return;
1145 }
1146 if (state->type == XML_REGEXP_START_STATE)
1147 fprintf(output, "START ");
1148 if (state->type == XML_REGEXP_FINAL_STATE)
1149 fprintf(output, "FINAL ");
1150
1151 fprintf(output, "%d, %d transitions:\n", state->no, state->nbTrans);
1152 for (i = 0;i < state->nbTrans; i++) {
1153 xmlRegPrintTrans(output, &(state->trans[i]));
1154 }
1155}
1156
1157#ifdef DEBUG_REGEXP_GRAPH
1158static void
1159xmlRegPrintCtxt(FILE *output, xmlRegParserCtxtPtr ctxt) {
1160 int i;
1161
1162 fprintf(output, " ctxt: ");
1163 if (ctxt == NULL) {
1164 fprintf(output, "NULL\n");
1165 return;
1166 }
1167 fprintf(output, "'%s' ", ctxt->string);
1168 if (ctxt->error)
1169 fprintf(output, "error ");
1170 if (ctxt->neg)
1171 fprintf(output, "neg ");
1172 fprintf(output, "\n");
1173 fprintf(output, "%d atoms:\n", ctxt->nbAtoms);
1174 for (i = 0;i < ctxt->nbAtoms; i++) {
1175 fprintf(output, " %02d ", i);
1176 xmlRegPrintAtom(output, ctxt->atoms[i]);
1177 }
1178 if (ctxt->atom != NULL) {
1179 fprintf(output, "current atom:\n");
1180 xmlRegPrintAtom(output, ctxt->atom);
1181 }
1182 fprintf(output, "%d states:", ctxt->nbStates);
1183 if (ctxt->start != NULL)
1184 fprintf(output, " start: %d", ctxt->start->no);
1185 if (ctxt->end != NULL)
1186 fprintf(output, " end: %d", ctxt->end->no);
1187 fprintf(output, "\n");
1188 for (i = 0;i < ctxt->nbStates; i++) {
1189 xmlRegPrintState(output, ctxt->states[i]);
1190 }
1191 fprintf(output, "%d counters:\n", ctxt->nbCounters);
1192 for (i = 0;i < ctxt->nbCounters; i++) {
1193 fprintf(output, " %d: min %d max %d\n", i, ctxt->counters[i].min,
1194 ctxt->counters[i].max);
1195 }
1196}
1197#endif
1198
1199/************************************************************************
1200 * *
1201 * Finite Automata structures manipulations *
1202 * *
1203 ************************************************************************/
1204
1205static void
1206xmlRegAtomAddRange(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom,
1207 int neg, xmlRegAtomType type, int start, int end,
1208 xmlChar *blockName) {
1209 xmlRegRangePtr range;
1210
1211 if (atom == NULL) {
1212 ERROR("add range: atom is NULL");
1213 return;
1214 }
1215 if (atom->type != XML_REGEXP_RANGES) {
1216 ERROR("add range: atom is not ranges");
1217 return;
1218 }
1219 if (atom->maxRanges == 0) {
1220 atom->maxRanges = 4;
1221 atom->ranges = (xmlRegRangePtr *) xmlMalloc(atom->maxRanges *
1222 sizeof(xmlRegRangePtr));
1223 if (atom->ranges == NULL) {
1224 xmlRegexpErrMemory(ctxt, "adding ranges");
1225 atom->maxRanges = 0;
1226 return;
1227 }
1228 } else if (atom->nbRanges >= atom->maxRanges) {
1229 xmlRegRangePtr *tmp;
1230 atom->maxRanges *= 2;
1231 tmp = (xmlRegRangePtr *) xmlRealloc(atom->ranges, atom->maxRanges *
1232 sizeof(xmlRegRangePtr));
1233 if (tmp == NULL) {
1234 xmlRegexpErrMemory(ctxt, "adding ranges");
1235 atom->maxRanges /= 2;
1236 return;
1237 }
1238 atom->ranges = tmp;
1239 }
1240 range = xmlRegNewRange(ctxt, neg, type, start, end);
1241 if (range == NULL)
1242 return;
1243 range->blockName = blockName;
1244 atom->ranges[atom->nbRanges++] = range;
1245
1246}
1247
1248static int
1249xmlRegGetCounter(xmlRegParserCtxtPtr ctxt) {
1250 if (ctxt->maxCounters == 0) {
1251 ctxt->maxCounters = 4;
1252 ctxt->counters = (xmlRegCounter *) xmlMalloc(ctxt->maxCounters *
1253 sizeof(xmlRegCounter));
1254 if (ctxt->counters == NULL) {
1255 xmlRegexpErrMemory(ctxt, "allocating counter");
1256 ctxt->maxCounters = 0;
1257 return(-1);
1258 }
1259 } else if (ctxt->nbCounters >= ctxt->maxCounters) {
1260 xmlRegCounter *tmp;
1261 ctxt->maxCounters *= 2;
1262 tmp = (xmlRegCounter *) xmlRealloc(ctxt->counters, ctxt->maxCounters *
1263 sizeof(xmlRegCounter));
1264 if (tmp == NULL) {
1265 xmlRegexpErrMemory(ctxt, "allocating counter");
1266 ctxt->maxCounters /= 2;
1267 return(-1);
1268 }
1269 ctxt->counters = tmp;
1270 }
1271 ctxt->counters[ctxt->nbCounters].min = -1;
1272 ctxt->counters[ctxt->nbCounters].max = -1;
1273 return(ctxt->nbCounters++);
1274}
1275
1276static int
1277xmlRegAtomPush(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom) {
1278 if (atom == NULL) {
1279 ERROR("atom push: atom is NULL");
1280 return(-1);
1281 }
1282 if (ctxt->maxAtoms == 0) {
1283 ctxt->maxAtoms = 4;
1284 ctxt->atoms = (xmlRegAtomPtr *) xmlMalloc(ctxt->maxAtoms *
1285 sizeof(xmlRegAtomPtr));
1286 if (ctxt->atoms == NULL) {
1287 xmlRegexpErrMemory(ctxt, "pushing atom");
1288 ctxt->maxAtoms = 0;
1289 return(-1);
1290 }
1291 } else if (ctxt->nbAtoms >= ctxt->maxAtoms) {
1292 xmlRegAtomPtr *tmp;
1293 ctxt->maxAtoms *= 2;
1294 tmp = (xmlRegAtomPtr *) xmlRealloc(ctxt->atoms, ctxt->maxAtoms *
1295 sizeof(xmlRegAtomPtr));
1296 if (tmp == NULL) {
1297 xmlRegexpErrMemory(ctxt, "allocating counter");
1298 ctxt->maxAtoms /= 2;
1299 return(-1);
1300 }
1301 ctxt->atoms = tmp;
1302 }
1303 atom->no = ctxt->nbAtoms;
1304 ctxt->atoms[ctxt->nbAtoms++] = atom;
1305 return(0);
1306}
1307
1308static void
1309xmlRegStateAddTransTo(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr target,
1310 int from) {
1311 if (target->maxTransTo == 0) {
1312 target->maxTransTo = 8;
1313 target->transTo = (int *) xmlMalloc(target->maxTransTo *
1314 sizeof(int));
1315 if (target->transTo == NULL) {
1316 xmlRegexpErrMemory(ctxt, "adding transition");
1317 target->maxTransTo = 0;
1318 return;
1319 }
1320 } else if (target->nbTransTo >= target->maxTransTo) {
1321 int *tmp;
1322 target->maxTransTo *= 2;
1323 tmp = (int *) xmlRealloc(target->transTo, target->maxTransTo *
1324 sizeof(int));
1325 if (tmp == NULL) {
1326 xmlRegexpErrMemory(ctxt, "adding transition");
1327 target->maxTransTo /= 2;
1328 return;
1329 }
1330 target->transTo = tmp;
1331 }
1332 target->transTo[target->nbTransTo] = from;
1333 target->nbTransTo++;
1334}
1335
1336static void
1337xmlRegStateAddTrans(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
1338 xmlRegAtomPtr atom, xmlRegStatePtr target,
1339 int counter, int count) {
1340
1341 int nrtrans;
1342
1343 if (state == NULL) {
1344 ERROR("add state: state is NULL");
1345 return;
1346 }
1347 if (target == NULL) {
1348 ERROR("add state: target is NULL");
1349 return;
1350 }
1351 /*
1352 * Other routines follow the philosophy 'When in doubt, add a transition'
1353 * so we check here whether such a transition is already present and, if
1354 * so, silently ignore this request.
1355 */
1356
1357 for (nrtrans = state->nbTrans - 1; nrtrans >= 0; nrtrans--) {
1358 xmlRegTransPtr trans = &(state->trans[nrtrans]);
1359 if ((trans->atom == atom) &&
1360 (trans->to == target->no) &&
1361 (trans->counter == counter) &&
1362 (trans->count == count)) {
1363#ifdef DEBUG_REGEXP_GRAPH
1364 printf("Ignoring duplicate transition from %d to %d\n",
1365 state->no, target->no);
1366#endif
1367 return;
1368 }
1369 }
1370
1371 if (state->maxTrans == 0) {
1372 state->maxTrans = 8;
1373 state->trans = (xmlRegTrans *) xmlMalloc(state->maxTrans *
1374 sizeof(xmlRegTrans));
1375 if (state->trans == NULL) {
1376 xmlRegexpErrMemory(ctxt, "adding transition");
1377 state->maxTrans = 0;
1378 return;
1379 }
1380 } else if (state->nbTrans >= state->maxTrans) {
1381 xmlRegTrans *tmp;
1382 state->maxTrans *= 2;
1383 tmp = (xmlRegTrans *) xmlRealloc(state->trans, state->maxTrans *
1384 sizeof(xmlRegTrans));
1385 if (tmp == NULL) {
1386 xmlRegexpErrMemory(ctxt, "adding transition");
1387 state->maxTrans /= 2;
1388 return;
1389 }
1390 state->trans = tmp;
1391 }
1392#ifdef DEBUG_REGEXP_GRAPH
1393 printf("Add trans from %d to %d ", state->no, target->no);
1394 if (count == REGEXP_ALL_COUNTER)
1395 printf("all transition\n");
1396 else if (count >= 0)
1397 printf("count based %d\n", count);
1398 else if (counter >= 0)
1399 printf("counted %d\n", counter);
1400 else if (atom == NULL)
1401 printf("epsilon transition\n");
1402 else if (atom != NULL)
1403 xmlRegPrintAtom(stdout, atom);
1404#endif
1405
1406 state->trans[state->nbTrans].atom = atom;
1407 state->trans[state->nbTrans].to = target->no;
1408 state->trans[state->nbTrans].counter = counter;
1409 state->trans[state->nbTrans].count = count;
1410 state->trans[state->nbTrans].nd = 0;
1411 state->nbTrans++;
1412 xmlRegStateAddTransTo(ctxt, target, state->no);
1413}
1414
1415static int
1416xmlRegStatePush(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state) {
1417 if (state == NULL) return(-1);
1418 if (ctxt->maxStates == 0) {
1419 ctxt->maxStates = 4;
1420 ctxt->states = (xmlRegStatePtr *) xmlMalloc(ctxt->maxStates *
1421 sizeof(xmlRegStatePtr));
1422 if (ctxt->states == NULL) {
1423 xmlRegexpErrMemory(ctxt, "adding state");
1424 ctxt->maxStates = 0;
1425 return(-1);
1426 }
1427 } else if (ctxt->nbStates >= ctxt->maxStates) {
1428 xmlRegStatePtr *tmp;
1429 ctxt->maxStates *= 2;
1430 tmp = (xmlRegStatePtr *) xmlRealloc(ctxt->states, ctxt->maxStates *
1431 sizeof(xmlRegStatePtr));
1432 if (tmp == NULL) {
1433 xmlRegexpErrMemory(ctxt, "adding state");
1434 ctxt->maxStates /= 2;
1435 return(-1);
1436 }
1437 ctxt->states = tmp;
1438 }
1439 state->no = ctxt->nbStates;
1440 ctxt->states[ctxt->nbStates++] = state;
1441 return(0);
1442}
1443
1444/**
1445 * xmlFAGenerateAllTransition:
1446 * @ctxt: a regexp parser context
1447 * @from: the from state
1448 * @to: the target state or NULL for building a new one
1449 * @lax:
1450 *
1451 */
1452static void
1453xmlFAGenerateAllTransition(xmlRegParserCtxtPtr ctxt,
1454 xmlRegStatePtr from, xmlRegStatePtr to,
1455 int lax) {
1456 if (to == NULL) {
1457 to = xmlRegNewState(ctxt);
1458 xmlRegStatePush(ctxt, to);
1459 ctxt->state = to;
1460 }
1461 if (lax)
1462 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_LAX_COUNTER);
1463 else
1464 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_COUNTER);
1465}
1466
1467/**
1468 * xmlFAGenerateEpsilonTransition:
1469 * @ctxt: a regexp parser context
1470 * @from: the from state
1471 * @to: the target state or NULL for building a new one
1472 *
1473 */
1474static void
1475xmlFAGenerateEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1476 xmlRegStatePtr from, xmlRegStatePtr to) {
1477 if (to == NULL) {
1478 to = xmlRegNewState(ctxt);
1479 xmlRegStatePush(ctxt, to);
1480 ctxt->state = to;
1481 }
1482 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, -1);
1483}
1484
1485/**
1486 * xmlFAGenerateCountedEpsilonTransition:
1487 * @ctxt: a regexp parser context
1488 * @from: the from state
1489 * @to: the target state or NULL for building a new one
1490 * counter: the counter for that transition
1491 *
1492 */
1493static void
1494xmlFAGenerateCountedEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1495 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1496 if (to == NULL) {
1497 to = xmlRegNewState(ctxt);
1498 xmlRegStatePush(ctxt, to);
1499 ctxt->state = to;
1500 }
1501 xmlRegStateAddTrans(ctxt, from, NULL, to, counter, -1);
1502}
1503
1504/**
1505 * xmlFAGenerateCountedTransition:
1506 * @ctxt: a regexp parser context
1507 * @from: the from state
1508 * @to: the target state or NULL for building a new one
1509 * counter: the counter for that transition
1510 *
1511 */
1512static void
1513xmlFAGenerateCountedTransition(xmlRegParserCtxtPtr ctxt,
1514 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1515 if (to == NULL) {
1516 to = xmlRegNewState(ctxt);
1517 xmlRegStatePush(ctxt, to);
1518 ctxt->state = to;
1519 }
1520 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, counter);
1521}
1522
1523/**
1524 * xmlFAGenerateTransitions:
1525 * @ctxt: a regexp parser context
1526 * @from: the from state
1527 * @to: the target state or NULL for building a new one
1528 * @atom: the atom generating the transition
1529 *
1530 * Returns 0 if success and -1 in case of error.
1531 */
1532static int
1533xmlFAGenerateTransitions(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr from,
1534 xmlRegStatePtr to, xmlRegAtomPtr atom) {
1535 if (atom == NULL) {
1536 ERROR("genrate transition: atom == NULL");
1537 return(-1);
1538 }
1539 if (atom->type == XML_REGEXP_SUBREG) {
1540 /*
1541 * this is a subexpression handling one should not need to
1542 * create a new node except for XML_REGEXP_QUANT_RANGE.
1543 */
1544 if (xmlRegAtomPush(ctxt, atom) < 0) {
1545 return(-1);
1546 }
1547 if ((to != NULL) && (atom->stop != to) &&
1548 (atom->quant != XML_REGEXP_QUANT_RANGE)) {
1549 /*
1550 * Generate an epsilon transition to link to the target
1551 */
1552 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, to);
1553#ifdef DV
1554 } else if ((to == NULL) && (atom->quant != XML_REGEXP_QUANT_RANGE) &&
1555 (atom->quant != XML_REGEXP_QUANT_ONCE)) {
1556 to = xmlRegNewState(ctxt);
1557 xmlRegStatePush(ctxt, to);
1558 ctxt->state = to;
1559 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, to);
1560#endif
1561 }
1562 switch (atom->quant) {
1563 case XML_REGEXP_QUANT_OPT:
1564 atom->quant = XML_REGEXP_QUANT_ONCE;
1565 /*
1566 * transition done to the state after end of atom.
1567 * 1. set transition from atom start to new state
1568 * 2. set transition from atom end to this state.
1569 */
1570 xmlFAGenerateEpsilonTransition(ctxt, atom->start, 0);
1571 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, ctxt->state);
1572 break;
1573 case XML_REGEXP_QUANT_MULT:
1574 atom->quant = XML_REGEXP_QUANT_ONCE;
1575 xmlFAGenerateEpsilonTransition(ctxt, atom->start, atom->stop);
1576 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1577 break;
1578 case XML_REGEXP_QUANT_PLUS:
1579 atom->quant = XML_REGEXP_QUANT_ONCE;
1580 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1581 break;
1582 case XML_REGEXP_QUANT_RANGE: {
1583 int counter;
1584 xmlRegStatePtr inter, newstate;
1585
1586 /*
1587 * create the final state now if needed
1588 */
1589 if (to != NULL) {
1590 newstate = to;
1591 } else {
1592 newstate = xmlRegNewState(ctxt);
1593 xmlRegStatePush(ctxt, newstate);
1594 }
1595
1596 /*
1597 * The principle here is to use counted transition
1598 * to avoid explosion in the number of states in the
1599 * graph. This is clearly more complex but should not
1600 * be exploitable at runtime.
1601 */
1602 if ((atom->min == 0) && (atom->start0 == NULL)) {
1603 xmlRegAtomPtr copy;
1604 /*
1605 * duplicate a transition based on atom to count next
1606 * occurences after 1. We cannot loop to atom->start
1607 * directly because we need an epsilon transition to
1608 * newstate.
1609 */
1610 /* ???? For some reason it seems we never reach that
1611 case, I suppose this got optimized out before when
1612 building the automata */
1613 copy = xmlRegCopyAtom(ctxt, atom);
1614 if (copy == NULL)
1615 return(-1);
1616 copy->quant = XML_REGEXP_QUANT_ONCE;
1617 copy->min = 0;
1618 copy->max = 0;
1619
1620 if (xmlFAGenerateTransitions(ctxt, atom->start, NULL, copy)
1621 < 0)
1622 return(-1);
1623 inter = ctxt->state;
1624 counter = xmlRegGetCounter(ctxt);
1625 ctxt->counters[counter].min = atom->min - 1;
1626 ctxt->counters[counter].max = atom->max - 1;
1627 /* count the number of times we see it again */
1628 xmlFAGenerateCountedEpsilonTransition(ctxt, inter,
1629 atom->stop, counter);
1630 /* allow a way out based on the count */
1631 xmlFAGenerateCountedTransition(ctxt, inter,
1632 newstate, counter);
1633 /* and also allow a direct exit for 0 */
1634 xmlFAGenerateEpsilonTransition(ctxt, atom->start,
1635 newstate);
1636 } else {
1637 /*
1638 * either we need the atom at least once or there
1639 * is an atom->start0 allowing to easilly plug the
1640 * epsilon transition.
1641 */
1642 counter = xmlRegGetCounter(ctxt);
1643 ctxt->counters[counter].min = atom->min - 1;
1644 ctxt->counters[counter].max = atom->max - 1;
1645 /* count the number of times we see it again */
1646 xmlFAGenerateCountedEpsilonTransition(ctxt, atom->stop,
1647 atom->start, counter);
1648 /* allow a way out based on the count */
1649 xmlFAGenerateCountedTransition(ctxt, atom->stop,
1650 newstate, counter);
1651 /* and if needed allow a direct exit for 0 */
1652 if (atom->min == 0)
1653 xmlFAGenerateEpsilonTransition(ctxt, atom->start0,
1654 newstate);
1655
1656 }
1657 atom->min = 0;
1658 atom->max = 0;
1659 atom->quant = XML_REGEXP_QUANT_ONCE;
1660 ctxt->state = newstate;
1661 }
1662 default:
1663 break;
1664 }
1665 return(0);
1666 }
1667 if ((atom->min == 0) && (atom->max == 0) &&
1668 (atom->quant == XML_REGEXP_QUANT_RANGE)) {
1669 /*
1670 * we can discard the atom and generate an epsilon transition instead
1671 */
1672 if (to == NULL) {
1673 to = xmlRegNewState(ctxt);
1674 if (to != NULL)
1675 xmlRegStatePush(ctxt, to);
1676 else {
1677 return(-1);
1678 }
1679 }
1680 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1681 ctxt->state = to;
1682 xmlRegFreeAtom(atom);
1683 return(0);
1684 }
1685 if (to == NULL) {
1686 to = xmlRegNewState(ctxt);
1687 if (to != NULL)
1688 xmlRegStatePush(ctxt, to);
1689 else {
1690 return(-1);
1691 }
1692 }
1693 if (xmlRegAtomPush(ctxt, atom) < 0) {
1694 return(-1);
1695 }
1696 xmlRegStateAddTrans(ctxt, from, atom, to, -1, -1);
1697 ctxt->state = to;
1698 switch (atom->quant) {
1699 case XML_REGEXP_QUANT_OPT:
1700 atom->quant = XML_REGEXP_QUANT_ONCE;
1701 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1702 break;
1703 case XML_REGEXP_QUANT_MULT:
1704 atom->quant = XML_REGEXP_QUANT_ONCE;
1705 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1706 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1707 break;
1708 case XML_REGEXP_QUANT_PLUS:
1709 atom->quant = XML_REGEXP_QUANT_ONCE;
1710 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1711 break;
1712 case XML_REGEXP_QUANT_RANGE:
1713#if DV_test
1714 if (atom->min == 0) {
1715 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1716 }
1717#endif
1718 break;
1719 default:
1720 break;
1721 }
1722 return(0);
1723}
1724
1725/**
1726 * xmlFAReduceEpsilonTransitions:
1727 * @ctxt: a regexp parser context
1728 * @fromnr: the from state
1729 * @tonr: the to state
1730 * @counter: should that transition be associated to a counted
1731 *
1732 */
1733static void
1734xmlFAReduceEpsilonTransitions(xmlRegParserCtxtPtr ctxt, int fromnr,
1735 int tonr, int counter) {
1736 int transnr;
1737 xmlRegStatePtr from;
1738 xmlRegStatePtr to;
1739
1740#ifdef DEBUG_REGEXP_GRAPH
1741 printf("xmlFAReduceEpsilonTransitions(%d, %d)\n", fromnr, tonr);
1742#endif
1743 from = ctxt->states[fromnr];
1744 if (from == NULL)
1745 return;
1746 to = ctxt->states[tonr];
1747 if (to == NULL)
1748 return;
1749 if ((to->mark == XML_REGEXP_MARK_START) ||
1750 (to->mark == XML_REGEXP_MARK_VISITED))
1751 return;
1752
1753 to->mark = XML_REGEXP_MARK_VISITED;
1754 if (to->type == XML_REGEXP_FINAL_STATE) {
1755#ifdef DEBUG_REGEXP_GRAPH
1756 printf("State %d is final, so %d becomes final\n", tonr, fromnr);
1757#endif
1758 from->type = XML_REGEXP_FINAL_STATE;
1759 }
1760 for (transnr = 0;transnr < to->nbTrans;transnr++) {
1761 if (to->trans[transnr].to < 0)
1762 continue;
1763 if (to->trans[transnr].atom == NULL) {
1764 /*
1765 * Don't remove counted transitions
1766 * Don't loop either
1767 */
1768 if (to->trans[transnr].to != fromnr) {
1769 if (to->trans[transnr].count >= 0) {
1770 int newto = to->trans[transnr].to;
1771
1772 xmlRegStateAddTrans(ctxt, from, NULL,
1773 ctxt->states[newto],
1774 -1, to->trans[transnr].count);
1775 } else {
1776#ifdef DEBUG_REGEXP_GRAPH
1777 printf("Found epsilon trans %d from %d to %d\n",
1778 transnr, tonr, to->trans[transnr].to);
1779#endif
1780 if (to->trans[transnr].counter >= 0) {
1781 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1782 to->trans[transnr].to,
1783 to->trans[transnr].counter);
1784 } else {
1785 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1786 to->trans[transnr].to,
1787 counter);
1788 }
1789 }
1790 }
1791 } else {
1792 int newto = to->trans[transnr].to;
1793
1794 if (to->trans[transnr].counter >= 0) {
1795 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1796 ctxt->states[newto],
1797 to->trans[transnr].counter, -1);
1798 } else {
1799 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1800 ctxt->states[newto], counter, -1);
1801 }
1802 }
1803 }
1804 to->mark = XML_REGEXP_MARK_NORMAL;
1805}
1806
1807/**
1808 * xmlFAEliminateSimpleEpsilonTransitions:
1809 * @ctxt: a regexp parser context
1810 *
1811 * Eliminating general epsilon transitions can get costly in the general
1812 * algorithm due to the large amount of generated new transitions and
1813 * associated comparisons. However for simple epsilon transition used just
1814 * to separate building blocks when generating the automata this can be
1815 * reduced to state elimination:
1816 * - if there exists an epsilon from X to Y
1817 * - if there is no other transition from X
1818 * then X and Y are semantically equivalent and X can be eliminated
1819 * If X is the start state then make Y the start state, else replace the
1820 * target of all transitions to X by transitions to Y.
1821 */
1822static void
1823xmlFAEliminateSimpleEpsilonTransitions(xmlRegParserCtxtPtr ctxt) {
1824 int statenr, i, j, newto;
1825 xmlRegStatePtr state, tmp;
1826
1827 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1828 state = ctxt->states[statenr];
1829 if (state == NULL)
1830 continue;
1831 if (state->nbTrans != 1)
1832 continue;
1833 if (state->type == XML_REGEXP_UNREACH_STATE)
1834 continue;
1835 /* is the only transition out a basic transition */
1836 if ((state->trans[0].atom == NULL) &&
1837 (state->trans[0].to >= 0) &&
1838 (state->trans[0].to != statenr) &&
1839 (state->trans[0].counter < 0) &&
1840 (state->trans[0].count < 0)) {
1841 newto = state->trans[0].to;
1842
1843 if (state->type == XML_REGEXP_START_STATE) {
1844#ifdef DEBUG_REGEXP_GRAPH
1845 printf("Found simple epsilon trans from start %d to %d\n",
1846 statenr, newto);
1847#endif
1848 } else {
1849#ifdef DEBUG_REGEXP_GRAPH
1850 printf("Found simple epsilon trans from %d to %d\n",
1851 statenr, newto);
1852#endif
1853 for (i = 0;i < state->nbTransTo;i++) {
1854 tmp = ctxt->states[state->transTo[i]];
1855 for (j = 0;j < tmp->nbTrans;j++) {
1856 if (tmp->trans[j].to == statenr) {
1857#ifdef DEBUG_REGEXP_GRAPH
1858 printf("Changed transition %d on %d to go to %d\n",
1859 j, tmp->no, newto);
1860#endif
1861 tmp->trans[j].to = -1;
1862 xmlRegStateAddTrans(ctxt, tmp, tmp->trans[j].atom,
1863 ctxt->states[newto],
1864 tmp->trans[j].counter,
1865 tmp->trans[j].count);
1866 }
1867 }
1868 }
1869 if (state->type == XML_REGEXP_FINAL_STATE)
1870 ctxt->states[newto]->type = XML_REGEXP_FINAL_STATE;
1871 /* eliminate the transition completely */
1872 state->nbTrans = 0;
1873
1874 state->type = XML_REGEXP_UNREACH_STATE;
1875
1876 }
1877
1878 }
1879 }
1880}
1881/**
1882 * xmlFAEliminateEpsilonTransitions:
1883 * @ctxt: a regexp parser context
1884 *
1885 */
1886static void
1887xmlFAEliminateEpsilonTransitions(xmlRegParserCtxtPtr ctxt) {
1888 int statenr, transnr;
1889 xmlRegStatePtr state;
1890 int has_epsilon;
1891
1892 if (ctxt->states == NULL) return;
1893
1894 /*
1895 * Eliminate simple epsilon transition and the associated unreachable
1896 * states.
1897 */
1898 xmlFAEliminateSimpleEpsilonTransitions(ctxt);
1899 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1900 state = ctxt->states[statenr];
1901 if ((state != NULL) && (state->type == XML_REGEXP_UNREACH_STATE)) {
1902#ifdef DEBUG_REGEXP_GRAPH
1903 printf("Removed unreachable state %d\n", statenr);
1904#endif
1905 xmlRegFreeState(state);
1906 ctxt->states[statenr] = NULL;
1907 }
1908 }
1909
1910 has_epsilon = 0;
1911
1912 /*
1913 * Build the completed transitions bypassing the epsilons
1914 * Use a marking algorithm to avoid loops
1915 * Mark sink states too.
1916 * Process from the latests states backward to the start when
1917 * there is long cascading epsilon chains this minimize the
1918 * recursions and transition compares when adding the new ones
1919 */
1920 for (statenr = ctxt->nbStates - 1;statenr >= 0;statenr--) {
1921 state = ctxt->states[statenr];
1922 if (state == NULL)
1923 continue;
1924 if ((state->nbTrans == 0) &&
1925 (state->type != XML_REGEXP_FINAL_STATE)) {
1926 state->type = XML_REGEXP_SINK_STATE;
1927 }
1928 for (transnr = 0;transnr < state->nbTrans;transnr++) {
1929 if ((state->trans[transnr].atom == NULL) &&
1930 (state->trans[transnr].to >= 0)) {
1931 if (state->trans[transnr].to == statenr) {
1932 state->trans[transnr].to = -1;
1933#ifdef DEBUG_REGEXP_GRAPH
1934 printf("Removed loopback epsilon trans %d on %d\n",
1935 transnr, statenr);
1936#endif
1937 } else if (state->trans[transnr].count < 0) {
1938 int newto = state->trans[transnr].to;
1939
1940#ifdef DEBUG_REGEXP_GRAPH
1941 printf("Found epsilon trans %d from %d to %d\n",
1942 transnr, statenr, newto);
1943#endif
1944 has_epsilon = 1;
1945 state->trans[transnr].to = -2;
1946 state->mark = XML_REGEXP_MARK_START;
1947 xmlFAReduceEpsilonTransitions(ctxt, statenr,
1948 newto, state->trans[transnr].counter);
1949 state->mark = XML_REGEXP_MARK_NORMAL;
1950#ifdef DEBUG_REGEXP_GRAPH
1951 } else {
1952 printf("Found counted transition %d on %d\n",
1953 transnr, statenr);
1954#endif
1955 }
1956 }
1957 }
1958 }
1959 /*
1960 * Eliminate the epsilon transitions
1961 */
1962 if (has_epsilon) {
1963 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1964 state = ctxt->states[statenr];
1965 if (state == NULL)
1966 continue;
1967 for (transnr = 0;transnr < state->nbTrans;transnr++) {
1968 xmlRegTransPtr trans = &(state->trans[transnr]);
1969 if ((trans->atom == NULL) &&
1970 (trans->count < 0) &&
1971 (trans->to >= 0)) {
1972 trans->to = -1;
1973 }
1974 }
1975 }
1976 }
1977
1978 /*
1979 * Use this pass to detect unreachable states too
1980 */
1981 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1982 state = ctxt->states[statenr];
1983 if (state != NULL)
1984 state->reached = XML_REGEXP_MARK_NORMAL;
1985 }
1986 state = ctxt->states[0];
1987 if (state != NULL)
1988 state->reached = XML_REGEXP_MARK_START;
1989 while (state != NULL) {
1990 xmlRegStatePtr target = NULL;
1991 state->reached = XML_REGEXP_MARK_VISITED;
1992 /*
1993 * Mark all states reachable from the current reachable state
1994 */
1995 for (transnr = 0;transnr < state->nbTrans;transnr++) {
1996 if ((state->trans[transnr].to >= 0) &&
1997 ((state->trans[transnr].atom != NULL) ||
1998 (state->trans[transnr].count >= 0))) {
1999 int newto = state->trans[transnr].to;
2000
2001 if (ctxt->states[newto] == NULL)
2002 continue;
2003 if (ctxt->states[newto]->reached == XML_REGEXP_MARK_NORMAL) {
2004 ctxt->states[newto]->reached = XML_REGEXP_MARK_START;
2005 target = ctxt->states[newto];
2006 }
2007 }
2008 }
2009
2010 /*
2011 * find the next accessible state not explored
2012 */
2013 if (target == NULL) {
2014 for (statenr = 1;statenr < ctxt->nbStates;statenr++) {
2015 state = ctxt->states[statenr];
2016 if ((state != NULL) && (state->reached ==
2017 XML_REGEXP_MARK_START)) {
2018 target = state;
2019 break;
2020 }
2021 }
2022 }
2023 state = target;
2024 }
2025 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2026 state = ctxt->states[statenr];
2027 if ((state != NULL) && (state->reached == XML_REGEXP_MARK_NORMAL)) {
2028#ifdef DEBUG_REGEXP_GRAPH
2029 printf("Removed unreachable state %d\n", statenr);
2030#endif
2031 xmlRegFreeState(state);
2032 ctxt->states[statenr] = NULL;
2033 }
2034 }
2035
2036}
2037
2038static int
2039xmlFACompareRanges(xmlRegRangePtr range1, xmlRegRangePtr range2) {
2040 int ret = 0;
2041
2042 if ((range1->type == XML_REGEXP_RANGES) ||
2043 (range2->type == XML_REGEXP_RANGES) ||
2044 (range2->type == XML_REGEXP_SUBREG) ||
2045 (range1->type == XML_REGEXP_SUBREG) ||
2046 (range1->type == XML_REGEXP_STRING) ||
2047 (range2->type == XML_REGEXP_STRING))
2048 return(-1);
2049
2050 /* put them in order */
2051 if (range1->type > range2->type) {
2052 xmlRegRangePtr tmp;
2053
2054 tmp = range1;
2055 range1 = range2;
2056 range2 = tmp;
2057 }
2058 if ((range1->type == XML_REGEXP_ANYCHAR) ||
2059 (range2->type == XML_REGEXP_ANYCHAR)) {
2060 ret = 1;
2061 } else if ((range1->type == XML_REGEXP_EPSILON) ||
2062 (range2->type == XML_REGEXP_EPSILON)) {
2063 return(0);
2064 } else if (range1->type == range2->type) {
2065 if ((range1->type != XML_REGEXP_CHARVAL) ||
2066 (range1->end < range2->start) ||
2067 (range2->end < range1->start))
2068 ret = 1;
2069 else
2070 ret = 0;
2071 } else if (range1->type == XML_REGEXP_CHARVAL) {
2072 int codepoint;
2073 int neg = 0;
2074
2075 /*
2076 * just check all codepoints in the range for acceptance,
2077 * this is usually way cheaper since done only once at
2078 * compilation than testing over and over at runtime or
2079 * pushing too many states when evaluating.
2080 */
2081 if (((range1->neg == 0) && (range2->neg != 0)) ||
2082 ((range1->neg != 0) && (range2->neg == 0)))
2083 neg = 1;
2084
2085 for (codepoint = range1->start;codepoint <= range1->end ;codepoint++) {
2086 ret = xmlRegCheckCharacterRange(range2->type, codepoint,
2087 0, range2->start, range2->end,
2088 range2->blockName);
2089 if (ret < 0)
2090 return(-1);
2091 if (((neg == 1) && (ret == 0)) ||
2092 ((neg == 0) && (ret == 1)))
2093 return(1);
2094 }
2095 return(0);
2096 } else if ((range1->type == XML_REGEXP_BLOCK_NAME) ||
2097 (range2->type == XML_REGEXP_BLOCK_NAME)) {
2098 if (range1->type == range2->type) {
2099 ret = xmlStrEqual(range1->blockName, range2->blockName);
2100 } else {
2101 /*
2102 * comparing a block range with anything else is way
2103 * too costly, and maintining the table is like too much
2104 * memory too, so let's force the automata to save state
2105 * here.
2106 */
2107 return(1);
2108 }
2109 } else if ((range1->type < XML_REGEXP_LETTER) ||
2110 (range2->type < XML_REGEXP_LETTER)) {
2111 if ((range1->type == XML_REGEXP_ANYSPACE) &&
2112 (range2->type == XML_REGEXP_NOTSPACE))
2113 ret = 0;
2114 else if ((range1->type == XML_REGEXP_INITNAME) &&
2115 (range2->type == XML_REGEXP_NOTINITNAME))
2116 ret = 0;
2117 else if ((range1->type == XML_REGEXP_NAMECHAR) &&
2118 (range2->type == XML_REGEXP_NOTNAMECHAR))
2119 ret = 0;
2120 else if ((range1->type == XML_REGEXP_DECIMAL) &&
2121 (range2->type == XML_REGEXP_NOTDECIMAL))
2122 ret = 0;
2123 else if ((range1->type == XML_REGEXP_REALCHAR) &&
2124 (range2->type == XML_REGEXP_NOTREALCHAR))
2125 ret = 0;
2126 else {
2127 /* same thing to limit complexity */
2128 return(1);
2129 }
2130 } else {
2131 ret = 0;
2132 /* range1->type < range2->type here */
2133 switch (range1->type) {
2134 case XML_REGEXP_LETTER:
2135 /* all disjoint except in the subgroups */
2136 if ((range2->type == XML_REGEXP_LETTER_UPPERCASE) ||
2137 (range2->type == XML_REGEXP_LETTER_LOWERCASE) ||
2138 (range2->type == XML_REGEXP_LETTER_TITLECASE) ||
2139 (range2->type == XML_REGEXP_LETTER_MODIFIER) ||
2140 (range2->type == XML_REGEXP_LETTER_OTHERS))
2141 ret = 1;
2142 break;
2143 case XML_REGEXP_MARK:
2144 if ((range2->type == XML_REGEXP_MARK_NONSPACING) ||
2145 (range2->type == XML_REGEXP_MARK_SPACECOMBINING) ||
2146 (range2->type == XML_REGEXP_MARK_ENCLOSING))
2147 ret = 1;
2148 break;
2149 case XML_REGEXP_NUMBER:
2150 if ((range2->type == XML_REGEXP_NUMBER_DECIMAL) ||
2151 (range2->type == XML_REGEXP_NUMBER_LETTER) ||
2152 (range2->type == XML_REGEXP_NUMBER_OTHERS))
2153 ret = 1;
2154 break;
2155 case XML_REGEXP_PUNCT:
2156 if ((range2->type == XML_REGEXP_PUNCT_CONNECTOR) ||
2157 (range2->type == XML_REGEXP_PUNCT_DASH) ||
2158 (range2->type == XML_REGEXP_PUNCT_OPEN) ||
2159 (range2->type == XML_REGEXP_PUNCT_CLOSE) ||
2160 (range2->type == XML_REGEXP_PUNCT_INITQUOTE) ||
2161 (range2->type == XML_REGEXP_PUNCT_FINQUOTE) ||
2162 (range2->type == XML_REGEXP_PUNCT_OTHERS))
2163 ret = 1;
2164 break;
2165 case XML_REGEXP_SEPAR:
2166 if ((range2->type == XML_REGEXP_SEPAR_SPACE) ||
2167 (range2->type == XML_REGEXP_SEPAR_LINE) ||
2168 (range2->type == XML_REGEXP_SEPAR_PARA))
2169 ret = 1;
2170 break;
2171 case XML_REGEXP_SYMBOL:
2172 if ((range2->type == XML_REGEXP_SYMBOL_MATH) ||
2173 (range2->type == XML_REGEXP_SYMBOL_CURRENCY) ||
2174 (range2->type == XML_REGEXP_SYMBOL_MODIFIER) ||
2175 (range2->type == XML_REGEXP_SYMBOL_OTHERS))
2176 ret = 1;
2177 break;
2178 case XML_REGEXP_OTHER:
2179 if ((range2->type == XML_REGEXP_OTHER_CONTROL) ||
2180 (range2->type == XML_REGEXP_OTHER_FORMAT) ||
2181 (range2->type == XML_REGEXP_OTHER_PRIVATE))
2182 ret = 1;
2183 break;
2184 default:
2185 if ((range2->type >= XML_REGEXP_LETTER) &&
2186 (range2->type < XML_REGEXP_BLOCK_NAME))
2187 ret = 0;
2188 else {
2189 /* safety net ! */
2190 return(1);
2191 }
2192 }
2193 }
2194 if (((range1->neg == 0) && (range2->neg != 0)) ||
2195 ((range1->neg != 0) && (range2->neg == 0)))
2196 ret = !ret;
2197 return(1);
2198}
2199
2200/**
2201 * xmlFACompareAtomTypes:
2202 * @type1: an atom type
2203 * @type2: an atom type
2204 *
2205 * Compares two atoms type to check whether they intersect in some ways,
2206 * this is used by xmlFACompareAtoms only
2207 *
2208 * Returns 1 if they may intersect and 0 otherwise
2209 */
2210static int
2211xmlFACompareAtomTypes(xmlRegAtomType type1, xmlRegAtomType type2) {
2212 if ((type1 == XML_REGEXP_EPSILON) ||
2213 (type1 == XML_REGEXP_CHARVAL) ||
2214 (type1 == XML_REGEXP_RANGES) ||
2215 (type1 == XML_REGEXP_SUBREG) ||
2216 (type1 == XML_REGEXP_STRING) ||
2217 (type1 == XML_REGEXP_ANYCHAR))
2218 return(1);
2219 if ((type2 == XML_REGEXP_EPSILON) ||
2220 (type2 == XML_REGEXP_CHARVAL) ||
2221 (type2 == XML_REGEXP_RANGES) ||
2222 (type2 == XML_REGEXP_SUBREG) ||
2223 (type2 == XML_REGEXP_STRING) ||
2224 (type2 == XML_REGEXP_ANYCHAR))
2225 return(1);
2226
2227 if (type1 == type2) return(1);
2228
2229 /* simplify subsequent compares by making sure type1 < type2 */
2230 if (type1 > type2) {
2231 xmlRegAtomType tmp = type1;
2232 type1 = type2;
2233 type2 = tmp;
2234 }
2235 switch (type1) {
2236 case XML_REGEXP_ANYSPACE: /* \s */
2237 /* can't be a letter, number, mark, pontuation, symbol */
2238 if ((type2 == XML_REGEXP_NOTSPACE) ||
2239 ((type2 >= XML_REGEXP_LETTER) &&
2240 (type2 <= XML_REGEXP_LETTER_OTHERS)) ||
2241 ((type2 >= XML_REGEXP_NUMBER) &&
2242 (type2 <= XML_REGEXP_NUMBER_OTHERS)) ||
2243 ((type2 >= XML_REGEXP_MARK) &&
2244 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2245 ((type2 >= XML_REGEXP_PUNCT) &&
2246 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2247 ((type2 >= XML_REGEXP_SYMBOL) &&
2248 (type2 <= XML_REGEXP_SYMBOL_OTHERS))
2249 ) return(0);
2250 break;
2251 case XML_REGEXP_NOTSPACE: /* \S */
2252 break;
2253 case XML_REGEXP_INITNAME: /* \l */
2254 /* can't be a number, mark, separator, pontuation, symbol or other */
2255 if ((type2 == XML_REGEXP_NOTINITNAME) ||
2256 ((type2 >= XML_REGEXP_NUMBER) &&
2257 (type2 <= XML_REGEXP_NUMBER_OTHERS)) ||
2258 ((type2 >= XML_REGEXP_MARK) &&
2259 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2260 ((type2 >= XML_REGEXP_SEPAR) &&
2261 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2262 ((type2 >= XML_REGEXP_PUNCT) &&
2263 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2264 ((type2 >= XML_REGEXP_SYMBOL) &&
2265 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2266 ((type2 >= XML_REGEXP_OTHER) &&
2267 (type2 <= XML_REGEXP_OTHER_NA))
2268 ) return(0);
2269 break;
2270 case XML_REGEXP_NOTINITNAME: /* \L */
2271 break;
2272 case XML_REGEXP_NAMECHAR: /* \c */
2273 /* can't be a mark, separator, pontuation, symbol or other */
2274 if ((type2 == XML_REGEXP_NOTNAMECHAR) ||
2275 ((type2 >= XML_REGEXP_MARK) &&
2276 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2277 ((type2 >= XML_REGEXP_PUNCT) &&
2278 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2279 ((type2 >= XML_REGEXP_SEPAR) &&
2280 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2281 ((type2 >= XML_REGEXP_SYMBOL) &&
2282 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2283 ((type2 >= XML_REGEXP_OTHER) &&
2284 (type2 <= XML_REGEXP_OTHER_NA))
2285 ) return(0);
2286 break;
2287 case XML_REGEXP_NOTNAMECHAR: /* \C */
2288 break;
2289 case XML_REGEXP_DECIMAL: /* \d */
2290 /* can't be a letter, mark, separator, pontuation, symbol or other */
2291 if ((type2 == XML_REGEXP_NOTDECIMAL) ||
2292 (type2 == XML_REGEXP_REALCHAR) ||
2293 ((type2 >= XML_REGEXP_LETTER) &&
2294 (type2 <= XML_REGEXP_LETTER_OTHERS)) ||
2295 ((type2 >= XML_REGEXP_MARK) &&
2296 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2297 ((type2 >= XML_REGEXP_PUNCT) &&
2298 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2299 ((type2 >= XML_REGEXP_SEPAR) &&
2300 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2301 ((type2 >= XML_REGEXP_SYMBOL) &&
2302 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2303 ((type2 >= XML_REGEXP_OTHER) &&
2304 (type2 <= XML_REGEXP_OTHER_NA))
2305 )return(0);
2306 break;
2307 case XML_REGEXP_NOTDECIMAL: /* \D */
2308 break;
2309 case XML_REGEXP_REALCHAR: /* \w */
2310 /* can't be a mark, separator, pontuation, symbol or other */
2311 if ((type2 == XML_REGEXP_NOTDECIMAL) ||
2312 ((type2 >= XML_REGEXP_MARK) &&
2313 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2314 ((type2 >= XML_REGEXP_PUNCT) &&
2315 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2316 ((type2 >= XML_REGEXP_SEPAR) &&
2317 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2318 ((type2 >= XML_REGEXP_SYMBOL) &&
2319 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2320 ((type2 >= XML_REGEXP_OTHER) &&
2321 (type2 <= XML_REGEXP_OTHER_NA))
2322 )return(0);
2323 break;
2324 case XML_REGEXP_NOTREALCHAR: /* \W */
2325 break;
2326 /*
2327 * at that point we know both type 1 and type2 are from
2328 * character categories are ordered and are different,
2329 * it becomes simple because this is a partition
2330 */
2331 case XML_REGEXP_LETTER:
2332 if (type2 <= XML_REGEXP_LETTER_OTHERS)
2333 return(1);
2334 return(0);
2335 case XML_REGEXP_LETTER_UPPERCASE:
2336 case XML_REGEXP_LETTER_LOWERCASE:
2337 case XML_REGEXP_LETTER_TITLECASE:
2338 case XML_REGEXP_LETTER_MODIFIER:
2339 case XML_REGEXP_LETTER_OTHERS:
2340 return(0);
2341 case XML_REGEXP_MARK:
2342 if (type2 <= XML_REGEXP_MARK_ENCLOSING)
2343 return(1);
2344 return(0);
2345 case XML_REGEXP_MARK_NONSPACING:
2346 case XML_REGEXP_MARK_SPACECOMBINING:
2347 case XML_REGEXP_MARK_ENCLOSING:
2348 return(0);
2349 case XML_REGEXP_NUMBER:
2350 if (type2 <= XML_REGEXP_NUMBER_OTHERS)
2351 return(1);
2352 return(0);
2353 case XML_REGEXP_NUMBER_DECIMAL:
2354 case XML_REGEXP_NUMBER_LETTER:
2355 case XML_REGEXP_NUMBER_OTHERS:
2356 return(0);
2357 case XML_REGEXP_PUNCT:
2358 if (type2 <= XML_REGEXP_PUNCT_OTHERS)
2359 return(1);
2360 return(0);
2361 case XML_REGEXP_PUNCT_CONNECTOR:
2362 case XML_REGEXP_PUNCT_DASH:
2363 case XML_REGEXP_PUNCT_OPEN:
2364 case XML_REGEXP_PUNCT_CLOSE:
2365 case XML_REGEXP_PUNCT_INITQUOTE:
2366 case XML_REGEXP_PUNCT_FINQUOTE:
2367 case XML_REGEXP_PUNCT_OTHERS:
2368 return(0);
2369 case XML_REGEXP_SEPAR:
2370 if (type2 <= XML_REGEXP_SEPAR_PARA)
2371 return(1);
2372 return(0);
2373 case XML_REGEXP_SEPAR_SPACE:
2374 case XML_REGEXP_SEPAR_LINE:
2375 case XML_REGEXP_SEPAR_PARA:
2376 return(0);
2377 case XML_REGEXP_SYMBOL:
2378 if (type2 <= XML_REGEXP_SYMBOL_OTHERS)
2379 return(1);
2380 return(0);
2381 case XML_REGEXP_SYMBOL_MATH:
2382 case XML_REGEXP_SYMBOL_CURRENCY:
2383 case XML_REGEXP_SYMBOL_MODIFIER:
2384 case XML_REGEXP_SYMBOL_OTHERS:
2385 return(0);
2386 case XML_REGEXP_OTHER:
2387 if (type2 <= XML_REGEXP_OTHER_NA)
2388 return(1);
2389 return(0);
2390 case XML_REGEXP_OTHER_CONTROL:
2391 case XML_REGEXP_OTHER_FORMAT:
2392 case XML_REGEXP_OTHER_PRIVATE:
2393 case XML_REGEXP_OTHER_NA:
2394 return(0);
2395 default:
2396 break;
2397 }
2398 return(1);
2399}
2400
2401/**
2402 * xmlFAEqualAtoms:
2403 * @atom1: an atom
2404 * @atom2: an atom
2405 *
2406 * Compares two atoms to check whether they are the same exactly
2407 * this is used to remove equivalent transitions
2408 *
2409 * Returns 1 if same and 0 otherwise
2410 */
2411static int
2412xmlFAEqualAtoms(xmlRegAtomPtr atom1, xmlRegAtomPtr atom2) {
2413 int ret = 0;
2414
2415 if (atom1 == atom2)
2416 return(1);
2417 if ((atom1 == NULL) || (atom2 == NULL))
2418 return(0);
2419
2420 if (atom1->type != atom2->type)
2421 return(0);
2422 switch (atom1->type) {
2423 case XML_REGEXP_EPSILON:
2424 ret = 0;
2425 break;
2426 case XML_REGEXP_STRING:
2427 ret = xmlStrEqual((xmlChar *)atom1->valuep,
2428 (xmlChar *)atom2->valuep);
2429 break;
2430 case XML_REGEXP_CHARVAL:
2431 ret = (atom1->codepoint == atom2->codepoint);
2432 break;
2433 case XML_REGEXP_RANGES:
2434 /* too hard to do in the general case */
2435 ret = 0;
2436 default:
2437 break;
2438 }
2439 return(ret);
2440}
2441
2442/**
2443 * xmlFACompareAtoms:
2444 * @atom1: an atom
2445 * @atom2: an atom
2446 *
2447 * Compares two atoms to check whether they intersect in some ways,
2448 * this is used by xmlFAComputesDeterminism and xmlFARecurseDeterminism only
2449 *
2450 * Returns 1 if yes and 0 otherwise
2451 */
2452static int
2453xmlFACompareAtoms(xmlRegAtomPtr atom1, xmlRegAtomPtr atom2) {
2454 int ret = 1;
2455
2456 if (atom1 == atom2)
2457 return(1);
2458 if ((atom1 == NULL) || (atom2 == NULL))
2459 return(0);
2460
2461 if ((atom1->type == XML_REGEXP_ANYCHAR) ||
2462 (atom2->type == XML_REGEXP_ANYCHAR))
2463 return(1);
2464
2465 if (atom1->type > atom2->type) {
2466 xmlRegAtomPtr tmp;
2467 tmp = atom1;
2468 atom1 = atom2;
2469 atom2 = tmp;
2470 }
2471 if (atom1->type != atom2->type) {
2472 ret = xmlFACompareAtomTypes(atom1->type, atom2->type);
2473 /* if they can't intersect at the type level break now */
2474 if (ret == 0)
2475 return(0);
2476 }
2477 switch (atom1->type) {
2478 case XML_REGEXP_STRING:
2479 ret = xmlRegStrEqualWildcard((xmlChar *)atom1->valuep,
2480 (xmlChar *)atom2->valuep);
2481 break;
2482 case XML_REGEXP_EPSILON:
2483 goto not_determinist;
2484 case XML_REGEXP_CHARVAL:
2485 if (atom2->type == XML_REGEXP_CHARVAL) {
2486 ret = (atom1->codepoint == atom2->codepoint);
2487 } else {
2488 ret = xmlRegCheckCharacter(atom2, atom1->codepoint);
2489 if (ret < 0)
2490 ret = 1;
2491 }
2492 break;
2493 case XML_REGEXP_RANGES:
2494 if (atom2->type == XML_REGEXP_RANGES) {
2495 int i, j, res;
2496 xmlRegRangePtr r1, r2;
2497
2498 /*
2499 * need to check that none of the ranges eventually matches
2500 */
2501 for (i = 0;i < atom1->nbRanges;i++) {
2502 for (j = 0;j < atom2->nbRanges;j++) {
2503 r1 = atom1->ranges[i];
2504 r2 = atom2->ranges[j];
2505 res = xmlFACompareRanges(r1, r2);
2506 if (res == 1) {
2507 ret = 1;
2508 goto done;
2509 }
2510 }
2511 }
2512 ret = 0;
2513 }
2514 break;
2515 default:
2516 goto not_determinist;
2517 }
2518done:
2519 if (atom1->neg != atom2->neg) {
2520 ret = !ret;
2521 }
2522 if (ret == 0)
2523 return(0);
2524not_determinist:
2525 return(1);
2526}
2527
2528/**
2529 * xmlFARecurseDeterminism:
2530 * @ctxt: a regexp parser context
2531 *
2532 * Check whether the associated regexp is determinist,
2533 * should be called after xmlFAEliminateEpsilonTransitions()
2534 *
2535 */
2536static int
2537xmlFARecurseDeterminism(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
2538 int to, xmlRegAtomPtr atom) {
2539 int ret = 1;
2540 int res;
2541 int transnr, nbTrans;
2542 xmlRegTransPtr t1;
2543
2544 if (state == NULL)
2545 return(ret);
2546 /*
2547 * don't recurse on transitions potentially added in the course of
2548 * the elimination.
2549 */
2550 nbTrans = state->nbTrans;
2551 for (transnr = 0;transnr < nbTrans;transnr++) {
2552 t1 = &(state->trans[transnr]);
2553 /*
2554 * check transitions conflicting with the one looked at
2555 */
2556 if (t1->atom == NULL) {
2557 if (t1->to < 0)
2558 continue;
2559 res = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
2560 to, atom);
2561 if (res == 0) {
2562 ret = 0;
2563 /* t1->nd = 1; */
2564 }
2565 continue;
2566 }
2567 if (t1->to != to)
2568 continue;
2569 if (xmlFACompareAtoms(t1->atom, atom)) {
2570 ret = 0;
2571 /* mark the transition as non-deterministic */
2572 t1->nd = 1;
2573 }
2574 }
2575 return(ret);
2576}
2577
2578/**
2579 * xmlFAComputesDeterminism:
2580 * @ctxt: a regexp parser context
2581 *
2582 * Check whether the associated regexp is determinist,
2583 * should be called after xmlFAEliminateEpsilonTransitions()
2584 *
2585 */
2586static int
2587xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt) {
2588 int statenr, transnr;
2589 xmlRegStatePtr state;
2590 xmlRegTransPtr t1, t2, last;
2591 int i;
2592 int ret = 1;
2593
2594#ifdef DEBUG_REGEXP_GRAPH
2595 printf("xmlFAComputesDeterminism\n");
2596 xmlRegPrintCtxt(stdout, ctxt);
2597#endif
2598 if (ctxt->determinist != -1)
2599 return(ctxt->determinist);
2600
2601 /*
2602 * First cleanup the automata removing cancelled transitions
2603 */
2604 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2605 state = ctxt->states[statenr];
2606 if (state == NULL)
2607 continue;
2608 if (state->nbTrans < 2)
2609 continue;
2610 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2611 t1 = &(state->trans[transnr]);
2612 /*
2613 * Determinism checks in case of counted or all transitions
2614 * will have to be handled separately
2615 */
2616 if (t1->atom == NULL) {
2617 /* t1->nd = 1; */
2618 continue;
2619 }
2620 if (t1->to == -1) /* eliminated */
2621 continue;
2622 for (i = 0;i < transnr;i++) {
2623 t2 = &(state->trans[i]);
2624 if (t2->to == -1) /* eliminated */
2625 continue;
2626 if (t2->atom != NULL) {
2627 if (t1->to == t2->to) {
2628 if (xmlFAEqualAtoms(t1->atom, t2->atom))
2629 t2->to = -1; /* eliminated */
2630 }
2631 }
2632 }
2633 }
2634 }
2635
2636 /*
2637 * Check for all states that there aren't 2 transitions
2638 * with the same atom and a different target.
2639 */
2640 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2641 state = ctxt->states[statenr];
2642 if (state == NULL)
2643 continue;
2644 if (state->nbTrans < 2)
2645 continue;
2646 last = NULL;
2647 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2648 t1 = &(state->trans[transnr]);
2649 /*
2650 * Determinism checks in case of counted or all transitions
2651 * will have to be handled separately
2652 */
2653 if (t1->atom == NULL) {
2654 continue;
2655 }
2656 if (t1->to == -1) /* eliminated */
2657 continue;
2658 for (i = 0;i < transnr;i++) {
2659 t2 = &(state->trans[i]);
2660 if (t2->to == -1) /* eliminated */
2661 continue;
2662 if (t2->atom != NULL) {
2663 /* not determinist ! */
2664 if (xmlFACompareAtoms(t1->atom, t2->atom)) {
2665 ret = 0;
2666 /* mark the transitions as non-deterministic ones */
2667 t1->nd = 1;
2668 t2->nd = 1;
2669 last = t1;
2670 }
2671 } else if (t1->to != -1) {
2672 /*
2673 * do the closure in case of remaining specific
2674 * epsilon transitions like choices or all
2675 */
2676 ret = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
2677 t2->to, t2->atom);
2678 /* don't shortcut the computation so all non deterministic
2679 transition get marked down
2680 if (ret == 0)
2681 return(0);
2682 */
2683 if (ret == 0) {
2684 t1->nd = 1;
2685 /* t2->nd = 1; */
2686 last = t1;
2687 }
2688 }
2689 }
2690 /* don't shortcut the computation so all non deterministic
2691 transition get marked down
2692 if (ret == 0)
2693 break; */
2694 }
2695
2696 /*
2697 * mark specifically the last non-deterministic transition
2698 * from a state since there is no need to set-up rollback
2699 * from it
2700 */
2701 if (last != NULL) {
2702 last->nd = 2;
2703 }
2704
2705 /* don't shortcut the computation so all non deterministic
2706 transition get marked down
2707 if (ret == 0)
2708 break; */
2709 }
2710
2711 ctxt->determinist = ret;
2712 return(ret);
2713}
2714
2715/************************************************************************
2716 * *
2717 * Routines to check input against transition atoms *
2718 * *
2719 ************************************************************************/
2720
2721static int
2722xmlRegCheckCharacterRange(xmlRegAtomType type, int codepoint, int neg,
2723 int start, int end, const xmlChar *blockName) {
2724 int ret = 0;
2725
2726 switch (type) {
2727 case XML_REGEXP_STRING:
2728 case XML_REGEXP_SUBREG:
2729 case XML_REGEXP_RANGES:
2730 case XML_REGEXP_EPSILON:
2731 return(-1);
2732 case XML_REGEXP_ANYCHAR:
2733 ret = ((codepoint != '\n') && (codepoint != '\r'));
2734 break;
2735 case XML_REGEXP_CHARVAL:
2736 ret = ((codepoint >= start) && (codepoint <= end));
2737 break;
2738 case XML_REGEXP_NOTSPACE:
2739 neg = !neg;
2740 case XML_REGEXP_ANYSPACE:
2741 ret = ((codepoint == '\n') || (codepoint == '\r') ||
2742 (codepoint == '\t') || (codepoint == ' '));
2743 break;
2744 case XML_REGEXP_NOTINITNAME:
2745 neg = !neg;
2746 case XML_REGEXP_INITNAME:
2747 ret = (IS_LETTER(codepoint) ||
2748 (codepoint == '_') || (codepoint == ':'));
2749 break;
2750 case XML_REGEXP_NOTNAMECHAR:
2751 neg = !neg;
2752 case XML_REGEXP_NAMECHAR:
2753 ret = (IS_LETTER(codepoint) || IS_DIGIT(codepoint) ||
2754 (codepoint == '.') || (codepoint == '-') ||
2755 (codepoint == '_') || (codepoint == ':') ||
2756 IS_COMBINING(codepoint) || IS_EXTENDER(codepoint));
2757 break;
2758 case XML_REGEXP_NOTDECIMAL:
2759 neg = !neg;
2760 case XML_REGEXP_DECIMAL:
2761 ret = xmlUCSIsCatNd(codepoint);
2762 break;
2763 case XML_REGEXP_REALCHAR:
2764 neg = !neg;
2765 case XML_REGEXP_NOTREALCHAR:
2766 ret = xmlUCSIsCatP(codepoint);
2767 if (ret == 0)
2768 ret = xmlUCSIsCatZ(codepoint);
2769 if (ret == 0)
2770 ret = xmlUCSIsCatC(codepoint);
2771 break;
2772 case XML_REGEXP_LETTER:
2773 ret = xmlUCSIsCatL(codepoint);
2774 break;
2775 case XML_REGEXP_LETTER_UPPERCASE:
2776 ret = xmlUCSIsCatLu(codepoint);
2777 break;
2778 case XML_REGEXP_LETTER_LOWERCASE:
2779 ret = xmlUCSIsCatLl(codepoint);
2780 break;
2781 case XML_REGEXP_LETTER_TITLECASE:
2782 ret = xmlUCSIsCatLt(codepoint);
2783 break;
2784 case XML_REGEXP_LETTER_MODIFIER:
2785 ret = xmlUCSIsCatLm(codepoint);
2786 break;
2787 case XML_REGEXP_LETTER_OTHERS:
2788 ret = xmlUCSIsCatLo(codepoint);
2789 break;
2790 case XML_REGEXP_MARK:
2791 ret = xmlUCSIsCatM(codepoint);
2792 break;
2793 case XML_REGEXP_MARK_NONSPACING:
2794 ret = xmlUCSIsCatMn(codepoint);
2795 break;
2796 case XML_REGEXP_MARK_SPACECOMBINING:
2797 ret = xmlUCSIsCatMc(codepoint);
2798 break;
2799 case XML_REGEXP_MARK_ENCLOSING:
2800 ret = xmlUCSIsCatMe(codepoint);
2801 break;
2802 case XML_REGEXP_NUMBER:
2803 ret = xmlUCSIsCatN(codepoint);
2804 break;
2805 case XML_REGEXP_NUMBER_DECIMAL:
2806 ret = xmlUCSIsCatNd(codepoint);
2807 break;
2808 case XML_REGEXP_NUMBER_LETTER:
2809 ret = xmlUCSIsCatNl(codepoint);
2810 break;
2811 case XML_REGEXP_NUMBER_OTHERS:
2812 ret = xmlUCSIsCatNo(codepoint);
2813 break;
2814 case XML_REGEXP_PUNCT:
2815 ret = xmlUCSIsCatP(codepoint);
2816 break;
2817 case XML_REGEXP_PUNCT_CONNECTOR:
2818 ret = xmlUCSIsCatPc(codepoint);
2819 break;
2820 case XML_REGEXP_PUNCT_DASH:
2821 ret = xmlUCSIsCatPd(codepoint);
2822 break;
2823 case XML_REGEXP_PUNCT_OPEN:
2824 ret = xmlUCSIsCatPs(codepoint);
2825 break;
2826 case XML_REGEXP_PUNCT_CLOSE:
2827 ret = xmlUCSIsCatPe(codepoint);
2828 break;
2829 case XML_REGEXP_PUNCT_INITQUOTE:
2830 ret = xmlUCSIsCatPi(codepoint);
2831 break;
2832 case XML_REGEXP_PUNCT_FINQUOTE:
2833 ret = xmlUCSIsCatPf(codepoint);
2834 break;
2835 case XML_REGEXP_PUNCT_OTHERS:
2836 ret = xmlUCSIsCatPo(codepoint);
2837 break;
2838 case XML_REGEXP_SEPAR:
2839 ret = xmlUCSIsCatZ(codepoint);
2840 break;
2841 case XML_REGEXP_SEPAR_SPACE:
2842 ret = xmlUCSIsCatZs(codepoint);
2843 break;
2844 case XML_REGEXP_SEPAR_LINE:
2845 ret = xmlUCSIsCatZl(codepoint);
2846 break;
2847 case XML_REGEXP_SEPAR_PARA:
2848 ret = xmlUCSIsCatZp(codepoint);
2849 break;
2850 case XML_REGEXP_SYMBOL:
2851 ret = xmlUCSIsCatS(codepoint);
2852 break;
2853 case XML_REGEXP_SYMBOL_MATH:
2854 ret = xmlUCSIsCatSm(codepoint);
2855 break;
2856 case XML_REGEXP_SYMBOL_CURRENCY:
2857 ret = xmlUCSIsCatSc(codepoint);
2858 break;
2859 case XML_REGEXP_SYMBOL_MODIFIER:
2860 ret = xmlUCSIsCatSk(codepoint);
2861 break;
2862 case XML_REGEXP_SYMBOL_OTHERS:
2863 ret = xmlUCSIsCatSo(codepoint);
2864 break;
2865 case XML_REGEXP_OTHER:
2866 ret = xmlUCSIsCatC(codepoint);
2867 break;
2868 case XML_REGEXP_OTHER_CONTROL:
2869 ret = xmlUCSIsCatCc(codepoint);
2870 break;
2871 case XML_REGEXP_OTHER_FORMAT:
2872 ret = xmlUCSIsCatCf(codepoint);
2873 break;
2874 case XML_REGEXP_OTHER_PRIVATE:
2875 ret = xmlUCSIsCatCo(codepoint);
2876 break;
2877 case XML_REGEXP_OTHER_NA:
2878 /* ret = xmlUCSIsCatCn(codepoint); */
2879 /* Seems it doesn't exist anymore in recent Unicode releases */
2880 ret = 0;
2881 break;
2882 case XML_REGEXP_BLOCK_NAME:
2883 ret = xmlUCSIsBlock(codepoint, (const char *) blockName);
2884 break;
2885 }
2886 if (neg)
2887 return(!ret);
2888 return(ret);
2889}
2890
2891static int
2892xmlRegCheckCharacter(xmlRegAtomPtr atom, int codepoint) {
2893 int i, ret = 0;
2894 xmlRegRangePtr range;
2895
2896 if ((atom == NULL) || (!IS_CHAR(codepoint)))
2897 return(-1);
2898
2899 switch (atom->type) {
2900 case XML_REGEXP_SUBREG:
2901 case XML_REGEXP_EPSILON:
2902 return(-1);
2903 case XML_REGEXP_CHARVAL:
2904 return(codepoint == atom->codepoint);
2905 case XML_REGEXP_RANGES: {
2906 int accept = 0;
2907
2908 for (i = 0;i < atom->nbRanges;i++) {
2909 range = atom->ranges[i];
2910 if (range->neg == 2) {
2911 ret = xmlRegCheckCharacterRange(range->type, codepoint,
2912 0, range->start, range->end,
2913 range->blockName);
2914 if (ret != 0)
2915 return(0); /* excluded char */
2916 } else if (range->neg) {
2917 ret = xmlRegCheckCharacterRange(range->type, codepoint,
2918 0, range->start, range->end,
2919 range->blockName);
2920 if (ret == 0)
2921 accept = 1;
2922 else
2923 return(0);
2924 } else {
2925 ret = xmlRegCheckCharacterRange(range->type, codepoint,
2926 0, range->start, range->end,
2927 range->blockName);
2928 if (ret != 0)
2929 accept = 1; /* might still be excluded */
2930 }
2931 }
2932 return(accept);
2933 }
2934 case XML_REGEXP_STRING:
2935 printf("TODO: XML_REGEXP_STRING\n");
2936 return(-1);
2937 case XML_REGEXP_ANYCHAR:
2938 case XML_REGEXP_ANYSPACE:
2939 case XML_REGEXP_NOTSPACE:
2940 case XML_REGEXP_INITNAME:
2941 case XML_REGEXP_NOTINITNAME:
2942 case XML_REGEXP_NAMECHAR:
2943 case XML_REGEXP_NOTNAMECHAR:
2944 case XML_REGEXP_DECIMAL:
2945 case XML_REGEXP_NOTDECIMAL:
2946 case XML_REGEXP_REALCHAR:
2947 case XML_REGEXP_NOTREALCHAR:
2948 case XML_REGEXP_LETTER:
2949 case XML_REGEXP_LETTER_UPPERCASE:
2950 case XML_REGEXP_LETTER_LOWERCASE:
2951 case XML_REGEXP_LETTER_TITLECASE:
2952 case XML_REGEXP_LETTER_MODIFIER:
2953 case XML_REGEXP_LETTER_OTHERS:
2954 case XML_REGEXP_MARK:
2955 case XML_REGEXP_MARK_NONSPACING:
2956 case XML_REGEXP_MARK_SPACECOMBINING:
2957 case XML_REGEXP_MARK_ENCLOSING:
2958 case XML_REGEXP_NUMBER:
2959 case XML_REGEXP_NUMBER_DECIMAL:
2960 case XML_REGEXP_NUMBER_LETTER:
2961 case XML_REGEXP_NUMBER_OTHERS:
2962 case XML_REGEXP_PUNCT:
2963 case XML_REGEXP_PUNCT_CONNECTOR:
2964 case XML_REGEXP_PUNCT_DASH:
2965 case XML_REGEXP_PUNCT_OPEN:
2966 case XML_REGEXP_PUNCT_CLOSE:
2967 case XML_REGEXP_PUNCT_INITQUOTE:
2968 case XML_REGEXP_PUNCT_FINQUOTE:
2969 case XML_REGEXP_PUNCT_OTHERS:
2970 case XML_REGEXP_SEPAR:
2971 case XML_REGEXP_SEPAR_SPACE:
2972 case XML_REGEXP_SEPAR_LINE:
2973 case XML_REGEXP_SEPAR_PARA:
2974 case XML_REGEXP_SYMBOL:
2975 case XML_REGEXP_SYMBOL_MATH:
2976 case XML_REGEXP_SYMBOL_CURRENCY:
2977 case XML_REGEXP_SYMBOL_MODIFIER:
2978 case XML_REGEXP_SYMBOL_OTHERS:
2979 case XML_REGEXP_OTHER:
2980 case XML_REGEXP_OTHER_CONTROL:
2981 case XML_REGEXP_OTHER_FORMAT:
2982 case XML_REGEXP_OTHER_PRIVATE:
2983 case XML_REGEXP_OTHER_NA:
2984 case XML_REGEXP_BLOCK_NAME:
2985 ret = xmlRegCheckCharacterRange(atom->type, codepoint, 0, 0, 0,
2986 (const xmlChar *)atom->valuep);
2987 if (atom->neg)
2988 ret = !ret;
2989 break;
2990 }
2991 return(ret);
2992}
2993
2994/************************************************************************
2995 * *
2996 * Saving and restoring state of an execution context *
2997 * *
2998 ************************************************************************/
2999
3000#ifdef DEBUG_REGEXP_EXEC
3001static void
3002xmlFARegDebugExec(xmlRegExecCtxtPtr exec) {
3003 printf("state: %d:%d:idx %d", exec->state->no, exec->transno, exec->index);
3004 if (exec->inputStack != NULL) {
3005 int i;
3006 printf(": ");
3007 for (i = 0;(i < 3) && (i < exec->inputStackNr);i++)
3008 printf("%s ", (const char *)
3009 exec->inputStack[exec->inputStackNr - (i + 1)].value);
3010 } else {
3011 printf(": %s", &(exec->inputString[exec->index]));
3012 }
3013 printf("\n");
3014}
3015#endif
3016
3017static void
3018xmlFARegExecSave(xmlRegExecCtxtPtr exec) {
3019#ifdef DEBUG_REGEXP_EXEC
3020 printf("saving ");
3021 exec->transno++;
3022 xmlFARegDebugExec(exec);
3023 exec->transno--;
3024#endif
3025#ifdef MAX_PUSH
3026 if (exec->nbPush > MAX_PUSH) {
3027 return;
3028 }
3029 exec->nbPush++;
3030#endif
3031
3032 if (exec->maxRollbacks == 0) {
3033 exec->maxRollbacks = 4;
3034 exec->rollbacks = (xmlRegExecRollback *) xmlMalloc(exec->maxRollbacks *
3035 sizeof(xmlRegExecRollback));
3036 if (exec->rollbacks == NULL) {
3037 xmlRegexpErrMemory(NULL, "saving regexp");
3038 exec->maxRollbacks = 0;
3039 return;
3040 }
3041 memset(exec->rollbacks, 0,
3042 exec->maxRollbacks * sizeof(xmlRegExecRollback));
3043 } else if (exec->nbRollbacks >= exec->maxRollbacks) {
3044 xmlRegExecRollback *tmp;
3045 int len = exec->maxRollbacks;
3046
3047 exec->maxRollbacks *= 2;
3048 tmp = (xmlRegExecRollback *) xmlRealloc(exec->rollbacks,
3049 exec->maxRollbacks * sizeof(xmlRegExecRollback));
3050 if (tmp == NULL) {
3051 xmlRegexpErrMemory(NULL, "saving regexp");
3052 exec->maxRollbacks /= 2;
3053 return;
3054 }
3055 exec->rollbacks = tmp;
3056 tmp = &exec->rollbacks[len];
3057 memset(tmp, 0, (exec->maxRollbacks - len) * sizeof(xmlRegExecRollback));
3058 }
3059 exec->rollbacks[exec->nbRollbacks].state = exec->state;
3060 exec->rollbacks[exec->nbRollbacks].index = exec->index;
3061 exec->rollbacks[exec->nbRollbacks].nextbranch = exec->transno + 1;
3062 if (exec->comp->nbCounters > 0) {
3063 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3064 exec->rollbacks[exec->nbRollbacks].counts = (int *)
3065 xmlMalloc(exec->comp->nbCounters * sizeof(int));
3066 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3067 xmlRegexpErrMemory(NULL, "saving regexp");
3068 exec->status = -5;
3069 return;
3070 }
3071 }
3072 memcpy(exec->rollbacks[exec->nbRollbacks].counts, exec->counts,
3073 exec->comp->nbCounters * sizeof(int));
3074 }
3075 exec->nbRollbacks++;
3076}
3077
3078static void
3079xmlFARegExecRollBack(xmlRegExecCtxtPtr exec) {
3080 if (exec->nbRollbacks <= 0) {
3081 exec->status = -1;
3082#ifdef DEBUG_REGEXP_EXEC
3083 printf("rollback failed on empty stack\n");
3084#endif
3085 return;
3086 }
3087 exec->nbRollbacks--;
3088 exec->state = exec->rollbacks[exec->nbRollbacks].state;
3089 exec->index = exec->rollbacks[exec->nbRollbacks].index;
3090 exec->transno = exec->rollbacks[exec->nbRollbacks].nextbranch;
3091 if (exec->comp->nbCounters > 0) {
3092 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3093 fprintf(stderr, "exec save: allocation failed");
3094 exec->status = -6;
3095 return;
3096 }
3097 memcpy(exec->counts, exec->rollbacks[exec->nbRollbacks].counts,
3098 exec->comp->nbCounters * sizeof(int));
3099 }
3100
3101#ifdef DEBUG_REGEXP_EXEC
3102 printf("restored ");
3103 xmlFARegDebugExec(exec);
3104#endif
3105}
3106
3107/************************************************************************
3108 * *
3109 * Verifier, running an input against a compiled regexp *
3110 * *
3111 ************************************************************************/
3112
3113static int
3114xmlFARegExec(xmlRegexpPtr comp, const xmlChar *content) {
3115 xmlRegExecCtxt execval;
3116 xmlRegExecCtxtPtr exec = &execval;
3117 int ret, codepoint = 0, len, deter;
3118
3119 exec->inputString = content;
3120 exec->index = 0;
3121 exec->nbPush = 0;
3122 exec->determinist = 1;
3123 exec->maxRollbacks = 0;
3124 exec->nbRollbacks = 0;
3125 exec->rollbacks = NULL;
3126 exec->status = 0;
3127 exec->comp = comp;
3128 exec->state = comp->states[0];
3129 exec->transno = 0;
3130 exec->transcount = 0;
3131 exec->inputStack = NULL;
3132 exec->inputStackMax = 0;
3133 if (comp->nbCounters > 0) {
3134 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int));
3135 if (exec->counts == NULL) {
3136 xmlRegexpErrMemory(NULL, "running regexp");
3137 return(-1);
3138 }
3139 memset(exec->counts, 0, comp->nbCounters * sizeof(int));
3140 } else
3141 exec->counts = NULL;
3142 while ((exec->status == 0) &&
3143 ((exec->inputString[exec->index] != 0) ||
3144 (exec->state->type != XML_REGEXP_FINAL_STATE))) {
3145 xmlRegTransPtr trans;
3146 xmlRegAtomPtr atom;
3147
3148 /*
3149 * If end of input on non-terminal state, rollback, however we may
3150 * still have epsilon like transition for counted transitions
3151 * on counters, in that case don't break too early. Additionally,
3152 * if we are working on a range like "AB{0,2}", where B is not present,
3153 * we don't want to break.
3154 */
3155 len = 1;
3156 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL)) {
3157 /*
3158 * if there is a transition, we must check if
3159 * atom allows minOccurs of 0
3160 */
3161 if (exec->transno < exec->state->nbTrans) {
3162 trans = &exec->state->trans[exec->transno];
3163 if (trans->to >=0) {
3164 atom = trans->atom;
3165 if (!((atom->min == 0) && (atom->max > 0)))
3166 goto rollback;
3167 }
3168 } else
3169 goto rollback;
3170 }
3171
3172 exec->transcount = 0;
3173 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
3174 trans = &exec->state->trans[exec->transno];
3175 if (trans->to < 0)
3176 continue;
3177 atom = trans->atom;
3178 ret = 0;
3179 deter = 1;
3180 if (trans->count >= 0) {
3181 int count;
3182 xmlRegCounterPtr counter;
3183
3184 if (exec->counts == NULL) {
3185 exec->status = -1;
3186 goto error;
3187 }
3188 /*
3189 * A counted transition.
3190 */
3191
3192 count = exec->counts[trans->count];
3193 counter = &exec->comp->counters[trans->count];
3194#ifdef DEBUG_REGEXP_EXEC
3195 printf("testing count %d: val %d, min %d, max %d\n",
3196 trans->count, count, counter->min, counter->max);
3197#endif
3198 ret = ((count >= counter->min) && (count <= counter->max));
3199 if ((ret) && (counter->min != counter->max))
3200 deter = 0;
3201 } else if (atom == NULL) {
3202 fprintf(stderr, "epsilon transition left at runtime\n");
3203 exec->status = -2;
3204 break;
3205 } else if (exec->inputString[exec->index] != 0) {
3206 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
3207 ret = xmlRegCheckCharacter(atom, codepoint);
3208 if ((ret == 1) && (atom->min >= 0) && (atom->max > 0)) {
3209 xmlRegStatePtr to = comp->states[trans->to];
3210
3211 /*
3212 * this is a multiple input sequence
3213 * If there is a counter associated increment it now.
3214 * before potentially saving and rollback
3215 * do not increment if the counter is already over the
3216 * maximum limit in which case get to next transition
3217 */
3218 if (trans->counter >= 0) {
3219 xmlRegCounterPtr counter;
3220
3221 if ((exec->counts == NULL) ||
3222 (exec->comp == NULL) ||
3223 (exec->comp->counters == NULL)) {
3224 exec->status = -1;
3225 goto error;
3226 }
3227 counter = &exec->comp->counters[trans->counter];
3228 if (exec->counts[trans->counter] >= counter->max)
3229 continue; /* for loop on transitions */
3230
3231#ifdef DEBUG_REGEXP_EXEC
3232 printf("Increasing count %d\n", trans->counter);
3233#endif
3234 exec->counts[trans->counter]++;
3235 }
3236 if (exec->state->nbTrans > exec->transno + 1) {
3237 xmlFARegExecSave(exec);
3238 }
3239 exec->transcount = 1;
3240 do {
3241 /*
3242 * Try to progress as much as possible on the input
3243 */
3244 if (exec->transcount == atom->max) {
3245 break;
3246 }
3247 exec->index += len;
3248 /*
3249 * End of input: stop here
3250 */
3251 if (exec->inputString[exec->index] == 0) {
3252 exec->index -= len;
3253 break;
3254 }
3255 if (exec->transcount >= atom->min) {
3256 int transno = exec->transno;
3257 xmlRegStatePtr state = exec->state;
3258
3259 /*
3260 * The transition is acceptable save it
3261 */
3262 exec->transno = -1; /* trick */
3263 exec->state = to;
3264 xmlFARegExecSave(exec);
3265 exec->transno = transno;
3266 exec->state = state;
3267 }
3268 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
3269 len);
3270 ret = xmlRegCheckCharacter(atom, codepoint);
3271 exec->transcount++;
3272 } while (ret == 1);
3273 if (exec->transcount < atom->min)
3274 ret = 0;
3275
3276 /*
3277 * If the last check failed but one transition was found
3278 * possible, rollback
3279 */
3280 if (ret < 0)
3281 ret = 0;
3282 if (ret == 0) {
3283 goto rollback;
3284 }
3285 if (trans->counter >= 0) {
3286 if (exec->counts == NULL) {
3287 exec->status = -1;
3288 goto error;
3289 }
3290#ifdef DEBUG_REGEXP_EXEC
3291 printf("Decreasing count %d\n", trans->counter);
3292#endif
3293 exec->counts[trans->counter]--;
3294 }
3295 } else if ((ret == 0) && (atom->min == 0) && (atom->max > 0)) {
3296 /*
3297 * we don't match on the codepoint, but minOccurs of 0
3298 * says that's ok. Setting len to 0 inhibits stepping
3299 * over the codepoint.
3300 */
3301 exec->transcount = 1;
3302 len = 0;
3303 ret = 1;
3304 }
3305 } else if ((atom->min == 0) && (atom->max > 0)) {
3306 /* another spot to match when minOccurs is 0 */
3307 exec->transcount = 1;
3308 len = 0;
3309 ret = 1;
3310 }
3311 if (ret == 1) {
3312 if ((trans->nd == 1) ||
3313 ((trans->count >= 0) && (deter == 0) &&
3314 (exec->state->nbTrans > exec->transno + 1))) {
3315#ifdef DEBUG_REGEXP_EXEC
3316 if (trans->nd == 1)
3317 printf("Saving on nd transition atom %d for %c at %d\n",
3318 trans->atom->no, codepoint, exec->index);
3319 else
3320 printf("Saving on counted transition count %d for %c at %d\n",
3321 trans->count, codepoint, exec->index);
3322#endif
3323 xmlFARegExecSave(exec);
3324 }
3325 if (trans->counter >= 0) {
3326 xmlRegCounterPtr counter;
3327
3328 /* make sure we don't go over the counter maximum value */
3329 if ((exec->counts == NULL) ||
3330 (exec->comp == NULL) ||
3331 (exec->comp->counters == NULL)) {
3332 exec->status = -1;
3333 goto error;
3334 }
3335 counter = &exec->comp->counters[trans->counter];
3336 if (exec->counts[trans->counter] >= counter->max)
3337 continue; /* for loop on transitions */
3338#ifdef DEBUG_REGEXP_EXEC
3339 printf("Increasing count %d\n", trans->counter);
3340#endif
3341 exec->counts[trans->counter]++;
3342 }
3343 if ((trans->count >= 0) &&
3344 (trans->count < REGEXP_ALL_COUNTER)) {
3345 if (exec->counts == NULL) {
3346 exec->status = -1;
3347 goto error;
3348 }
3349#ifdef DEBUG_REGEXP_EXEC
3350 printf("resetting count %d on transition\n",
3351 trans->count);
3352#endif
3353 exec->counts[trans->count] = 0;
3354 }
3355#ifdef DEBUG_REGEXP_EXEC
3356 printf("entering state %d\n", trans->to);
3357#endif
3358 exec->state = comp->states[trans->to];
3359 exec->transno = 0;
3360 if (trans->atom != NULL) {
3361 exec->index += len;
3362 }
3363 goto progress;
3364 } else if (ret < 0) {
3365 exec->status = -4;
3366 break;
3367 }
3368 }
3369 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
3370rollback:
3371 /*
3372 * Failed to find a way out
3373 */
3374 exec->determinist = 0;
3375#ifdef DEBUG_REGEXP_EXEC
3376 printf("rollback from state %d on %d:%c\n", exec->state->no,
3377 codepoint,codepoint);
3378#endif
3379 xmlFARegExecRollBack(exec);
3380 }
3381progress:
3382 continue;
3383 }
3384error:
3385 if (exec->rollbacks != NULL) {
3386 if (exec->counts != NULL) {
3387 int i;
3388
3389 for (i = 0;i < exec->maxRollbacks;i++)
3390 if (exec->rollbacks[i].counts != NULL)
3391 xmlFree(exec->rollbacks[i].counts);
3392 }
3393 xmlFree(exec->rollbacks);
3394 }
3395 if (exec->counts != NULL)
3396 xmlFree(exec->counts);
3397 if (exec->status == 0)
3398 return(1);
3399 if (exec->status == -1) {
3400 if (exec->nbPush > MAX_PUSH)
3401 return(-1);
3402 return(0);
3403 }
3404 return(exec->status);
3405}
3406
3407/************************************************************************
3408 * *
3409 * Progressive interface to the verifier one atom at a time *
3410 * *
3411 ************************************************************************/
3412#ifdef DEBUG_ERR
3413static void testerr(xmlRegExecCtxtPtr exec);
3414#endif
3415
3416/**
3417 * xmlRegNewExecCtxt:
3418 * @comp: a precompiled regular expression
3419 * @callback: a callback function used for handling progresses in the
3420 * automata matching phase
3421 * @data: the context data associated to the callback in this context
3422 *
3423 * Build a context used for progressive evaluation of a regexp.
3424 *
3425 * Returns the new context
3426 */
3427xmlRegExecCtxtPtr
3428xmlRegNewExecCtxt(xmlRegexpPtr comp, xmlRegExecCallbacks callback, void *data) {
3429 xmlRegExecCtxtPtr exec;
3430
3431 if (comp == NULL)
3432 return(NULL);
3433 if ((comp->compact == NULL) && (comp->states == NULL))
3434 return(NULL);
3435 exec = (xmlRegExecCtxtPtr) xmlMalloc(sizeof(xmlRegExecCtxt));
3436 if (exec == NULL) {
3437 xmlRegexpErrMemory(NULL, "creating execution context");
3438 return(NULL);
3439 }
3440 memset(exec, 0, sizeof(xmlRegExecCtxt));
3441 exec->inputString = NULL;
3442 exec->index = 0;
3443 exec->determinist = 1;
3444 exec->maxRollbacks = 0;
3445 exec->nbRollbacks = 0;
3446 exec->rollbacks = NULL;
3447 exec->status = 0;
3448 exec->comp = comp;
3449 if (comp->compact == NULL)
3450 exec->state = comp->states[0];
3451 exec->transno = 0;
3452 exec->transcount = 0;
3453 exec->callback = callback;
3454 exec->data = data;
3455 if (comp->nbCounters > 0) {
3456 /*
3457 * For error handling, exec->counts is allocated twice the size
3458 * the second half is used to store the data in case of rollback
3459 */
3460 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int)
3461 * 2);
3462 if (exec->counts == NULL) {
3463 xmlRegexpErrMemory(NULL, "creating execution context");
3464 xmlFree(exec);
3465 return(NULL);
3466 }
3467 memset(exec->counts, 0, comp->nbCounters * sizeof(int) * 2);
3468 exec->errCounts = &exec->counts[comp->nbCounters];
3469 } else {
3470 exec->counts = NULL;
3471 exec->errCounts = NULL;
3472 }
3473 exec->inputStackMax = 0;
3474 exec->inputStackNr = 0;
3475 exec->inputStack = NULL;
3476 exec->errStateNo = -1;
3477 exec->errString = NULL;
3478 exec->nbPush = 0;
3479 return(exec);
3480}
3481
3482/**
3483 * xmlRegFreeExecCtxt:
3484 * @exec: a regular expression evaulation context
3485 *
3486 * Free the structures associated to a regular expression evaulation context.
3487 */
3488void
3489xmlRegFreeExecCtxt(xmlRegExecCtxtPtr exec) {
3490 if (exec == NULL)
3491 return;
3492
3493 if (exec->rollbacks != NULL) {
3494 if (exec->counts != NULL) {
3495 int i;
3496
3497 for (i = 0;i < exec->maxRollbacks;i++)
3498 if (exec->rollbacks[i].counts != NULL)
3499 xmlFree(exec->rollbacks[i].counts);
3500 }
3501 xmlFree(exec->rollbacks);
3502 }
3503 if (exec->counts != NULL)
3504 xmlFree(exec->counts);
3505 if (exec->inputStack != NULL) {
3506 int i;
3507
3508 for (i = 0;i < exec->inputStackNr;i++) {
3509 if (exec->inputStack[i].value != NULL)
3510 xmlFree(exec->inputStack[i].value);
3511 }
3512 xmlFree(exec->inputStack);
3513 }
3514 if (exec->errString != NULL)
3515 xmlFree(exec->errString);
3516 xmlFree(exec);
3517}
3518
3519static void
3520xmlFARegExecSaveInputString(xmlRegExecCtxtPtr exec, const xmlChar *value,
3521 void *data) {
3522#ifdef DEBUG_PUSH
3523 printf("saving value: %d:%s\n", exec->inputStackNr, value);
3524#endif
3525 if (exec->inputStackMax == 0) {
3526 exec->inputStackMax = 4;
3527 exec->inputStack = (xmlRegInputTokenPtr)
3528 xmlMalloc(exec->inputStackMax * sizeof(xmlRegInputToken));
3529 if (exec->inputStack == NULL) {
3530 xmlRegexpErrMemory(NULL, "pushing input string");
3531 exec->inputStackMax = 0;
3532 return;
3533 }
3534 } else if (exec->inputStackNr + 1 >= exec->inputStackMax) {
3535 xmlRegInputTokenPtr tmp;
3536
3537 exec->inputStackMax *= 2;
3538 tmp = (xmlRegInputTokenPtr) xmlRealloc(exec->inputStack,
3539 exec->inputStackMax * sizeof(xmlRegInputToken));
3540 if (tmp == NULL) {
3541 xmlRegexpErrMemory(NULL, "pushing input string");
3542 exec->inputStackMax /= 2;
3543 return;
3544 }
3545 exec->inputStack = tmp;
3546 }
3547 exec->inputStack[exec->inputStackNr].value = xmlStrdup(value);
3548 exec->inputStack[exec->inputStackNr].data = data;
3549 exec->inputStackNr++;
3550 exec->inputStack[exec->inputStackNr].value = NULL;
3551 exec->inputStack[exec->inputStackNr].data = NULL;
3552}
3553
3554/**
3555 * xmlRegStrEqualWildcard:
3556 * @expStr: the string to be evaluated
3557 * @valStr: the validation string
3558 *
3559 * Checks if both strings are equal or have the same content. "*"
3560 * can be used as a wildcard in @valStr; "|" is used as a seperator of
3561 * substrings in both @expStr and @valStr.
3562 *
3563 * Returns 1 if the comparison is satisfied and the number of substrings
3564 * is equal, 0 otherwise.
3565 */
3566
3567static int
3568xmlRegStrEqualWildcard(const xmlChar *expStr, const xmlChar *valStr) {
3569 if (expStr == valStr) return(1);
3570 if (expStr == NULL) return(0);
3571 if (valStr == NULL) return(0);
3572 do {
3573 /*
3574 * Eval if we have a wildcard for the current item.
3575 */
3576 if (*expStr != *valStr) {
3577 /* if one of them starts with a wildcard make valStr be it */
3578 if (*valStr == '*') {
3579 const xmlChar *tmp;
3580
3581 tmp = valStr;
3582 valStr = expStr;
3583 expStr = tmp;
3584 }
3585 if ((*valStr != 0) && (*expStr != 0) && (*expStr++ == '*')) {
3586 do {
3587 if (*valStr == XML_REG_STRING_SEPARATOR)
3588 break;
3589 valStr++;
3590 } while (*valStr != 0);
3591 continue;
3592 } else
3593 return(0);
3594 }
3595 expStr++;
3596 valStr++;
3597 } while (*valStr != 0);
3598 if (*expStr != 0)
3599 return (0);
3600 else
3601 return (1);
3602}
3603
3604/**
3605 * xmlRegCompactPushString:
3606 * @exec: a regexp execution context
3607 * @comp: the precompiled exec with a compact table
3608 * @value: a string token input
3609 * @data: data associated to the token to reuse in callbacks
3610 *
3611 * Push one input token in the execution context
3612 *
3613 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
3614 * a negative value in case of error.
3615 */
3616static int
3617xmlRegCompactPushString(xmlRegExecCtxtPtr exec,
3618 xmlRegexpPtr comp,
3619 const xmlChar *value,
3620 void *data) {
3621 int state = exec->index;
3622 int i, target;
3623
3624 if ((comp == NULL) || (comp->compact == NULL) || (comp->stringMap == NULL))
3625 return(-1);
3626
3627 if (value == NULL) {
3628 /*
3629 * are we at a final state ?
3630 */
3631 if (comp->compact[state * (comp->nbstrings + 1)] ==
3632 XML_REGEXP_FINAL_STATE)
3633 return(1);
3634 return(0);
3635 }
3636
3637#ifdef DEBUG_PUSH
3638 printf("value pushed: %s\n", value);
3639#endif
3640
3641 /*
3642 * Examine all outside transitions from current state
3643 */
3644 for (i = 0;i < comp->nbstrings;i++) {
3645 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
3646 if ((target > 0) && (target <= comp->nbstates)) {
3647 target--; /* to avoid 0 */
3648 if (xmlRegStrEqualWildcard(comp->stringMap[i], value)) {
3649 exec->index = target;
3650 if ((exec->callback != NULL) && (comp->transdata != NULL)) {
3651 exec->callback(exec->data, value,
3652 comp->transdata[state * comp->nbstrings + i], data);
3653 }
3654#ifdef DEBUG_PUSH
3655 printf("entering state %d\n", target);
3656#endif
3657 if (comp->compact[target * (comp->nbstrings + 1)] ==
3658 XML_REGEXP_SINK_STATE)
3659 goto error;
3660
3661 if (comp->compact[target * (comp->nbstrings + 1)] ==
3662 XML_REGEXP_FINAL_STATE)
3663 return(1);
3664 return(0);
3665 }
3666 }
3667 }
3668 /*
3669 * Failed to find an exit transition out from current state for the
3670 * current token
3671 */
3672#ifdef DEBUG_PUSH
3673 printf("failed to find a transition for %s on state %d\n", value, state);
3674#endif
3675error:
3676 if (exec->errString != NULL)
3677 xmlFree(exec->errString);
3678 exec->errString = xmlStrdup(value);
3679 exec->errStateNo = state;
3680 exec->status = -1;
3681#ifdef DEBUG_ERR
3682 testerr(exec);
3683#endif
3684 return(-1);
3685}
3686
3687/**
3688 * xmlRegExecPushStringInternal:
3689 * @exec: a regexp execution context or NULL to indicate the end
3690 * @value: a string token input
3691 * @data: data associated to the token to reuse in callbacks
3692 * @compound: value was assembled from 2 strings
3693 *
3694 * Push one input token in the execution context
3695 *
3696 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
3697 * a negative value in case of error.
3698 */
3699static int
3700xmlRegExecPushStringInternal(xmlRegExecCtxtPtr exec, const xmlChar *value,
3701 void *data, int compound) {
3702 xmlRegTransPtr trans;
3703 xmlRegAtomPtr atom;
3704 int ret;
3705 int final = 0;
3706 int progress = 1;
3707
3708 if (exec == NULL)
3709 return(-1);
3710 if (exec->comp == NULL)
3711 return(-1);
3712 if (exec->status != 0)
3713 return(exec->status);
3714
3715 if (exec->comp->compact != NULL)
3716 return(xmlRegCompactPushString(exec, exec->comp, value, data));
3717
3718 if (value == NULL) {
3719 if (exec->state->type == XML_REGEXP_FINAL_STATE)
3720 return(1);
3721 final = 1;
3722 }
3723
3724#ifdef DEBUG_PUSH
3725 printf("value pushed: %s\n", value);
3726#endif
3727 /*
3728 * If we have an active rollback stack push the new value there
3729 * and get back to where we were left
3730 */
3731 if ((value != NULL) && (exec->inputStackNr > 0)) {
3732 xmlFARegExecSaveInputString(exec, value, data);
3733 value = exec->inputStack[exec->index].value;
3734 data = exec->inputStack[exec->index].data;
3735#ifdef DEBUG_PUSH
3736 printf("value loaded: %s\n", value);
3737#endif
3738 }
3739
3740 while ((exec->status == 0) &&
3741 ((value != NULL) ||
3742 ((final == 1) &&
3743 (exec->state->type != XML_REGEXP_FINAL_STATE)))) {
3744
3745 /*
3746 * End of input on non-terminal state, rollback, however we may
3747 * still have epsilon like transition for counted transitions
3748 * on counters, in that case don't break too early.
3749 */
3750 if ((value == NULL) && (exec->counts == NULL))
3751 goto rollback;
3752
3753 exec->transcount = 0;
3754 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
3755 trans = &exec->state->trans[exec->transno];
3756 if (trans->to < 0)
3757 continue;
3758 atom = trans->atom;
3759 ret = 0;
3760 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
3761 int i;
3762 int count;
3763 xmlRegTransPtr t;
3764 xmlRegCounterPtr counter;
3765
3766 ret = 0;
3767
3768#ifdef DEBUG_PUSH
3769 printf("testing all lax %d\n", trans->count);
3770#endif
3771 /*
3772 * Check all counted transitions from the current state
3773 */
3774 if ((value == NULL) && (final)) {
3775 ret = 1;
3776 } else if (value != NULL) {
3777 for (i = 0;i < exec->state->nbTrans;i++) {
3778 t = &exec->state->trans[i];
3779 if ((t->counter < 0) || (t == trans))
3780 continue;
3781 counter = &exec->comp->counters[t->counter];
3782 count = exec->counts[t->counter];
3783 if ((count < counter->max) &&
3784 (t->atom != NULL) &&
3785 (xmlStrEqual(value, t->atom->valuep))) {
3786 ret = 0;
3787 break;
3788 }
3789 if ((count >= counter->min) &&
3790 (count < counter->max) &&
3791 (t->atom != NULL) &&
3792 (xmlStrEqual(value, t->atom->valuep))) {
3793 ret = 1;
3794 break;
3795 }
3796 }
3797 }
3798 } else if (trans->count == REGEXP_ALL_COUNTER) {
3799 int i;
3800 int count;
3801 xmlRegTransPtr t;
3802 xmlRegCounterPtr counter;
3803
3804 ret = 1;
3805
3806#ifdef DEBUG_PUSH
3807 printf("testing all %d\n", trans->count);
3808#endif
3809 /*
3810 * Check all counted transitions from the current state
3811 */
3812 for (i = 0;i < exec->state->nbTrans;i++) {
3813 t = &exec->state->trans[i];
3814 if ((t->counter < 0) || (t == trans))
3815 continue;
3816 counter = &exec->comp->counters[t->counter];
3817 count = exec->counts[t->counter];
3818 if ((count < counter->min) || (count > counter->max)) {
3819 ret = 0;
3820 break;
3821 }
3822 }
3823 } else if (trans->count >= 0) {
3824 int count;
3825 xmlRegCounterPtr counter;
3826
3827 /*
3828 * A counted transition.
3829 */
3830
3831 count = exec->counts[trans->count];
3832 counter = &exec->comp->counters[trans->count];
3833#ifdef DEBUG_PUSH
3834 printf("testing count %d: val %d, min %d, max %d\n",
3835 trans->count, count, counter->min, counter->max);
3836#endif
3837 ret = ((count >= counter->min) && (count <= counter->max));
3838 } else if (atom == NULL) {
3839 fprintf(stderr, "epsilon transition left at runtime\n");
3840 exec->status = -2;
3841 break;
3842 } else if (value != NULL) {
3843 ret = xmlRegStrEqualWildcard(atom->valuep, value);
3844 if (atom->neg) {
3845 ret = !ret;
3846 if (!compound)
3847 ret = 0;
3848 }
3849 if ((ret == 1) && (trans->counter >= 0)) {
3850 xmlRegCounterPtr counter;
3851 int count;
3852
3853 count = exec->counts[trans->counter];
3854 counter = &exec->comp->counters[trans->counter];
3855 if (count >= counter->max)
3856 ret = 0;
3857 }
3858
3859 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
3860 xmlRegStatePtr to = exec->comp->states[trans->to];
3861
3862 /*
3863 * this is a multiple input sequence
3864 */
3865 if (exec->state->nbTrans > exec->transno + 1) {
3866 if (exec->inputStackNr <= 0) {
3867 xmlFARegExecSaveInputString(exec, value, data);
3868 }
3869 xmlFARegExecSave(exec);
3870 }
3871 exec->transcount = 1;
3872 do {
3873 /*
3874 * Try to progress as much as possible on the input
3875 */
3876 if (exec->transcount == atom->max) {
3877 break;
3878 }
3879 exec->index++;
3880 value = exec->inputStack[exec->index].value;
3881 data = exec->inputStack[exec->index].data;
3882#ifdef DEBUG_PUSH
3883 printf("value loaded: %s\n", value);
3884#endif
3885
3886 /*
3887 * End of input: stop here
3888 */
3889 if (value == NULL) {
3890 exec->index --;
3891 break;
3892 }
3893 if (exec->transcount >= atom->min) {
3894 int transno = exec->transno;
3895 xmlRegStatePtr state = exec->state;
3896
3897 /*
3898 * The transition is acceptable save it
3899 */
3900 exec->transno = -1; /* trick */
3901 exec->state = to;
3902 if (exec->inputStackNr <= 0) {
3903 xmlFARegExecSaveInputString(exec, value, data);
3904 }
3905 xmlFARegExecSave(exec);
3906 exec->transno = transno;
3907 exec->state = state;
3908 }
3909 ret = xmlStrEqual(value, atom->valuep);
3910 exec->transcount++;
3911 } while (ret == 1);
3912 if (exec->transcount < atom->min)
3913 ret = 0;
3914
3915 /*
3916 * If the last check failed but one transition was found
3917 * possible, rollback
3918 */
3919 if (ret < 0)
3920 ret = 0;
3921 if (ret == 0) {
3922 goto rollback;
3923 }
3924 }
3925 }
3926 if (ret == 1) {
3927 if ((exec->callback != NULL) && (atom != NULL) &&
3928 (data != NULL)) {
3929 exec->callback(exec->data, atom->valuep,
3930 atom->data, data);
3931 }
3932 if (exec->state->nbTrans > exec->transno + 1) {
3933 if (exec->inputStackNr <= 0) {
3934 xmlFARegExecSaveInputString(exec, value, data);
3935 }
3936 xmlFARegExecSave(exec);
3937 }
3938 if (trans->counter >= 0) {
3939#ifdef DEBUG_PUSH
3940 printf("Increasing count %d\n", trans->counter);
3941#endif
3942 exec->counts[trans->counter]++;
3943 }
3944 if ((trans->count >= 0) &&
3945 (trans->count < REGEXP_ALL_COUNTER)) {
3946#ifdef DEBUG_REGEXP_EXEC
3947 printf("resetting count %d on transition\n",
3948 trans->count);
3949#endif
3950 exec->counts[trans->count] = 0;
3951 }
3952#ifdef DEBUG_PUSH
3953 printf("entering state %d\n", trans->to);
3954#endif
3955 if ((exec->comp->states[trans->to] != NULL) &&
3956 (exec->comp->states[trans->to]->type ==
3957 XML_REGEXP_SINK_STATE)) {
3958 /*
3959 * entering a sink state, save the current state as error
3960 * state.
3961 */
3962 if (exec->errString != NULL)
3963 xmlFree(exec->errString);
3964 exec->errString = xmlStrdup(value);
3965 exec->errState = exec->state;
3966 memcpy(exec->errCounts, exec->counts,
3967 exec->comp->nbCounters * sizeof(int));
3968 }
3969 exec->state = exec->comp->states[trans->to];
3970 exec->transno = 0;
3971 if (trans->atom != NULL) {
3972 if (exec->inputStack != NULL) {
3973 exec->index++;
3974 if (exec->index < exec->inputStackNr) {
3975 value = exec->inputStack[exec->index].value;
3976 data = exec->inputStack[exec->index].data;
3977#ifdef DEBUG_PUSH
3978 printf("value loaded: %s\n", value);
3979#endif
3980 } else {
3981 value = NULL;
3982 data = NULL;
3983#ifdef DEBUG_PUSH
3984 printf("end of input\n");
3985#endif
3986 }
3987 } else {
3988 value = NULL;
3989 data = NULL;
3990#ifdef DEBUG_PUSH
3991 printf("end of input\n");
3992#endif
3993 }
3994 }
3995 goto progress;
3996 } else if (ret < 0) {
3997 exec->status = -4;
3998 break;
3999 }
4000 }
4001 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
4002rollback:
4003 /*
4004 * if we didn't yet rollback on the current input
4005 * store the current state as the error state.
4006 */
4007 if ((progress) && (exec->state != NULL) &&
4008 (exec->state->type != XML_REGEXP_SINK_STATE)) {
4009 progress = 0;
4010 if (exec->errString != NULL)
4011 xmlFree(exec->errString);
4012 exec->errString = xmlStrdup(value);
4013 exec->errState = exec->state;
4014 memcpy(exec->errCounts, exec->counts,
4015 exec->comp->nbCounters * sizeof(int));
4016 }
4017
4018 /*
4019 * Failed to find a way out
4020 */
4021 exec->determinist = 0;
4022 xmlFARegExecRollBack(exec);
4023 if (exec->status == 0) {
4024 value = exec->inputStack[exec->index].value;
4025 data = exec->inputStack[exec->index].data;
4026#ifdef DEBUG_PUSH
4027 printf("value loaded: %s\n", value);
4028#endif
4029 }
4030 }
4031 continue;
4032progress:
4033 progress = 1;
4034 continue;
4035 }
4036 if (exec->status == 0) {
4037 return(exec->state->type == XML_REGEXP_FINAL_STATE);
4038 }
4039#ifdef DEBUG_ERR
4040 if (exec->status < 0) {
4041 testerr(exec);
4042 }
4043#endif
4044 return(exec->status);
4045}
4046
4047/**
4048 * xmlRegExecPushString:
4049 * @exec: a regexp execution context or NULL to indicate the end
4050 * @value: a string token input
4051 * @data: data associated to the token to reuse in callbacks
4052 *
4053 * Push one input token in the execution context
4054 *
4055 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
4056 * a negative value in case of error.
4057 */
4058int
4059xmlRegExecPushString(xmlRegExecCtxtPtr exec, const xmlChar *value,
4060 void *data) {
4061 return(xmlRegExecPushStringInternal(exec, value, data, 0));
4062}
4063
4064/**
4065 * xmlRegExecPushString2:
4066 * @exec: a regexp execution context or NULL to indicate the end
4067 * @value: the first string token input
4068 * @value2: the second string token input
4069 * @data: data associated to the token to reuse in callbacks
4070 *
4071 * Push one input token in the execution context
4072 *
4073 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
4074 * a negative value in case of error.
4075 */
4076int
4077xmlRegExecPushString2(xmlRegExecCtxtPtr exec, const xmlChar *value,
4078 const xmlChar *value2, void *data) {
4079 xmlChar buf[150];
4080 int lenn, lenp, ret;
4081 xmlChar *str;
4082
4083 if (exec == NULL)
4084 return(-1);
4085 if (exec->comp == NULL)
4086 return(-1);
4087 if (exec->status != 0)
4088 return(exec->status);
4089
4090 if (value2 == NULL)
4091 return(xmlRegExecPushString(exec, value, data));
4092
4093 lenn = strlen((char *) value2);
4094 lenp = strlen((char *) value);
4095
4096 if (150 < lenn + lenp + 2) {
4097 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
4098 if (str == NULL) {
4099 exec->status = -1;
4100 return(-1);
4101 }
4102 } else {
4103 str = buf;
4104 }
4105 memcpy(&str[0], value, lenp);
4106 str[lenp] = XML_REG_STRING_SEPARATOR;
4107 memcpy(&str[lenp + 1], value2, lenn);
4108 str[lenn + lenp + 1] = 0;
4109
4110 if (exec->comp->compact != NULL)
4111 ret = xmlRegCompactPushString(exec, exec->comp, str, data);
4112 else
4113 ret = xmlRegExecPushStringInternal(exec, str, data, 1);
4114
4115 if (str != buf)
4116 xmlFree(str);
4117 return(ret);
4118}
4119
4120/**
4121 * xmlRegExecGetValues:
4122 * @exec: a regexp execution context
4123 * @err: error extraction or normal one
4124 * @nbval: pointer to the number of accepted values IN/OUT
4125 * @nbneg: return number of negative transitions
4126 * @values: pointer to the array of acceptable values
4127 * @terminal: return value if this was a terminal state
4128 *
4129 * Extract informations from the regexp execution, internal routine to
4130 * implement xmlRegExecNextValues() and xmlRegExecErrInfo()
4131 *
4132 * Returns: 0 in case of success or -1 in case of error.
4133 */
4134static int
4135xmlRegExecGetValues(xmlRegExecCtxtPtr exec, int err,
4136 int *nbval, int *nbneg,
4137 xmlChar **values, int *terminal) {
4138 int maxval;
4139 int nb = 0;
4140
4141 if ((exec == NULL) || (nbval == NULL) || (nbneg == NULL) ||
4142 (values == NULL) || (*nbval <= 0))
4143 return(-1);
4144
4145 maxval = *nbval;
4146 *nbval = 0;
4147 *nbneg = 0;
4148 if ((exec->comp != NULL) && (exec->comp->compact != NULL)) {
4149 xmlRegexpPtr comp;
4150 int target, i, state;
4151
4152 comp = exec->comp;
4153
4154 if (err) {
4155 if (exec->errStateNo == -1) return(-1);
4156 state = exec->errStateNo;
4157 } else {
4158 state = exec->index;
4159 }
4160 if (terminal != NULL) {
4161 if (comp->compact[state * (comp->nbstrings + 1)] ==
4162 XML_REGEXP_FINAL_STATE)
4163 *terminal = 1;
4164 else
4165 *terminal = 0;
4166 }
4167 for (i = 0;(i < comp->nbstrings) && (nb < maxval);i++) {
4168 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
4169 if ((target > 0) && (target <= comp->nbstates) &&
4170 (comp->compact[(target - 1) * (comp->nbstrings + 1)] !=
4171 XML_REGEXP_SINK_STATE)) {
4172 values[nb++] = comp->stringMap[i];
4173 (*nbval)++;
4174 }
4175 }
4176 for (i = 0;(i < comp->nbstrings) && (nb < maxval);i++) {
4177 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
4178 if ((target > 0) && (target <= comp->nbstates) &&
4179 (comp->compact[(target - 1) * (comp->nbstrings + 1)] ==
4180 XML_REGEXP_SINK_STATE)) {
4181 values[nb++] = comp->stringMap[i];
4182 (*nbneg)++;
4183 }
4184 }
4185 } else {
4186 int transno;
4187 xmlRegTransPtr trans;
4188 xmlRegAtomPtr atom;
4189 xmlRegStatePtr state;
4190
4191 if (terminal != NULL) {
4192 if (exec->state->type == XML_REGEXP_FINAL_STATE)
4193 *terminal = 1;
4194 else
4195 *terminal = 0;
4196 }
4197
4198 if (err) {
4199 if (exec->errState == NULL) return(-1);
4200 state = exec->errState;
4201 } else {
4202 if (exec->state == NULL) return(-1);
4203 state = exec->state;
4204 }
4205 for (transno = 0;
4206 (transno < state->nbTrans) && (nb < maxval);
4207 transno++) {
4208 trans = &state->trans[transno];
4209 if (trans->to < 0)
4210 continue;
4211 atom = trans->atom;
4212 if ((atom == NULL) || (atom->valuep == NULL))
4213 continue;
4214 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
4215 /* this should not be reached but ... */
4216 TODO;
4217 } else if (trans->count == REGEXP_ALL_COUNTER) {
4218 /* this should not be reached but ... */
4219 TODO;
4220 } else if (trans->counter >= 0) {
4221 xmlRegCounterPtr counter = NULL;
4222 int count;
4223
4224 if (err)
4225 count = exec->errCounts[trans->counter];
4226 else
4227 count = exec->counts[trans->counter];
4228 if (exec->comp != NULL)
4229 counter = &exec->comp->counters[trans->counter];
4230 if ((counter == NULL) || (count < counter->max)) {
4231 if (atom->neg)
4232 values[nb++] = (xmlChar *) atom->valuep2;
4233 else
4234 values[nb++] = (xmlChar *) atom->valuep;
4235 (*nbval)++;
4236 }
4237 } else {
4238 if ((exec->comp->states[trans->to] != NULL) &&
4239 (exec->comp->states[trans->to]->type !=
4240 XML_REGEXP_SINK_STATE)) {
4241 if (atom->neg)
4242 values[nb++] = (xmlChar *) atom->valuep2;
4243 else
4244 values[nb++] = (xmlChar *) atom->valuep;
4245 (*nbval)++;
4246 }
4247 }
4248 }
4249 for (transno = 0;
4250 (transno < state->nbTrans) && (nb < maxval);
4251 transno++) {
4252 trans = &state->trans[transno];
4253 if (trans->to < 0)
4254 continue;
4255 atom = trans->atom;
4256 if ((atom == NULL) || (atom->valuep == NULL))
4257 continue;
4258 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
4259 continue;
4260 } else if (trans->count == REGEXP_ALL_COUNTER) {
4261 continue;
4262 } else if (trans->counter >= 0) {
4263 continue;
4264 } else {
4265 if ((exec->comp->states[trans->to] != NULL) &&
4266 (exec->comp->states[trans->to]->type ==
4267 XML_REGEXP_SINK_STATE)) {
4268 if (atom->neg)
4269 values[nb++] = (xmlChar *) atom->valuep2;
4270 else
4271 values[nb++] = (xmlChar *) atom->valuep;
4272 (*nbneg)++;
4273 }
4274 }
4275 }
4276 }
4277 return(0);
4278}
4279
4280/**
4281 * xmlRegExecNextValues:
4282 * @exec: a regexp execution context
4283 * @nbval: pointer to the number of accepted values IN/OUT
4284 * @nbneg: return number of negative transitions
4285 * @values: pointer to the array of acceptable values
4286 * @terminal: return value if this was a terminal state
4287 *
4288 * Extract informations from the regexp execution,
4289 * the parameter @values must point to an array of @nbval string pointers
4290 * on return nbval will contain the number of possible strings in that
4291 * state and the @values array will be updated with them. The string values
4292 * returned will be freed with the @exec context and don't need to be
4293 * deallocated.
4294 *
4295 * Returns: 0 in case of success or -1 in case of error.
4296 */
4297int
4298xmlRegExecNextValues(xmlRegExecCtxtPtr exec, int *nbval, int *nbneg,
4299 xmlChar **values, int *terminal) {
4300 return(xmlRegExecGetValues(exec, 0, nbval, nbneg, values, terminal));
4301}
4302
4303/**
4304 * xmlRegExecErrInfo:
4305 * @exec: a regexp execution context generating an error
4306 * @string: return value for the error string
4307 * @nbval: pointer to the number of accepted values IN/OUT
4308 * @nbneg: return number of negative transitions
4309 * @values: pointer to the array of acceptable values
4310 * @terminal: return value if this was a terminal state
4311 *
4312 * Extract error informations from the regexp execution, the parameter
4313 * @string will be updated with the value pushed and not accepted,
4314 * the parameter @values must point to an array of @nbval string pointers
4315 * on return nbval will contain the number of possible strings in that
4316 * state and the @values array will be updated with them. The string values
4317 * returned will be freed with the @exec context and don't need to be
4318 * deallocated.
4319 *
4320 * Returns: 0 in case of success or -1 in case of error.
4321 */
4322int
4323xmlRegExecErrInfo(xmlRegExecCtxtPtr exec, const xmlChar **string,
4324 int *nbval, int *nbneg, xmlChar **values, int *terminal) {
4325 if (exec == NULL)
4326 return(-1);
4327 if (string != NULL) {
4328 if (exec->status != 0)
4329 *string = exec->errString;
4330 else
4331 *string = NULL;
4332 }
4333 return(xmlRegExecGetValues(exec, 1, nbval, nbneg, values, terminal));
4334}
4335
4336#ifdef DEBUG_ERR
4337static void testerr(xmlRegExecCtxtPtr exec) {
4338 const xmlChar *string;
4339 xmlChar *values[5];
4340 int nb = 5;
4341 int nbneg;
4342 int terminal;
4343 xmlRegExecErrInfo(exec, &string, &nb, &nbneg, &values[0], &terminal);
4344}
4345#endif
4346
4347#if 0
4348static int
4349xmlRegExecPushChar(xmlRegExecCtxtPtr exec, int UCS) {
4350 xmlRegTransPtr trans;
4351 xmlRegAtomPtr atom;
4352 int ret;
4353 int codepoint, len;
4354
4355 if (exec == NULL)
4356 return(-1);
4357 if (exec->status != 0)
4358 return(exec->status);
4359
4360 while ((exec->status == 0) &&
4361 ((exec->inputString[exec->index] != 0) ||
4362 (exec->state->type != XML_REGEXP_FINAL_STATE))) {
4363
4364 /*
4365 * End of input on non-terminal state, rollback, however we may
4366 * still have epsilon like transition for counted transitions
4367 * on counters, in that case don't break too early.
4368 */
4369 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL))
4370 goto rollback;
4371
4372 exec->transcount = 0;
4373 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
4374 trans = &exec->state->trans[exec->transno];
4375 if (trans->to < 0)
4376 continue;
4377 atom = trans->atom;
4378 ret = 0;
4379 if (trans->count >= 0) {
4380 int count;
4381 xmlRegCounterPtr counter;
4382
4383 /*
4384 * A counted transition.
4385 */
4386
4387 count = exec->counts[trans->count];
4388 counter = &exec->comp->counters[trans->count];
4389#ifdef DEBUG_REGEXP_EXEC
4390 printf("testing count %d: val %d, min %d, max %d\n",
4391 trans->count, count, counter->min, counter->max);
4392#endif
4393 ret = ((count >= counter->min) && (count <= counter->max));
4394 } else if (atom == NULL) {
4395 fprintf(stderr, "epsilon transition left at runtime\n");
4396 exec->status = -2;
4397 break;
4398 } else if (exec->inputString[exec->index] != 0) {
4399 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
4400 ret = xmlRegCheckCharacter(atom, codepoint);
4401 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
4402 xmlRegStatePtr to = exec->comp->states[trans->to];
4403
4404 /*
4405 * this is a multiple input sequence
4406 */
4407 if (exec->state->nbTrans > exec->transno + 1) {
4408 xmlFARegExecSave(exec);
4409 }
4410 exec->transcount = 1;
4411 do {
4412 /*
4413 * Try to progress as much as possible on the input
4414 */
4415 if (exec->transcount == atom->max) {
4416 break;
4417 }
4418 exec->index += len;
4419 /*
4420 * End of input: stop here
4421 */
4422 if (exec->inputString[exec->index] == 0) {
4423 exec->index -= len;
4424 break;
4425 }
4426 if (exec->transcount >= atom->min) {
4427 int transno = exec->transno;
4428 xmlRegStatePtr state = exec->state;
4429
4430 /*
4431 * The transition is acceptable save it
4432 */
4433 exec->transno = -1; /* trick */
4434 exec->state = to;
4435 xmlFARegExecSave(exec);
4436 exec->transno = transno;
4437 exec->state = state;
4438 }
4439 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
4440 len);
4441 ret = xmlRegCheckCharacter(atom, codepoint);
4442 exec->transcount++;
4443 } while (ret == 1);
4444 if (exec->transcount < atom->min)
4445 ret = 0;
4446
4447 /*
4448 * If the last check failed but one transition was found
4449 * possible, rollback
4450 */
4451 if (ret < 0)
4452 ret = 0;
4453 if (ret == 0) {
4454 goto rollback;
4455 }
4456 }
4457 }
4458 if (ret == 1) {
4459 if (exec->state->nbTrans > exec->transno + 1) {
4460 xmlFARegExecSave(exec);
4461 }
4462 /*
4463 * restart count for expressions like this ((abc){2})*
4464 */
4465 if (trans->count >= 0) {
4466#ifdef DEBUG_REGEXP_EXEC
4467 printf("Reset count %d\n", trans->count);
4468#endif
4469 exec->counts[trans->count] = 0;
4470 }
4471 if (trans->counter >= 0) {
4472#ifdef DEBUG_REGEXP_EXEC
4473 printf("Increasing count %d\n", trans->counter);
4474#endif
4475 exec->counts[trans->counter]++;
4476 }
4477#ifdef DEBUG_REGEXP_EXEC
4478 printf("entering state %d\n", trans->to);
4479#endif
4480 exec->state = exec->comp->states[trans->to];
4481 exec->transno = 0;
4482 if (trans->atom != NULL) {
4483 exec->index += len;
4484 }
4485 goto progress;
4486 } else if (ret < 0) {
4487 exec->status = -4;
4488 break;
4489 }
4490 }
4491 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
4492rollback:
4493 /*
4494 * Failed to find a way out
4495 */
4496 exec->determinist = 0;
4497 xmlFARegExecRollBack(exec);
4498 }
4499progress:
4500 continue;
4501 }
4502}
4503#endif
4504/************************************************************************
4505 * *
4506 * Parser for the Schemas Datatype Regular Expressions *
4507 * http://www.w3.org/TR/2001/REC-xmlschema-2-20010502/#regexs *
4508 * *
4509 ************************************************************************/
4510
4511/**
4512 * xmlFAIsChar:
4513 * @ctxt: a regexp parser context
4514 *
4515 * [10] Char ::= [^.\?*+()|#x5B#x5D]
4516 */
4517static int
4518xmlFAIsChar(xmlRegParserCtxtPtr ctxt) {
4519 int cur;
4520 int len;
4521
4522 cur = CUR_SCHAR(ctxt->cur, len);
4523 if ((cur == '.') || (cur == '\\') || (cur == '?') ||
4524 (cur == '*') || (cur == '+') || (cur == '(') ||
4525 (cur == ')') || (cur == '|') || (cur == 0x5B) ||
4526 (cur == 0x5D) || (cur == 0))
4527 return(-1);
4528 return(cur);
4529}
4530
4531/**
4532 * xmlFAParseCharProp:
4533 * @ctxt: a regexp parser context
4534 *
4535 * [27] charProp ::= IsCategory | IsBlock
4536 * [28] IsCategory ::= Letters | Marks | Numbers | Punctuation |
4537 * Separators | Symbols | Others
4538 * [29] Letters ::= 'L' [ultmo]?
4539 * [30] Marks ::= 'M' [nce]?
4540 * [31] Numbers ::= 'N' [dlo]?
4541 * [32] Punctuation ::= 'P' [cdseifo]?
4542 * [33] Separators ::= 'Z' [slp]?
4543 * [34] Symbols ::= 'S' [mcko]?
4544 * [35] Others ::= 'C' [cfon]?
4545 * [36] IsBlock ::= 'Is' [a-zA-Z0-9#x2D]+
4546 */
4547static void
4548xmlFAParseCharProp(xmlRegParserCtxtPtr ctxt) {
4549 int cur;
4550 xmlRegAtomType type = (xmlRegAtomType) 0;
4551 xmlChar *blockName = NULL;
4552
4553 cur = CUR;
4554 if (cur == 'L') {
4555 NEXT;
4556 cur = CUR;
4557 if (cur == 'u') {
4558 NEXT;
4559 type = XML_REGEXP_LETTER_UPPERCASE;
4560 } else if (cur == 'l') {
4561 NEXT;
4562 type = XML_REGEXP_LETTER_LOWERCASE;
4563 } else if (cur == 't') {
4564 NEXT;
4565 type = XML_REGEXP_LETTER_TITLECASE;
4566 } else if (cur == 'm') {
4567 NEXT;
4568 type = XML_REGEXP_LETTER_MODIFIER;
4569 } else if (cur == 'o') {
4570 NEXT;
4571 type = XML_REGEXP_LETTER_OTHERS;
4572 } else {
4573 type = XML_REGEXP_LETTER;
4574 }
4575 } else if (cur == 'M') {
4576 NEXT;
4577 cur = CUR;
4578 if (cur == 'n') {
4579 NEXT;
4580 /* nonspacing */
4581 type = XML_REGEXP_MARK_NONSPACING;
4582 } else if (cur == 'c') {
4583 NEXT;
4584 /* spacing combining */
4585 type = XML_REGEXP_MARK_SPACECOMBINING;
4586 } else if (cur == 'e') {
4587 NEXT;
4588 /* enclosing */
4589 type = XML_REGEXP_MARK_ENCLOSING;
4590 } else {
4591 /* all marks */
4592 type = XML_REGEXP_MARK;
4593 }
4594 } else if (cur == 'N') {
4595 NEXT;
4596 cur = CUR;
4597 if (cur == 'd') {
4598 NEXT;
4599 /* digital */
4600 type = XML_REGEXP_NUMBER_DECIMAL;
4601 } else if (cur == 'l') {
4602 NEXT;
4603 /* letter */
4604 type = XML_REGEXP_NUMBER_LETTER;
4605 } else if (cur == 'o') {
4606 NEXT;
4607 /* other */
4608 type = XML_REGEXP_NUMBER_OTHERS;
4609 } else {
4610 /* all numbers */
4611 type = XML_REGEXP_NUMBER;
4612 }
4613 } else if (cur == 'P') {
4614 NEXT;
4615 cur = CUR;
4616 if (cur == 'c') {
4617 NEXT;
4618 /* connector */
4619 type = XML_REGEXP_PUNCT_CONNECTOR;
4620 } else if (cur == 'd') {
4621 NEXT;
4622 /* dash */
4623 type = XML_REGEXP_PUNCT_DASH;
4624 } else if (cur == 's') {
4625 NEXT;
4626 /* open */
4627 type = XML_REGEXP_PUNCT_OPEN;
4628 } else if (cur == 'e') {
4629 NEXT;
4630 /* close */
4631 type = XML_REGEXP_PUNCT_CLOSE;
4632 } else if (cur == 'i') {
4633 NEXT;
4634 /* initial quote */
4635 type = XML_REGEXP_PUNCT_INITQUOTE;
4636 } else if (cur == 'f') {
4637 NEXT;
4638 /* final quote */
4639 type = XML_REGEXP_PUNCT_FINQUOTE;
4640 } else if (cur == 'o') {
4641 NEXT;
4642 /* other */
4643 type = XML_REGEXP_PUNCT_OTHERS;
4644 } else {
4645 /* all punctuation */
4646 type = XML_REGEXP_PUNCT;
4647 }
4648 } else if (cur == 'Z') {
4649 NEXT;
4650 cur = CUR;
4651 if (cur == 's') {
4652 NEXT;
4653 /* space */
4654 type = XML_REGEXP_SEPAR_SPACE;
4655 } else if (cur == 'l') {
4656 NEXT;
4657 /* line */
4658 type = XML_REGEXP_SEPAR_LINE;
4659 } else if (cur == 'p') {
4660 NEXT;
4661 /* paragraph */
4662 type = XML_REGEXP_SEPAR_PARA;
4663 } else {
4664 /* all separators */
4665 type = XML_REGEXP_SEPAR;
4666 }
4667 } else if (cur == 'S') {
4668 NEXT;
4669 cur = CUR;
4670 if (cur == 'm') {
4671 NEXT;
4672 type = XML_REGEXP_SYMBOL_MATH;
4673 /* math */
4674 } else if (cur == 'c') {
4675 NEXT;
4676 type = XML_REGEXP_SYMBOL_CURRENCY;
4677 /* currency */
4678 } else if (cur == 'k') {
4679 NEXT;
4680 type = XML_REGEXP_SYMBOL_MODIFIER;
4681 /* modifiers */
4682 } else if (cur == 'o') {
4683 NEXT;
4684 type = XML_REGEXP_SYMBOL_OTHERS;
4685 /* other */
4686 } else {
4687 /* all symbols */
4688 type = XML_REGEXP_SYMBOL;
4689 }
4690 } else if (cur == 'C') {
4691 NEXT;
4692 cur = CUR;
4693 if (cur == 'c') {
4694 NEXT;
4695 /* control */
4696 type = XML_REGEXP_OTHER_CONTROL;
4697 } else if (cur == 'f') {
4698 NEXT;
4699 /* format */
4700 type = XML_REGEXP_OTHER_FORMAT;
4701 } else if (cur == 'o') {
4702 NEXT;
4703 /* private use */
4704 type = XML_REGEXP_OTHER_PRIVATE;
4705 } else if (cur == 'n') {
4706 NEXT;
4707 /* not assigned */
4708 type = XML_REGEXP_OTHER_NA;
4709 } else {
4710 /* all others */
4711 type = XML_REGEXP_OTHER;
4712 }
4713 } else if (cur == 'I') {
4714 const xmlChar *start;
4715 NEXT;
4716 cur = CUR;
4717 if (cur != 's') {
4718 ERROR("IsXXXX expected");
4719 return;
4720 }
4721 NEXT;
4722 start = ctxt->cur;
4723 cur = CUR;
4724 if (((cur >= 'a') && (cur <= 'z')) ||
4725 ((cur >= 'A') && (cur <= 'Z')) ||
4726 ((cur >= '0') && (cur <= '9')) ||
4727 (cur == 0x2D)) {
4728 NEXT;
4729 cur = CUR;
4730 while (((cur >= 'a') && (cur <= 'z')) ||
4731 ((cur >= 'A') && (cur <= 'Z')) ||
4732 ((cur >= '0') && (cur <= '9')) ||
4733 (cur == 0x2D)) {
4734 NEXT;
4735 cur = CUR;
4736 }
4737 }
4738 type = XML_REGEXP_BLOCK_NAME;
4739 blockName = xmlStrndup(start, ctxt->cur - start);
4740 } else {
4741 ERROR("Unknown char property");
4742 return;
4743 }
4744 if (ctxt->atom == NULL) {
4745 ctxt->atom = xmlRegNewAtom(ctxt, type);
4746 if (ctxt->atom != NULL)
4747 ctxt->atom->valuep = blockName;
4748 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4749 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4750 type, 0, 0, blockName);
4751 }
4752}
4753
4754/**
4755 * xmlFAParseCharClassEsc:
4756 * @ctxt: a regexp parser context
4757 *
4758 * [23] charClassEsc ::= ( SingleCharEsc | MultiCharEsc | catEsc | complEsc )
4759 * [24] SingleCharEsc ::= '\' [nrt\|.?*+(){}#x2D#x5B#x5D#x5E]
4760 * [25] catEsc ::= '\p{' charProp '}'
4761 * [26] complEsc ::= '\P{' charProp '}'
4762 * [37] MultiCharEsc ::= '.' | ('\' [sSiIcCdDwW])
4763 */
4764static void
4765xmlFAParseCharClassEsc(xmlRegParserCtxtPtr ctxt) {
4766 int cur;
4767
4768 if (CUR == '.') {
4769 if (ctxt->atom == NULL) {
4770 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_ANYCHAR);
4771 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4772 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4773 XML_REGEXP_ANYCHAR, 0, 0, NULL);
4774 }
4775 NEXT;
4776 return;
4777 }
4778 if (CUR != '\\') {
4779 ERROR("Escaped sequence: expecting \\");
4780 return;
4781 }
4782 NEXT;
4783 cur = CUR;
4784 if (cur == 'p') {
4785 NEXT;
4786 if (CUR != '{') {
4787 ERROR("Expecting '{'");
4788 return;
4789 }
4790 NEXT;
4791 xmlFAParseCharProp(ctxt);
4792 if (CUR != '}') {
4793 ERROR("Expecting '}'");
4794 return;
4795 }
4796 NEXT;
4797 } else if (cur == 'P') {
4798 NEXT;
4799 if (CUR != '{') {
4800 ERROR("Expecting '{'");
4801 return;
4802 }
4803 NEXT;
4804 xmlFAParseCharProp(ctxt);
4805 ctxt->atom->neg = 1;
4806 if (CUR != '}') {
4807 ERROR("Expecting '}'");
4808 return;
4809 }
4810 NEXT;
4811 } else if ((cur == 'n') || (cur == 'r') || (cur == 't') || (cur == '\\') ||
4812 (cur == '|') || (cur == '.') || (cur == '?') || (cur == '*') ||
4813 (cur == '+') || (cur == '(') || (cur == ')') || (cur == '{') ||
4814 (cur == '}') || (cur == 0x2D) || (cur == 0x5B) || (cur == 0x5D) ||
4815 (cur == 0x5E)) {
4816 if (ctxt->atom == NULL) {
4817 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
4818 if (ctxt->atom != NULL) {
4819 switch (cur) {
4820 case 'n':
4821 ctxt->atom->codepoint = '\n';
4822 break;
4823 case 'r':
4824 ctxt->atom->codepoint = '\r';
4825 break;
4826 case 't':
4827 ctxt->atom->codepoint = '\t';
4828 break;
4829 default:
4830 ctxt->atom->codepoint = cur;
4831 }
4832 }
4833 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4834 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4835 XML_REGEXP_CHARVAL, cur, cur, NULL);
4836 }
4837 NEXT;
4838 } else if ((cur == 's') || (cur == 'S') || (cur == 'i') || (cur == 'I') ||
4839 (cur == 'c') || (cur == 'C') || (cur == 'd') || (cur == 'D') ||
4840 (cur == 'w') || (cur == 'W')) {
4841 xmlRegAtomType type = XML_REGEXP_ANYSPACE;
4842
4843 switch (cur) {
4844 case 's':
4845 type = XML_REGEXP_ANYSPACE;
4846 break;
4847 case 'S':
4848 type = XML_REGEXP_NOTSPACE;
4849 break;
4850 case 'i':
4851 type = XML_REGEXP_INITNAME;
4852 break;
4853 case 'I':
4854 type = XML_REGEXP_NOTINITNAME;
4855 break;
4856 case 'c':
4857 type = XML_REGEXP_NAMECHAR;
4858 break;
4859 case 'C':
4860 type = XML_REGEXP_NOTNAMECHAR;
4861 break;
4862 case 'd':
4863 type = XML_REGEXP_DECIMAL;
4864 break;
4865 case 'D':
4866 type = XML_REGEXP_NOTDECIMAL;
4867 break;
4868 case 'w':
4869 type = XML_REGEXP_REALCHAR;
4870 break;
4871 case 'W':
4872 type = XML_REGEXP_NOTREALCHAR;
4873 break;
4874 }
4875 NEXT;
4876 if (ctxt->atom == NULL) {
4877 ctxt->atom = xmlRegNewAtom(ctxt, type);
4878 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4879 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4880 type, 0, 0, NULL);
4881 }
4882 } else {
4883 ERROR("Wrong escape sequence, misuse of character '\\'");
4884 }
4885}
4886
4887/**
4888 * xmlFAParseCharRef:
4889 * @ctxt: a regexp parser context
4890 *
4891 * [19] XmlCharRef ::= ( '&#' [0-9]+ ';' ) | (' &#x' [0-9a-fA-F]+ ';' )
4892 */
4893static int
4894xmlFAParseCharRef(xmlRegParserCtxtPtr ctxt) {
4895 int ret = 0, cur;
4896
4897 if ((CUR != '&') || (NXT(1) != '#'))
4898 return(-1);
4899 NEXT;
4900 NEXT;
4901 cur = CUR;
4902 if (cur == 'x') {
4903 NEXT;
4904 cur = CUR;
4905 if (((cur >= '0') && (cur <= '9')) ||
4906 ((cur >= 'a') && (cur <= 'f')) ||
4907 ((cur >= 'A') && (cur <= 'F'))) {
4908 while (((cur >= '0') && (cur <= '9')) ||
4909 ((cur >= 'a') && (cur <= 'f')) ||
4910 ((cur >= 'A') && (cur <= 'F'))) {
4911 if ((cur >= '0') && (cur <= '9'))
4912 ret = ret * 16 + cur - '0';
4913 else if ((cur >= 'a') && (cur <= 'f'))
4914 ret = ret * 16 + 10 + (cur - 'a');
4915 else
4916 ret = ret * 16 + 10 + (cur - 'A');
4917 NEXT;
4918 cur = CUR;
4919 }
4920 } else {
4921 ERROR("Char ref: expecting [0-9A-F]");
4922 return(-1);
4923 }
4924 } else {
4925 if ((cur >= '0') && (cur <= '9')) {
4926 while ((cur >= '0') && (cur <= '9')) {
4927 ret = ret * 10 + cur - '0';
4928 NEXT;
4929 cur = CUR;
4930 }
4931 } else {
4932 ERROR("Char ref: expecting [0-9]");
4933 return(-1);
4934 }
4935 }
4936 if (cur != ';') {
4937 ERROR("Char ref: expecting ';'");
4938 return(-1);
4939 } else {
4940 NEXT;
4941 }
4942 return(ret);
4943}
4944
4945/**
4946 * xmlFAParseCharRange:
4947 * @ctxt: a regexp parser context
4948 *
4949 * [17] charRange ::= seRange | XmlCharRef | XmlCharIncDash
4950 * [18] seRange ::= charOrEsc '-' charOrEsc
4951 * [20] charOrEsc ::= XmlChar | SingleCharEsc
4952 * [21] XmlChar ::= [^\#x2D#x5B#x5D]
4953 * [22] XmlCharIncDash ::= [^\#x5B#x5D]
4954 */
4955static void
4956xmlFAParseCharRange(xmlRegParserCtxtPtr ctxt) {
4957 int cur, len;
4958 int start = -1;
4959 int end = -1;
4960
4961 if (CUR == '\0') {
4962 ERROR("Expecting ']'");
4963 return;
4964 }
4965
4966 if ((CUR == '&') && (NXT(1) == '#')) {
4967 end = start = xmlFAParseCharRef(ctxt);
4968 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4969 XML_REGEXP_CHARVAL, start, end, NULL);
4970 return;
4971 }
4972 cur = CUR;
4973 if (cur == '\\') {
4974 NEXT;
4975 cur = CUR;
4976 switch (cur) {
4977 case 'n': start = 0xA; break;
4978 case 'r': start = 0xD; break;
4979 case 't': start = 0x9; break;
4980 case '\\': case '|': case '.': case '-': case '^': case '?':
4981 case '*': case '+': case '{': case '}': case '(': case ')':
4982 case '[': case ']':
4983 start = cur; break;
4984 default:
4985 ERROR("Invalid escape value");
4986 return;
4987 }
4988 end = start;
4989 len = 1;
4990 } else if ((cur != 0x5B) && (cur != 0x5D)) {
4991 end = start = CUR_SCHAR(ctxt->cur, len);
4992 } else {
4993 ERROR("Expecting a char range");
4994 return;
4995 }
4996 /*
4997 * Since we are "inside" a range, we can assume ctxt->cur is past
4998 * the start of ctxt->string, and PREV should be safe
4999 */
5000 if ((start == '-') && (NXT(1) != ']') && (PREV != '[') && (PREV != '^')) {
5001 NEXTL(len);
5002 return;
5003 }
5004 NEXTL(len);
5005 cur = CUR;
5006 if ((cur != '-') || (NXT(1) == ']')) {
5007 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5008 XML_REGEXP_CHARVAL, start, end, NULL);
5009 return;
5010 }
5011 NEXT;
5012 cur = CUR;
5013 if (cur == '\\') {
5014 NEXT;
5015 cur = CUR;
5016 switch (cur) {
5017 case 'n': end = 0xA; break;
5018 case 'r': end = 0xD; break;
5019 case 't': end = 0x9; break;
5020 case '\\': case '|': case '.': case '-': case '^': case '?':
5021 case '*': case '+': case '{': case '}': case '(': case ')':
5022 case '[': case ']':
5023 end = cur; break;
5024 default:
5025 ERROR("Invalid escape value");
5026 return;
5027 }
5028 len = 1;
5029 } else if ((cur != 0x5B) && (cur != 0x5D)) {
5030 end = CUR_SCHAR(ctxt->cur, len);
5031 } else {
5032 ERROR("Expecting the end of a char range");
5033 return;
5034 }
5035 NEXTL(len);
5036 /* TODO check that the values are acceptable character ranges for XML */
5037 if (end < start) {
5038 ERROR("End of range is before start of range");
5039 } else {
5040 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5041 XML_REGEXP_CHARVAL, start, end, NULL);
5042 }
5043 return;
5044}
5045
5046/**
5047 * xmlFAParsePosCharGroup:
5048 * @ctxt: a regexp parser context
5049 *
5050 * [14] posCharGroup ::= ( charRange | charClassEsc )+
5051 */
5052static void
5053xmlFAParsePosCharGroup(xmlRegParserCtxtPtr ctxt) {
5054 do {
5055 if ((CUR == '\\') || (CUR == '.')) {
5056 xmlFAParseCharClassEsc(ctxt);
5057 } else {
5058 xmlFAParseCharRange(ctxt);
5059 }
5060 } while ((CUR != ']') && (CUR != '^') && (CUR != '-') &&
5061 (CUR != 0) && (ctxt->error == 0));
5062}
5063
5064/**
5065 * xmlFAParseCharGroup:
5066 * @ctxt: a regexp parser context
5067 *
5068 * [13] charGroup ::= posCharGroup | negCharGroup | charClassSub
5069 * [15] negCharGroup ::= '^' posCharGroup
5070 * [16] charClassSub ::= ( posCharGroup | negCharGroup ) '-' charClassExpr
5071 * [12] charClassExpr ::= '[' charGroup ']'
5072 */
5073static void
5074xmlFAParseCharGroup(xmlRegParserCtxtPtr ctxt) {
5075 int n = ctxt->neg;
5076 while ((CUR != ']') && (ctxt->error == 0)) {
5077 if (CUR == '^') {
5078 int neg = ctxt->neg;
5079
5080 NEXT;
5081 ctxt->neg = !ctxt->neg;
5082 xmlFAParsePosCharGroup(ctxt);
5083 ctxt->neg = neg;
5084 } else if ((CUR == '-') && (NXT(1) == '[')) {
5085 int neg = ctxt->neg;
5086 ctxt->neg = 2;
5087 NEXT; /* eat the '-' */
5088 NEXT; /* eat the '[' */
5089 xmlFAParseCharGroup(ctxt);
5090 if (CUR == ']') {
5091 NEXT;
5092 } else {
5093 ERROR("charClassExpr: ']' expected");
5094 break;
5095 }
5096 ctxt->neg = neg;
5097 break;
5098 } else if (CUR != ']') {
5099 xmlFAParsePosCharGroup(ctxt);
5100 }
5101 }
5102 ctxt->neg = n;
5103}
5104
5105/**
5106 * xmlFAParseCharClass:
5107 * @ctxt: a regexp parser context
5108 *
5109 * [11] charClass ::= charClassEsc | charClassExpr
5110 * [12] charClassExpr ::= '[' charGroup ']'
5111 */
5112static void
5113xmlFAParseCharClass(xmlRegParserCtxtPtr ctxt) {
5114 if (CUR == '[') {
5115 NEXT;
5116 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_RANGES);
5117 if (ctxt->atom == NULL)
5118 return;
5119 xmlFAParseCharGroup(ctxt);
5120 if (CUR == ']') {
5121 NEXT;
5122 } else {
5123 ERROR("xmlFAParseCharClass: ']' expected");
5124 }
5125 } else {
5126 xmlFAParseCharClassEsc(ctxt);
5127 }
5128}
5129
5130/**
5131 * xmlFAParseQuantExact:
5132 * @ctxt: a regexp parser context
5133 *
5134 * [8] QuantExact ::= [0-9]+
5135 *
5136 * Returns 0 if success or -1 in case of error
5137 */
5138static int
5139xmlFAParseQuantExact(xmlRegParserCtxtPtr ctxt) {
5140 int ret = 0;
5141 int ok = 0;
5142
5143 while ((CUR >= '0') && (CUR <= '9')) {
5144 ret = ret * 10 + (CUR - '0');
5145 ok = 1;
5146 NEXT;
5147 }
5148 if (ok != 1) {
5149 return(-1);
5150 }
5151 return(ret);
5152}
5153
5154/**
5155 * xmlFAParseQuantifier:
5156 * @ctxt: a regexp parser context
5157 *
5158 * [4] quantifier ::= [?*+] | ( '{' quantity '}' )
5159 * [5] quantity ::= quantRange | quantMin | QuantExact
5160 * [6] quantRange ::= QuantExact ',' QuantExact
5161 * [7] quantMin ::= QuantExact ','
5162 * [8] QuantExact ::= [0-9]+
5163 */
5164static int
5165xmlFAParseQuantifier(xmlRegParserCtxtPtr ctxt) {
5166 int cur;
5167
5168 cur = CUR;
5169 if ((cur == '?') || (cur == '*') || (cur == '+')) {
5170 if (ctxt->atom != NULL) {
5171 if (cur == '?')
5172 ctxt->atom->quant = XML_REGEXP_QUANT_OPT;
5173 else if (cur == '*')
5174 ctxt->atom->quant = XML_REGEXP_QUANT_MULT;
5175 else if (cur == '+')
5176 ctxt->atom->quant = XML_REGEXP_QUANT_PLUS;
5177 }
5178 NEXT;
5179 return(1);
5180 }
5181 if (cur == '{') {
5182 int min = 0, max = 0;
5183
5184 NEXT;
5185 cur = xmlFAParseQuantExact(ctxt);
5186 if (cur >= 0)
5187 min = cur;
5188 if (CUR == ',') {
5189 NEXT;
5190 if (CUR == '}')
5191 max = INT_MAX;
5192 else {
5193 cur = xmlFAParseQuantExact(ctxt);
5194 if (cur >= 0)
5195 max = cur;
5196 else {
5197 ERROR("Improper quantifier");
5198 }
5199 }
5200 }
5201 if (CUR == '}') {
5202 NEXT;
5203 } else {
5204 ERROR("Unterminated quantifier");
5205 }
5206 if (max == 0)
5207 max = min;
5208 if (ctxt->atom != NULL) {
5209 ctxt->atom->quant = XML_REGEXP_QUANT_RANGE;
5210 ctxt->atom->min = min;
5211 ctxt->atom->max = max;
5212 }
5213 return(1);
5214 }
5215 return(0);
5216}
5217
5218/**
5219 * xmlFAParseAtom:
5220 * @ctxt: a regexp parser context
5221 *
5222 * [9] atom ::= Char | charClass | ( '(' regExp ')' )
5223 */
5224static int
5225xmlFAParseAtom(xmlRegParserCtxtPtr ctxt) {
5226 int codepoint, len;
5227
5228 codepoint = xmlFAIsChar(ctxt);
5229 if (codepoint > 0) {
5230 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
5231 if (ctxt->atom == NULL)
5232 return(-1);
5233 codepoint = CUR_SCHAR(ctxt->cur, len);
5234 ctxt->atom->codepoint = codepoint;
5235 NEXTL(len);
5236 return(1);
5237 } else if (CUR == '|') {
5238 return(0);
5239 } else if (CUR == 0) {
5240 return(0);
5241 } else if (CUR == ')') {
5242 return(0);
5243 } else if (CUR == '(') {
5244 xmlRegStatePtr start, oldend, start0;
5245
5246 NEXT;
5247 /*
5248 * this extra Epsilon transition is needed if we count with 0 allowed
5249 * unfortunately this can't be known at that point
5250 */
5251 xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, NULL);
5252 start0 = ctxt->state;
5253 xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, NULL);
5254 start = ctxt->state;
5255 oldend = ctxt->end;
5256 ctxt->end = NULL;
5257 ctxt->atom = NULL;
5258 xmlFAParseRegExp(ctxt, 0);
5259 if (CUR == ')') {
5260 NEXT;
5261 } else {
5262 ERROR("xmlFAParseAtom: expecting ')'");
5263 }
5264 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_SUBREG);
5265 if (ctxt->atom == NULL)
5266 return(-1);
5267 ctxt->atom->start = start;
5268 ctxt->atom->start0 = start0;
5269 ctxt->atom->stop = ctxt->state;
5270 ctxt->end = oldend;
5271 return(1);
5272 } else if ((CUR == '[') || (CUR == '\\') || (CUR == '.')) {
5273 xmlFAParseCharClass(ctxt);
5274 return(1);
5275 }
5276 return(0);
5277}
5278
5279/**
5280 * xmlFAParsePiece:
5281 * @ctxt: a regexp parser context
5282 *
5283 * [3] piece ::= atom quantifier?
5284 */
5285static int
5286xmlFAParsePiece(xmlRegParserCtxtPtr ctxt) {
5287 int ret;
5288
5289 ctxt->atom = NULL;
5290 ret = xmlFAParseAtom(ctxt);
5291 if (ret == 0)
5292 return(0);
5293 if (ctxt->atom == NULL) {
5294 ERROR("internal: no atom generated");
5295 }
5296 xmlFAParseQuantifier(ctxt);
5297 return(1);
5298}
5299
5300/**
5301 * xmlFAParseBranch:
5302 * @ctxt: a regexp parser context
5303 * @to: optional target to the end of the branch
5304 *
5305 * @to is used to optimize by removing duplicate path in automata
5306 * in expressions like (a|b)(c|d)
5307 *
5308 * [2] branch ::= piece*
5309 */
5310static int
5311xmlFAParseBranch(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr to) {
5312 xmlRegStatePtr previous;
5313 int ret;
5314
5315 previous = ctxt->state;
5316 ret = xmlFAParsePiece(ctxt);
5317 if (ret != 0) {
5318 if (xmlFAGenerateTransitions(ctxt, previous,
5319 (CUR=='|' || CUR==')') ? to : NULL, ctxt->atom) < 0)
5320 return(-1);
5321 previous = ctxt->state;
5322 ctxt->atom = NULL;
5323 }
5324 while ((ret != 0) && (ctxt->error == 0)) {
5325 ret = xmlFAParsePiece(ctxt);
5326 if (ret != 0) {
5327 if (xmlFAGenerateTransitions(ctxt, previous,
5328 (CUR=='|' || CUR==')') ? to : NULL, ctxt->atom) < 0)
5329 return(-1);
5330 previous = ctxt->state;
5331 ctxt->atom = NULL;
5332 }
5333 }
5334 return(0);
5335}
5336
5337/**
5338 * xmlFAParseRegExp:
5339 * @ctxt: a regexp parser context
5340 * @top: is this the top-level expression ?
5341 *
5342 * [1] regExp ::= branch ( '|' branch )*
5343 */
5344static void
5345xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top) {
5346 xmlRegStatePtr start, end;
5347
5348 /* if not top start should have been generated by an epsilon trans */
5349 start = ctxt->state;
5350 ctxt->end = NULL;
5351 xmlFAParseBranch(ctxt, NULL);
5352 if (top) {
5353#ifdef DEBUG_REGEXP_GRAPH
5354 printf("State %d is final\n", ctxt->state->no);
5355#endif
5356 ctxt->state->type = XML_REGEXP_FINAL_STATE;
5357 }
5358 if (CUR != '|') {
5359 ctxt->end = ctxt->state;
5360 return;
5361 }
5362 end = ctxt->state;
5363 while ((CUR == '|') && (ctxt->error == 0)) {
5364 NEXT;
5365 ctxt->state = start;
5366 ctxt->end = NULL;
5367 xmlFAParseBranch(ctxt, end);
5368 }
5369 if (!top) {
5370 ctxt->state = end;
5371 ctxt->end = end;
5372 }
5373}
5374
5375/************************************************************************
5376 * *
5377 * The basic API *
5378 * *
5379 ************************************************************************/
5380
5381/**
5382 * xmlRegexpPrint:
5383 * @output: the file for the output debug
5384 * @regexp: the compiled regexp
5385 *
5386 * Print the content of the compiled regular expression
5387 */
5388void
5389xmlRegexpPrint(FILE *output, xmlRegexpPtr regexp) {
5390 int i;
5391
5392 if (output == NULL)
5393 return;
5394 fprintf(output, " regexp: ");
5395 if (regexp == NULL) {
5396 fprintf(output, "NULL\n");
5397 return;
5398 }
5399 fprintf(output, "'%s' ", regexp->string);
5400 fprintf(output, "\n");
5401 fprintf(output, "%d atoms:\n", regexp->nbAtoms);
5402 for (i = 0;i < regexp->nbAtoms; i++) {
5403 fprintf(output, " %02d ", i);
5404 xmlRegPrintAtom(output, regexp->atoms[i]);
5405 }
5406 fprintf(output, "%d states:", regexp->nbStates);
5407 fprintf(output, "\n");
5408 for (i = 0;i < regexp->nbStates; i++) {
5409 xmlRegPrintState(output, regexp->states[i]);
5410 }
5411 fprintf(output, "%d counters:\n", regexp->nbCounters);
5412 for (i = 0;i < regexp->nbCounters; i++) {
5413 fprintf(output, " %d: min %d max %d\n", i, regexp->counters[i].min,
5414 regexp->counters[i].max);
5415 }
5416}
5417
5418/**
5419 * xmlRegexpCompile:
5420 * @regexp: a regular expression string
5421 *
5422 * Parses a regular expression conforming to XML Schemas Part 2 Datatype
5423 * Appendix F and builds an automata suitable for testing strings against
5424 * that regular expression
5425 *
5426 * Returns the compiled expression or NULL in case of error
5427 */
5428xmlRegexpPtr
5429xmlRegexpCompile(const xmlChar *regexp) {
5430 xmlRegexpPtr ret;
5431 xmlRegParserCtxtPtr ctxt;
5432
5433 ctxt = xmlRegNewParserCtxt(regexp);
5434 if (ctxt == NULL)
5435 return(NULL);
5436
5437 /* initialize the parser */
5438 ctxt->end = NULL;
5439 ctxt->start = ctxt->state = xmlRegNewState(ctxt);
5440 xmlRegStatePush(ctxt, ctxt->start);
5441
5442 /* parse the expression building an automata */
5443 xmlFAParseRegExp(ctxt, 1);
5444 if (CUR != 0) {
5445 ERROR("xmlFAParseRegExp: extra characters");
5446 }
5447 if (ctxt->error != 0) {
5448 xmlRegFreeParserCtxt(ctxt);
5449 return(NULL);
5450 }
5451 ctxt->end = ctxt->state;
5452 ctxt->start->type = XML_REGEXP_START_STATE;
5453 ctxt->end->type = XML_REGEXP_FINAL_STATE;
5454
5455 /* remove the Epsilon except for counted transitions */
5456 xmlFAEliminateEpsilonTransitions(ctxt);
5457
5458
5459 if (ctxt->error != 0) {
5460 xmlRegFreeParserCtxt(ctxt);
5461 return(NULL);
5462 }
5463 ret = xmlRegEpxFromParse(ctxt);
5464 xmlRegFreeParserCtxt(ctxt);
5465 return(ret);
5466}
5467
5468/**
5469 * xmlRegexpExec:
5470 * @comp: the compiled regular expression
5471 * @content: the value to check against the regular expression
5472 *
5473 * Check if the regular expression generates the value
5474 *
5475 * Returns 1 if it matches, 0 if not and a negative value in case of error
5476 */
5477int
5478xmlRegexpExec(xmlRegexpPtr comp, const xmlChar *content) {
5479 if ((comp == NULL) || (content == NULL))
5480 return(-1);
5481 return(xmlFARegExec(comp, content));
5482}
5483
5484/**
5485 * xmlRegexpIsDeterminist:
5486 * @comp: the compiled regular expression
5487 *
5488 * Check if the regular expression is determinist
5489 *
5490 * Returns 1 if it yes, 0 if not and a negative value in case of error
5491 */
5492int
5493xmlRegexpIsDeterminist(xmlRegexpPtr comp) {
5494 xmlAutomataPtr am;
5495 int ret;
5496
5497 if (comp == NULL)
5498 return(-1);
5499 if (comp->determinist != -1)
5500 return(comp->determinist);
5501
5502 am = xmlNewAutomata();
5503 if (am->states != NULL) {
5504 int i;
5505
5506 for (i = 0;i < am->nbStates;i++)
5507 xmlRegFreeState(am->states[i]);
5508 xmlFree(am->states);
5509 }
5510 am->nbAtoms = comp->nbAtoms;
5511 am->atoms = comp->atoms;
5512 am->nbStates = comp->nbStates;
5513 am->states = comp->states;
5514 am->determinist = -1;
5515 ret = xmlFAComputesDeterminism(am);
5516 am->atoms = NULL;
5517 am->states = NULL;
5518 xmlFreeAutomata(am);
5519 return(ret);
5520}
5521
5522/**
5523 * xmlRegFreeRegexp:
5524 * @regexp: the regexp
5525 *
5526 * Free a regexp
5527 */
5528void
5529xmlRegFreeRegexp(xmlRegexpPtr regexp) {
5530 int i;
5531 if (regexp == NULL)
5532 return;
5533
5534 if (regexp->string != NULL)
5535 xmlFree(regexp->string);
5536 if (regexp->states != NULL) {
5537 for (i = 0;i < regexp->nbStates;i++)
5538 xmlRegFreeState(regexp->states[i]);
5539 xmlFree(regexp->states);
5540 }
5541 if (regexp->atoms != NULL) {
5542 for (i = 0;i < regexp->nbAtoms;i++)
5543 xmlRegFreeAtom(regexp->atoms[i]);
5544 xmlFree(regexp->atoms);
5545 }
5546 if (regexp->counters != NULL)
5547 xmlFree(regexp->counters);
5548 if (regexp->compact != NULL)
5549 xmlFree(regexp->compact);
5550 if (regexp->transdata != NULL)
5551 xmlFree(regexp->transdata);
5552 if (regexp->stringMap != NULL) {
5553 for (i = 0; i < regexp->nbstrings;i++)
5554 xmlFree(regexp->stringMap[i]);
5555 xmlFree(regexp->stringMap);
5556 }
5557
5558 xmlFree(regexp);
5559}
5560
5561#ifdef LIBXML_AUTOMATA_ENABLED
5562/************************************************************************
5563 * *
5564 * The Automata interface *
5565 * *
5566 ************************************************************************/
5567
5568/**
5569 * xmlNewAutomata:
5570 *
5571 * Create a new automata
5572 *
5573 * Returns the new object or NULL in case of failure
5574 */
5575xmlAutomataPtr
5576xmlNewAutomata(void) {
5577 xmlAutomataPtr ctxt;
5578
5579 ctxt = xmlRegNewParserCtxt(NULL);
5580 if (ctxt == NULL)
5581 return(NULL);
5582
5583 /* initialize the parser */
5584 ctxt->end = NULL;
5585 ctxt->start = ctxt->state = xmlRegNewState(ctxt);
5586 if (ctxt->start == NULL) {
5587 xmlFreeAutomata(ctxt);
5588 return(NULL);
5589 }
5590 ctxt->start->type = XML_REGEXP_START_STATE;
5591 if (xmlRegStatePush(ctxt, ctxt->start) < 0) {
5592 xmlRegFreeState(ctxt->start);
5593 xmlFreeAutomata(ctxt);
5594 return(NULL);
5595 }
5596
5597 return(ctxt);
5598}
5599
5600/**
5601 * xmlFreeAutomata:
5602 * @am: an automata
5603 *
5604 * Free an automata
5605 */
5606void
5607xmlFreeAutomata(xmlAutomataPtr am) {
5608 if (am == NULL)
5609 return;
5610 xmlRegFreeParserCtxt(am);
5611}
5612
5613/**
5614 * xmlAutomataGetInitState:
5615 * @am: an automata
5616 *
5617 * Initial state lookup
5618 *
5619 * Returns the initial state of the automata
5620 */
5621xmlAutomataStatePtr
5622xmlAutomataGetInitState(xmlAutomataPtr am) {
5623 if (am == NULL)
5624 return(NULL);
5625 return(am->start);
5626}
5627
5628/**
5629 * xmlAutomataSetFinalState:
5630 * @am: an automata
5631 * @state: a state in this automata
5632 *
5633 * Makes that state a final state
5634 *
5635 * Returns 0 or -1 in case of error
5636 */
5637int
5638xmlAutomataSetFinalState(xmlAutomataPtr am, xmlAutomataStatePtr state) {
5639 if ((am == NULL) || (state == NULL))
5640 return(-1);
5641 state->type = XML_REGEXP_FINAL_STATE;
5642 return(0);
5643}
5644
5645/**
5646 * xmlAutomataNewTransition:
5647 * @am: an automata
5648 * @from: the starting point of the transition
5649 * @to: the target point of the transition or NULL
5650 * @token: the input string associated to that transition
5651 * @data: data passed to the callback function if the transition is activated
5652 *
5653 * If @to is NULL, this creates first a new target state in the automata
5654 * and then adds a transition from the @from state to the target state
5655 * activated by the value of @token
5656 *
5657 * Returns the target state or NULL in case of error
5658 */
5659xmlAutomataStatePtr
5660xmlAutomataNewTransition(xmlAutomataPtr am, xmlAutomataStatePtr from,
5661 xmlAutomataStatePtr to, const xmlChar *token,
5662 void *data) {
5663 xmlRegAtomPtr atom;
5664
5665 if ((am == NULL) || (from == NULL) || (token == NULL))
5666 return(NULL);
5667 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5668 if (atom == NULL)
5669 return(NULL);
5670 atom->data = data;
5671 if (atom == NULL)
5672 return(NULL);
5673 atom->valuep = xmlStrdup(token);
5674
5675 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5676 xmlRegFreeAtom(atom);
5677 return(NULL);
5678 }
5679 if (to == NULL)
5680 return(am->state);
5681 return(to);
5682}
5683
5684/**
5685 * xmlAutomataNewTransition2:
5686 * @am: an automata
5687 * @from: the starting point of the transition
5688 * @to: the target point of the transition or NULL
5689 * @token: the first input string associated to that transition
5690 * @token2: the second input string associated to that transition
5691 * @data: data passed to the callback function if the transition is activated
5692 *
5693 * If @to is NULL, this creates first a new target state in the automata
5694 * and then adds a transition from the @from state to the target state
5695 * activated by the value of @token
5696 *
5697 * Returns the target state or NULL in case of error
5698 */
5699xmlAutomataStatePtr
5700xmlAutomataNewTransition2(xmlAutomataPtr am, xmlAutomataStatePtr from,
5701 xmlAutomataStatePtr to, const xmlChar *token,
5702 const xmlChar *token2, void *data) {
5703 xmlRegAtomPtr atom;
5704
5705 if ((am == NULL) || (from == NULL) || (token == NULL))
5706 return(NULL);
5707 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5708 if (atom == NULL)
5709 return(NULL);
5710 atom->data = data;
5711 if ((token2 == NULL) || (*token2 == 0)) {
5712 atom->valuep = xmlStrdup(token);
5713 } else {
5714 int lenn, lenp;
5715 xmlChar *str;
5716
5717 lenn = strlen((char *) token2);
5718 lenp = strlen((char *) token);
5719
5720 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5721 if (str == NULL) {
5722 xmlRegFreeAtom(atom);
5723 return(NULL);
5724 }
5725 memcpy(&str[0], token, lenp);
5726 str[lenp] = '|';
5727 memcpy(&str[lenp + 1], token2, lenn);
5728 str[lenn + lenp + 1] = 0;
5729
5730 atom->valuep = str;
5731 }
5732
5733 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5734 xmlRegFreeAtom(atom);
5735 return(NULL);
5736 }
5737 if (to == NULL)
5738 return(am->state);
5739 return(to);
5740}
5741
5742/**
5743 * xmlAutomataNewNegTrans:
5744 * @am: an automata
5745 * @from: the starting point of the transition
5746 * @to: the target point of the transition or NULL
5747 * @token: the first input string associated to that transition
5748 * @token2: the second input string associated to that transition
5749 * @data: data passed to the callback function if the transition is activated
5750 *
5751 * If @to is NULL, this creates first a new target state in the automata
5752 * and then adds a transition from the @from state to the target state
5753 * activated by any value except (@token,@token2)
5754 * Note that if @token2 is not NULL, then (X, NULL) won't match to follow
5755 # the semantic of XSD ##other
5756 *
5757 * Returns the target state or NULL in case of error
5758 */
5759xmlAutomataStatePtr
5760xmlAutomataNewNegTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
5761 xmlAutomataStatePtr to, const xmlChar *token,
5762 const xmlChar *token2, void *data) {
5763 xmlRegAtomPtr atom;
5764 xmlChar err_msg[200];
5765
5766 if ((am == NULL) || (from == NULL) || (token == NULL))
5767 return(NULL);
5768 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5769 if (atom == NULL)
5770 return(NULL);
5771 atom->data = data;
5772 atom->neg = 1;
5773 if ((token2 == NULL) || (*token2 == 0)) {
5774 atom->valuep = xmlStrdup(token);
5775 } else {
5776 int lenn, lenp;
5777 xmlChar *str;
5778
5779 lenn = strlen((char *) token2);
5780 lenp = strlen((char *) token);
5781
5782 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5783 if (str == NULL) {
5784 xmlRegFreeAtom(atom);
5785 return(NULL);
5786 }
5787 memcpy(&str[0], token, lenp);
5788 str[lenp] = '|';
5789 memcpy(&str[lenp + 1], token2, lenn);
5790 str[lenn + lenp + 1] = 0;
5791
5792 atom->valuep = str;
5793 }
5794 snprintf((char *) err_msg, 199, "not %s", (const char *) atom->valuep);
5795 err_msg[199] = 0;
5796 atom->valuep2 = xmlStrdup(err_msg);
5797
5798 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5799 xmlRegFreeAtom(atom);
5800 return(NULL);
5801 }
5802 am->negs++;
5803 if (to == NULL)
5804 return(am->state);
5805 return(to);
5806}
5807
5808/**
5809 * xmlAutomataNewCountTrans2:
5810 * @am: an automata
5811 * @from: the starting point of the transition
5812 * @to: the target point of the transition or NULL
5813 * @token: the input string associated to that transition
5814 * @token2: the second input string associated to that transition
5815 * @min: the minimum successive occurences of token
5816 * @max: the maximum successive occurences of token
5817 * @data: data associated to the transition
5818 *
5819 * If @to is NULL, this creates first a new target state in the automata
5820 * and then adds a transition from the @from state to the target state
5821 * activated by a succession of input of value @token and @token2 and
5822 * whose number is between @min and @max
5823 *
5824 * Returns the target state or NULL in case of error
5825 */
5826xmlAutomataStatePtr
5827xmlAutomataNewCountTrans2(xmlAutomataPtr am, xmlAutomataStatePtr from,
5828 xmlAutomataStatePtr to, const xmlChar *token,
5829 const xmlChar *token2,
5830 int min, int max, void *data) {
5831 xmlRegAtomPtr atom;
5832 int counter;
5833
5834 if ((am == NULL) || (from == NULL) || (token == NULL))
5835 return(NULL);
5836 if (min < 0)
5837 return(NULL);
5838 if ((max < min) || (max < 1))
5839 return(NULL);
5840 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5841 if (atom == NULL)
5842 return(NULL);
5843 if ((token2 == NULL) || (*token2 == 0)) {
5844 atom->valuep = xmlStrdup(token);
5845 } else {
5846 int lenn, lenp;
5847 xmlChar *str;
5848
5849 lenn = strlen((char *) token2);
5850 lenp = strlen((char *) token);
5851
5852 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5853 if (str == NULL) {
5854 xmlRegFreeAtom(atom);
5855 return(NULL);
5856 }
5857 memcpy(&str[0], token, lenp);
5858 str[lenp] = '|';
5859 memcpy(&str[lenp + 1], token2, lenn);
5860 str[lenn + lenp + 1] = 0;
5861
5862 atom->valuep = str;
5863 }
5864 atom->data = data;
5865 if (min == 0)
5866 atom->min = 1;
5867 else
5868 atom->min = min;
5869 atom->max = max;
5870
5871 /*
5872 * associate a counter to the transition.
5873 */
5874 counter = xmlRegGetCounter(am);
5875 am->counters[counter].min = min;
5876 am->counters[counter].max = max;
5877
5878 /* xmlFAGenerateTransitions(am, from, to, atom); */
5879 if (to == NULL) {
5880 to = xmlRegNewState(am);
5881 xmlRegStatePush(am, to);
5882 }
5883 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
5884 xmlRegAtomPush(am, atom);
5885 am->state = to;
5886
5887 if (to == NULL)
5888 to = am->state;
5889 if (to == NULL)
5890 return(NULL);
5891 if (min == 0)
5892 xmlFAGenerateEpsilonTransition(am, from, to);
5893 return(to);
5894}
5895
5896/**
5897 * xmlAutomataNewCountTrans:
5898 * @am: an automata
5899 * @from: the starting point of the transition
5900 * @to: the target point of the transition or NULL
5901 * @token: the input string associated to that transition
5902 * @min: the minimum successive occurences of token
5903 * @max: the maximum successive occurences of token
5904 * @data: data associated to the transition
5905 *
5906 * If @to is NULL, this creates first a new target state in the automata
5907 * and then adds a transition from the @from state to the target state
5908 * activated by a succession of input of value @token and whose number
5909 * is between @min and @max
5910 *
5911 * Returns the target state or NULL in case of error
5912 */
5913xmlAutomataStatePtr
5914xmlAutomataNewCountTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
5915 xmlAutomataStatePtr to, const xmlChar *token,
5916 int min, int max, void *data) {
5917 xmlRegAtomPtr atom;
5918 int counter;
5919
5920 if ((am == NULL) || (from == NULL) || (token == NULL))
5921 return(NULL);
5922 if (min < 0)
5923 return(NULL);
5924 if ((max < min) || (max < 1))
5925 return(NULL);
5926 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5927 if (atom == NULL)
5928 return(NULL);
5929 atom->valuep = xmlStrdup(token);
5930 atom->data = data;
5931 if (min == 0)
5932 atom->min = 1;
5933 else
5934 atom->min = min;
5935 atom->max = max;
5936
5937 /*
5938 * associate a counter to the transition.
5939 */
5940 counter = xmlRegGetCounter(am);
5941 am->counters[counter].min = min;
5942 am->counters[counter].max = max;
5943
5944 /* xmlFAGenerateTransitions(am, from, to, atom); */
5945 if (to == NULL) {
5946 to = xmlRegNewState(am);
5947 xmlRegStatePush(am, to);
5948 }
5949 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
5950 xmlRegAtomPush(am, atom);
5951 am->state = to;
5952
5953 if (to == NULL)
5954 to = am->state;
5955 if (to == NULL)
5956 return(NULL);
5957 if (min == 0)
5958 xmlFAGenerateEpsilonTransition(am, from, to);
5959 return(to);
5960}
5961
5962/**
5963 * xmlAutomataNewOnceTrans2:
5964 * @am: an automata
5965 * @from: the starting point of the transition
5966 * @to: the target point of the transition or NULL
5967 * @token: the input string associated to that transition
5968 * @token2: the second input string associated to that transition
5969 * @min: the minimum successive occurences of token
5970 * @max: the maximum successive occurences of token
5971 * @data: data associated to the transition
5972 *
5973 * If @to is NULL, this creates first a new target state in the automata
5974 * and then adds a transition from the @from state to the target state
5975 * activated by a succession of input of value @token and @token2 and whose
5976 * number is between @min and @max, moreover that transition can only be
5977 * crossed once.
5978 *
5979 * Returns the target state or NULL in case of error
5980 */
5981xmlAutomataStatePtr
5982xmlAutomataNewOnceTrans2(xmlAutomataPtr am, xmlAutomataStatePtr from,
5983 xmlAutomataStatePtr to, const xmlChar *token,
5984 const xmlChar *token2,
5985 int min, int max, void *data) {
5986 xmlRegAtomPtr atom;
5987 int counter;
5988
5989 if ((am == NULL) || (from == NULL) || (token == NULL))
5990 return(NULL);
5991 if (min < 1)
5992 return(NULL);
5993 if ((max < min) || (max < 1))
5994 return(NULL);
5995 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5996 if (atom == NULL)
5997 return(NULL);
5998 if ((token2 == NULL) || (*token2 == 0)) {
5999 atom->valuep = xmlStrdup(token);
6000 } else {
6001 int lenn, lenp;
6002 xmlChar *str;
6003
6004 lenn = strlen((char *) token2);
6005 lenp = strlen((char *) token);
6006
6007 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
6008 if (str == NULL) {
6009 xmlRegFreeAtom(atom);
6010 return(NULL);
6011 }
6012 memcpy(&str[0], token, lenp);
6013 str[lenp] = '|';
6014 memcpy(&str[lenp + 1], token2, lenn);
6015 str[lenn + lenp + 1] = 0;
6016
6017 atom->valuep = str;
6018 }
6019 atom->data = data;
6020 atom->quant = XML_REGEXP_QUANT_ONCEONLY;
6021 atom->min = min;
6022 atom->max = max;
6023 /*
6024 * associate a counter to the transition.
6025 */
6026 counter = xmlRegGetCounter(am);
6027 am->counters[counter].min = 1;
6028 am->counters[counter].max = 1;
6029
6030 /* xmlFAGenerateTransitions(am, from, to, atom); */
6031 if (to == NULL) {
6032 to = xmlRegNewState(am);
6033 xmlRegStatePush(am, to);
6034 }
6035 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6036 xmlRegAtomPush(am, atom);
6037 am->state = to;
6038 return(to);
6039}
6040
6041
6042
6043/**
6044 * xmlAutomataNewOnceTrans:
6045 * @am: an automata
6046 * @from: the starting point of the transition
6047 * @to: the target point of the transition or NULL
6048 * @token: the input string associated to that transition
6049 * @min: the minimum successive occurences of token
6050 * @max: the maximum successive occurences of token
6051 * @data: data associated to the transition
6052 *
6053 * If @to is NULL, this creates first a new target state in the automata
6054 * and then adds a transition from the @from state to the target state
6055 * activated by a succession of input of value @token and whose number
6056 * is between @min and @max, moreover that transition can only be crossed
6057 * once.
6058 *
6059 * Returns the target state or NULL in case of error
6060 */
6061xmlAutomataStatePtr
6062xmlAutomataNewOnceTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6063 xmlAutomataStatePtr to, const xmlChar *token,
6064 int min, int max, void *data) {
6065 xmlRegAtomPtr atom;
6066 int counter;
6067
6068 if ((am == NULL) || (from == NULL) || (token == NULL))
6069 return(NULL);
6070 if (min < 1)
6071 return(NULL);
6072 if ((max < min) || (max < 1))
6073 return(NULL);
6074 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6075 if (atom == NULL)
6076 return(NULL);
6077 atom->valuep = xmlStrdup(token);
6078 atom->data = data;
6079 atom->quant = XML_REGEXP_QUANT_ONCEONLY;
6080 atom->min = min;
6081 atom->max = max;
6082 /*
6083 * associate a counter to the transition.
6084 */
6085 counter = xmlRegGetCounter(am);
6086 am->counters[counter].min = 1;
6087 am->counters[counter].max = 1;
6088
6089 /* xmlFAGenerateTransitions(am, from, to, atom); */
6090 if (to == NULL) {
6091 to = xmlRegNewState(am);
6092 xmlRegStatePush(am, to);
6093 }
6094 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6095 xmlRegAtomPush(am, atom);
6096 am->state = to;
6097 return(to);
6098}
6099
6100/**
6101 * xmlAutomataNewState:
6102 * @am: an automata
6103 *
6104 * Create a new disconnected state in the automata
6105 *
6106 * Returns the new state or NULL in case of error
6107 */
6108xmlAutomataStatePtr
6109xmlAutomataNewState(xmlAutomataPtr am) {
6110 xmlAutomataStatePtr to;
6111
6112 if (am == NULL)
6113 return(NULL);
6114 to = xmlRegNewState(am);
6115 xmlRegStatePush(am, to);
6116 return(to);
6117}
6118
6119/**
6120 * xmlAutomataNewEpsilon:
6121 * @am: an automata
6122 * @from: the starting point of the transition
6123 * @to: the target point of the transition or NULL
6124 *
6125 * If @to is NULL, this creates first a new target state in the automata
6126 * and then adds an epsilon transition from the @from state to the
6127 * target state
6128 *
6129 * Returns the target state or NULL in case of error
6130 */
6131xmlAutomataStatePtr
6132xmlAutomataNewEpsilon(xmlAutomataPtr am, xmlAutomataStatePtr from,
6133 xmlAutomataStatePtr to) {
6134 if ((am == NULL) || (from == NULL))
6135 return(NULL);
6136 xmlFAGenerateEpsilonTransition(am, from, to);
6137 if (to == NULL)
6138 return(am->state);
6139 return(to);
6140}
6141
6142/**
6143 * xmlAutomataNewAllTrans:
6144 * @am: an automata
6145 * @from: the starting point of the transition
6146 * @to: the target point of the transition or NULL
6147 * @lax: allow to transition if not all all transitions have been activated
6148 *
6149 * If @to is NULL, this creates first a new target state in the automata
6150 * and then adds a an ALL transition from the @from state to the
6151 * target state. That transition is an epsilon transition allowed only when
6152 * all transitions from the @from node have been activated.
6153 *
6154 * Returns the target state or NULL in case of error
6155 */
6156xmlAutomataStatePtr
6157xmlAutomataNewAllTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6158 xmlAutomataStatePtr to, int lax) {
6159 if ((am == NULL) || (from == NULL))
6160 return(NULL);
6161 xmlFAGenerateAllTransition(am, from, to, lax);
6162 if (to == NULL)
6163 return(am->state);
6164 return(to);
6165}
6166
6167/**
6168 * xmlAutomataNewCounter:
6169 * @am: an automata
6170 * @min: the minimal value on the counter
6171 * @max: the maximal value on the counter
6172 *
6173 * Create a new counter
6174 *
6175 * Returns the counter number or -1 in case of error
6176 */
6177int
6178xmlAutomataNewCounter(xmlAutomataPtr am, int min, int max) {
6179 int ret;
6180
6181 if (am == NULL)
6182 return(-1);
6183
6184 ret = xmlRegGetCounter(am);
6185 if (ret < 0)
6186 return(-1);
6187 am->counters[ret].min = min;
6188 am->counters[ret].max = max;
6189 return(ret);
6190}
6191
6192/**
6193 * xmlAutomataNewCountedTrans:
6194 * @am: an automata
6195 * @from: the starting point of the transition
6196 * @to: the target point of the transition or NULL
6197 * @counter: the counter associated to that transition
6198 *
6199 * If @to is NULL, this creates first a new target state in the automata
6200 * and then adds an epsilon transition from the @from state to the target state
6201 * which will increment the counter provided
6202 *
6203 * Returns the target state or NULL in case of error
6204 */
6205xmlAutomataStatePtr
6206xmlAutomataNewCountedTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6207 xmlAutomataStatePtr to, int counter) {
6208 if ((am == NULL) || (from == NULL) || (counter < 0))
6209 return(NULL);
6210 xmlFAGenerateCountedEpsilonTransition(am, from, to, counter);
6211 if (to == NULL)
6212 return(am->state);
6213 return(to);
6214}
6215
6216/**
6217 * xmlAutomataNewCounterTrans:
6218 * @am: an automata
6219 * @from: the starting point of the transition
6220 * @to: the target point of the transition or NULL
6221 * @counter: the counter associated to that transition
6222 *
6223 * If @to is NULL, this creates first a new target state in the automata
6224 * and then adds an epsilon transition from the @from state to the target state
6225 * which will be allowed only if the counter is within the right range.
6226 *
6227 * Returns the target state or NULL in case of error
6228 */
6229xmlAutomataStatePtr
6230xmlAutomataNewCounterTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6231 xmlAutomataStatePtr to, int counter) {
6232 if ((am == NULL) || (from == NULL) || (counter < 0))
6233 return(NULL);
6234 xmlFAGenerateCountedTransition(am, from, to, counter);
6235 if (to == NULL)
6236 return(am->state);
6237 return(to);
6238}
6239
6240/**
6241 * xmlAutomataCompile:
6242 * @am: an automata
6243 *
6244 * Compile the automata into a Reg Exp ready for being executed.
6245 * The automata should be free after this point.
6246 *
6247 * Returns the compiled regexp or NULL in case of error
6248 */
6249xmlRegexpPtr
6250xmlAutomataCompile(xmlAutomataPtr am) {
6251 xmlRegexpPtr ret;
6252
6253 if ((am == NULL) || (am->error != 0)) return(NULL);
6254 xmlFAEliminateEpsilonTransitions(am);
6255 /* xmlFAComputesDeterminism(am); */
6256 ret = xmlRegEpxFromParse(am);
6257
6258 return(ret);
6259}
6260
6261/**
6262 * xmlAutomataIsDeterminist:
6263 * @am: an automata
6264 *
6265 * Checks if an automata is determinist.
6266 *
6267 * Returns 1 if true, 0 if not, and -1 in case of error
6268 */
6269int
6270xmlAutomataIsDeterminist(xmlAutomataPtr am) {
6271 int ret;
6272
6273 if (am == NULL)
6274 return(-1);
6275
6276 ret = xmlFAComputesDeterminism(am);
6277 return(ret);
6278}
6279#endif /* LIBXML_AUTOMATA_ENABLED */
6280
6281#ifdef LIBXML_EXPR_ENABLED
6282/************************************************************************
6283 * *
6284 * Formal Expression handling code *
6285 * *
6286 ************************************************************************/
6287/************************************************************************
6288 * *
6289 * Expression handling context *
6290 * *
6291 ************************************************************************/
6292
6293struct _xmlExpCtxt {
6294 xmlDictPtr dict;
6295 xmlExpNodePtr *table;
6296 int size;
6297 int nbElems;
6298 int nb_nodes;
6299 const char *expr;
6300 const char *cur;
6301 int nb_cons;
6302 int tabSize;
6303};
6304
6305/**
6306 * xmlExpNewCtxt:
6307 * @maxNodes: the maximum number of nodes
6308 * @dict: optional dictionnary to use internally
6309 *
6310 * Creates a new context for manipulating expressions
6311 *
6312 * Returns the context or NULL in case of error
6313 */
6314xmlExpCtxtPtr
6315xmlExpNewCtxt(int maxNodes, xmlDictPtr dict) {
6316 xmlExpCtxtPtr ret;
6317 int size = 256;
6318
6319 if (maxNodes <= 4096)
6320 maxNodes = 4096;
6321
6322 ret = (xmlExpCtxtPtr) xmlMalloc(sizeof(xmlExpCtxt));
6323 if (ret == NULL)
6324 return(NULL);
6325 memset(ret, 0, sizeof(xmlExpCtxt));
6326 ret->size = size;
6327 ret->nbElems = 0;
6328 ret->table = xmlMalloc(size * sizeof(xmlExpNodePtr));
6329 if (ret->table == NULL) {
6330 xmlFree(ret);
6331 return(NULL);
6332 }
6333 memset(ret->table, 0, size * sizeof(xmlExpNodePtr));
6334 if (dict == NULL) {
6335 ret->dict = xmlDictCreate();
6336 if (ret->dict == NULL) {
6337 xmlFree(ret->table);
6338 xmlFree(ret);
6339 return(NULL);
6340 }
6341 } else {
6342 ret->dict = dict;
6343 xmlDictReference(ret->dict);
6344 }
6345 return(ret);
6346}
6347
6348/**
6349 * xmlExpFreeCtxt:
6350 * @ctxt: an expression context
6351 *
6352 * Free an expression context
6353 */
6354void
6355xmlExpFreeCtxt(xmlExpCtxtPtr ctxt) {
6356 if (ctxt == NULL)
6357 return;
6358 xmlDictFree(ctxt->dict);
6359 if (ctxt->table != NULL)
6360 xmlFree(ctxt->table);
6361 xmlFree(ctxt);
6362}
6363
6364/************************************************************************
6365 * *
6366 * Structure associated to an expression node *
6367 * *
6368 ************************************************************************/
6369#define MAX_NODES 10000
6370
6371/* #define DEBUG_DERIV */
6372
6373/*
6374 * TODO:
6375 * - Wildcards
6376 * - public API for creation
6377 *
6378 * Started
6379 * - regression testing
6380 *
6381 * Done
6382 * - split into module and test tool
6383 * - memleaks
6384 */
6385
6386typedef enum {
6387 XML_EXP_NILABLE = (1 << 0)
6388} xmlExpNodeInfo;
6389
6390#define IS_NILLABLE(node) ((node)->info & XML_EXP_NILABLE)
6391
6392struct _xmlExpNode {
6393 unsigned char type;/* xmlExpNodeType */
6394 unsigned char info;/* OR of xmlExpNodeInfo */
6395 unsigned short key; /* the hash key */
6396 unsigned int ref; /* The number of references */
6397 int c_max; /* the maximum length it can consume */
6398 xmlExpNodePtr exp_left;
6399 xmlExpNodePtr next;/* the next node in the hash table or free list */
6400 union {
6401 struct {
6402 int f_min;
6403 int f_max;
6404 } count;
6405 struct {
6406 xmlExpNodePtr f_right;
6407 } children;
6408 const xmlChar *f_str;
6409 } field;
6410};
6411
6412#define exp_min field.count.f_min
6413#define exp_max field.count.f_max
6414/* #define exp_left field.children.f_left */
6415#define exp_right field.children.f_right
6416#define exp_str field.f_str
6417
6418static xmlExpNodePtr xmlExpNewNode(xmlExpCtxtPtr ctxt, xmlExpNodeType type);
6419static xmlExpNode forbiddenExpNode = {
6420 XML_EXP_FORBID, 0, 0, 0, 0, NULL, NULL, {{ 0, 0}}
6421};
6422xmlExpNodePtr forbiddenExp = &forbiddenExpNode;
6423static xmlExpNode emptyExpNode = {
6424 XML_EXP_EMPTY, 1, 0, 0, 0, NULL, NULL, {{ 0, 0}}
6425};
6426xmlExpNodePtr emptyExp = &emptyExpNode;
6427
6428/************************************************************************
6429 * *
6430 * The custom hash table for unicity and canonicalization *
6431 * of sub-expressions pointers *
6432 * *
6433 ************************************************************************/
6434/*
6435 * xmlExpHashNameComputeKey:
6436 * Calculate the hash key for a token
6437 */
6438static unsigned short
6439xmlExpHashNameComputeKey(const xmlChar *name) {
6440 unsigned short value = 0L;
6441 char ch;
6442
6443 if (name != NULL) {
6444 value += 30 * (*name);
6445 while ((ch = *name++) != 0) {
6446 value = value ^ ((value << 5) + (value >> 3) + (unsigned long)ch);
6447 }
6448 }
6449 return (value);
6450}
6451
6452/*
6453 * xmlExpHashComputeKey:
6454 * Calculate the hash key for a compound expression
6455 */
6456static unsigned short
6457xmlExpHashComputeKey(xmlExpNodeType type, xmlExpNodePtr left,
6458 xmlExpNodePtr right) {
6459 unsigned long value;
6460 unsigned short ret;
6461
6462 switch (type) {
6463 case XML_EXP_SEQ:
6464 value = left->key;
6465 value += right->key;
6466 value *= 3;
6467 ret = (unsigned short) value;
6468 break;
6469 case XML_EXP_OR:
6470 value = left->key;
6471 value += right->key;
6472 value *= 7;
6473 ret = (unsigned short) value;
6474 break;
6475 case XML_EXP_COUNT:
6476 value = left->key;
6477 value += right->key;
6478 ret = (unsigned short) value;
6479 break;
6480 default:
6481 ret = 0;
6482 }
6483 return(ret);
6484}
6485
6486
6487static xmlExpNodePtr
6488xmlExpNewNode(xmlExpCtxtPtr ctxt, xmlExpNodeType type) {
6489 xmlExpNodePtr ret;
6490
6491 if (ctxt->nb_nodes >= MAX_NODES)
6492 return(NULL);
6493 ret = (xmlExpNodePtr) xmlMalloc(sizeof(xmlExpNode));
6494 if (ret == NULL)
6495 return(NULL);
6496 memset(ret, 0, sizeof(xmlExpNode));
6497 ret->type = type;
6498 ret->next = NULL;
6499 ctxt->nb_nodes++;
6500 ctxt->nb_cons++;
6501 return(ret);
6502}
6503
6504/**
6505 * xmlExpHashGetEntry:
6506 * @table: the hash table
6507 *
6508 * Get the unique entry from the hash table. The entry is created if
6509 * needed. @left and @right are consumed, i.e. their ref count will
6510 * be decremented by the operation.
6511 *
6512 * Returns the pointer or NULL in case of error
6513 */
6514static xmlExpNodePtr
6515xmlExpHashGetEntry(xmlExpCtxtPtr ctxt, xmlExpNodeType type,
6516 xmlExpNodePtr left, xmlExpNodePtr right,
6517 const xmlChar *name, int min, int max) {
6518 unsigned short kbase, key;
6519 xmlExpNodePtr entry;
6520 xmlExpNodePtr insert;
6521
6522 if (ctxt == NULL)
6523 return(NULL);
6524
6525 /*
6526 * Check for duplicate and insertion location.
6527 */
6528 if (type == XML_EXP_ATOM) {
6529 kbase = xmlExpHashNameComputeKey(name);
6530 } else if (type == XML_EXP_COUNT) {
6531 /* COUNT reduction rule 1 */
6532 /* a{1} -> a */
6533 if (min == max) {
6534 if (min == 1) {
6535 return(left);
6536 }
6537 if (min == 0) {
6538 xmlExpFree(ctxt, left);
6539 return(emptyExp);
6540 }
6541 }
6542 if (min < 0) {
6543 xmlExpFree(ctxt, left);
6544 return(forbiddenExp);
6545 }
6546 if (max == -1)
6547 kbase = min + 79;
6548 else
6549 kbase = max - min;
6550 kbase += left->key;
6551 } else if (type == XML_EXP_OR) {
6552 /* Forbid reduction rules */
6553 if (left->type == XML_EXP_FORBID) {
6554 xmlExpFree(ctxt, left);
6555 return(right);
6556 }
6557 if (right->type == XML_EXP_FORBID) {
6558 xmlExpFree(ctxt, right);
6559 return(left);
6560 }
6561
6562 /* OR reduction rule 1 */
6563 /* a | a reduced to a */
6564 if (left == right) {
6565 left->ref--;
6566 return(left);
6567 }
6568 /* OR canonicalization rule 1 */
6569 /* linearize (a | b) | c into a | (b | c) */
6570 if ((left->type == XML_EXP_OR) && (right->type != XML_EXP_OR)) {
6571 xmlExpNodePtr tmp = left;
6572 left = right;
6573 right = tmp;
6574 }
6575 /* OR reduction rule 2 */
6576 /* a | (a | b) and b | (a | b) are reduced to a | b */
6577 if (right->type == XML_EXP_OR) {
6578 if ((left == right->exp_left) ||
6579 (left == right->exp_right)) {
6580 xmlExpFree(ctxt, left);
6581 return(right);
6582 }
6583 }
6584 /* OR canonicalization rule 2 */
6585 /* linearize (a | b) | c into a | (b | c) */
6586 if (left->type == XML_EXP_OR) {
6587 xmlExpNodePtr tmp;
6588
6589 /* OR canonicalization rule 2 */
6590 if ((left->exp_right->type != XML_EXP_OR) &&
6591 (left->exp_right->key < left->exp_left->key)) {
6592 tmp = left->exp_right;
6593 left->exp_right = left->exp_left;
6594 left->exp_left = tmp;
6595 }
6596 left->exp_right->ref++;
6597 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left->exp_right, right,
6598 NULL, 0, 0);
6599 left->exp_left->ref++;
6600 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left->exp_left, tmp,
6601 NULL, 0, 0);
6602
6603 xmlExpFree(ctxt, left);
6604 return(tmp);
6605 }
6606 if (right->type == XML_EXP_OR) {
6607 /* Ordering in the tree */
6608 /* C | (A | B) -> A | (B | C) */
6609 if (left->key > right->exp_right->key) {
6610 xmlExpNodePtr tmp;
6611 right->exp_right->ref++;
6612 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_right,
6613 left, NULL, 0, 0);
6614 right->exp_left->ref++;
6615 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_left,
6616 tmp, NULL, 0, 0);
6617 xmlExpFree(ctxt, right);
6618 return(tmp);
6619 }
6620 /* Ordering in the tree */
6621 /* B | (A | C) -> A | (B | C) */
6622 if (left->key > right->exp_left->key) {
6623 xmlExpNodePtr tmp;
6624 right->exp_right->ref++;
6625 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left,
6626 right->exp_right, NULL, 0, 0);
6627 right->exp_left->ref++;
6628 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_left,
6629 tmp, NULL, 0, 0);
6630 xmlExpFree(ctxt, right);
6631 return(tmp);
6632 }
6633 }
6634 /* we know both types are != XML_EXP_OR here */
6635 else if (left->key > right->key) {
6636 xmlExpNodePtr tmp = left;
6637 left = right;
6638 right = tmp;
6639 }
6640 kbase = xmlExpHashComputeKey(type, left, right);
6641 } else if (type == XML_EXP_SEQ) {
6642 /* Forbid reduction rules */
6643 if (left->type == XML_EXP_FORBID) {
6644 xmlExpFree(ctxt, right);
6645 return(left);
6646 }
6647 if (right->type == XML_EXP_FORBID) {
6648 xmlExpFree(ctxt, left);
6649 return(right);
6650 }
6651 /* Empty reduction rules */
6652 if (right->type == XML_EXP_EMPTY) {
6653 return(left);
6654 }
6655 if (left->type == XML_EXP_EMPTY) {
6656 return(right);
6657 }
6658 kbase = xmlExpHashComputeKey(type, left, right);
6659 } else
6660 return(NULL);
6661
6662 key = kbase % ctxt->size;
6663 if (ctxt->table[key] != NULL) {
6664 for (insert = ctxt->table[key]; insert != NULL;
6665 insert = insert->next) {
6666 if ((insert->key == kbase) &&
6667 (insert->type == type)) {
6668 if (type == XML_EXP_ATOM) {
6669 if (name == insert->exp_str) {
6670 insert->ref++;
6671 return(insert);
6672 }
6673 } else if (type == XML_EXP_COUNT) {
6674 if ((insert->exp_min == min) && (insert->exp_max == max) &&
6675 (insert->exp_left == left)) {
6676 insert->ref++;
6677 left->ref--;
6678 return(insert);
6679 }
6680 } else if ((insert->exp_left == left) &&
6681 (insert->exp_right == right)) {
6682 insert->ref++;
6683 left->ref--;
6684 right->ref--;
6685 return(insert);
6686 }
6687 }
6688 }
6689 }
6690
6691 entry = xmlExpNewNode(ctxt, type);
6692 if (entry == NULL)
6693 return(NULL);
6694 entry->key = kbase;
6695 if (type == XML_EXP_ATOM) {
6696 entry->exp_str = name;
6697 entry->c_max = 1;
6698 } else if (type == XML_EXP_COUNT) {
6699 entry->exp_min = min;
6700 entry->exp_max = max;
6701 entry->exp_left = left;
6702 if ((min == 0) || (IS_NILLABLE(left)))
6703 entry->info |= XML_EXP_NILABLE;
6704 if (max < 0)
6705 entry->c_max = -1;
6706 else
6707 entry->c_max = max * entry->exp_left->c_max;
6708 } else {
6709 entry->exp_left = left;
6710 entry->exp_right = right;
6711 if (type == XML_EXP_OR) {
6712 if ((IS_NILLABLE(left)) || (IS_NILLABLE(right)))
6713 entry->info |= XML_EXP_NILABLE;
6714 if ((entry->exp_left->c_max == -1) ||
6715 (entry->exp_right->c_max == -1))
6716 entry->c_max = -1;
6717 else if (entry->exp_left->c_max > entry->exp_right->c_max)
6718 entry->c_max = entry->exp_left->c_max;
6719 else
6720 entry->c_max = entry->exp_right->c_max;
6721 } else {
6722 if ((IS_NILLABLE(left)) && (IS_NILLABLE(right)))
6723 entry->info |= XML_EXP_NILABLE;
6724 if ((entry->exp_left->c_max == -1) ||
6725 (entry->exp_right->c_max == -1))
6726 entry->c_max = -1;
6727 else
6728 entry->c_max = entry->exp_left->c_max + entry->exp_right->c_max;
6729 }
6730 }
6731 entry->ref = 1;
6732 if (ctxt->table[key] != NULL)
6733 entry->next = ctxt->table[key];
6734
6735 ctxt->table[key] = entry;
6736 ctxt->nbElems++;
6737
6738 return(entry);
6739}
6740
6741/**
6742 * xmlExpFree:
6743 * @ctxt: the expression context
6744 * @exp: the expression
6745 *
6746 * Dereference the expression
6747 */
6748void
6749xmlExpFree(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp) {
6750 if ((exp == NULL) || (exp == forbiddenExp) || (exp == emptyExp))
6751 return;
6752 exp->ref--;
6753 if (exp->ref == 0) {
6754 unsigned short key;
6755
6756 /* Unlink it first from the hash table */
6757 key = exp->key % ctxt->size;
6758 if (ctxt->table[key] == exp) {
6759 ctxt->table[key] = exp->next;
6760 } else {
6761 xmlExpNodePtr tmp;
6762
6763 tmp = ctxt->table[key];
6764 while (tmp != NULL) {
6765 if (tmp->next == exp) {
6766 tmp->next = exp->next;
6767 break;
6768 }
6769 tmp = tmp->next;
6770 }
6771 }
6772
6773 if ((exp->type == XML_EXP_SEQ) || (exp->type == XML_EXP_OR)) {
6774 xmlExpFree(ctxt, exp->exp_left);
6775 xmlExpFree(ctxt, exp->exp_right);
6776 } else if (exp->type == XML_EXP_COUNT) {
6777 xmlExpFree(ctxt, exp->exp_left);
6778 }
6779 xmlFree(exp);
6780 ctxt->nb_nodes--;
6781 }
6782}
6783
6784/**
6785 * xmlExpRef:
6786 * @exp: the expression
6787 *
6788 * Increase the reference count of the expression
6789 */
6790void
6791xmlExpRef(xmlExpNodePtr exp) {
6792 if (exp != NULL)
6793 exp->ref++;
6794}
6795
6796/**
6797 * xmlExpNewAtom:
6798 * @ctxt: the expression context
6799 * @name: the atom name
6800 * @len: the atom name lenght in byte (or -1);
6801 *
6802 * Get the atom associated to this name from that context
6803 *
6804 * Returns the node or NULL in case of error
6805 */
6806xmlExpNodePtr
6807xmlExpNewAtom(xmlExpCtxtPtr ctxt, const xmlChar *name, int len) {
6808 if ((ctxt == NULL) || (name == NULL))
6809 return(NULL);
6810 name = xmlDictLookup(ctxt->dict, name, len);
6811 if (name == NULL)
6812 return(NULL);
6813 return(xmlExpHashGetEntry(ctxt, XML_EXP_ATOM, NULL, NULL, name, 0, 0));
6814}
6815
6816/**
6817 * xmlExpNewOr:
6818 * @ctxt: the expression context
6819 * @left: left expression
6820 * @right: right expression
6821 *
6822 * Get the atom associated to the choice @left | @right
6823 * Note that @left and @right are consumed in the operation, to keep
6824 * an handle on them use xmlExpRef() and use xmlExpFree() to release them,
6825 * this is true even in case of failure (unless ctxt == NULL).
6826 *
6827 * Returns the node or NULL in case of error
6828 */
6829xmlExpNodePtr
6830xmlExpNewOr(xmlExpCtxtPtr ctxt, xmlExpNodePtr left, xmlExpNodePtr right) {
6831 if (ctxt == NULL)
6832 return(NULL);
6833 if ((left == NULL) || (right == NULL)) {
6834 xmlExpFree(ctxt, left);
6835 xmlExpFree(ctxt, right);
6836 return(NULL);
6837 }
6838 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, left, right, NULL, 0, 0));
6839}
6840
6841/**
6842 * xmlExpNewSeq:
6843 * @ctxt: the expression context
6844 * @left: left expression
6845 * @right: right expression
6846 *
6847 * Get the atom associated to the sequence @left , @right
6848 * Note that @left and @right are consumed in the operation, to keep
6849 * an handle on them use xmlExpRef() and use xmlExpFree() to release them,
6850 * this is true even in case of failure (unless ctxt == NULL).
6851 *
6852 * Returns the node or NULL in case of error
6853 */
6854xmlExpNodePtr
6855xmlExpNewSeq(xmlExpCtxtPtr ctxt, xmlExpNodePtr left, xmlExpNodePtr right) {
6856 if (ctxt == NULL)
6857 return(NULL);
6858 if ((left == NULL) || (right == NULL)) {
6859 xmlExpFree(ctxt, left);
6860 xmlExpFree(ctxt, right);
6861 return(NULL);
6862 }
6863 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, left, right, NULL, 0, 0));
6864}
6865
6866/**
6867 * xmlExpNewRange:
6868 * @ctxt: the expression context
6869 * @subset: the expression to be repeated
6870 * @min: the lower bound for the repetition
6871 * @max: the upper bound for the repetition, -1 means infinite
6872 *
6873 * Get the atom associated to the range (@subset){@min, @max}
6874 * Note that @subset is consumed in the operation, to keep
6875 * an handle on it use xmlExpRef() and use xmlExpFree() to release it,
6876 * this is true even in case of failure (unless ctxt == NULL).
6877 *
6878 * Returns the node or NULL in case of error
6879 */
6880xmlExpNodePtr
6881xmlExpNewRange(xmlExpCtxtPtr ctxt, xmlExpNodePtr subset, int min, int max) {
6882 if (ctxt == NULL)
6883 return(NULL);
6884 if ((subset == NULL) || (min < 0) || (max < -1) ||
6885 ((max >= 0) && (min > max))) {
6886 xmlExpFree(ctxt, subset);
6887 return(NULL);
6888 }
6889 return(xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, subset,
6890 NULL, NULL, min, max));
6891}
6892
6893/************************************************************************
6894 * *
6895 * Public API for operations on expressions *
6896 * *
6897 ************************************************************************/
6898
6899static int
6900xmlExpGetLanguageInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
6901 const xmlChar**list, int len, int nb) {
6902 int tmp, tmp2;
6903tail:
6904 switch (exp->type) {
6905 case XML_EXP_EMPTY:
6906 return(0);
6907 case XML_EXP_ATOM:
6908 for (tmp = 0;tmp < nb;tmp++)
6909 if (list[tmp] == exp->exp_str)
6910 return(0);
6911 if (nb >= len)
6912 return(-2);
6913 list[nb++] = exp->exp_str;
6914 return(1);
6915 case XML_EXP_COUNT:
6916 exp = exp->exp_left;
6917 goto tail;
6918 case XML_EXP_SEQ:
6919 case XML_EXP_OR:
6920 tmp = xmlExpGetLanguageInt(ctxt, exp->exp_left, list, len, nb);
6921 if (tmp < 0)
6922 return(tmp);
6923 tmp2 = xmlExpGetLanguageInt(ctxt, exp->exp_right, list, len,
6924 nb + tmp);
6925 if (tmp2 < 0)
6926 return(tmp2);
6927 return(tmp + tmp2);
6928 }
6929 return(-1);
6930}
6931
6932/**
6933 * xmlExpGetLanguage:
6934 * @ctxt: the expression context
6935 * @exp: the expression
6936 * @langList: where to store the tokens
6937 * @len: the allocated lenght of @list
6938 *
6939 * Find all the strings used in @exp and store them in @list
6940 *
6941 * Returns the number of unique strings found, -1 in case of errors and
6942 * -2 if there is more than @len strings
6943 */
6944int
6945xmlExpGetLanguage(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
6946 const xmlChar**langList, int len) {
6947 if ((ctxt == NULL) || (exp == NULL) || (langList == NULL) || (len <= 0))
6948 return(-1);
6949 return(xmlExpGetLanguageInt(ctxt, exp, langList, len, 0));
6950}
6951
6952static int
6953xmlExpGetStartInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
6954 const xmlChar**list, int len, int nb) {
6955 int tmp, tmp2;
6956tail:
6957 switch (exp->type) {
6958 case XML_EXP_FORBID:
6959 return(0);
6960 case XML_EXP_EMPTY:
6961 return(0);
6962 case XML_EXP_ATOM:
6963 for (tmp = 0;tmp < nb;tmp++)
6964 if (list[tmp] == exp->exp_str)
6965 return(0);
6966 if (nb >= len)
6967 return(-2);
6968 list[nb++] = exp->exp_str;
6969 return(1);
6970 case XML_EXP_COUNT:
6971 exp = exp->exp_left;
6972 goto tail;
6973 case XML_EXP_SEQ:
6974 tmp = xmlExpGetStartInt(ctxt, exp->exp_left, list, len, nb);
6975 if (tmp < 0)
6976 return(tmp);
6977 if (IS_NILLABLE(exp->exp_left)) {
6978 tmp2 = xmlExpGetStartInt(ctxt, exp->exp_right, list, len,
6979 nb + tmp);
6980 if (tmp2 < 0)
6981 return(tmp2);
6982 tmp += tmp2;
6983 }
6984 return(tmp);
6985 case XML_EXP_OR:
6986 tmp = xmlExpGetStartInt(ctxt, exp->exp_left, list, len, nb);
6987 if (tmp < 0)
6988 return(tmp);
6989 tmp2 = xmlExpGetStartInt(ctxt, exp->exp_right, list, len,
6990 nb + tmp);
6991 if (tmp2 < 0)
6992 return(tmp2);
6993 return(tmp + tmp2);
6994 }
6995 return(-1);
6996}
6997
6998/**
6999 * xmlExpGetStart:
7000 * @ctxt: the expression context
7001 * @exp: the expression
7002 * @tokList: where to store the tokens
7003 * @len: the allocated lenght of @list
7004 *
7005 * Find all the strings that appears at the start of the languages
7006 * accepted by @exp and store them in @list. E.g. for (a, b) | c
7007 * it will return the list [a, c]
7008 *
7009 * Returns the number of unique strings found, -1 in case of errors and
7010 * -2 if there is more than @len strings
7011 */
7012int
7013xmlExpGetStart(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7014 const xmlChar**tokList, int len) {
7015 if ((ctxt == NULL) || (exp == NULL) || (tokList == NULL) || (len <= 0))
7016 return(-1);
7017 return(xmlExpGetStartInt(ctxt, exp, tokList, len, 0));
7018}
7019
7020/**
7021 * xmlExpIsNillable:
7022 * @exp: the expression
7023 *
7024 * Finds if the expression is nillable, i.e. if it accepts the empty sequqnce
7025 *
7026 * Returns 1 if nillable, 0 if not and -1 in case of error
7027 */
7028int
7029xmlExpIsNillable(xmlExpNodePtr exp) {
7030 if (exp == NULL)
7031 return(-1);
7032 return(IS_NILLABLE(exp) != 0);
7033}
7034
7035static xmlExpNodePtr
7036xmlExpStringDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, const xmlChar *str)
7037{
7038 xmlExpNodePtr ret;
7039
7040 switch (exp->type) {
7041 case XML_EXP_EMPTY:
7042 return(forbiddenExp);
7043 case XML_EXP_FORBID:
7044 return(forbiddenExp);
7045 case XML_EXP_ATOM:
7046 if (exp->exp_str == str) {
7047#ifdef DEBUG_DERIV
7048 printf("deriv atom: equal => Empty\n");
7049#endif
7050 ret = emptyExp;
7051 } else {
7052#ifdef DEBUG_DERIV
7053 printf("deriv atom: mismatch => forbid\n");
7054#endif
7055 /* TODO wildcards here */
7056 ret = forbiddenExp;
7057 }
7058 return(ret);
7059 case XML_EXP_OR: {
7060 xmlExpNodePtr tmp;
7061
7062#ifdef DEBUG_DERIV
7063 printf("deriv or: => or(derivs)\n");
7064#endif
7065 tmp = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7066 if (tmp == NULL) {
7067 return(NULL);
7068 }
7069 ret = xmlExpStringDeriveInt(ctxt, exp->exp_right, str);
7070 if (ret == NULL) {
7071 xmlExpFree(ctxt, tmp);
7072 return(NULL);
7073 }
7074 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, tmp, ret,
7075 NULL, 0, 0);
7076 return(ret);
7077 }
7078 case XML_EXP_SEQ:
7079#ifdef DEBUG_DERIV
7080 printf("deriv seq: starting with left\n");
7081#endif
7082 ret = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7083 if (ret == NULL) {
7084 return(NULL);
7085 } else if (ret == forbiddenExp) {
7086 if (IS_NILLABLE(exp->exp_left)) {
7087#ifdef DEBUG_DERIV
7088 printf("deriv seq: left failed but nillable\n");
7089#endif
7090 ret = xmlExpStringDeriveInt(ctxt, exp->exp_right, str);
7091 }
7092 } else {
7093#ifdef DEBUG_DERIV
7094 printf("deriv seq: left match => sequence\n");
7095#endif
7096 exp->exp_right->ref++;
7097 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, exp->exp_right,
7098 NULL, 0, 0);
7099 }
7100 return(ret);
7101 case XML_EXP_COUNT: {
7102 int min, max;
7103 xmlExpNodePtr tmp;
7104
7105 if (exp->exp_max == 0)
7106 return(forbiddenExp);
7107 ret = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7108 if (ret == NULL)
7109 return(NULL);
7110 if (ret == forbiddenExp) {
7111#ifdef DEBUG_DERIV
7112 printf("deriv count: pattern mismatch => forbid\n");
7113#endif
7114 return(ret);
7115 }
7116 if (exp->exp_max == 1)
7117 return(ret);
7118 if (exp->exp_max < 0) /* unbounded */
7119 max = -1;
7120 else
7121 max = exp->exp_max - 1;
7122 if (exp->exp_min > 0)
7123 min = exp->exp_min - 1;
7124 else
7125 min = 0;
7126 exp->exp_left->ref++;
7127 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left, NULL,
7128 NULL, min, max);
7129 if (ret == emptyExp) {
7130#ifdef DEBUG_DERIV
7131 printf("deriv count: match to empty => new count\n");
7132#endif
7133 return(tmp);
7134 }
7135#ifdef DEBUG_DERIV
7136 printf("deriv count: match => sequence with new count\n");
7137#endif
7138 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, tmp,
7139 NULL, 0, 0));
7140 }
7141 }
7142 return(NULL);
7143}
7144
7145/**
7146 * xmlExpStringDerive:
7147 * @ctxt: the expression context
7148 * @exp: the expression
7149 * @str: the string
7150 * @len: the string len in bytes if available
7151 *
7152 * Do one step of Brzozowski derivation of the expression @exp with
7153 * respect to the input string
7154 *
7155 * Returns the resulting expression or NULL in case of internal error
7156 */
7157xmlExpNodePtr
7158xmlExpStringDerive(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7159 const xmlChar *str, int len) {
7160 const xmlChar *input;
7161
7162 if ((exp == NULL) || (ctxt == NULL) || (str == NULL)) {
7163 return(NULL);
7164 }
7165 /*
7166 * check the string is in the dictionnary, if yes use an interned
7167 * copy, otherwise we know it's not an acceptable input
7168 */
7169 input = xmlDictExists(ctxt->dict, str, len);
7170 if (input == NULL) {
7171 return(forbiddenExp);
7172 }
7173 return(xmlExpStringDeriveInt(ctxt, exp, input));
7174}
7175
7176static int
7177xmlExpCheckCard(xmlExpNodePtr exp, xmlExpNodePtr sub) {
7178 int ret = 1;
7179
7180 if (sub->c_max == -1) {
7181 if (exp->c_max != -1)
7182 ret = 0;
7183 } else if ((exp->c_max >= 0) && (exp->c_max < sub->c_max)) {
7184 ret = 0;
7185 }
7186#if 0
7187 if ((IS_NILLABLE(sub)) && (!IS_NILLABLE(exp)))
7188 ret = 0;
7189#endif
7190 return(ret);
7191}
7192
7193static xmlExpNodePtr xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7194 xmlExpNodePtr sub);
7195/**
7196 * xmlExpDivide:
7197 * @ctxt: the expressions context
7198 * @exp: the englobing expression
7199 * @sub: the subexpression
7200 * @mult: the multiple expression
7201 * @remain: the remain from the derivation of the multiple
7202 *
7203 * Check if exp is a multiple of sub, i.e. if there is a finite number n
7204 * so that sub{n} subsume exp
7205 *
7206 * Returns the multiple value if successful, 0 if it is not a multiple
7207 * and -1 in case of internel error.
7208 */
7209
7210static int
7211xmlExpDivide(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub,
7212 xmlExpNodePtr *mult, xmlExpNodePtr *remain) {
7213 int i;
7214 xmlExpNodePtr tmp, tmp2;
7215
7216 if (mult != NULL) *mult = NULL;
7217 if (remain != NULL) *remain = NULL;
7218 if (exp->c_max == -1) return(0);
7219 if (IS_NILLABLE(exp) && (!IS_NILLABLE(sub))) return(0);
7220
7221 for (i = 1;i <= exp->c_max;i++) {
7222 sub->ref++;
7223 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT,
7224 sub, NULL, NULL, i, i);
7225 if (tmp == NULL) {
7226 return(-1);
7227 }
7228 if (!xmlExpCheckCard(tmp, exp)) {
7229 xmlExpFree(ctxt, tmp);
7230 continue;
7231 }
7232 tmp2 = xmlExpExpDeriveInt(ctxt, tmp, exp);
7233 if (tmp2 == NULL) {
7234 xmlExpFree(ctxt, tmp);
7235 return(-1);
7236 }
7237 if ((tmp2 != forbiddenExp) && (IS_NILLABLE(tmp2))) {
7238 if (remain != NULL)
7239 *remain = tmp2;
7240 else
7241 xmlExpFree(ctxt, tmp2);
7242 if (mult != NULL)
7243 *mult = tmp;
7244 else
7245 xmlExpFree(ctxt, tmp);
7246#ifdef DEBUG_DERIV
7247 printf("Divide succeeded %d\n", i);
7248#endif
7249 return(i);
7250 }
7251 xmlExpFree(ctxt, tmp);
7252 xmlExpFree(ctxt, tmp2);
7253 }
7254#ifdef DEBUG_DERIV
7255 printf("Divide failed\n");
7256#endif
7257 return(0);
7258}
7259
7260/**
7261 * xmlExpExpDeriveInt:
7262 * @ctxt: the expressions context
7263 * @exp: the englobing expression
7264 * @sub: the subexpression
7265 *
7266 * Try to do a step of Brzozowski derivation but at a higher level
7267 * the input being a subexpression.
7268 *
7269 * Returns the resulting expression or NULL in case of internal error
7270 */
7271static xmlExpNodePtr
7272xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7273 xmlExpNodePtr ret, tmp, tmp2, tmp3;
7274 const xmlChar **tab;
7275 int len, i;
7276
7277 /*
7278 * In case of equality and if the expression can only consume a finite
7279 * amount, then the derivation is empty
7280 */
7281 if ((exp == sub) && (exp->c_max >= 0)) {
7282#ifdef DEBUG_DERIV
7283 printf("Equal(exp, sub) and finite -> Empty\n");
7284#endif
7285 return(emptyExp);
7286 }
7287 /*
7288 * decompose sub sequence first
7289 */
7290 if (sub->type == XML_EXP_EMPTY) {
7291#ifdef DEBUG_DERIV
7292 printf("Empty(sub) -> Empty\n");
7293#endif
7294 exp->ref++;
7295 return(exp);
7296 }
7297 if (sub->type == XML_EXP_SEQ) {
7298#ifdef DEBUG_DERIV
7299 printf("Seq(sub) -> decompose\n");
7300#endif
7301 tmp = xmlExpExpDeriveInt(ctxt, exp, sub->exp_left);
7302 if (tmp == NULL)
7303 return(NULL);
7304 if (tmp == forbiddenExp)
7305 return(tmp);
7306 ret = xmlExpExpDeriveInt(ctxt, tmp, sub->exp_right);
7307 xmlExpFree(ctxt, tmp);
7308 return(ret);
7309 }
7310 if (sub->type == XML_EXP_OR) {
7311#ifdef DEBUG_DERIV
7312 printf("Or(sub) -> decompose\n");
7313#endif
7314 tmp = xmlExpExpDeriveInt(ctxt, exp, sub->exp_left);
7315 if (tmp == forbiddenExp)
7316 return(tmp);
7317 if (tmp == NULL)
7318 return(NULL);
7319 ret = xmlExpExpDeriveInt(ctxt, exp, sub->exp_right);
7320 if ((ret == NULL) || (ret == forbiddenExp)) {
7321 xmlExpFree(ctxt, tmp);
7322 return(ret);
7323 }
7324 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, tmp, ret, NULL, 0, 0));
7325 }
7326 if (!xmlExpCheckCard(exp, sub)) {
7327#ifdef DEBUG_DERIV
7328 printf("CheckCard(exp, sub) failed -> Forbid\n");
7329#endif
7330 return(forbiddenExp);
7331 }
7332 switch (exp->type) {
7333 case XML_EXP_EMPTY:
7334 if (sub == emptyExp)
7335 return(emptyExp);
7336#ifdef DEBUG_DERIV
7337 printf("Empty(exp) -> Forbid\n");
7338#endif
7339 return(forbiddenExp);
7340 case XML_EXP_FORBID:
7341#ifdef DEBUG_DERIV
7342 printf("Forbid(exp) -> Forbid\n");
7343#endif
7344 return(forbiddenExp);
7345 case XML_EXP_ATOM:
7346 if (sub->type == XML_EXP_ATOM) {
7347 /* TODO: handle wildcards */
7348 if (exp->exp_str == sub->exp_str) {
7349#ifdef DEBUG_DERIV
7350 printf("Atom match -> Empty\n");
7351#endif
7352 return(emptyExp);
7353 }
7354#ifdef DEBUG_DERIV
7355 printf("Atom mismatch -> Forbid\n");
7356#endif
7357 return(forbiddenExp);
7358 }
7359 if ((sub->type == XML_EXP_COUNT) &&
7360 (sub->exp_max == 1) &&
7361 (sub->exp_left->type == XML_EXP_ATOM)) {
7362 /* TODO: handle wildcards */
7363 if (exp->exp_str == sub->exp_left->exp_str) {
7364#ifdef DEBUG_DERIV
7365 printf("Atom match -> Empty\n");
7366#endif
7367 return(emptyExp);
7368 }
7369#ifdef DEBUG_DERIV
7370 printf("Atom mismatch -> Forbid\n");
7371#endif
7372 return(forbiddenExp);
7373 }
7374#ifdef DEBUG_DERIV
7375 printf("Compex exp vs Atom -> Forbid\n");
7376#endif
7377 return(forbiddenExp);
7378 case XML_EXP_SEQ:
7379 /* try to get the sequence consumed only if possible */
7380 if (xmlExpCheckCard(exp->exp_left, sub)) {
7381 /* See if the sequence can be consumed directly */
7382#ifdef DEBUG_DERIV
7383 printf("Seq trying left only\n");
7384#endif
7385 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7386 if ((ret != forbiddenExp) && (ret != NULL)) {
7387#ifdef DEBUG_DERIV
7388 printf("Seq trying left only worked\n");
7389#endif
7390 /*
7391 * TODO: assumption here that we are determinist
7392 * i.e. we won't get to a nillable exp left
7393 * subset which could be matched by the right
7394 * part too.
7395 * e.g.: (a | b)+,(a | c) and 'a+,a'
7396 */
7397 exp->exp_right->ref++;
7398 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret,
7399 exp->exp_right, NULL, 0, 0));
7400 }
7401#ifdef DEBUG_DERIV
7402 } else {
7403 printf("Seq: left too short\n");
7404#endif
7405 }
7406 /* Try instead to decompose */
7407 if (sub->type == XML_EXP_COUNT) {
7408 int min, max;
7409
7410#ifdef DEBUG_DERIV
7411 printf("Seq: sub is a count\n");
7412#endif
7413 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub->exp_left);
7414 if (ret == NULL)
7415 return(NULL);
7416 if (ret != forbiddenExp) {
7417#ifdef DEBUG_DERIV
7418 printf("Seq , Count match on left\n");
7419#endif
7420 if (sub->exp_max < 0)
7421 max = -1;
7422 else
7423 max = sub->exp_max -1;
7424 if (sub->exp_min > 0)
7425 min = sub->exp_min -1;
7426 else
7427 min = 0;
7428 exp->exp_right->ref++;
7429 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret,
7430 exp->exp_right, NULL, 0, 0);
7431 if (tmp == NULL)
7432 return(NULL);
7433
7434 sub->exp_left->ref++;
7435 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT,
7436 sub->exp_left, NULL, NULL, min, max);
7437 if (tmp2 == NULL) {
7438 xmlExpFree(ctxt, tmp);
7439 return(NULL);
7440 }
7441 ret = xmlExpExpDeriveInt(ctxt, tmp, tmp2);
7442 xmlExpFree(ctxt, tmp);
7443 xmlExpFree(ctxt, tmp2);
7444 return(ret);
7445 }
7446 }
7447 /* we made no progress on structured operations */
7448 break;
7449 case XML_EXP_OR:
7450#ifdef DEBUG_DERIV
7451 printf("Or , trying both side\n");
7452#endif
7453 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7454 if (ret == NULL)
7455 return(NULL);
7456 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_right, sub);
7457 if (tmp == NULL) {
7458 xmlExpFree(ctxt, ret);
7459 return(NULL);
7460 }
7461 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, tmp, NULL, 0, 0));
7462 case XML_EXP_COUNT: {
7463 int min, max;
7464
7465 if (sub->type == XML_EXP_COUNT) {
7466 /*
7467 * Try to see if the loop is completely subsumed
7468 */
7469 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub->exp_left);
7470 if (tmp == NULL)
7471 return(NULL);
7472 if (tmp == forbiddenExp) {
7473 int mult;
7474
7475#ifdef DEBUG_DERIV
7476 printf("Count, Count inner don't subsume\n");
7477#endif
7478 mult = xmlExpDivide(ctxt, sub->exp_left, exp->exp_left,
7479 NULL, &tmp);
7480 if (mult <= 0) {
7481#ifdef DEBUG_DERIV
7482 printf("Count, Count not multiple => forbidden\n");
7483#endif
7484 return(forbiddenExp);
7485 }
7486 if (sub->exp_max == -1) {
7487 max = -1;
7488 if (exp->exp_max == -1) {
7489 if (exp->exp_min <= sub->exp_min * mult)
7490 min = 0;
7491 else
7492 min = exp->exp_min - sub->exp_min * mult;
7493 } else {
7494#ifdef DEBUG_DERIV
7495 printf("Count, Count finite can't subsume infinite\n");
7496#endif
7497 xmlExpFree(ctxt, tmp);
7498 return(forbiddenExp);
7499 }
7500 } else {
7501 if (exp->exp_max == -1) {
7502#ifdef DEBUG_DERIV
7503 printf("Infinite loop consume mult finite loop\n");
7504#endif
7505 if (exp->exp_min > sub->exp_min * mult) {
7506 max = -1;
7507 min = exp->exp_min - sub->exp_min * mult;
7508 } else {
7509 max = -1;
7510 min = 0;
7511 }
7512 } else {
7513 if (exp->exp_max < sub->exp_max * mult) {
7514#ifdef DEBUG_DERIV
7515 printf("loops max mult mismatch => forbidden\n");
7516#endif
7517 xmlExpFree(ctxt, tmp);
7518 return(forbiddenExp);
7519 }
7520 if (sub->exp_max * mult > exp->exp_min)
7521 min = 0;
7522 else
7523 min = exp->exp_min - sub->exp_max * mult;
7524 max = exp->exp_max - sub->exp_max * mult;
7525 }
7526 }
7527 } else if (!IS_NILLABLE(tmp)) {
7528 /*
7529 * TODO: loop here to try to grow if working on finite
7530 * blocks.
7531 */
7532#ifdef DEBUG_DERIV
7533 printf("Count, Count remain not nillable => forbidden\n");
7534#endif
7535 xmlExpFree(ctxt, tmp);
7536 return(forbiddenExp);
7537 } else if (sub->exp_max == -1) {
7538 if (exp->exp_max == -1) {
7539 if (exp->exp_min <= sub->exp_min) {
7540#ifdef DEBUG_DERIV
7541 printf("Infinite loops Okay => COUNT(0,Inf)\n");
7542#endif
7543 max = -1;
7544 min = 0;
7545 } else {
7546#ifdef DEBUG_DERIV
7547 printf("Infinite loops min => Count(X,Inf)\n");
7548#endif
7549 max = -1;
7550 min = exp->exp_min - sub->exp_min;
7551 }
7552 } else if (exp->exp_min > sub->exp_min) {
7553#ifdef DEBUG_DERIV
7554 printf("loops min mismatch 1 => forbidden ???\n");
7555#endif
7556 xmlExpFree(ctxt, tmp);
7557 return(forbiddenExp);
7558 } else {
7559 max = -1;
7560 min = 0;
7561 }
7562 } else {
7563 if (exp->exp_max == -1) {
7564#ifdef DEBUG_DERIV
7565 printf("Infinite loop consume finite loop\n");
7566#endif
7567 if (exp->exp_min > sub->exp_min) {
7568 max = -1;
7569 min = exp->exp_min - sub->exp_min;
7570 } else {
7571 max = -1;
7572 min = 0;
7573 }
7574 } else {
7575 if (exp->exp_max < sub->exp_max) {
7576#ifdef DEBUG_DERIV
7577 printf("loops max mismatch => forbidden\n");
7578#endif
7579 xmlExpFree(ctxt, tmp);
7580 return(forbiddenExp);
7581 }
7582 if (sub->exp_max > exp->exp_min)
7583 min = 0;
7584 else
7585 min = exp->exp_min - sub->exp_max;
7586 max = exp->exp_max - sub->exp_max;
7587 }
7588 }
7589#ifdef DEBUG_DERIV
7590 printf("loops match => SEQ(COUNT())\n");
7591#endif
7592 exp->exp_left->ref++;
7593 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left,
7594 NULL, NULL, min, max);
7595 if (tmp2 == NULL) {
7596 return(NULL);
7597 }
7598 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, tmp, tmp2,
7599 NULL, 0, 0);
7600 return(ret);
7601 }
7602 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7603 if (tmp == NULL)
7604 return(NULL);
7605 if (tmp == forbiddenExp) {
7606#ifdef DEBUG_DERIV
7607 printf("loop mismatch => forbidden\n");
7608#endif
7609 return(forbiddenExp);
7610 }
7611 if (exp->exp_min > 0)
7612 min = exp->exp_min - 1;
7613 else
7614 min = 0;
7615 if (exp->exp_max < 0)
7616 max = -1;
7617 else
7618 max = exp->exp_max - 1;
7619
7620#ifdef DEBUG_DERIV
7621 printf("loop match => SEQ(COUNT())\n");
7622#endif
7623 exp->exp_left->ref++;
7624 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left,
7625 NULL, NULL, min, max);
7626 if (tmp2 == NULL)
7627 return(NULL);
7628 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, tmp, tmp2,
7629 NULL, 0, 0);
7630 return(ret);
7631 }
7632 }
7633
7634#ifdef DEBUG_DERIV
7635 printf("Fallback to derivative\n");
7636#endif
7637 if (IS_NILLABLE(sub)) {
7638 if (!(IS_NILLABLE(exp)))
7639 return(forbiddenExp);
7640 else
7641 ret = emptyExp;
7642 } else
7643 ret = NULL;
7644 /*
7645 * here the structured derivation made no progress so
7646 * we use the default token based derivation to force one more step
7647 */
7648 if (ctxt->tabSize == 0)
7649 ctxt->tabSize = 40;
7650
7651 tab = (const xmlChar **) xmlMalloc(ctxt->tabSize *
7652 sizeof(const xmlChar *));
7653 if (tab == NULL) {
7654 return(NULL);
7655 }
7656
7657 /*
7658 * collect all the strings accepted by the subexpression on input
7659 */
7660 len = xmlExpGetStartInt(ctxt, sub, tab, ctxt->tabSize, 0);
7661 while (len < 0) {
7662 const xmlChar **temp;
7663 temp = (const xmlChar **) xmlRealloc((xmlChar **) tab, ctxt->tabSize * 2 *
7664 sizeof(const xmlChar *));
7665 if (temp == NULL) {
7666 xmlFree((xmlChar **) tab);
7667 return(NULL);
7668 }
7669 tab = temp;
7670 ctxt->tabSize *= 2;
7671 len = xmlExpGetStartInt(ctxt, sub, tab, ctxt->tabSize, 0);
7672 }
7673 for (i = 0;i < len;i++) {
7674 tmp = xmlExpStringDeriveInt(ctxt, exp, tab[i]);
7675 if ((tmp == NULL) || (tmp == forbiddenExp)) {
7676 xmlExpFree(ctxt, ret);
7677 xmlFree((xmlChar **) tab);
7678 return(tmp);
7679 }
7680 tmp2 = xmlExpStringDeriveInt(ctxt, sub, tab[i]);
7681 if ((tmp2 == NULL) || (tmp2 == forbiddenExp)) {
7682 xmlExpFree(ctxt, tmp);
7683 xmlExpFree(ctxt, ret);
7684 xmlFree((xmlChar **) tab);
7685 return(tmp);
7686 }
7687 tmp3 = xmlExpExpDeriveInt(ctxt, tmp, tmp2);
7688 xmlExpFree(ctxt, tmp);
7689 xmlExpFree(ctxt, tmp2);
7690
7691 if ((tmp3 == NULL) || (tmp3 == forbiddenExp)) {
7692 xmlExpFree(ctxt, ret);
7693 xmlFree((xmlChar **) tab);
7694 return(tmp3);
7695 }
7696
7697 if (ret == NULL)
7698 ret = tmp3;
7699 else {
7700 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, tmp3, NULL, 0, 0);
7701 if (ret == NULL) {
7702 xmlFree((xmlChar **) tab);
7703 return(NULL);
7704 }
7705 }
7706 }
7707 xmlFree((xmlChar **) tab);
7708 return(ret);
7709}
7710
7711/**
7712 * xmlExpExpDerive:
7713 * @ctxt: the expressions context
7714 * @exp: the englobing expression
7715 * @sub: the subexpression
7716 *
7717 * Evaluates the expression resulting from @exp consuming a sub expression @sub
7718 * Based on algebraic derivation and sometimes direct Brzozowski derivation
7719 * it usually tatkes less than linear time and can handle expressions generating
7720 * infinite languages.
7721 *
7722 * Returns the resulting expression or NULL in case of internal error, the
7723 * result must be freed
7724 */
7725xmlExpNodePtr
7726xmlExpExpDerive(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7727 if ((exp == NULL) || (ctxt == NULL) || (sub == NULL))
7728 return(NULL);
7729
7730 /*
7731 * O(1) speedups
7732 */
7733 if (IS_NILLABLE(sub) && (!IS_NILLABLE(exp))) {
7734#ifdef DEBUG_DERIV
7735 printf("Sub nillable and not exp : can't subsume\n");
7736#endif
7737 return(forbiddenExp);
7738 }
7739 if (xmlExpCheckCard(exp, sub) == 0) {
7740#ifdef DEBUG_DERIV
7741 printf("sub generate longuer sequances than exp : can't subsume\n");
7742#endif
7743 return(forbiddenExp);
7744 }
7745 return(xmlExpExpDeriveInt(ctxt, exp, sub));
7746}
7747
7748/**
7749 * xmlExpSubsume:
7750 * @ctxt: the expressions context
7751 * @exp: the englobing expression
7752 * @sub: the subexpression
7753 *
7754 * Check whether @exp accepts all the languages accexpted by @sub
7755 * the input being a subexpression.
7756 *
7757 * Returns 1 if true 0 if false and -1 in case of failure.
7758 */
7759int
7760xmlExpSubsume(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7761 xmlExpNodePtr tmp;
7762
7763 if ((exp == NULL) || (ctxt == NULL) || (sub == NULL))
7764 return(-1);
7765
7766 /*
7767 * TODO: speedup by checking the language of sub is a subset of the
7768 * language of exp
7769 */
7770 /*
7771 * O(1) speedups
7772 */
7773 if (IS_NILLABLE(sub) && (!IS_NILLABLE(exp))) {
7774#ifdef DEBUG_DERIV
7775 printf("Sub nillable and not exp : can't subsume\n");
7776#endif
7777 return(0);
7778 }
7779 if (xmlExpCheckCard(exp, sub) == 0) {
7780#ifdef DEBUG_DERIV
7781 printf("sub generate longuer sequances than exp : can't subsume\n");
7782#endif
7783 return(0);
7784 }
7785 tmp = xmlExpExpDeriveInt(ctxt, exp, sub);
7786#ifdef DEBUG_DERIV
7787 printf("Result derivation :\n");
7788 PRINT_EXP(tmp);
7789#endif
7790 if (tmp == NULL)
7791 return(-1);
7792 if (tmp == forbiddenExp)
7793 return(0);
7794 if (tmp == emptyExp)
7795 return(1);
7796 if ((tmp != NULL) && (IS_NILLABLE(tmp))) {
7797 xmlExpFree(ctxt, tmp);
7798 return(1);
7799 }
7800 xmlExpFree(ctxt, tmp);
7801 return(0);
7802}
7803
7804/************************************************************************
7805 * *
7806 * Parsing expression *
7807 * *
7808 ************************************************************************/
7809
7810static xmlExpNodePtr xmlExpParseExpr(xmlExpCtxtPtr ctxt);
7811
7812#undef CUR
7813#define CUR (*ctxt->cur)
7814#undef NEXT
7815#define NEXT ctxt->cur++;
7816#undef IS_BLANK
7817#define IS_BLANK(c) ((c == ' ') || (c == '\n') || (c == '\r') || (c == '\t'))
7818#define SKIP_BLANKS while (IS_BLANK(*ctxt->cur)) ctxt->cur++;
7819
7820static int
7821xmlExpParseNumber(xmlExpCtxtPtr ctxt) {
7822 int ret = 0;
7823
7824 SKIP_BLANKS
7825 if (CUR == '*') {
7826 NEXT
7827 return(-1);
7828 }
7829 if ((CUR < '0') || (CUR > '9'))
7830 return(-1);
7831 while ((CUR >= '0') && (CUR <= '9')) {
7832 ret = ret * 10 + (CUR - '0');
7833 NEXT
7834 }
7835 return(ret);
7836}
7837
7838static xmlExpNodePtr
7839xmlExpParseOr(xmlExpCtxtPtr ctxt) {
7840 const char *base;
7841 xmlExpNodePtr ret;
7842 const xmlChar *val;
7843
7844 SKIP_BLANKS
7845 base = ctxt->cur;
7846 if (*ctxt->cur == '(') {
7847 NEXT
7848 ret = xmlExpParseExpr(ctxt);
7849 SKIP_BLANKS
7850 if (*ctxt->cur != ')') {
7851 fprintf(stderr, "unbalanced '(' : %s\n", base);
7852 xmlExpFree(ctxt, ret);
7853 return(NULL);
7854 }
7855 NEXT;
7856 SKIP_BLANKS
7857 goto parse_quantifier;
7858 }
7859 while ((CUR != 0) && (!(IS_BLANK(CUR))) && (CUR != '(') &&
7860 (CUR != ')') && (CUR != '|') && (CUR != ',') && (CUR != '{') &&
7861 (CUR != '*') && (CUR != '+') && (CUR != '?') && (CUR != '}'))
7862 NEXT;
7863 val = xmlDictLookup(ctxt->dict, BAD_CAST base, ctxt->cur - base);
7864 if (val == NULL)
7865 return(NULL);
7866 ret = xmlExpHashGetEntry(ctxt, XML_EXP_ATOM, NULL, NULL, val, 0, 0);
7867 if (ret == NULL)
7868 return(NULL);
7869 SKIP_BLANKS
7870parse_quantifier:
7871 if (CUR == '{') {
7872 int min, max;
7873
7874 NEXT
7875 min = xmlExpParseNumber(ctxt);
7876 if (min < 0) {
7877 xmlExpFree(ctxt, ret);
7878 return(NULL);
7879 }
7880 SKIP_BLANKS
7881 if (CUR == ',') {
7882 NEXT
7883 max = xmlExpParseNumber(ctxt);
7884 SKIP_BLANKS
7885 } else
7886 max = min;
7887 if (CUR != '}') {
7888 xmlExpFree(ctxt, ret);
7889 return(NULL);
7890 }
7891 NEXT
7892 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7893 min, max);
7894 SKIP_BLANKS
7895 } else if (CUR == '?') {
7896 NEXT
7897 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7898 0, 1);
7899 SKIP_BLANKS
7900 } else if (CUR == '+') {
7901 NEXT
7902 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7903 1, -1);
7904 SKIP_BLANKS
7905 } else if (CUR == '*') {
7906 NEXT
7907 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7908 0, -1);
7909 SKIP_BLANKS
7910 }
7911 return(ret);
7912}
7913
7914
7915static xmlExpNodePtr
7916xmlExpParseSeq(xmlExpCtxtPtr ctxt) {
7917 xmlExpNodePtr ret, right;
7918
7919 ret = xmlExpParseOr(ctxt);
7920 SKIP_BLANKS
7921 while (CUR == '|') {
7922 NEXT
7923 right = xmlExpParseOr(ctxt);
7924 if (right == NULL) {
7925 xmlExpFree(ctxt, ret);
7926 return(NULL);
7927 }
7928 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, right, NULL, 0, 0);
7929 if (ret == NULL)
7930 return(NULL);
7931 }
7932 return(ret);
7933}
7934
7935static xmlExpNodePtr
7936xmlExpParseExpr(xmlExpCtxtPtr ctxt) {
7937 xmlExpNodePtr ret, right;
7938
7939 ret = xmlExpParseSeq(ctxt);
7940 SKIP_BLANKS
7941 while (CUR == ',') {
7942 NEXT
7943 right = xmlExpParseSeq(ctxt);
7944 if (right == NULL) {
7945 xmlExpFree(ctxt, ret);
7946 return(NULL);
7947 }
7948 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, right, NULL, 0, 0);
7949 if (ret == NULL)
7950 return(NULL);
7951 }
7952 return(ret);
7953}
7954
7955/**
7956 * xmlExpParse:
7957 * @ctxt: the expressions context
7958 * @expr: the 0 terminated string
7959 *
7960 * Minimal parser for regexps, it understand the following constructs
7961 * - string terminals
7962 * - choice operator |
7963 * - sequence operator ,
7964 * - subexpressions (...)
7965 * - usual cardinality operators + * and ?
7966 * - finite sequences { min, max }
7967 * - infinite sequences { min, * }
7968 * There is minimal checkings made especially no checking on strings values
7969 *
7970 * Returns a new expression or NULL in case of failure
7971 */
7972xmlExpNodePtr
7973xmlExpParse(xmlExpCtxtPtr ctxt, const char *expr) {
7974 xmlExpNodePtr ret;
7975
7976 ctxt->expr = expr;
7977 ctxt->cur = expr;
7978
7979 ret = xmlExpParseExpr(ctxt);
7980 SKIP_BLANKS
7981 if (*ctxt->cur != 0) {
7982 xmlExpFree(ctxt, ret);
7983 return(NULL);
7984 }
7985 return(ret);
7986}
7987
7988static void
7989xmlExpDumpInt(xmlBufferPtr buf, xmlExpNodePtr expr, int glob) {
7990 xmlExpNodePtr c;
7991
7992 if (expr == NULL) return;
7993 if (glob) xmlBufferWriteChar(buf, "(");
7994 switch (expr->type) {
7995 case XML_EXP_EMPTY:
7996 xmlBufferWriteChar(buf, "empty");
7997 break;
7998 case XML_EXP_FORBID:
7999 xmlBufferWriteChar(buf, "forbidden");
8000 break;
8001 case XML_EXP_ATOM:
8002 xmlBufferWriteCHAR(buf, expr->exp_str);
8003 break;
8004 case XML_EXP_SEQ:
8005 c = expr->exp_left;
8006 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8007 xmlExpDumpInt(buf, c, 1);
8008 else
8009 xmlExpDumpInt(buf, c, 0);
8010 xmlBufferWriteChar(buf, " , ");
8011 c = expr->exp_right;
8012 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8013 xmlExpDumpInt(buf, c, 1);
8014 else
8015 xmlExpDumpInt(buf, c, 0);
8016 break;
8017 case XML_EXP_OR:
8018 c = expr->exp_left;
8019 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8020 xmlExpDumpInt(buf, c, 1);
8021 else
8022 xmlExpDumpInt(buf, c, 0);
8023 xmlBufferWriteChar(buf, " | ");
8024 c = expr->exp_right;
8025 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8026 xmlExpDumpInt(buf, c, 1);
8027 else
8028 xmlExpDumpInt(buf, c, 0);
8029 break;
8030 case XML_EXP_COUNT: {
8031 char rep[40];
8032
8033 c = expr->exp_left;
8034 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8035 xmlExpDumpInt(buf, c, 1);
8036 else
8037 xmlExpDumpInt(buf, c, 0);
8038 if ((expr->exp_min == 0) && (expr->exp_max == 1)) {
8039 rep[0] = '?';
8040 rep[1] = 0;
8041 } else if ((expr->exp_min == 0) && (expr->exp_max == -1)) {
8042 rep[0] = '*';
8043 rep[1] = 0;
8044 } else if ((expr->exp_min == 1) && (expr->exp_max == -1)) {
8045 rep[0] = '+';
8046 rep[1] = 0;
8047 } else if (expr->exp_max == expr->exp_min) {
8048 snprintf(rep, 39, "{%d}", expr->exp_min);
8049 } else if (expr->exp_max < 0) {
8050 snprintf(rep, 39, "{%d,inf}", expr->exp_min);
8051 } else {
8052 snprintf(rep, 39, "{%d,%d}", expr->exp_min, expr->exp_max);
8053 }
8054 rep[39] = 0;
8055 xmlBufferWriteChar(buf, rep);
8056 break;
8057 }
8058 default:
8059 fprintf(stderr, "Error in tree\n");
8060 }
8061 if (glob)
8062 xmlBufferWriteChar(buf, ")");
8063}
8064/**
8065 * xmlExpDump:
8066 * @buf: a buffer to receive the output
8067 * @expr: the compiled expression
8068 *
8069 * Serialize the expression as compiled to the buffer
8070 */
8071void
8072xmlExpDump(xmlBufferPtr buf, xmlExpNodePtr expr) {
8073 if ((buf == NULL) || (expr == NULL))
8074 return;
8075 xmlExpDumpInt(buf, expr, 0);
8076}
8077
8078/**
8079 * xmlExpMaxToken:
8080 * @expr: a compiled expression
8081 *
8082 * Indicate the maximum number of input a expression can accept
8083 *
8084 * Returns the maximum length or -1 in case of error
8085 */
8086int
8087xmlExpMaxToken(xmlExpNodePtr expr) {
8088 if (expr == NULL)
8089 return(-1);
8090 return(expr->c_max);
8091}
8092
8093/**
8094 * xmlExpCtxtNbNodes:
8095 * @ctxt: an expression context
8096 *
8097 * Debugging facility provides the number of allocated nodes at a that point
8098 *
8099 * Returns the number of nodes in use or -1 in case of error
8100 */
8101int
8102xmlExpCtxtNbNodes(xmlExpCtxtPtr ctxt) {
8103 if (ctxt == NULL)
8104 return(-1);
8105 return(ctxt->nb_nodes);
8106}
8107
8108/**
8109 * xmlExpCtxtNbCons:
8110 * @ctxt: an expression context
8111 *
8112 * Debugging facility provides the number of allocated nodes over lifetime
8113 *
8114 * Returns the number of nodes ever allocated or -1 in case of error
8115 */
8116int
8117xmlExpCtxtNbCons(xmlExpCtxtPtr ctxt) {
8118 if (ctxt == NULL)
8119 return(-1);
8120 return(ctxt->nb_cons);
8121}
8122
8123#endif /* LIBXML_EXPR_ENABLED */
8124#define bottom_xmlregexp
8125#include "elfgcchack.h"
8126#endif /* LIBXML_REGEXP_ENABLED */
Note: See TracBrowser for help on using the repository browser.

© 2024 Oracle Support Privacy / Do Not Sell My Info Terms of Use Trademark Policy Automated Access Etiquette