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/schemas 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 <veillard@redhat.com>
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 #include <limits.h>
27
28 #include <libxml/tree.h>
29 #include <libxml/parserInternals.h>
30 #include <libxml/xmlregexp.h>
31 #include <libxml/xmlautomata.h>
32 #include <libxml/xmlunicode.h>
33
34 #include "private/error.h"
35 #include "private/regexp.h"
36
37 #ifndef SIZE_MAX
38 #define SIZE_MAX ((size_t) -1)
39 #endif
40
41 /* #define DEBUG_REGEXP_GRAPH */
42 /* #define DEBUG_REGEXP_EXEC */
43 /* #define DEBUG_PUSH */
44 /* #define DEBUG_COMPACTION */
45
46 #define MAX_PUSH 10000000
47
48 #ifdef ERROR
49 #undef ERROR
50 #endif
51 #define ERROR(str) \
52 ctxt->error = XML_REGEXP_COMPILE_ERROR; \
53 xmlRegexpErrCompile(ctxt, str);
54 #define NEXT ctxt->cur++
55 #define CUR (*(ctxt->cur))
56 #define NXT(index) (ctxt->cur[index])
57
58 #define CUR_SCHAR(s, l) xmlStringCurrentChar(NULL, s, &l)
59 #define NEXTL(l) ctxt->cur += l;
60 #define XML_REG_STRING_SEPARATOR '|'
61 /*
62 * Need PREV to check on a '-' within a Character Group. May only be used
63 * when it's guaranteed that cur is not at the beginning of ctxt->string!
64 */
65 #define PREV (ctxt->cur[-1])
66
67 /**
68 * TODO:
69 *
70 * macro to flag unimplemented blocks
71 */
72 #define TODO \
73 xmlGenericError(xmlGenericErrorContext, \
74 "Unimplemented block at %s:%d\n", \
75 __FILE__, __LINE__);
76
77 /************************************************************************
78 * *
79 * Datatypes and structures *
80 * *
81 ************************************************************************/
82
83 /*
84 * Note: the order of the enums below is significant, do not shuffle
85 */
86 typedef enum {
87 XML_REGEXP_EPSILON = 1,
88 XML_REGEXP_CHARVAL,
89 XML_REGEXP_RANGES,
90 XML_REGEXP_SUBREG, /* used for () sub regexps */
91 XML_REGEXP_STRING,
92 XML_REGEXP_ANYCHAR, /* . */
93 XML_REGEXP_ANYSPACE, /* \s */
94 XML_REGEXP_NOTSPACE, /* \S */
95 XML_REGEXP_INITNAME, /* \l */
96 XML_REGEXP_NOTINITNAME, /* \L */
97 XML_REGEXP_NAMECHAR, /* \c */
98 XML_REGEXP_NOTNAMECHAR, /* \C */
99 XML_REGEXP_DECIMAL, /* \d */
100 XML_REGEXP_NOTDECIMAL, /* \D */
101 XML_REGEXP_REALCHAR, /* \w */
102 XML_REGEXP_NOTREALCHAR, /* \W */
103 XML_REGEXP_LETTER = 100,
104 XML_REGEXP_LETTER_UPPERCASE,
105 XML_REGEXP_LETTER_LOWERCASE,
106 XML_REGEXP_LETTER_TITLECASE,
107 XML_REGEXP_LETTER_MODIFIER,
108 XML_REGEXP_LETTER_OTHERS,
109 XML_REGEXP_MARK,
110 XML_REGEXP_MARK_NONSPACING,
111 XML_REGEXP_MARK_SPACECOMBINING,
112 XML_REGEXP_MARK_ENCLOSING,
113 XML_REGEXP_NUMBER,
114 XML_REGEXP_NUMBER_DECIMAL,
115 XML_REGEXP_NUMBER_LETTER,
116 XML_REGEXP_NUMBER_OTHERS,
117 XML_REGEXP_PUNCT,
118 XML_REGEXP_PUNCT_CONNECTOR,
119 XML_REGEXP_PUNCT_DASH,
120 XML_REGEXP_PUNCT_OPEN,
121 XML_REGEXP_PUNCT_CLOSE,
122 XML_REGEXP_PUNCT_INITQUOTE,
123 XML_REGEXP_PUNCT_FINQUOTE,
124 XML_REGEXP_PUNCT_OTHERS,
125 XML_REGEXP_SEPAR,
126 XML_REGEXP_SEPAR_SPACE,
127 XML_REGEXP_SEPAR_LINE,
128 XML_REGEXP_SEPAR_PARA,
129 XML_REGEXP_SYMBOL,
130 XML_REGEXP_SYMBOL_MATH,
131 XML_REGEXP_SYMBOL_CURRENCY,
132 XML_REGEXP_SYMBOL_MODIFIER,
133 XML_REGEXP_SYMBOL_OTHERS,
134 XML_REGEXP_OTHER,
135 XML_REGEXP_OTHER_CONTROL,
136 XML_REGEXP_OTHER_FORMAT,
137 XML_REGEXP_OTHER_PRIVATE,
138 XML_REGEXP_OTHER_NA,
139 XML_REGEXP_BLOCK_NAME
140 } xmlRegAtomType;
141
142 typedef enum {
143 XML_REGEXP_QUANT_EPSILON = 1,
144 XML_REGEXP_QUANT_ONCE,
145 XML_REGEXP_QUANT_OPT,
146 XML_REGEXP_QUANT_MULT,
147 XML_REGEXP_QUANT_PLUS,
148 XML_REGEXP_QUANT_ONCEONLY,
149 XML_REGEXP_QUANT_ALL,
150 XML_REGEXP_QUANT_RANGE
151 } xmlRegQuantType;
152
153 typedef enum {
154 XML_REGEXP_START_STATE = 1,
155 XML_REGEXP_FINAL_STATE,
156 XML_REGEXP_TRANS_STATE,
157 XML_REGEXP_SINK_STATE,
158 XML_REGEXP_UNREACH_STATE
159 } xmlRegStateType;
160
161 typedef enum {
162 XML_REGEXP_MARK_NORMAL = 0,
163 XML_REGEXP_MARK_START,
164 XML_REGEXP_MARK_VISITED
165 } xmlRegMarkedType;
166
167 typedef struct _xmlRegRange xmlRegRange;
168 typedef xmlRegRange *xmlRegRangePtr;
169
170 struct _xmlRegRange {
171 int neg; /* 0 normal, 1 not, 2 exclude */
172 xmlRegAtomType type;
173 int start;
174 int end;
175 xmlChar *blockName;
176 };
177
178 typedef struct _xmlRegAtom xmlRegAtom;
179 typedef xmlRegAtom *xmlRegAtomPtr;
180
181 typedef struct _xmlAutomataState xmlRegState;
182 typedef xmlRegState *xmlRegStatePtr;
183
184 struct _xmlRegAtom {
185 int no;
186 xmlRegAtomType type;
187 xmlRegQuantType quant;
188 int min;
189 int max;
190
191 void *valuep;
192 void *valuep2;
193 int neg;
194 int codepoint;
195 xmlRegStatePtr start;
196 xmlRegStatePtr start0;
197 xmlRegStatePtr stop;
198 int maxRanges;
199 int nbRanges;
200 xmlRegRangePtr *ranges;
201 void *data;
202 };
203
204 typedef struct _xmlRegCounter xmlRegCounter;
205 typedef xmlRegCounter *xmlRegCounterPtr;
206
207 struct _xmlRegCounter {
208 int min;
209 int max;
210 };
211
212 typedef struct _xmlRegTrans xmlRegTrans;
213 typedef xmlRegTrans *xmlRegTransPtr;
214
215 struct _xmlRegTrans {
216 xmlRegAtomPtr atom;
217 int to;
218 int counter;
219 int count;
220 int nd;
221 };
222
223 struct _xmlAutomataState {
224 xmlRegStateType type;
225 xmlRegMarkedType mark;
226 xmlRegMarkedType markd;
227 xmlRegMarkedType reached;
228 int no;
229 int maxTrans;
230 int nbTrans;
231 xmlRegTrans *trans;
232 /* knowing states pointing to us can speed things up */
233 int maxTransTo;
234 int nbTransTo;
235 int *transTo;
236 };
237
238 typedef struct _xmlAutomata xmlRegParserCtxt;
239 typedef xmlRegParserCtxt *xmlRegParserCtxtPtr;
240
241 #define AM_AUTOMATA_RNG 1
242
243 struct _xmlAutomata {
244 xmlChar *string;
245 xmlChar *cur;
246
247 int error;
248 int neg;
249
250 xmlRegStatePtr start;
251 xmlRegStatePtr end;
252 xmlRegStatePtr state;
253
254 xmlRegAtomPtr atom;
255
256 int maxAtoms;
257 int nbAtoms;
258 xmlRegAtomPtr *atoms;
259
260 int maxStates;
261 int nbStates;
262 xmlRegStatePtr *states;
263
264 int maxCounters;
265 int nbCounters;
266 xmlRegCounter *counters;
267
268 int determinist;
269 int negs;
270 int flags;
271
272 int depth;
273 };
274
275 struct _xmlRegexp {
276 xmlChar *string;
277 int nbStates;
278 xmlRegStatePtr *states;
279 int nbAtoms;
280 xmlRegAtomPtr *atoms;
281 int nbCounters;
282 xmlRegCounter *counters;
283 int determinist;
284 int flags;
285 /*
286 * That's the compact form for determinists automatas
287 */
288 int nbstates;
289 int *compact;
290 void **transdata;
291 int nbstrings;
292 xmlChar **stringMap;
293 };
294
295 typedef struct _xmlRegExecRollback xmlRegExecRollback;
296 typedef xmlRegExecRollback *xmlRegExecRollbackPtr;
297
298 struct _xmlRegExecRollback {
299 xmlRegStatePtr state;/* the current state */
300 int index; /* the index in the input stack */
301 int nextbranch; /* the next transition to explore in that state */
302 int *counts; /* save the automata state if it has some */
303 };
304
305 typedef struct _xmlRegInputToken xmlRegInputToken;
306 typedef xmlRegInputToken *xmlRegInputTokenPtr;
307
308 struct _xmlRegInputToken {
309 xmlChar *value;
310 void *data;
311 };
312
313 struct _xmlRegExecCtxt {
314 int status; /* execution status != 0 indicate an error */
315 int determinist; /* did we find an indeterministic behaviour */
316 xmlRegexpPtr comp; /* the compiled regexp */
317 xmlRegExecCallbacks callback;
318 void *data;
319
320 xmlRegStatePtr state;/* the current state */
321 int transno; /* the current transition on that state */
322 int transcount; /* the number of chars in char counted transitions */
323
324 /*
325 * A stack of rollback states
326 */
327 int maxRollbacks;
328 int nbRollbacks;
329 xmlRegExecRollback *rollbacks;
330
331 /*
332 * The state of the automata if any
333 */
334 int *counts;
335
336 /*
337 * The input stack
338 */
339 int inputStackMax;
340 int inputStackNr;
341 int index;
342 int *charStack;
343 const xmlChar *inputString; /* when operating on characters */
344 xmlRegInputTokenPtr inputStack;/* when operating on strings */
345
346 /*
347 * error handling
348 */
349 int errStateNo; /* the error state number */
350 xmlRegStatePtr errState; /* the error state */
351 xmlChar *errString; /* the string raising the error */
352 int *errCounts; /* counters at the error state */
353 int nbPush;
354 };
355
356 #define REGEXP_ALL_COUNTER 0x123456
357 #define REGEXP_ALL_LAX_COUNTER 0x123457
358
359 static void xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top);
360 static void xmlRegFreeState(xmlRegStatePtr state);
361 static void xmlRegFreeAtom(xmlRegAtomPtr atom);
362 static int xmlRegStrEqualWildcard(const xmlChar *expStr, const xmlChar *valStr);
363 static int xmlRegCheckCharacter(xmlRegAtomPtr atom, int codepoint);
364 static int xmlRegCheckCharacterRange(xmlRegAtomType type, int codepoint,
365 int neg, int start, int end, const xmlChar *blockName);
366
367 /************************************************************************
368 * *
369 * Regexp memory error handler *
370 * *
371 ************************************************************************/
372 /**
373 * xmlRegexpErrMemory:
374 * @extra: extra information
375 *
376 * Handle an out of memory condition
377 */
378 static void
xmlRegexpErrMemory(xmlRegParserCtxtPtr ctxt,const char * extra)379 xmlRegexpErrMemory(xmlRegParserCtxtPtr ctxt, const char *extra)
380 {
381 const char *regexp = NULL;
382 if (ctxt != NULL) {
383 regexp = (const char *) ctxt->string;
384 ctxt->error = XML_ERR_NO_MEMORY;
385 }
386 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_REGEXP,
387 XML_ERR_NO_MEMORY, XML_ERR_FATAL, NULL, 0, extra,
388 regexp, NULL, 0, 0,
389 "Memory allocation failed : %s\n", extra);
390 }
391
392 /**
393 * xmlRegexpErrCompile:
394 * @extra: extra information
395 *
396 * Handle a compilation failure
397 */
398 static void
xmlRegexpErrCompile(xmlRegParserCtxtPtr ctxt,const char * extra)399 xmlRegexpErrCompile(xmlRegParserCtxtPtr ctxt, const char *extra)
400 {
401 const char *regexp = NULL;
402 int idx = 0;
403
404 if (ctxt != NULL) {
405 regexp = (const char *) ctxt->string;
406 idx = ctxt->cur - ctxt->string;
407 ctxt->error = XML_REGEXP_COMPILE_ERROR;
408 }
409 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_REGEXP,
410 XML_REGEXP_COMPILE_ERROR, XML_ERR_FATAL, NULL, 0, extra,
411 regexp, NULL, idx, 0,
412 "failed to compile: %s\n", extra);
413 }
414
415 /************************************************************************
416 * *
417 * Allocation/Deallocation *
418 * *
419 ************************************************************************/
420
421 static int xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt);
422
423 /**
424 * xmlRegCalloc2:
425 * @dim1: size of first dimension
426 * @dim2: size of second dimension
427 * @elemSize: size of element
428 *
429 * Allocate a two-dimensional array and set all elements to zero.
430 *
431 * Returns the new array or NULL in case of error.
432 */
433 static void*
xmlRegCalloc2(size_t dim1,size_t dim2,size_t elemSize)434 xmlRegCalloc2(size_t dim1, size_t dim2, size_t elemSize) {
435 size_t totalSize;
436 void *ret;
437
438 /* Check for overflow */
439 if ((dim2 == 0) || (elemSize == 0) ||
440 (dim1 > SIZE_MAX / dim2 / elemSize))
441 return (NULL);
442 totalSize = dim1 * dim2 * elemSize;
443 ret = xmlMalloc(totalSize);
444 if (ret != NULL)
445 memset(ret, 0, totalSize);
446 return (ret);
447 }
448
449 /**
450 * xmlRegEpxFromParse:
451 * @ctxt: the parser context used to build it
452 *
453 * Allocate a new regexp and fill it with the result from the parser
454 *
455 * Returns the new regexp or NULL in case of error
456 */
457 static xmlRegexpPtr
xmlRegEpxFromParse(xmlRegParserCtxtPtr ctxt)458 xmlRegEpxFromParse(xmlRegParserCtxtPtr ctxt) {
459 xmlRegexpPtr ret;
460
461 ret = (xmlRegexpPtr) xmlMalloc(sizeof(xmlRegexp));
462 if (ret == NULL) {
463 xmlRegexpErrMemory(ctxt, "compiling regexp");
464 return(NULL);
465 }
466 memset(ret, 0, sizeof(xmlRegexp));
467 ret->string = ctxt->string;
468 ret->nbStates = ctxt->nbStates;
469 ret->states = ctxt->states;
470 ret->nbAtoms = ctxt->nbAtoms;
471 ret->atoms = ctxt->atoms;
472 ret->nbCounters = ctxt->nbCounters;
473 ret->counters = ctxt->counters;
474 ret->determinist = ctxt->determinist;
475 ret->flags = ctxt->flags;
476 if (ret->determinist == -1) {
477 xmlRegexpIsDeterminist(ret);
478 }
479
480 if ((ret->determinist != 0) &&
481 (ret->nbCounters == 0) &&
482 (ctxt->negs == 0) &&
483 (ret->atoms != NULL) &&
484 (ret->atoms[0] != NULL) &&
485 (ret->atoms[0]->type == XML_REGEXP_STRING)) {
486 int i, j, nbstates = 0, nbatoms = 0;
487 int *stateRemap;
488 int *stringRemap;
489 int *transitions;
490 void **transdata;
491 xmlChar **stringMap;
492 xmlChar *value;
493
494 /*
495 * Switch to a compact representation
496 * 1/ counting the effective number of states left
497 * 2/ counting the unique number of atoms, and check that
498 * they are all of the string type
499 * 3/ build a table state x atom for the transitions
500 */
501
502 stateRemap = xmlMalloc(ret->nbStates * sizeof(int));
503 if (stateRemap == NULL) {
504 xmlRegexpErrMemory(ctxt, "compiling regexp");
505 xmlFree(ret);
506 return(NULL);
507 }
508 for (i = 0;i < ret->nbStates;i++) {
509 if (ret->states[i] != NULL) {
510 stateRemap[i] = nbstates;
511 nbstates++;
512 } else {
513 stateRemap[i] = -1;
514 }
515 }
516 #ifdef DEBUG_COMPACTION
517 printf("Final: %d states\n", nbstates);
518 #endif
519 stringMap = xmlMalloc(ret->nbAtoms * sizeof(char *));
520 if (stringMap == NULL) {
521 xmlRegexpErrMemory(ctxt, "compiling regexp");
522 xmlFree(stateRemap);
523 xmlFree(ret);
524 return(NULL);
525 }
526 stringRemap = xmlMalloc(ret->nbAtoms * sizeof(int));
527 if (stringRemap == NULL) {
528 xmlRegexpErrMemory(ctxt, "compiling regexp");
529 xmlFree(stringMap);
530 xmlFree(stateRemap);
531 xmlFree(ret);
532 return(NULL);
533 }
534 for (i = 0;i < ret->nbAtoms;i++) {
535 if ((ret->atoms[i]->type == XML_REGEXP_STRING) &&
536 (ret->atoms[i]->quant == XML_REGEXP_QUANT_ONCE)) {
537 value = ret->atoms[i]->valuep;
538 for (j = 0;j < nbatoms;j++) {
539 if (xmlStrEqual(stringMap[j], value)) {
540 stringRemap[i] = j;
541 break;
542 }
543 }
544 if (j >= nbatoms) {
545 stringRemap[i] = nbatoms;
546 stringMap[nbatoms] = xmlStrdup(value);
547 if (stringMap[nbatoms] == NULL) {
548 for (i = 0;i < nbatoms;i++)
549 xmlFree(stringMap[i]);
550 xmlFree(stringRemap);
551 xmlFree(stringMap);
552 xmlFree(stateRemap);
553 xmlFree(ret);
554 return(NULL);
555 }
556 nbatoms++;
557 }
558 } else {
559 xmlFree(stateRemap);
560 xmlFree(stringRemap);
561 for (i = 0;i < nbatoms;i++)
562 xmlFree(stringMap[i]);
563 xmlFree(stringMap);
564 xmlFree(ret);
565 return(NULL);
566 }
567 }
568 #ifdef DEBUG_COMPACTION
569 printf("Final: %d atoms\n", nbatoms);
570 #endif
571 transitions = (int *) xmlRegCalloc2(nbstates + 1, nbatoms + 1,
572 sizeof(int));
573 if (transitions == NULL) {
574 xmlFree(stateRemap);
575 xmlFree(stringRemap);
576 for (i = 0;i < nbatoms;i++)
577 xmlFree(stringMap[i]);
578 xmlFree(stringMap);
579 xmlFree(ret);
580 return(NULL);
581 }
582
583 /*
584 * Allocate the transition table. The first entry for each
585 * state corresponds to the state type.
586 */
587 transdata = NULL;
588
589 for (i = 0;i < ret->nbStates;i++) {
590 int stateno, atomno, targetno, prev;
591 xmlRegStatePtr state;
592 xmlRegTransPtr trans;
593
594 stateno = stateRemap[i];
595 if (stateno == -1)
596 continue;
597 state = ret->states[i];
598
599 transitions[stateno * (nbatoms + 1)] = state->type;
600
601 for (j = 0;j < state->nbTrans;j++) {
602 trans = &(state->trans[j]);
603 if ((trans->to == -1) || (trans->atom == NULL))
604 continue;
605 atomno = stringRemap[trans->atom->no];
606 if ((trans->atom->data != NULL) && (transdata == NULL)) {
607 transdata = (void **) xmlRegCalloc2(nbstates, nbatoms,
608 sizeof(void *));
609 if (transdata == NULL) {
610 xmlRegexpErrMemory(ctxt, "compiling regexp");
611 break;
612 }
613 }
614 targetno = stateRemap[trans->to];
615 /*
616 * if the same atom can generate transitions to 2 different
617 * states then it means the automata is not deterministic and
618 * the compact form can't be used !
619 */
620 prev = transitions[stateno * (nbatoms + 1) + atomno + 1];
621 if (prev != 0) {
622 if (prev != targetno + 1) {
623 ret->determinist = 0;
624 #ifdef DEBUG_COMPACTION
625 printf("Indet: state %d trans %d, atom %d to %d : %d to %d\n",
626 i, j, trans->atom->no, trans->to, atomno, targetno);
627 printf(" previous to is %d\n", prev);
628 #endif
629 if (transdata != NULL)
630 xmlFree(transdata);
631 xmlFree(transitions);
632 xmlFree(stateRemap);
633 xmlFree(stringRemap);
634 for (i = 0;i < nbatoms;i++)
635 xmlFree(stringMap[i]);
636 xmlFree(stringMap);
637 goto not_determ;
638 }
639 } else {
640 #if 0
641 printf("State %d trans %d: atom %d to %d : %d to %d\n",
642 i, j, trans->atom->no, trans->to, atomno, targetno);
643 #endif
644 transitions[stateno * (nbatoms + 1) + atomno + 1] =
645 targetno + 1; /* to avoid 0 */
646 if (transdata != NULL)
647 transdata[stateno * nbatoms + atomno] =
648 trans->atom->data;
649 }
650 }
651 }
652 ret->determinist = 1;
653 #ifdef DEBUG_COMPACTION
654 /*
655 * Debug
656 */
657 for (i = 0;i < nbstates;i++) {
658 for (j = 0;j < nbatoms + 1;j++) {
659 printf("%02d ", transitions[i * (nbatoms + 1) + j]);
660 }
661 printf("\n");
662 }
663 printf("\n");
664 #endif
665 /*
666 * Cleanup of the old data
667 */
668 if (ret->states != NULL) {
669 for (i = 0;i < ret->nbStates;i++)
670 xmlRegFreeState(ret->states[i]);
671 xmlFree(ret->states);
672 }
673 ret->states = NULL;
674 ret->nbStates = 0;
675 if (ret->atoms != NULL) {
676 for (i = 0;i < ret->nbAtoms;i++)
677 xmlRegFreeAtom(ret->atoms[i]);
678 xmlFree(ret->atoms);
679 }
680 ret->atoms = NULL;
681 ret->nbAtoms = 0;
682
683 ret->compact = transitions;
684 ret->transdata = transdata;
685 ret->stringMap = stringMap;
686 ret->nbstrings = nbatoms;
687 ret->nbstates = nbstates;
688 xmlFree(stateRemap);
689 xmlFree(stringRemap);
690 }
691 not_determ:
692 ctxt->string = NULL;
693 ctxt->nbStates = 0;
694 ctxt->states = NULL;
695 ctxt->nbAtoms = 0;
696 ctxt->atoms = NULL;
697 ctxt->nbCounters = 0;
698 ctxt->counters = NULL;
699 return(ret);
700 }
701
702 /**
703 * xmlRegNewParserCtxt:
704 * @string: the string to parse
705 *
706 * Allocate a new regexp parser context
707 *
708 * Returns the new context or NULL in case of error
709 */
710 static xmlRegParserCtxtPtr
xmlRegNewParserCtxt(const xmlChar * string)711 xmlRegNewParserCtxt(const xmlChar *string) {
712 xmlRegParserCtxtPtr ret;
713
714 ret = (xmlRegParserCtxtPtr) xmlMalloc(sizeof(xmlRegParserCtxt));
715 if (ret == NULL)
716 return(NULL);
717 memset(ret, 0, sizeof(xmlRegParserCtxt));
718 if (string != NULL)
719 ret->string = xmlStrdup(string);
720 ret->cur = ret->string;
721 ret->neg = 0;
722 ret->negs = 0;
723 ret->error = 0;
724 ret->determinist = -1;
725 return(ret);
726 }
727
728 /**
729 * xmlRegNewRange:
730 * @ctxt: the regexp parser context
731 * @neg: is that negative
732 * @type: the type of range
733 * @start: the start codepoint
734 * @end: the end codepoint
735 *
736 * Allocate a new regexp range
737 *
738 * Returns the new range or NULL in case of error
739 */
740 static xmlRegRangePtr
xmlRegNewRange(xmlRegParserCtxtPtr ctxt,int neg,xmlRegAtomType type,int start,int end)741 xmlRegNewRange(xmlRegParserCtxtPtr ctxt,
742 int neg, xmlRegAtomType type, int start, int end) {
743 xmlRegRangePtr ret;
744
745 ret = (xmlRegRangePtr) xmlMalloc(sizeof(xmlRegRange));
746 if (ret == NULL) {
747 xmlRegexpErrMemory(ctxt, "allocating range");
748 return(NULL);
749 }
750 ret->neg = neg;
751 ret->type = type;
752 ret->start = start;
753 ret->end = end;
754 return(ret);
755 }
756
757 /**
758 * xmlRegFreeRange:
759 * @range: the regexp range
760 *
761 * Free a regexp range
762 */
763 static void
xmlRegFreeRange(xmlRegRangePtr range)764 xmlRegFreeRange(xmlRegRangePtr range) {
765 if (range == NULL)
766 return;
767
768 if (range->blockName != NULL)
769 xmlFree(range->blockName);
770 xmlFree(range);
771 }
772
773 /**
774 * xmlRegCopyRange:
775 * @range: the regexp range
776 *
777 * Copy a regexp range
778 *
779 * Returns the new copy or NULL in case of error.
780 */
781 static xmlRegRangePtr
xmlRegCopyRange(xmlRegParserCtxtPtr ctxt,xmlRegRangePtr range)782 xmlRegCopyRange(xmlRegParserCtxtPtr ctxt, xmlRegRangePtr range) {
783 xmlRegRangePtr ret;
784
785 if (range == NULL)
786 return(NULL);
787
788 ret = xmlRegNewRange(ctxt, range->neg, range->type, range->start,
789 range->end);
790 if (ret == NULL)
791 return(NULL);
792 if (range->blockName != NULL) {
793 ret->blockName = xmlStrdup(range->blockName);
794 if (ret->blockName == NULL) {
795 xmlRegexpErrMemory(ctxt, "allocating range");
796 xmlRegFreeRange(ret);
797 return(NULL);
798 }
799 }
800 return(ret);
801 }
802
803 /**
804 * xmlRegNewAtom:
805 * @ctxt: the regexp parser context
806 * @type: the type of atom
807 *
808 * Allocate a new atom
809 *
810 * Returns the new atom or NULL in case of error
811 */
812 static xmlRegAtomPtr
xmlRegNewAtom(xmlRegParserCtxtPtr ctxt,xmlRegAtomType type)813 xmlRegNewAtom(xmlRegParserCtxtPtr ctxt, xmlRegAtomType type) {
814 xmlRegAtomPtr ret;
815
816 ret = (xmlRegAtomPtr) xmlMalloc(sizeof(xmlRegAtom));
817 if (ret == NULL) {
818 xmlRegexpErrMemory(ctxt, "allocating atom");
819 return(NULL);
820 }
821 memset(ret, 0, sizeof(xmlRegAtom));
822 ret->type = type;
823 ret->quant = XML_REGEXP_QUANT_ONCE;
824 ret->min = 0;
825 ret->max = 0;
826 return(ret);
827 }
828
829 /**
830 * xmlRegFreeAtom:
831 * @atom: the regexp atom
832 *
833 * Free a regexp atom
834 */
835 static void
xmlRegFreeAtom(xmlRegAtomPtr atom)836 xmlRegFreeAtom(xmlRegAtomPtr atom) {
837 int i;
838
839 if (atom == NULL)
840 return;
841
842 for (i = 0;i < atom->nbRanges;i++)
843 xmlRegFreeRange(atom->ranges[i]);
844 if (atom->ranges != NULL)
845 xmlFree(atom->ranges);
846 if ((atom->type == XML_REGEXP_STRING) && (atom->valuep != NULL))
847 xmlFree(atom->valuep);
848 if ((atom->type == XML_REGEXP_STRING) && (atom->valuep2 != NULL))
849 xmlFree(atom->valuep2);
850 if ((atom->type == XML_REGEXP_BLOCK_NAME) && (atom->valuep != NULL))
851 xmlFree(atom->valuep);
852 xmlFree(atom);
853 }
854
855 /**
856 * xmlRegCopyAtom:
857 * @ctxt: the regexp parser context
858 * @atom: the original atom
859 *
860 * Allocate a new regexp range
861 *
862 * Returns the new atom or NULL in case of error
863 */
864 static xmlRegAtomPtr
xmlRegCopyAtom(xmlRegParserCtxtPtr ctxt,xmlRegAtomPtr atom)865 xmlRegCopyAtom(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom) {
866 xmlRegAtomPtr ret;
867
868 ret = (xmlRegAtomPtr) xmlMalloc(sizeof(xmlRegAtom));
869 if (ret == NULL) {
870 xmlRegexpErrMemory(ctxt, "copying atom");
871 return(NULL);
872 }
873 memset(ret, 0, sizeof(xmlRegAtom));
874 ret->type = atom->type;
875 ret->quant = atom->quant;
876 ret->min = atom->min;
877 ret->max = atom->max;
878 if (atom->nbRanges > 0) {
879 int i;
880
881 ret->ranges = (xmlRegRangePtr *) xmlMalloc(sizeof(xmlRegRangePtr) *
882 atom->nbRanges);
883 if (ret->ranges == NULL) {
884 xmlRegexpErrMemory(ctxt, "copying atom");
885 goto error;
886 }
887 for (i = 0;i < atom->nbRanges;i++) {
888 ret->ranges[i] = xmlRegCopyRange(ctxt, atom->ranges[i]);
889 if (ret->ranges[i] == NULL)
890 goto error;
891 ret->nbRanges = i + 1;
892 }
893 }
894 return(ret);
895
896 error:
897 xmlRegFreeAtom(ret);
898 return(NULL);
899 }
900
901 static xmlRegStatePtr
xmlRegNewState(xmlRegParserCtxtPtr ctxt)902 xmlRegNewState(xmlRegParserCtxtPtr ctxt) {
903 xmlRegStatePtr ret;
904
905 ret = (xmlRegStatePtr) xmlMalloc(sizeof(xmlRegState));
906 if (ret == NULL) {
907 xmlRegexpErrMemory(ctxt, "allocating state");
908 return(NULL);
909 }
910 memset(ret, 0, sizeof(xmlRegState));
911 ret->type = XML_REGEXP_TRANS_STATE;
912 ret->mark = XML_REGEXP_MARK_NORMAL;
913 return(ret);
914 }
915
916 /**
917 * xmlRegFreeState:
918 * @state: the regexp state
919 *
920 * Free a regexp state
921 */
922 static void
xmlRegFreeState(xmlRegStatePtr state)923 xmlRegFreeState(xmlRegStatePtr state) {
924 if (state == NULL)
925 return;
926
927 if (state->trans != NULL)
928 xmlFree(state->trans);
929 if (state->transTo != NULL)
930 xmlFree(state->transTo);
931 xmlFree(state);
932 }
933
934 /**
935 * xmlRegFreeParserCtxt:
936 * @ctxt: the regexp parser context
937 *
938 * Free a regexp parser context
939 */
940 static void
xmlRegFreeParserCtxt(xmlRegParserCtxtPtr ctxt)941 xmlRegFreeParserCtxt(xmlRegParserCtxtPtr ctxt) {
942 int i;
943 if (ctxt == NULL)
944 return;
945
946 if (ctxt->string != NULL)
947 xmlFree(ctxt->string);
948 if (ctxt->states != NULL) {
949 for (i = 0;i < ctxt->nbStates;i++)
950 xmlRegFreeState(ctxt->states[i]);
951 xmlFree(ctxt->states);
952 }
953 if (ctxt->atoms != NULL) {
954 for (i = 0;i < ctxt->nbAtoms;i++)
955 xmlRegFreeAtom(ctxt->atoms[i]);
956 xmlFree(ctxt->atoms);
957 }
958 if (ctxt->counters != NULL)
959 xmlFree(ctxt->counters);
960 xmlFree(ctxt);
961 }
962
963 /************************************************************************
964 * *
965 * Display of Data structures *
966 * *
967 ************************************************************************/
968
969 static void
xmlRegPrintAtomType(FILE * output,xmlRegAtomType type)970 xmlRegPrintAtomType(FILE *output, xmlRegAtomType type) {
971 switch (type) {
972 case XML_REGEXP_EPSILON:
973 fprintf(output, "epsilon "); break;
974 case XML_REGEXP_CHARVAL:
975 fprintf(output, "charval "); break;
976 case XML_REGEXP_RANGES:
977 fprintf(output, "ranges "); break;
978 case XML_REGEXP_SUBREG:
979 fprintf(output, "subexpr "); break;
980 case XML_REGEXP_STRING:
981 fprintf(output, "string "); break;
982 case XML_REGEXP_ANYCHAR:
983 fprintf(output, "anychar "); break;
984 case XML_REGEXP_ANYSPACE:
985 fprintf(output, "anyspace "); break;
986 case XML_REGEXP_NOTSPACE:
987 fprintf(output, "notspace "); break;
988 case XML_REGEXP_INITNAME:
989 fprintf(output, "initname "); break;
990 case XML_REGEXP_NOTINITNAME:
991 fprintf(output, "notinitname "); break;
992 case XML_REGEXP_NAMECHAR:
993 fprintf(output, "namechar "); break;
994 case XML_REGEXP_NOTNAMECHAR:
995 fprintf(output, "notnamechar "); break;
996 case XML_REGEXP_DECIMAL:
997 fprintf(output, "decimal "); break;
998 case XML_REGEXP_NOTDECIMAL:
999 fprintf(output, "notdecimal "); break;
1000 case XML_REGEXP_REALCHAR:
1001 fprintf(output, "realchar "); break;
1002 case XML_REGEXP_NOTREALCHAR:
1003 fprintf(output, "notrealchar "); break;
1004 case XML_REGEXP_LETTER:
1005 fprintf(output, "LETTER "); break;
1006 case XML_REGEXP_LETTER_UPPERCASE:
1007 fprintf(output, "LETTER_UPPERCASE "); break;
1008 case XML_REGEXP_LETTER_LOWERCASE:
1009 fprintf(output, "LETTER_LOWERCASE "); break;
1010 case XML_REGEXP_LETTER_TITLECASE:
1011 fprintf(output, "LETTER_TITLECASE "); break;
1012 case XML_REGEXP_LETTER_MODIFIER:
1013 fprintf(output, "LETTER_MODIFIER "); break;
1014 case XML_REGEXP_LETTER_OTHERS:
1015 fprintf(output, "LETTER_OTHERS "); break;
1016 case XML_REGEXP_MARK:
1017 fprintf(output, "MARK "); break;
1018 case XML_REGEXP_MARK_NONSPACING:
1019 fprintf(output, "MARK_NONSPACING "); break;
1020 case XML_REGEXP_MARK_SPACECOMBINING:
1021 fprintf(output, "MARK_SPACECOMBINING "); break;
1022 case XML_REGEXP_MARK_ENCLOSING:
1023 fprintf(output, "MARK_ENCLOSING "); break;
1024 case XML_REGEXP_NUMBER:
1025 fprintf(output, "NUMBER "); break;
1026 case XML_REGEXP_NUMBER_DECIMAL:
1027 fprintf(output, "NUMBER_DECIMAL "); break;
1028 case XML_REGEXP_NUMBER_LETTER:
1029 fprintf(output, "NUMBER_LETTER "); break;
1030 case XML_REGEXP_NUMBER_OTHERS:
1031 fprintf(output, "NUMBER_OTHERS "); break;
1032 case XML_REGEXP_PUNCT:
1033 fprintf(output, "PUNCT "); break;
1034 case XML_REGEXP_PUNCT_CONNECTOR:
1035 fprintf(output, "PUNCT_CONNECTOR "); break;
1036 case XML_REGEXP_PUNCT_DASH:
1037 fprintf(output, "PUNCT_DASH "); break;
1038 case XML_REGEXP_PUNCT_OPEN:
1039 fprintf(output, "PUNCT_OPEN "); break;
1040 case XML_REGEXP_PUNCT_CLOSE:
1041 fprintf(output, "PUNCT_CLOSE "); break;
1042 case XML_REGEXP_PUNCT_INITQUOTE:
1043 fprintf(output, "PUNCT_INITQUOTE "); break;
1044 case XML_REGEXP_PUNCT_FINQUOTE:
1045 fprintf(output, "PUNCT_FINQUOTE "); break;
1046 case XML_REGEXP_PUNCT_OTHERS:
1047 fprintf(output, "PUNCT_OTHERS "); break;
1048 case XML_REGEXP_SEPAR:
1049 fprintf(output, "SEPAR "); break;
1050 case XML_REGEXP_SEPAR_SPACE:
1051 fprintf(output, "SEPAR_SPACE "); break;
1052 case XML_REGEXP_SEPAR_LINE:
1053 fprintf(output, "SEPAR_LINE "); break;
1054 case XML_REGEXP_SEPAR_PARA:
1055 fprintf(output, "SEPAR_PARA "); break;
1056 case XML_REGEXP_SYMBOL:
1057 fprintf(output, "SYMBOL "); break;
1058 case XML_REGEXP_SYMBOL_MATH:
1059 fprintf(output, "SYMBOL_MATH "); break;
1060 case XML_REGEXP_SYMBOL_CURRENCY:
1061 fprintf(output, "SYMBOL_CURRENCY "); break;
1062 case XML_REGEXP_SYMBOL_MODIFIER:
1063 fprintf(output, "SYMBOL_MODIFIER "); break;
1064 case XML_REGEXP_SYMBOL_OTHERS:
1065 fprintf(output, "SYMBOL_OTHERS "); break;
1066 case XML_REGEXP_OTHER:
1067 fprintf(output, "OTHER "); break;
1068 case XML_REGEXP_OTHER_CONTROL:
1069 fprintf(output, "OTHER_CONTROL "); break;
1070 case XML_REGEXP_OTHER_FORMAT:
1071 fprintf(output, "OTHER_FORMAT "); break;
1072 case XML_REGEXP_OTHER_PRIVATE:
1073 fprintf(output, "OTHER_PRIVATE "); break;
1074 case XML_REGEXP_OTHER_NA:
1075 fprintf(output, "OTHER_NA "); break;
1076 case XML_REGEXP_BLOCK_NAME:
1077 fprintf(output, "BLOCK "); break;
1078 }
1079 }
1080
1081 static void
xmlRegPrintQuantType(FILE * output,xmlRegQuantType type)1082 xmlRegPrintQuantType(FILE *output, xmlRegQuantType type) {
1083 switch (type) {
1084 case XML_REGEXP_QUANT_EPSILON:
1085 fprintf(output, "epsilon "); break;
1086 case XML_REGEXP_QUANT_ONCE:
1087 fprintf(output, "once "); break;
1088 case XML_REGEXP_QUANT_OPT:
1089 fprintf(output, "? "); break;
1090 case XML_REGEXP_QUANT_MULT:
1091 fprintf(output, "* "); break;
1092 case XML_REGEXP_QUANT_PLUS:
1093 fprintf(output, "+ "); break;
1094 case XML_REGEXP_QUANT_RANGE:
1095 fprintf(output, "range "); break;
1096 case XML_REGEXP_QUANT_ONCEONLY:
1097 fprintf(output, "onceonly "); break;
1098 case XML_REGEXP_QUANT_ALL:
1099 fprintf(output, "all "); break;
1100 }
1101 }
1102 static void
xmlRegPrintRange(FILE * output,xmlRegRangePtr range)1103 xmlRegPrintRange(FILE *output, xmlRegRangePtr range) {
1104 fprintf(output, " range: ");
1105 if (range->neg)
1106 fprintf(output, "negative ");
1107 xmlRegPrintAtomType(output, range->type);
1108 fprintf(output, "%c - %c\n", range->start, range->end);
1109 }
1110
1111 static void
xmlRegPrintAtom(FILE * output,xmlRegAtomPtr atom)1112 xmlRegPrintAtom(FILE *output, xmlRegAtomPtr atom) {
1113 fprintf(output, " atom: ");
1114 if (atom == NULL) {
1115 fprintf(output, "NULL\n");
1116 return;
1117 }
1118 if (atom->neg)
1119 fprintf(output, "not ");
1120 xmlRegPrintAtomType(output, atom->type);
1121 xmlRegPrintQuantType(output, atom->quant);
1122 if (atom->quant == XML_REGEXP_QUANT_RANGE)
1123 fprintf(output, "%d-%d ", atom->min, atom->max);
1124 if (atom->type == XML_REGEXP_STRING)
1125 fprintf(output, "'%s' ", (char *) atom->valuep);
1126 if (atom->type == XML_REGEXP_CHARVAL)
1127 fprintf(output, "char %c\n", atom->codepoint);
1128 else if (atom->type == XML_REGEXP_RANGES) {
1129 int i;
1130 fprintf(output, "%d entries\n", atom->nbRanges);
1131 for (i = 0; i < atom->nbRanges;i++)
1132 xmlRegPrintRange(output, atom->ranges[i]);
1133 } else if (atom->type == XML_REGEXP_SUBREG) {
1134 fprintf(output, "start %d end %d\n", atom->start->no, atom->stop->no);
1135 } else {
1136 fprintf(output, "\n");
1137 }
1138 }
1139
1140 static void
xmlRegPrintTrans(FILE * output,xmlRegTransPtr trans)1141 xmlRegPrintTrans(FILE *output, xmlRegTransPtr trans) {
1142 fprintf(output, " trans: ");
1143 if (trans == NULL) {
1144 fprintf(output, "NULL\n");
1145 return;
1146 }
1147 if (trans->to < 0) {
1148 fprintf(output, "removed\n");
1149 return;
1150 }
1151 if (trans->nd != 0) {
1152 if (trans->nd == 2)
1153 fprintf(output, "last not determinist, ");
1154 else
1155 fprintf(output, "not determinist, ");
1156 }
1157 if (trans->counter >= 0) {
1158 fprintf(output, "counted %d, ", trans->counter);
1159 }
1160 if (trans->count == REGEXP_ALL_COUNTER) {
1161 fprintf(output, "all transition, ");
1162 } else if (trans->count >= 0) {
1163 fprintf(output, "count based %d, ", trans->count);
1164 }
1165 if (trans->atom == NULL) {
1166 fprintf(output, "epsilon to %d\n", trans->to);
1167 return;
1168 }
1169 if (trans->atom->type == XML_REGEXP_CHARVAL)
1170 fprintf(output, "char %c ", trans->atom->codepoint);
1171 fprintf(output, "atom %d, to %d\n", trans->atom->no, trans->to);
1172 }
1173
1174 static void
xmlRegPrintState(FILE * output,xmlRegStatePtr state)1175 xmlRegPrintState(FILE *output, xmlRegStatePtr state) {
1176 int i;
1177
1178 fprintf(output, " state: ");
1179 if (state == NULL) {
1180 fprintf(output, "NULL\n");
1181 return;
1182 }
1183 if (state->type == XML_REGEXP_START_STATE)
1184 fprintf(output, "START ");
1185 if (state->type == XML_REGEXP_FINAL_STATE)
1186 fprintf(output, "FINAL ");
1187
1188 fprintf(output, "%d, %d transitions:\n", state->no, state->nbTrans);
1189 for (i = 0;i < state->nbTrans; i++) {
1190 xmlRegPrintTrans(output, &(state->trans[i]));
1191 }
1192 }
1193
1194 #ifdef DEBUG_REGEXP_GRAPH
1195 static void
xmlRegPrintCtxt(FILE * output,xmlRegParserCtxtPtr ctxt)1196 xmlRegPrintCtxt(FILE *output, xmlRegParserCtxtPtr ctxt) {
1197 int i;
1198
1199 fprintf(output, " ctxt: ");
1200 if (ctxt == NULL) {
1201 fprintf(output, "NULL\n");
1202 return;
1203 }
1204 fprintf(output, "'%s' ", ctxt->string);
1205 if (ctxt->error)
1206 fprintf(output, "error ");
1207 if (ctxt->neg)
1208 fprintf(output, "neg ");
1209 fprintf(output, "\n");
1210 fprintf(output, "%d atoms:\n", ctxt->nbAtoms);
1211 for (i = 0;i < ctxt->nbAtoms; i++) {
1212 fprintf(output, " %02d ", i);
1213 xmlRegPrintAtom(output, ctxt->atoms[i]);
1214 }
1215 if (ctxt->atom != NULL) {
1216 fprintf(output, "current atom:\n");
1217 xmlRegPrintAtom(output, ctxt->atom);
1218 }
1219 fprintf(output, "%d states:", ctxt->nbStates);
1220 if (ctxt->start != NULL)
1221 fprintf(output, " start: %d", ctxt->start->no);
1222 if (ctxt->end != NULL)
1223 fprintf(output, " end: %d", ctxt->end->no);
1224 fprintf(output, "\n");
1225 for (i = 0;i < ctxt->nbStates; i++) {
1226 xmlRegPrintState(output, ctxt->states[i]);
1227 }
1228 fprintf(output, "%d counters:\n", ctxt->nbCounters);
1229 for (i = 0;i < ctxt->nbCounters; i++) {
1230 fprintf(output, " %d: min %d max %d\n", i, ctxt->counters[i].min,
1231 ctxt->counters[i].max);
1232 }
1233 }
1234 #endif
1235
1236 /************************************************************************
1237 * *
1238 * Finite Automata structures manipulations *
1239 * *
1240 ************************************************************************/
1241
1242 static xmlRegRangePtr
xmlRegAtomAddRange(xmlRegParserCtxtPtr ctxt,xmlRegAtomPtr atom,int neg,xmlRegAtomType type,int start,int end,xmlChar * blockName)1243 xmlRegAtomAddRange(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom,
1244 int neg, xmlRegAtomType type, int start, int end,
1245 xmlChar *blockName) {
1246 xmlRegRangePtr range;
1247
1248 if (atom == NULL) {
1249 ERROR("add range: atom is NULL");
1250 return(NULL);
1251 }
1252 if (atom->type != XML_REGEXP_RANGES) {
1253 ERROR("add range: atom is not ranges");
1254 return(NULL);
1255 }
1256 if (atom->maxRanges == 0) {
1257 atom->maxRanges = 4;
1258 atom->ranges = (xmlRegRangePtr *) xmlMalloc(atom->maxRanges *
1259 sizeof(xmlRegRangePtr));
1260 if (atom->ranges == NULL) {
1261 xmlRegexpErrMemory(ctxt, "adding ranges");
1262 atom->maxRanges = 0;
1263 return(NULL);
1264 }
1265 } else if (atom->nbRanges >= atom->maxRanges) {
1266 xmlRegRangePtr *tmp;
1267 atom->maxRanges *= 2;
1268 tmp = (xmlRegRangePtr *) xmlRealloc(atom->ranges, atom->maxRanges *
1269 sizeof(xmlRegRangePtr));
1270 if (tmp == NULL) {
1271 xmlRegexpErrMemory(ctxt, "adding ranges");
1272 atom->maxRanges /= 2;
1273 return(NULL);
1274 }
1275 atom->ranges = tmp;
1276 }
1277 range = xmlRegNewRange(ctxt, neg, type, start, end);
1278 if (range == NULL)
1279 return(NULL);
1280 range->blockName = blockName;
1281 atom->ranges[atom->nbRanges++] = range;
1282
1283 return(range);
1284 }
1285
1286 static int
xmlRegGetCounter(xmlRegParserCtxtPtr ctxt)1287 xmlRegGetCounter(xmlRegParserCtxtPtr ctxt) {
1288 if (ctxt->maxCounters == 0) {
1289 ctxt->maxCounters = 4;
1290 ctxt->counters = (xmlRegCounter *) xmlMalloc(ctxt->maxCounters *
1291 sizeof(xmlRegCounter));
1292 if (ctxt->counters == NULL) {
1293 xmlRegexpErrMemory(ctxt, "allocating counter");
1294 ctxt->maxCounters = 0;
1295 return(-1);
1296 }
1297 } else if (ctxt->nbCounters >= ctxt->maxCounters) {
1298 xmlRegCounter *tmp;
1299 ctxt->maxCounters *= 2;
1300 tmp = (xmlRegCounter *) xmlRealloc(ctxt->counters, ctxt->maxCounters *
1301 sizeof(xmlRegCounter));
1302 if (tmp == NULL) {
1303 xmlRegexpErrMemory(ctxt, "allocating counter");
1304 ctxt->maxCounters /= 2;
1305 return(-1);
1306 }
1307 ctxt->counters = tmp;
1308 }
1309 ctxt->counters[ctxt->nbCounters].min = -1;
1310 ctxt->counters[ctxt->nbCounters].max = -1;
1311 return(ctxt->nbCounters++);
1312 }
1313
1314 static int
xmlRegAtomPush(xmlRegParserCtxtPtr ctxt,xmlRegAtomPtr atom)1315 xmlRegAtomPush(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom) {
1316 if (atom == NULL) {
1317 ERROR("atom push: atom is NULL");
1318 return(-1);
1319 }
1320 if (ctxt->nbAtoms >= ctxt->maxAtoms) {
1321 size_t newSize = ctxt->maxAtoms ? ctxt->maxAtoms * 2 : 4;
1322 xmlRegAtomPtr *tmp;
1323
1324 tmp = xmlRealloc(ctxt->atoms, newSize * sizeof(xmlRegAtomPtr));
1325 if (tmp == NULL) {
1326 xmlRegexpErrMemory(ctxt, "allocating counter");
1327 return(-1);
1328 }
1329 ctxt->atoms = tmp;
1330 ctxt->maxAtoms = newSize;
1331 }
1332 atom->no = ctxt->nbAtoms;
1333 ctxt->atoms[ctxt->nbAtoms++] = atom;
1334 return(0);
1335 }
1336
1337 static void
xmlRegStateAddTransTo(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr target,int from)1338 xmlRegStateAddTransTo(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr target,
1339 int from) {
1340 if (target->maxTransTo == 0) {
1341 target->maxTransTo = 8;
1342 target->transTo = (int *) xmlMalloc(target->maxTransTo *
1343 sizeof(int));
1344 if (target->transTo == NULL) {
1345 xmlRegexpErrMemory(ctxt, "adding transition");
1346 target->maxTransTo = 0;
1347 return;
1348 }
1349 } else if (target->nbTransTo >= target->maxTransTo) {
1350 int *tmp;
1351 target->maxTransTo *= 2;
1352 tmp = (int *) xmlRealloc(target->transTo, target->maxTransTo *
1353 sizeof(int));
1354 if (tmp == NULL) {
1355 xmlRegexpErrMemory(ctxt, "adding transition");
1356 target->maxTransTo /= 2;
1357 return;
1358 }
1359 target->transTo = tmp;
1360 }
1361 target->transTo[target->nbTransTo] = from;
1362 target->nbTransTo++;
1363 }
1364
1365 static void
xmlRegStateAddTrans(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr state,xmlRegAtomPtr atom,xmlRegStatePtr target,int counter,int count)1366 xmlRegStateAddTrans(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
1367 xmlRegAtomPtr atom, xmlRegStatePtr target,
1368 int counter, int count) {
1369
1370 int nrtrans;
1371
1372 if (state == NULL) {
1373 ERROR("add state: state is NULL");
1374 return;
1375 }
1376 if (target == NULL) {
1377 ERROR("add state: target is NULL");
1378 return;
1379 }
1380 /*
1381 * Other routines follow the philosophy 'When in doubt, add a transition'
1382 * so we check here whether such a transition is already present and, if
1383 * so, silently ignore this request.
1384 */
1385
1386 for (nrtrans = state->nbTrans - 1; nrtrans >= 0; nrtrans--) {
1387 xmlRegTransPtr trans = &(state->trans[nrtrans]);
1388 if ((trans->atom == atom) &&
1389 (trans->to == target->no) &&
1390 (trans->counter == counter) &&
1391 (trans->count == count)) {
1392 #ifdef DEBUG_REGEXP_GRAPH
1393 printf("Ignoring duplicate transition from %d to %d\n",
1394 state->no, target->no);
1395 #endif
1396 return;
1397 }
1398 }
1399
1400 if (state->maxTrans == 0) {
1401 state->maxTrans = 8;
1402 state->trans = (xmlRegTrans *) xmlMalloc(state->maxTrans *
1403 sizeof(xmlRegTrans));
1404 if (state->trans == NULL) {
1405 xmlRegexpErrMemory(ctxt, "adding transition");
1406 state->maxTrans = 0;
1407 return;
1408 }
1409 } else if (state->nbTrans >= state->maxTrans) {
1410 xmlRegTrans *tmp;
1411 state->maxTrans *= 2;
1412 tmp = (xmlRegTrans *) xmlRealloc(state->trans, state->maxTrans *
1413 sizeof(xmlRegTrans));
1414 if (tmp == NULL) {
1415 xmlRegexpErrMemory(ctxt, "adding transition");
1416 state->maxTrans /= 2;
1417 return;
1418 }
1419 state->trans = tmp;
1420 }
1421 #ifdef DEBUG_REGEXP_GRAPH
1422 printf("Add trans from %d to %d ", state->no, target->no);
1423 if (count == REGEXP_ALL_COUNTER)
1424 printf("all transition\n");
1425 else if (count >= 0)
1426 printf("count based %d\n", count);
1427 else if (counter >= 0)
1428 printf("counted %d\n", counter);
1429 else if (atom == NULL)
1430 printf("epsilon transition\n");
1431 else if (atom != NULL)
1432 xmlRegPrintAtom(stdout, atom);
1433 #endif
1434
1435 state->trans[state->nbTrans].atom = atom;
1436 state->trans[state->nbTrans].to = target->no;
1437 state->trans[state->nbTrans].counter = counter;
1438 state->trans[state->nbTrans].count = count;
1439 state->trans[state->nbTrans].nd = 0;
1440 state->nbTrans++;
1441 xmlRegStateAddTransTo(ctxt, target, state->no);
1442 }
1443
1444 static xmlRegStatePtr
xmlRegStatePush(xmlRegParserCtxtPtr ctxt)1445 xmlRegStatePush(xmlRegParserCtxtPtr ctxt) {
1446 xmlRegStatePtr state;
1447
1448 if (ctxt->nbStates >= ctxt->maxStates) {
1449 size_t newSize = ctxt->maxStates ? ctxt->maxStates * 2 : 4;
1450 xmlRegStatePtr *tmp;
1451
1452 tmp = xmlRealloc(ctxt->states, newSize * sizeof(tmp[0]));
1453 if (tmp == NULL) {
1454 xmlRegexpErrMemory(ctxt, "adding state");
1455 return(NULL);
1456 }
1457 ctxt->states = tmp;
1458 ctxt->maxStates = newSize;
1459 }
1460
1461 state = xmlRegNewState(ctxt);
1462 if (state == NULL)
1463 return(NULL);
1464
1465 state->no = ctxt->nbStates;
1466 ctxt->states[ctxt->nbStates++] = state;
1467
1468 return(state);
1469 }
1470
1471 /**
1472 * xmlFAGenerateAllTransition:
1473 * @ctxt: a regexp parser context
1474 * @from: the from state
1475 * @to: the target state or NULL for building a new one
1476 * @lax:
1477 *
1478 */
1479 static int
xmlFAGenerateAllTransition(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr from,xmlRegStatePtr to,int lax)1480 xmlFAGenerateAllTransition(xmlRegParserCtxtPtr ctxt,
1481 xmlRegStatePtr from, xmlRegStatePtr to,
1482 int lax) {
1483 if (to == NULL) {
1484 to = xmlRegStatePush(ctxt);
1485 if (to == NULL)
1486 return(-1);
1487 ctxt->state = to;
1488 }
1489 if (lax)
1490 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_LAX_COUNTER);
1491 else
1492 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_COUNTER);
1493 return(0);
1494 }
1495
1496 /**
1497 * xmlFAGenerateEpsilonTransition:
1498 * @ctxt: a regexp parser context
1499 * @from: the from state
1500 * @to: the target state or NULL for building a new one
1501 *
1502 */
1503 static int
xmlFAGenerateEpsilonTransition(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr from,xmlRegStatePtr to)1504 xmlFAGenerateEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1505 xmlRegStatePtr from, xmlRegStatePtr to) {
1506 if (to == NULL) {
1507 to = xmlRegStatePush(ctxt);
1508 if (to == NULL)
1509 return(-1);
1510 ctxt->state = to;
1511 }
1512 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, -1);
1513 return(0);
1514 }
1515
1516 /**
1517 * xmlFAGenerateCountedEpsilonTransition:
1518 * @ctxt: a regexp parser context
1519 * @from: the from state
1520 * @to: the target state or NULL for building a new one
1521 * counter: the counter for that transition
1522 *
1523 */
1524 static int
xmlFAGenerateCountedEpsilonTransition(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr from,xmlRegStatePtr to,int counter)1525 xmlFAGenerateCountedEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1526 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1527 if (to == NULL) {
1528 to = xmlRegStatePush(ctxt);
1529 if (to == NULL)
1530 return(-1);
1531 ctxt->state = to;
1532 }
1533 xmlRegStateAddTrans(ctxt, from, NULL, to, counter, -1);
1534 return(0);
1535 }
1536
1537 /**
1538 * xmlFAGenerateCountedTransition:
1539 * @ctxt: a regexp parser context
1540 * @from: the from state
1541 * @to: the target state or NULL for building a new one
1542 * counter: the counter for that transition
1543 *
1544 */
1545 static int
xmlFAGenerateCountedTransition(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr from,xmlRegStatePtr to,int counter)1546 xmlFAGenerateCountedTransition(xmlRegParserCtxtPtr ctxt,
1547 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1548 if (to == NULL) {
1549 to = xmlRegStatePush(ctxt);
1550 if (to == NULL)
1551 return(-1);
1552 ctxt->state = to;
1553 }
1554 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, counter);
1555 return(0);
1556 }
1557
1558 /**
1559 * xmlFAGenerateTransitions:
1560 * @ctxt: a regexp parser context
1561 * @from: the from state
1562 * @to: the target state or NULL for building a new one
1563 * @atom: the atom generating the transition
1564 *
1565 * Returns 0 if success and -1 in case of error.
1566 */
1567 static int
xmlFAGenerateTransitions(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr from,xmlRegStatePtr to,xmlRegAtomPtr atom)1568 xmlFAGenerateTransitions(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr from,
1569 xmlRegStatePtr to, xmlRegAtomPtr atom) {
1570 xmlRegStatePtr end;
1571 int nullable = 0;
1572
1573 if (atom == NULL) {
1574 ERROR("generate transition: atom == NULL");
1575 return(-1);
1576 }
1577 if (atom->type == XML_REGEXP_SUBREG) {
1578 /*
1579 * this is a subexpression handling one should not need to
1580 * create a new node except for XML_REGEXP_QUANT_RANGE.
1581 */
1582 if ((to != NULL) && (atom->stop != to) &&
1583 (atom->quant != XML_REGEXP_QUANT_RANGE)) {
1584 /*
1585 * Generate an epsilon transition to link to the target
1586 */
1587 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, to);
1588 #ifdef DV
1589 } else if ((to == NULL) && (atom->quant != XML_REGEXP_QUANT_RANGE) &&
1590 (atom->quant != XML_REGEXP_QUANT_ONCE)) {
1591 to = xmlRegStatePush(ctxt, to);
1592 if (to == NULL)
1593 return(-1);
1594 ctxt->state = to;
1595 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, to);
1596 #endif
1597 }
1598 switch (atom->quant) {
1599 case XML_REGEXP_QUANT_OPT:
1600 atom->quant = XML_REGEXP_QUANT_ONCE;
1601 /*
1602 * transition done to the state after end of atom.
1603 * 1. set transition from atom start to new state
1604 * 2. set transition from atom end to this state.
1605 */
1606 if (to == NULL) {
1607 xmlFAGenerateEpsilonTransition(ctxt, atom->start, 0);
1608 xmlFAGenerateEpsilonTransition(ctxt, atom->stop,
1609 ctxt->state);
1610 } else {
1611 xmlFAGenerateEpsilonTransition(ctxt, atom->start, to);
1612 }
1613 break;
1614 case XML_REGEXP_QUANT_MULT:
1615 atom->quant = XML_REGEXP_QUANT_ONCE;
1616 xmlFAGenerateEpsilonTransition(ctxt, atom->start, atom->stop);
1617 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1618 break;
1619 case XML_REGEXP_QUANT_PLUS:
1620 atom->quant = XML_REGEXP_QUANT_ONCE;
1621 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1622 break;
1623 case XML_REGEXP_QUANT_RANGE: {
1624 int counter;
1625 xmlRegStatePtr inter, newstate;
1626
1627 /*
1628 * create the final state now if needed
1629 */
1630 if (to != NULL) {
1631 newstate = to;
1632 } else {
1633 newstate = xmlRegStatePush(ctxt);
1634 if (newstate == NULL)
1635 return(-1);
1636 }
1637
1638 /*
1639 * The principle here is to use counted transition
1640 * to avoid explosion in the number of states in the
1641 * graph. This is clearly more complex but should not
1642 * be exploitable at runtime.
1643 */
1644 if ((atom->min == 0) && (atom->start0 == NULL)) {
1645 xmlRegAtomPtr copy;
1646 /*
1647 * duplicate a transition based on atom to count next
1648 * occurrences after 1. We cannot loop to atom->start
1649 * directly because we need an epsilon transition to
1650 * newstate.
1651 */
1652 /* ???? For some reason it seems we never reach that
1653 case, I suppose this got optimized out before when
1654 building the automata */
1655 copy = xmlRegCopyAtom(ctxt, atom);
1656 if (copy == NULL)
1657 return(-1);
1658 copy->quant = XML_REGEXP_QUANT_ONCE;
1659 copy->min = 0;
1660 copy->max = 0;
1661
1662 if (xmlFAGenerateTransitions(ctxt, atom->start, NULL, copy)
1663 < 0) {
1664 xmlRegFreeAtom(copy);
1665 return(-1);
1666 }
1667 inter = ctxt->state;
1668 counter = xmlRegGetCounter(ctxt);
1669 if (counter < 0)
1670 return(-1);
1671 ctxt->counters[counter].min = atom->min - 1;
1672 ctxt->counters[counter].max = atom->max - 1;
1673 /* count the number of times we see it again */
1674 xmlFAGenerateCountedEpsilonTransition(ctxt, inter,
1675 atom->stop, counter);
1676 /* allow a way out based on the count */
1677 xmlFAGenerateCountedTransition(ctxt, inter,
1678 newstate, counter);
1679 /* and also allow a direct exit for 0 */
1680 xmlFAGenerateEpsilonTransition(ctxt, atom->start,
1681 newstate);
1682 } else {
1683 /*
1684 * either we need the atom at least once or there
1685 * is an atom->start0 allowing to easily plug the
1686 * epsilon transition.
1687 */
1688 counter = xmlRegGetCounter(ctxt);
1689 if (counter < 0)
1690 return(-1);
1691 ctxt->counters[counter].min = atom->min - 1;
1692 ctxt->counters[counter].max = atom->max - 1;
1693 /* allow a way out based on the count */
1694 xmlFAGenerateCountedTransition(ctxt, atom->stop,
1695 newstate, counter);
1696 /* count the number of times we see it again */
1697 xmlFAGenerateCountedEpsilonTransition(ctxt, atom->stop,
1698 atom->start, counter);
1699 /* and if needed allow a direct exit for 0 */
1700 if (atom->min == 0)
1701 xmlFAGenerateEpsilonTransition(ctxt, atom->start0,
1702 newstate);
1703
1704 }
1705 atom->min = 0;
1706 atom->max = 0;
1707 atom->quant = XML_REGEXP_QUANT_ONCE;
1708 ctxt->state = newstate;
1709 }
1710 default:
1711 break;
1712 }
1713 if (xmlRegAtomPush(ctxt, atom) < 0)
1714 return(-1);
1715 return(0);
1716 }
1717 if ((atom->min == 0) && (atom->max == 0) &&
1718 (atom->quant == XML_REGEXP_QUANT_RANGE)) {
1719 /*
1720 * we can discard the atom and generate an epsilon transition instead
1721 */
1722 if (to == NULL) {
1723 to = xmlRegStatePush(ctxt);
1724 if (to == NULL)
1725 return(-1);
1726 }
1727 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1728 ctxt->state = to;
1729 xmlRegFreeAtom(atom);
1730 return(0);
1731 }
1732 if (to == NULL) {
1733 to = xmlRegStatePush(ctxt);
1734 if (to == NULL)
1735 return(-1);
1736 }
1737 end = to;
1738 if ((atom->quant == XML_REGEXP_QUANT_MULT) ||
1739 (atom->quant == XML_REGEXP_QUANT_PLUS)) {
1740 /*
1741 * Do not pollute the target state by adding transitions from
1742 * it as it is likely to be the shared target of multiple branches.
1743 * So isolate with an epsilon transition.
1744 */
1745 xmlRegStatePtr tmp;
1746
1747 tmp = xmlRegStatePush(ctxt);
1748 if (tmp == NULL)
1749 return(-1);
1750 xmlFAGenerateEpsilonTransition(ctxt, tmp, to);
1751 to = tmp;
1752 }
1753 if ((atom->quant == XML_REGEXP_QUANT_RANGE) &&
1754 (atom->min == 0) && (atom->max > 0)) {
1755 nullable = 1;
1756 atom->min = 1;
1757 if (atom->max == 1)
1758 atom->quant = XML_REGEXP_QUANT_OPT;
1759 }
1760 xmlRegStateAddTrans(ctxt, from, atom, to, -1, -1);
1761 ctxt->state = end;
1762 switch (atom->quant) {
1763 case XML_REGEXP_QUANT_OPT:
1764 atom->quant = XML_REGEXP_QUANT_ONCE;
1765 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1766 break;
1767 case XML_REGEXP_QUANT_MULT:
1768 atom->quant = XML_REGEXP_QUANT_ONCE;
1769 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1770 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1771 break;
1772 case XML_REGEXP_QUANT_PLUS:
1773 atom->quant = XML_REGEXP_QUANT_ONCE;
1774 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1775 break;
1776 case XML_REGEXP_QUANT_RANGE:
1777 if (nullable)
1778 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1779 break;
1780 default:
1781 break;
1782 }
1783 if (xmlRegAtomPush(ctxt, atom) < 0)
1784 return(-1);
1785 return(0);
1786 }
1787
1788 /**
1789 * xmlFAReduceEpsilonTransitions:
1790 * @ctxt: a regexp parser context
1791 * @fromnr: the from state
1792 * @tonr: the to state
1793 * @counter: should that transition be associated to a counted
1794 *
1795 */
1796 static void
xmlFAReduceEpsilonTransitions(xmlRegParserCtxtPtr ctxt,int fromnr,int tonr,int counter)1797 xmlFAReduceEpsilonTransitions(xmlRegParserCtxtPtr ctxt, int fromnr,
1798 int tonr, int counter) {
1799 int transnr;
1800 xmlRegStatePtr from;
1801 xmlRegStatePtr to;
1802
1803 #ifdef DEBUG_REGEXP_GRAPH
1804 printf("xmlFAReduceEpsilonTransitions(%d, %d)\n", fromnr, tonr);
1805 #endif
1806 from = ctxt->states[fromnr];
1807 if (from == NULL)
1808 return;
1809 to = ctxt->states[tonr];
1810 if (to == NULL)
1811 return;
1812 if ((to->mark == XML_REGEXP_MARK_START) ||
1813 (to->mark == XML_REGEXP_MARK_VISITED))
1814 return;
1815
1816 to->mark = XML_REGEXP_MARK_VISITED;
1817 if (to->type == XML_REGEXP_FINAL_STATE) {
1818 #ifdef DEBUG_REGEXP_GRAPH
1819 printf("State %d is final, so %d becomes final\n", tonr, fromnr);
1820 #endif
1821 from->type = XML_REGEXP_FINAL_STATE;
1822 }
1823 for (transnr = 0;transnr < to->nbTrans;transnr++) {
1824 if (to->trans[transnr].to < 0)
1825 continue;
1826 if (to->trans[transnr].atom == NULL) {
1827 /*
1828 * Don't remove counted transitions
1829 * Don't loop either
1830 */
1831 if (to->trans[transnr].to != fromnr) {
1832 if (to->trans[transnr].count >= 0) {
1833 int newto = to->trans[transnr].to;
1834
1835 xmlRegStateAddTrans(ctxt, from, NULL,
1836 ctxt->states[newto],
1837 -1, to->trans[transnr].count);
1838 } else {
1839 #ifdef DEBUG_REGEXP_GRAPH
1840 printf("Found epsilon trans %d from %d to %d\n",
1841 transnr, tonr, to->trans[transnr].to);
1842 #endif
1843 if (to->trans[transnr].counter >= 0) {
1844 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1845 to->trans[transnr].to,
1846 to->trans[transnr].counter);
1847 } else {
1848 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1849 to->trans[transnr].to,
1850 counter);
1851 }
1852 }
1853 }
1854 } else {
1855 int newto = to->trans[transnr].to;
1856
1857 if (to->trans[transnr].counter >= 0) {
1858 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1859 ctxt->states[newto],
1860 to->trans[transnr].counter, -1);
1861 } else {
1862 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1863 ctxt->states[newto], counter, -1);
1864 }
1865 }
1866 }
1867 to->mark = XML_REGEXP_MARK_NORMAL;
1868 }
1869
1870 /**
1871 * xmlFAEliminateSimpleEpsilonTransitions:
1872 * @ctxt: a regexp parser context
1873 *
1874 * Eliminating general epsilon transitions can get costly in the general
1875 * algorithm due to the large amount of generated new transitions and
1876 * associated comparisons. However for simple epsilon transition used just
1877 * to separate building blocks when generating the automata this can be
1878 * reduced to state elimination:
1879 * - if there exists an epsilon from X to Y
1880 * - if there is no other transition from X
1881 * then X and Y are semantically equivalent and X can be eliminated
1882 * If X is the start state then make Y the start state, else replace the
1883 * target of all transitions to X by transitions to Y.
1884 *
1885 * If X is a final state, skip it.
1886 * Otherwise it would be necessary to manipulate counters for this case when
1887 * eliminating state 2:
1888 * State 1 has a transition with an atom to state 2.
1889 * State 2 is final and has an epsilon transition to state 1.
1890 */
1891 static void
xmlFAEliminateSimpleEpsilonTransitions(xmlRegParserCtxtPtr ctxt)1892 xmlFAEliminateSimpleEpsilonTransitions(xmlRegParserCtxtPtr ctxt) {
1893 int statenr, i, j, newto;
1894 xmlRegStatePtr state, tmp;
1895
1896 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1897 state = ctxt->states[statenr];
1898 if (state == NULL)
1899 continue;
1900 if (state->nbTrans != 1)
1901 continue;
1902 if (state->type == XML_REGEXP_UNREACH_STATE ||
1903 state->type == XML_REGEXP_FINAL_STATE)
1904 continue;
1905 /* is the only transition out a basic transition */
1906 if ((state->trans[0].atom == NULL) &&
1907 (state->trans[0].to >= 0) &&
1908 (state->trans[0].to != statenr) &&
1909 (state->trans[0].counter < 0) &&
1910 (state->trans[0].count < 0)) {
1911 newto = state->trans[0].to;
1912
1913 if (state->type == XML_REGEXP_START_STATE) {
1914 #ifdef DEBUG_REGEXP_GRAPH
1915 printf("Found simple epsilon trans from start %d to %d\n",
1916 statenr, newto);
1917 #endif
1918 } else {
1919 #ifdef DEBUG_REGEXP_GRAPH
1920 printf("Found simple epsilon trans from %d to %d\n",
1921 statenr, newto);
1922 #endif
1923 for (i = 0;i < state->nbTransTo;i++) {
1924 tmp = ctxt->states[state->transTo[i]];
1925 for (j = 0;j < tmp->nbTrans;j++) {
1926 if (tmp->trans[j].to == statenr) {
1927 #ifdef DEBUG_REGEXP_GRAPH
1928 printf("Changed transition %d on %d to go to %d\n",
1929 j, tmp->no, newto);
1930 #endif
1931 tmp->trans[j].to = -1;
1932 xmlRegStateAddTrans(ctxt, tmp, tmp->trans[j].atom,
1933 ctxt->states[newto],
1934 tmp->trans[j].counter,
1935 tmp->trans[j].count);
1936 }
1937 }
1938 }
1939 if (state->type == XML_REGEXP_FINAL_STATE)
1940 ctxt->states[newto]->type = XML_REGEXP_FINAL_STATE;
1941 /* eliminate the transition completely */
1942 state->nbTrans = 0;
1943
1944 state->type = XML_REGEXP_UNREACH_STATE;
1945
1946 }
1947
1948 }
1949 }
1950 }
1951 /**
1952 * xmlFAEliminateEpsilonTransitions:
1953 * @ctxt: a regexp parser context
1954 *
1955 */
1956 static void
xmlFAEliminateEpsilonTransitions(xmlRegParserCtxtPtr ctxt)1957 xmlFAEliminateEpsilonTransitions(xmlRegParserCtxtPtr ctxt) {
1958 int statenr, transnr;
1959 xmlRegStatePtr state;
1960 int has_epsilon;
1961
1962 if (ctxt->states == NULL) return;
1963
1964 /*
1965 * Eliminate simple epsilon transition and the associated unreachable
1966 * states.
1967 */
1968 xmlFAEliminateSimpleEpsilonTransitions(ctxt);
1969 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1970 state = ctxt->states[statenr];
1971 if ((state != NULL) && (state->type == XML_REGEXP_UNREACH_STATE)) {
1972 #ifdef DEBUG_REGEXP_GRAPH
1973 printf("Removed unreachable state %d\n", statenr);
1974 #endif
1975 xmlRegFreeState(state);
1976 ctxt->states[statenr] = NULL;
1977 }
1978 }
1979
1980 has_epsilon = 0;
1981
1982 /*
1983 * Build the completed transitions bypassing the epsilons
1984 * Use a marking algorithm to avoid loops
1985 * Mark sink states too.
1986 * Process from the latest states backward to the start when
1987 * there is long cascading epsilon chains this minimize the
1988 * recursions and transition compares when adding the new ones
1989 */
1990 for (statenr = ctxt->nbStates - 1;statenr >= 0;statenr--) {
1991 state = ctxt->states[statenr];
1992 if (state == NULL)
1993 continue;
1994 if ((state->nbTrans == 0) &&
1995 (state->type != XML_REGEXP_FINAL_STATE)) {
1996 state->type = XML_REGEXP_SINK_STATE;
1997 }
1998 for (transnr = 0;transnr < state->nbTrans;transnr++) {
1999 if ((state->trans[transnr].atom == NULL) &&
2000 (state->trans[transnr].to >= 0)) {
2001 if (state->trans[transnr].to == statenr) {
2002 state->trans[transnr].to = -1;
2003 #ifdef DEBUG_REGEXP_GRAPH
2004 printf("Removed loopback epsilon trans %d on %d\n",
2005 transnr, statenr);
2006 #endif
2007 } else if (state->trans[transnr].count < 0) {
2008 int newto = state->trans[transnr].to;
2009
2010 #ifdef DEBUG_REGEXP_GRAPH
2011 printf("Found epsilon trans %d from %d to %d\n",
2012 transnr, statenr, newto);
2013 #endif
2014 has_epsilon = 1;
2015 state->trans[transnr].to = -2;
2016 state->mark = XML_REGEXP_MARK_START;
2017 xmlFAReduceEpsilonTransitions(ctxt, statenr,
2018 newto, state->trans[transnr].counter);
2019 state->mark = XML_REGEXP_MARK_NORMAL;
2020 #ifdef DEBUG_REGEXP_GRAPH
2021 } else {
2022 printf("Found counted transition %d on %d\n",
2023 transnr, statenr);
2024 #endif
2025 }
2026 }
2027 }
2028 }
2029 /*
2030 * Eliminate the epsilon transitions
2031 */
2032 if (has_epsilon) {
2033 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2034 state = ctxt->states[statenr];
2035 if (state == NULL)
2036 continue;
2037 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2038 xmlRegTransPtr trans = &(state->trans[transnr]);
2039 if ((trans->atom == NULL) &&
2040 (trans->count < 0) &&
2041 (trans->to >= 0)) {
2042 trans->to = -1;
2043 }
2044 }
2045 }
2046 }
2047
2048 /*
2049 * Use this pass to detect unreachable states too
2050 */
2051 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2052 state = ctxt->states[statenr];
2053 if (state != NULL)
2054 state->reached = XML_REGEXP_MARK_NORMAL;
2055 }
2056 state = ctxt->states[0];
2057 if (state != NULL)
2058 state->reached = XML_REGEXP_MARK_START;
2059 while (state != NULL) {
2060 xmlRegStatePtr target = NULL;
2061 state->reached = XML_REGEXP_MARK_VISITED;
2062 /*
2063 * Mark all states reachable from the current reachable state
2064 */
2065 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2066 if ((state->trans[transnr].to >= 0) &&
2067 ((state->trans[transnr].atom != NULL) ||
2068 (state->trans[transnr].count >= 0))) {
2069 int newto = state->trans[transnr].to;
2070
2071 if (ctxt->states[newto] == NULL)
2072 continue;
2073 if (ctxt->states[newto]->reached == XML_REGEXP_MARK_NORMAL) {
2074 ctxt->states[newto]->reached = XML_REGEXP_MARK_START;
2075 target = ctxt->states[newto];
2076 }
2077 }
2078 }
2079
2080 /*
2081 * find the next accessible state not explored
2082 */
2083 if (target == NULL) {
2084 for (statenr = 1;statenr < ctxt->nbStates;statenr++) {
2085 state = ctxt->states[statenr];
2086 if ((state != NULL) && (state->reached ==
2087 XML_REGEXP_MARK_START)) {
2088 target = state;
2089 break;
2090 }
2091 }
2092 }
2093 state = target;
2094 }
2095 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2096 state = ctxt->states[statenr];
2097 if ((state != NULL) && (state->reached == XML_REGEXP_MARK_NORMAL)) {
2098 #ifdef DEBUG_REGEXP_GRAPH
2099 printf("Removed unreachable state %d\n", statenr);
2100 #endif
2101 xmlRegFreeState(state);
2102 ctxt->states[statenr] = NULL;
2103 }
2104 }
2105
2106 }
2107
2108 static int
xmlFACompareRanges(xmlRegRangePtr range1,xmlRegRangePtr range2)2109 xmlFACompareRanges(xmlRegRangePtr range1, xmlRegRangePtr range2) {
2110 int ret = 0;
2111
2112 if ((range1->type == XML_REGEXP_RANGES) ||
2113 (range2->type == XML_REGEXP_RANGES) ||
2114 (range2->type == XML_REGEXP_SUBREG) ||
2115 (range1->type == XML_REGEXP_SUBREG) ||
2116 (range1->type == XML_REGEXP_STRING) ||
2117 (range2->type == XML_REGEXP_STRING))
2118 return(-1);
2119
2120 /* put them in order */
2121 if (range1->type > range2->type) {
2122 xmlRegRangePtr tmp;
2123
2124 tmp = range1;
2125 range1 = range2;
2126 range2 = tmp;
2127 }
2128 if ((range1->type == XML_REGEXP_ANYCHAR) ||
2129 (range2->type == XML_REGEXP_ANYCHAR)) {
2130 ret = 1;
2131 } else if ((range1->type == XML_REGEXP_EPSILON) ||
2132 (range2->type == XML_REGEXP_EPSILON)) {
2133 return(0);
2134 } else if (range1->type == range2->type) {
2135 if (range1->type != XML_REGEXP_CHARVAL)
2136 ret = 1;
2137 else if ((range1->end < range2->start) ||
2138 (range2->end < range1->start))
2139 ret = 0;
2140 else
2141 ret = 1;
2142 } else if (range1->type == XML_REGEXP_CHARVAL) {
2143 int codepoint;
2144 int neg = 0;
2145
2146 /*
2147 * just check all codepoints in the range for acceptance,
2148 * this is usually way cheaper since done only once at
2149 * compilation than testing over and over at runtime or
2150 * pushing too many states when evaluating.
2151 */
2152 if (((range1->neg == 0) && (range2->neg != 0)) ||
2153 ((range1->neg != 0) && (range2->neg == 0)))
2154 neg = 1;
2155
2156 for (codepoint = range1->start;codepoint <= range1->end ;codepoint++) {
2157 ret = xmlRegCheckCharacterRange(range2->type, codepoint,
2158 0, range2->start, range2->end,
2159 range2->blockName);
2160 if (ret < 0)
2161 return(-1);
2162 if (((neg == 1) && (ret == 0)) ||
2163 ((neg == 0) && (ret == 1)))
2164 return(1);
2165 }
2166 return(0);
2167 } else if ((range1->type == XML_REGEXP_BLOCK_NAME) ||
2168 (range2->type == XML_REGEXP_BLOCK_NAME)) {
2169 if (range1->type == range2->type) {
2170 ret = xmlStrEqual(range1->blockName, range2->blockName);
2171 } else {
2172 /*
2173 * comparing a block range with anything else is way
2174 * too costly, and maintaining the table is like too much
2175 * memory too, so let's force the automata to save state
2176 * here.
2177 */
2178 return(1);
2179 }
2180 } else if ((range1->type < XML_REGEXP_LETTER) ||
2181 (range2->type < XML_REGEXP_LETTER)) {
2182 if ((range1->type == XML_REGEXP_ANYSPACE) &&
2183 (range2->type == XML_REGEXP_NOTSPACE))
2184 ret = 0;
2185 else if ((range1->type == XML_REGEXP_INITNAME) &&
2186 (range2->type == XML_REGEXP_NOTINITNAME))
2187 ret = 0;
2188 else if ((range1->type == XML_REGEXP_NAMECHAR) &&
2189 (range2->type == XML_REGEXP_NOTNAMECHAR))
2190 ret = 0;
2191 else if ((range1->type == XML_REGEXP_DECIMAL) &&
2192 (range2->type == XML_REGEXP_NOTDECIMAL))
2193 ret = 0;
2194 else if ((range1->type == XML_REGEXP_REALCHAR) &&
2195 (range2->type == XML_REGEXP_NOTREALCHAR))
2196 ret = 0;
2197 else {
2198 /* same thing to limit complexity */
2199 return(1);
2200 }
2201 } else {
2202 ret = 0;
2203 /* range1->type < range2->type here */
2204 switch (range1->type) {
2205 case XML_REGEXP_LETTER:
2206 /* all disjoint except in the subgroups */
2207 if ((range2->type == XML_REGEXP_LETTER_UPPERCASE) ||
2208 (range2->type == XML_REGEXP_LETTER_LOWERCASE) ||
2209 (range2->type == XML_REGEXP_LETTER_TITLECASE) ||
2210 (range2->type == XML_REGEXP_LETTER_MODIFIER) ||
2211 (range2->type == XML_REGEXP_LETTER_OTHERS))
2212 ret = 1;
2213 break;
2214 case XML_REGEXP_MARK:
2215 if ((range2->type == XML_REGEXP_MARK_NONSPACING) ||
2216 (range2->type == XML_REGEXP_MARK_SPACECOMBINING) ||
2217 (range2->type == XML_REGEXP_MARK_ENCLOSING))
2218 ret = 1;
2219 break;
2220 case XML_REGEXP_NUMBER:
2221 if ((range2->type == XML_REGEXP_NUMBER_DECIMAL) ||
2222 (range2->type == XML_REGEXP_NUMBER_LETTER) ||
2223 (range2->type == XML_REGEXP_NUMBER_OTHERS))
2224 ret = 1;
2225 break;
2226 case XML_REGEXP_PUNCT:
2227 if ((range2->type == XML_REGEXP_PUNCT_CONNECTOR) ||
2228 (range2->type == XML_REGEXP_PUNCT_DASH) ||
2229 (range2->type == XML_REGEXP_PUNCT_OPEN) ||
2230 (range2->type == XML_REGEXP_PUNCT_CLOSE) ||
2231 (range2->type == XML_REGEXP_PUNCT_INITQUOTE) ||
2232 (range2->type == XML_REGEXP_PUNCT_FINQUOTE) ||
2233 (range2->type == XML_REGEXP_PUNCT_OTHERS))
2234 ret = 1;
2235 break;
2236 case XML_REGEXP_SEPAR:
2237 if ((range2->type == XML_REGEXP_SEPAR_SPACE) ||
2238 (range2->type == XML_REGEXP_SEPAR_LINE) ||
2239 (range2->type == XML_REGEXP_SEPAR_PARA))
2240 ret = 1;
2241 break;
2242 case XML_REGEXP_SYMBOL:
2243 if ((range2->type == XML_REGEXP_SYMBOL_MATH) ||
2244 (range2->type == XML_REGEXP_SYMBOL_CURRENCY) ||
2245 (range2->type == XML_REGEXP_SYMBOL_MODIFIER) ||
2246 (range2->type == XML_REGEXP_SYMBOL_OTHERS))
2247 ret = 1;
2248 break;
2249 case XML_REGEXP_OTHER:
2250 if ((range2->type == XML_REGEXP_OTHER_CONTROL) ||
2251 (range2->type == XML_REGEXP_OTHER_FORMAT) ||
2252 (range2->type == XML_REGEXP_OTHER_PRIVATE))
2253 ret = 1;
2254 break;
2255 default:
2256 if ((range2->type >= XML_REGEXP_LETTER) &&
2257 (range2->type < XML_REGEXP_BLOCK_NAME))
2258 ret = 0;
2259 else {
2260 /* safety net ! */
2261 return(1);
2262 }
2263 }
2264 }
2265 if (((range1->neg == 0) && (range2->neg != 0)) ||
2266 ((range1->neg != 0) && (range2->neg == 0)))
2267 ret = !ret;
2268 return(ret);
2269 }
2270
2271 /**
2272 * xmlFACompareAtomTypes:
2273 * @type1: an atom type
2274 * @type2: an atom type
2275 *
2276 * Compares two atoms type to check whether they intersect in some ways,
2277 * this is used by xmlFACompareAtoms only
2278 *
2279 * Returns 1 if they may intersect and 0 otherwise
2280 */
2281 static int
xmlFACompareAtomTypes(xmlRegAtomType type1,xmlRegAtomType type2)2282 xmlFACompareAtomTypes(xmlRegAtomType type1, xmlRegAtomType type2) {
2283 if ((type1 == XML_REGEXP_EPSILON) ||
2284 (type1 == XML_REGEXP_CHARVAL) ||
2285 (type1 == XML_REGEXP_RANGES) ||
2286 (type1 == XML_REGEXP_SUBREG) ||
2287 (type1 == XML_REGEXP_STRING) ||
2288 (type1 == XML_REGEXP_ANYCHAR))
2289 return(1);
2290 if ((type2 == XML_REGEXP_EPSILON) ||
2291 (type2 == XML_REGEXP_CHARVAL) ||
2292 (type2 == XML_REGEXP_RANGES) ||
2293 (type2 == XML_REGEXP_SUBREG) ||
2294 (type2 == XML_REGEXP_STRING) ||
2295 (type2 == XML_REGEXP_ANYCHAR))
2296 return(1);
2297
2298 if (type1 == type2) return(1);
2299
2300 /* simplify subsequent compares by making sure type1 < type2 */
2301 if (type1 > type2) {
2302 xmlRegAtomType tmp = type1;
2303 type1 = type2;
2304 type2 = tmp;
2305 }
2306 switch (type1) {
2307 case XML_REGEXP_ANYSPACE: /* \s */
2308 /* can't be a letter, number, mark, punctuation, symbol */
2309 if ((type2 == XML_REGEXP_NOTSPACE) ||
2310 ((type2 >= XML_REGEXP_LETTER) &&
2311 (type2 <= XML_REGEXP_LETTER_OTHERS)) ||
2312 ((type2 >= XML_REGEXP_NUMBER) &&
2313 (type2 <= XML_REGEXP_NUMBER_OTHERS)) ||
2314 ((type2 >= XML_REGEXP_MARK) &&
2315 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2316 ((type2 >= XML_REGEXP_PUNCT) &&
2317 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2318 ((type2 >= XML_REGEXP_SYMBOL) &&
2319 (type2 <= XML_REGEXP_SYMBOL_OTHERS))
2320 ) return(0);
2321 break;
2322 case XML_REGEXP_NOTSPACE: /* \S */
2323 break;
2324 case XML_REGEXP_INITNAME: /* \l */
2325 /* can't be a number, mark, separator, punctuation, symbol or other */
2326 if ((type2 == XML_REGEXP_NOTINITNAME) ||
2327 ((type2 >= XML_REGEXP_NUMBER) &&
2328 (type2 <= XML_REGEXP_NUMBER_OTHERS)) ||
2329 ((type2 >= XML_REGEXP_MARK) &&
2330 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2331 ((type2 >= XML_REGEXP_SEPAR) &&
2332 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2333 ((type2 >= XML_REGEXP_PUNCT) &&
2334 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2335 ((type2 >= XML_REGEXP_SYMBOL) &&
2336 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2337 ((type2 >= XML_REGEXP_OTHER) &&
2338 (type2 <= XML_REGEXP_OTHER_NA))
2339 ) return(0);
2340 break;
2341 case XML_REGEXP_NOTINITNAME: /* \L */
2342 break;
2343 case XML_REGEXP_NAMECHAR: /* \c */
2344 /* can't be a mark, separator, punctuation, symbol or other */
2345 if ((type2 == XML_REGEXP_NOTNAMECHAR) ||
2346 ((type2 >= XML_REGEXP_MARK) &&
2347 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2348 ((type2 >= XML_REGEXP_PUNCT) &&
2349 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2350 ((type2 >= XML_REGEXP_SEPAR) &&
2351 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2352 ((type2 >= XML_REGEXP_SYMBOL) &&
2353 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2354 ((type2 >= XML_REGEXP_OTHER) &&
2355 (type2 <= XML_REGEXP_OTHER_NA))
2356 ) return(0);
2357 break;
2358 case XML_REGEXP_NOTNAMECHAR: /* \C */
2359 break;
2360 case XML_REGEXP_DECIMAL: /* \d */
2361 /* can't be a letter, mark, separator, punctuation, symbol or other */
2362 if ((type2 == XML_REGEXP_NOTDECIMAL) ||
2363 (type2 == XML_REGEXP_REALCHAR) ||
2364 ((type2 >= XML_REGEXP_LETTER) &&
2365 (type2 <= XML_REGEXP_LETTER_OTHERS)) ||
2366 ((type2 >= XML_REGEXP_MARK) &&
2367 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2368 ((type2 >= XML_REGEXP_PUNCT) &&
2369 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2370 ((type2 >= XML_REGEXP_SEPAR) &&
2371 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2372 ((type2 >= XML_REGEXP_SYMBOL) &&
2373 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2374 ((type2 >= XML_REGEXP_OTHER) &&
2375 (type2 <= XML_REGEXP_OTHER_NA))
2376 )return(0);
2377 break;
2378 case XML_REGEXP_NOTDECIMAL: /* \D */
2379 break;
2380 case XML_REGEXP_REALCHAR: /* \w */
2381 /* can't be a mark, separator, punctuation, symbol or other */
2382 if ((type2 == XML_REGEXP_NOTDECIMAL) ||
2383 ((type2 >= XML_REGEXP_MARK) &&
2384 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2385 ((type2 >= XML_REGEXP_PUNCT) &&
2386 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2387 ((type2 >= XML_REGEXP_SEPAR) &&
2388 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2389 ((type2 >= XML_REGEXP_SYMBOL) &&
2390 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2391 ((type2 >= XML_REGEXP_OTHER) &&
2392 (type2 <= XML_REGEXP_OTHER_NA))
2393 )return(0);
2394 break;
2395 case XML_REGEXP_NOTREALCHAR: /* \W */
2396 break;
2397 /*
2398 * at that point we know both type 1 and type2 are from
2399 * character categories are ordered and are different,
2400 * it becomes simple because this is a partition
2401 */
2402 case XML_REGEXP_LETTER:
2403 if (type2 <= XML_REGEXP_LETTER_OTHERS)
2404 return(1);
2405 return(0);
2406 case XML_REGEXP_LETTER_UPPERCASE:
2407 case XML_REGEXP_LETTER_LOWERCASE:
2408 case XML_REGEXP_LETTER_TITLECASE:
2409 case XML_REGEXP_LETTER_MODIFIER:
2410 case XML_REGEXP_LETTER_OTHERS:
2411 return(0);
2412 case XML_REGEXP_MARK:
2413 if (type2 <= XML_REGEXP_MARK_ENCLOSING)
2414 return(1);
2415 return(0);
2416 case XML_REGEXP_MARK_NONSPACING:
2417 case XML_REGEXP_MARK_SPACECOMBINING:
2418 case XML_REGEXP_MARK_ENCLOSING:
2419 return(0);
2420 case XML_REGEXP_NUMBER:
2421 if (type2 <= XML_REGEXP_NUMBER_OTHERS)
2422 return(1);
2423 return(0);
2424 case XML_REGEXP_NUMBER_DECIMAL:
2425 case XML_REGEXP_NUMBER_LETTER:
2426 case XML_REGEXP_NUMBER_OTHERS:
2427 return(0);
2428 case XML_REGEXP_PUNCT:
2429 if (type2 <= XML_REGEXP_PUNCT_OTHERS)
2430 return(1);
2431 return(0);
2432 case XML_REGEXP_PUNCT_CONNECTOR:
2433 case XML_REGEXP_PUNCT_DASH:
2434 case XML_REGEXP_PUNCT_OPEN:
2435 case XML_REGEXP_PUNCT_CLOSE:
2436 case XML_REGEXP_PUNCT_INITQUOTE:
2437 case XML_REGEXP_PUNCT_FINQUOTE:
2438 case XML_REGEXP_PUNCT_OTHERS:
2439 return(0);
2440 case XML_REGEXP_SEPAR:
2441 if (type2 <= XML_REGEXP_SEPAR_PARA)
2442 return(1);
2443 return(0);
2444 case XML_REGEXP_SEPAR_SPACE:
2445 case XML_REGEXP_SEPAR_LINE:
2446 case XML_REGEXP_SEPAR_PARA:
2447 return(0);
2448 case XML_REGEXP_SYMBOL:
2449 if (type2 <= XML_REGEXP_SYMBOL_OTHERS)
2450 return(1);
2451 return(0);
2452 case XML_REGEXP_SYMBOL_MATH:
2453 case XML_REGEXP_SYMBOL_CURRENCY:
2454 case XML_REGEXP_SYMBOL_MODIFIER:
2455 case XML_REGEXP_SYMBOL_OTHERS:
2456 return(0);
2457 case XML_REGEXP_OTHER:
2458 if (type2 <= XML_REGEXP_OTHER_NA)
2459 return(1);
2460 return(0);
2461 case XML_REGEXP_OTHER_CONTROL:
2462 case XML_REGEXP_OTHER_FORMAT:
2463 case XML_REGEXP_OTHER_PRIVATE:
2464 case XML_REGEXP_OTHER_NA:
2465 return(0);
2466 default:
2467 break;
2468 }
2469 return(1);
2470 }
2471
2472 /**
2473 * xmlFAEqualAtoms:
2474 * @atom1: an atom
2475 * @atom2: an atom
2476 * @deep: if not set only compare string pointers
2477 *
2478 * Compares two atoms to check whether they are the same exactly
2479 * this is used to remove equivalent transitions
2480 *
2481 * Returns 1 if same and 0 otherwise
2482 */
2483 static int
xmlFAEqualAtoms(xmlRegAtomPtr atom1,xmlRegAtomPtr atom2,int deep)2484 xmlFAEqualAtoms(xmlRegAtomPtr atom1, xmlRegAtomPtr atom2, int deep) {
2485 int ret = 0;
2486
2487 if (atom1 == atom2)
2488 return(1);
2489 if ((atom1 == NULL) || (atom2 == NULL))
2490 return(0);
2491
2492 if (atom1->type != atom2->type)
2493 return(0);
2494 switch (atom1->type) {
2495 case XML_REGEXP_EPSILON:
2496 ret = 0;
2497 break;
2498 case XML_REGEXP_STRING:
2499 if (!deep)
2500 ret = (atom1->valuep == atom2->valuep);
2501 else
2502 ret = xmlStrEqual((xmlChar *)atom1->valuep,
2503 (xmlChar *)atom2->valuep);
2504 break;
2505 case XML_REGEXP_CHARVAL:
2506 ret = (atom1->codepoint == atom2->codepoint);
2507 break;
2508 case XML_REGEXP_RANGES:
2509 /* too hard to do in the general case */
2510 ret = 0;
2511 default:
2512 break;
2513 }
2514 return(ret);
2515 }
2516
2517 /**
2518 * xmlFACompareAtoms:
2519 * @atom1: an atom
2520 * @atom2: an atom
2521 * @deep: if not set only compare string pointers
2522 *
2523 * Compares two atoms to check whether they intersect in some ways,
2524 * this is used by xmlFAComputesDeterminism and xmlFARecurseDeterminism only
2525 *
2526 * Returns 1 if yes and 0 otherwise
2527 */
2528 static int
xmlFACompareAtoms(xmlRegAtomPtr atom1,xmlRegAtomPtr atom2,int deep)2529 xmlFACompareAtoms(xmlRegAtomPtr atom1, xmlRegAtomPtr atom2, int deep) {
2530 int ret = 1;
2531
2532 if (atom1 == atom2)
2533 return(1);
2534 if ((atom1 == NULL) || (atom2 == NULL))
2535 return(0);
2536
2537 if ((atom1->type == XML_REGEXP_ANYCHAR) ||
2538 (atom2->type == XML_REGEXP_ANYCHAR))
2539 return(1);
2540
2541 if (atom1->type > atom2->type) {
2542 xmlRegAtomPtr tmp;
2543 tmp = atom1;
2544 atom1 = atom2;
2545 atom2 = tmp;
2546 }
2547 if (atom1->type != atom2->type) {
2548 ret = xmlFACompareAtomTypes(atom1->type, atom2->type);
2549 /* if they can't intersect at the type level break now */
2550 if (ret == 0)
2551 return(0);
2552 }
2553 switch (atom1->type) {
2554 case XML_REGEXP_STRING:
2555 if (!deep)
2556 ret = (atom1->valuep != atom2->valuep);
2557 else {
2558 xmlChar *val1 = (xmlChar *)atom1->valuep;
2559 xmlChar *val2 = (xmlChar *)atom2->valuep;
2560 int compound1 = (xmlStrchr(val1, '|') != NULL);
2561 int compound2 = (xmlStrchr(val2, '|') != NULL);
2562
2563 /* Ignore negative match flag for ##other namespaces */
2564 if (compound1 != compound2)
2565 return(0);
2566
2567 ret = xmlRegStrEqualWildcard(val1, val2);
2568 }
2569 break;
2570 case XML_REGEXP_EPSILON:
2571 goto not_determinist;
2572 case XML_REGEXP_CHARVAL:
2573 if (atom2->type == XML_REGEXP_CHARVAL) {
2574 ret = (atom1->codepoint == atom2->codepoint);
2575 } else {
2576 ret = xmlRegCheckCharacter(atom2, atom1->codepoint);
2577 if (ret < 0)
2578 ret = 1;
2579 }
2580 break;
2581 case XML_REGEXP_RANGES:
2582 if (atom2->type == XML_REGEXP_RANGES) {
2583 int i, j, res;
2584 xmlRegRangePtr r1, r2;
2585
2586 /*
2587 * need to check that none of the ranges eventually matches
2588 */
2589 for (i = 0;i < atom1->nbRanges;i++) {
2590 for (j = 0;j < atom2->nbRanges;j++) {
2591 r1 = atom1->ranges[i];
2592 r2 = atom2->ranges[j];
2593 res = xmlFACompareRanges(r1, r2);
2594 if (res == 1) {
2595 ret = 1;
2596 goto done;
2597 }
2598 }
2599 }
2600 ret = 0;
2601 }
2602 break;
2603 default:
2604 goto not_determinist;
2605 }
2606 done:
2607 if (atom1->neg != atom2->neg) {
2608 ret = !ret;
2609 }
2610 if (ret == 0)
2611 return(0);
2612 not_determinist:
2613 return(1);
2614 }
2615
2616 /**
2617 * xmlFARecurseDeterminism:
2618 * @ctxt: a regexp parser context
2619 *
2620 * Check whether the associated regexp is determinist,
2621 * should be called after xmlFAEliminateEpsilonTransitions()
2622 *
2623 */
2624 static int
xmlFARecurseDeterminism(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr state,int to,xmlRegAtomPtr atom)2625 xmlFARecurseDeterminism(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
2626 int to, xmlRegAtomPtr atom) {
2627 int ret = 1;
2628 int res;
2629 int transnr, nbTrans;
2630 xmlRegTransPtr t1;
2631 int deep = 1;
2632
2633 if (state == NULL)
2634 return(ret);
2635 if (state->markd == XML_REGEXP_MARK_VISITED)
2636 return(ret);
2637
2638 if (ctxt->flags & AM_AUTOMATA_RNG)
2639 deep = 0;
2640
2641 /*
2642 * don't recurse on transitions potentially added in the course of
2643 * the elimination.
2644 */
2645 nbTrans = state->nbTrans;
2646 for (transnr = 0;transnr < nbTrans;transnr++) {
2647 t1 = &(state->trans[transnr]);
2648 /*
2649 * check transitions conflicting with the one looked at
2650 */
2651 if (t1->atom == NULL) {
2652 if (t1->to < 0)
2653 continue;
2654 state->markd = XML_REGEXP_MARK_VISITED;
2655 res = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
2656 to, atom);
2657 if (res == 0) {
2658 ret = 0;
2659 /* t1->nd = 1; */
2660 }
2661 continue;
2662 }
2663 if (t1->to != to)
2664 continue;
2665 if (xmlFACompareAtoms(t1->atom, atom, deep)) {
2666 ret = 0;
2667 /* mark the transition as non-deterministic */
2668 t1->nd = 1;
2669 }
2670 }
2671 return(ret);
2672 }
2673
2674 /**
2675 * xmlFAFinishRecurseDeterminism:
2676 * @ctxt: a regexp parser context
2677 *
2678 * Reset flags after checking determinism.
2679 */
2680 static void
xmlFAFinishRecurseDeterminism(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr state)2681 xmlFAFinishRecurseDeterminism(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state) {
2682 int transnr, nbTrans;
2683
2684 if (state == NULL)
2685 return;
2686 if (state->markd != XML_REGEXP_MARK_VISITED)
2687 return;
2688 state->markd = 0;
2689
2690 nbTrans = state->nbTrans;
2691 for (transnr = 0; transnr < nbTrans; transnr++) {
2692 xmlRegTransPtr t1 = &state->trans[transnr];
2693 if ((t1->atom == NULL) && (t1->to >= 0))
2694 xmlFAFinishRecurseDeterminism(ctxt, ctxt->states[t1->to]);
2695 }
2696 }
2697
2698 /**
2699 * xmlFAComputesDeterminism:
2700 * @ctxt: a regexp parser context
2701 *
2702 * Check whether the associated regexp is determinist,
2703 * should be called after xmlFAEliminateEpsilonTransitions()
2704 *
2705 */
2706 static int
xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt)2707 xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt) {
2708 int statenr, transnr;
2709 xmlRegStatePtr state;
2710 xmlRegTransPtr t1, t2, last;
2711 int i;
2712 int ret = 1;
2713 int deep = 1;
2714
2715 #ifdef DEBUG_REGEXP_GRAPH
2716 printf("xmlFAComputesDeterminism\n");
2717 xmlRegPrintCtxt(stdout, ctxt);
2718 #endif
2719 if (ctxt->determinist != -1)
2720 return(ctxt->determinist);
2721
2722 if (ctxt->flags & AM_AUTOMATA_RNG)
2723 deep = 0;
2724
2725 /*
2726 * First cleanup the automata removing cancelled transitions
2727 */
2728 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2729 state = ctxt->states[statenr];
2730 if (state == NULL)
2731 continue;
2732 if (state->nbTrans < 2)
2733 continue;
2734 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2735 t1 = &(state->trans[transnr]);
2736 /*
2737 * Determinism checks in case of counted or all transitions
2738 * will have to be handled separately
2739 */
2740 if (t1->atom == NULL) {
2741 /* t1->nd = 1; */
2742 continue;
2743 }
2744 if (t1->to == -1) /* eliminated */
2745 continue;
2746 for (i = 0;i < transnr;i++) {
2747 t2 = &(state->trans[i]);
2748 if (t2->to == -1) /* eliminated */
2749 continue;
2750 if (t2->atom != NULL) {
2751 if (t1->to == t2->to) {
2752 /*
2753 * Here we use deep because we want to keep the
2754 * transitions which indicate a conflict
2755 */
2756 if (xmlFAEqualAtoms(t1->atom, t2->atom, deep) &&
2757 (t1->counter == t2->counter) &&
2758 (t1->count == t2->count))
2759 t2->to = -1; /* eliminated */
2760 }
2761 }
2762 }
2763 }
2764 }
2765
2766 /*
2767 * Check for all states that there aren't 2 transitions
2768 * with the same atom and a different target.
2769 */
2770 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2771 state = ctxt->states[statenr];
2772 if (state == NULL)
2773 continue;
2774 if (state->nbTrans < 2)
2775 continue;
2776 last = NULL;
2777 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2778 t1 = &(state->trans[transnr]);
2779 /*
2780 * Determinism checks in case of counted or all transitions
2781 * will have to be handled separately
2782 */
2783 if (t1->atom == NULL) {
2784 continue;
2785 }
2786 if (t1->to == -1) /* eliminated */
2787 continue;
2788 for (i = 0;i < transnr;i++) {
2789 t2 = &(state->trans[i]);
2790 if (t2->to == -1) /* eliminated */
2791 continue;
2792 if (t2->atom != NULL) {
2793 /*
2794 * But here we don't use deep because we want to
2795 * find transitions which indicate a conflict
2796 */
2797 if (xmlFACompareAtoms(t1->atom, t2->atom, 1)) {
2798 ret = 0;
2799 /* mark the transitions as non-deterministic ones */
2800 t1->nd = 1;
2801 t2->nd = 1;
2802 last = t1;
2803 }
2804 } else if (t1->to != -1) {
2805 /*
2806 * do the closure in case of remaining specific
2807 * epsilon transitions like choices or all
2808 */
2809 ret = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
2810 t2->to, t2->atom);
2811 xmlFAFinishRecurseDeterminism(ctxt, ctxt->states[t1->to]);
2812 /* don't shortcut the computation so all non deterministic
2813 transition get marked down
2814 if (ret == 0)
2815 return(0);
2816 */
2817 if (ret == 0) {
2818 t1->nd = 1;
2819 /* t2->nd = 1; */
2820 last = t1;
2821 }
2822 }
2823 }
2824 /* don't shortcut the computation so all non deterministic
2825 transition get marked down
2826 if (ret == 0)
2827 break; */
2828 }
2829
2830 /*
2831 * mark specifically the last non-deterministic transition
2832 * from a state since there is no need to set-up rollback
2833 * from it
2834 */
2835 if (last != NULL) {
2836 last->nd = 2;
2837 }
2838
2839 /* don't shortcut the computation so all non deterministic
2840 transition get marked down
2841 if (ret == 0)
2842 break; */
2843 }
2844
2845 ctxt->determinist = ret;
2846 return(ret);
2847 }
2848
2849 /************************************************************************
2850 * *
2851 * Routines to check input against transition atoms *
2852 * *
2853 ************************************************************************/
2854
2855 static int
xmlRegCheckCharacterRange(xmlRegAtomType type,int codepoint,int neg,int start,int end,const xmlChar * blockName)2856 xmlRegCheckCharacterRange(xmlRegAtomType type, int codepoint, int neg,
2857 int start, int end, const xmlChar *blockName) {
2858 int ret = 0;
2859
2860 switch (type) {
2861 case XML_REGEXP_STRING:
2862 case XML_REGEXP_SUBREG:
2863 case XML_REGEXP_RANGES:
2864 case XML_REGEXP_EPSILON:
2865 return(-1);
2866 case XML_REGEXP_ANYCHAR:
2867 ret = ((codepoint != '\n') && (codepoint != '\r'));
2868 break;
2869 case XML_REGEXP_CHARVAL:
2870 ret = ((codepoint >= start) && (codepoint <= end));
2871 break;
2872 case XML_REGEXP_NOTSPACE:
2873 neg = !neg;
2874 /* Falls through. */
2875 case XML_REGEXP_ANYSPACE:
2876 ret = ((codepoint == '\n') || (codepoint == '\r') ||
2877 (codepoint == '\t') || (codepoint == ' '));
2878 break;
2879 case XML_REGEXP_NOTINITNAME:
2880 neg = !neg;
2881 /* Falls through. */
2882 case XML_REGEXP_INITNAME:
2883 ret = (IS_LETTER(codepoint) ||
2884 (codepoint == '_') || (codepoint == ':'));
2885 break;
2886 case XML_REGEXP_NOTNAMECHAR:
2887 neg = !neg;
2888 /* Falls through. */
2889 case XML_REGEXP_NAMECHAR:
2890 ret = (IS_LETTER(codepoint) || IS_DIGIT(codepoint) ||
2891 (codepoint == '.') || (codepoint == '-') ||
2892 (codepoint == '_') || (codepoint == ':') ||
2893 IS_COMBINING(codepoint) || IS_EXTENDER(codepoint));
2894 break;
2895 case XML_REGEXP_NOTDECIMAL:
2896 neg = !neg;
2897 /* Falls through. */
2898 case XML_REGEXP_DECIMAL:
2899 ret = xmlUCSIsCatNd(codepoint);
2900 break;
2901 case XML_REGEXP_REALCHAR:
2902 neg = !neg;
2903 /* Falls through. */
2904 case XML_REGEXP_NOTREALCHAR:
2905 ret = xmlUCSIsCatP(codepoint);
2906 if (ret == 0)
2907 ret = xmlUCSIsCatZ(codepoint);
2908 if (ret == 0)
2909 ret = xmlUCSIsCatC(codepoint);
2910 break;
2911 case XML_REGEXP_LETTER:
2912 ret = xmlUCSIsCatL(codepoint);
2913 break;
2914 case XML_REGEXP_LETTER_UPPERCASE:
2915 ret = xmlUCSIsCatLu(codepoint);
2916 break;
2917 case XML_REGEXP_LETTER_LOWERCASE:
2918 ret = xmlUCSIsCatLl(codepoint);
2919 break;
2920 case XML_REGEXP_LETTER_TITLECASE:
2921 ret = xmlUCSIsCatLt(codepoint);
2922 break;
2923 case XML_REGEXP_LETTER_MODIFIER:
2924 ret = xmlUCSIsCatLm(codepoint);
2925 break;
2926 case XML_REGEXP_LETTER_OTHERS:
2927 ret = xmlUCSIsCatLo(codepoint);
2928 break;
2929 case XML_REGEXP_MARK:
2930 ret = xmlUCSIsCatM(codepoint);
2931 break;
2932 case XML_REGEXP_MARK_NONSPACING:
2933 ret = xmlUCSIsCatMn(codepoint);
2934 break;
2935 case XML_REGEXP_MARK_SPACECOMBINING:
2936 ret = xmlUCSIsCatMc(codepoint);
2937 break;
2938 case XML_REGEXP_MARK_ENCLOSING:
2939 ret = xmlUCSIsCatMe(codepoint);
2940 break;
2941 case XML_REGEXP_NUMBER:
2942 ret = xmlUCSIsCatN(codepoint);
2943 break;
2944 case XML_REGEXP_NUMBER_DECIMAL:
2945 ret = xmlUCSIsCatNd(codepoint);
2946 break;
2947 case XML_REGEXP_NUMBER_LETTER:
2948 ret = xmlUCSIsCatNl(codepoint);
2949 break;
2950 case XML_REGEXP_NUMBER_OTHERS:
2951 ret = xmlUCSIsCatNo(codepoint);
2952 break;
2953 case XML_REGEXP_PUNCT:
2954 ret = xmlUCSIsCatP(codepoint);
2955 break;
2956 case XML_REGEXP_PUNCT_CONNECTOR:
2957 ret = xmlUCSIsCatPc(codepoint);
2958 break;
2959 case XML_REGEXP_PUNCT_DASH:
2960 ret = xmlUCSIsCatPd(codepoint);
2961 break;
2962 case XML_REGEXP_PUNCT_OPEN:
2963 ret = xmlUCSIsCatPs(codepoint);
2964 break;
2965 case XML_REGEXP_PUNCT_CLOSE:
2966 ret = xmlUCSIsCatPe(codepoint);
2967 break;
2968 case XML_REGEXP_PUNCT_INITQUOTE:
2969 ret = xmlUCSIsCatPi(codepoint);
2970 break;
2971 case XML_REGEXP_PUNCT_FINQUOTE:
2972 ret = xmlUCSIsCatPf(codepoint);
2973 break;
2974 case XML_REGEXP_PUNCT_OTHERS:
2975 ret = xmlUCSIsCatPo(codepoint);
2976 break;
2977 case XML_REGEXP_SEPAR:
2978 ret = xmlUCSIsCatZ(codepoint);
2979 break;
2980 case XML_REGEXP_SEPAR_SPACE:
2981 ret = xmlUCSIsCatZs(codepoint);
2982 break;
2983 case XML_REGEXP_SEPAR_LINE:
2984 ret = xmlUCSIsCatZl(codepoint);
2985 break;
2986 case XML_REGEXP_SEPAR_PARA:
2987 ret = xmlUCSIsCatZp(codepoint);
2988 break;
2989 case XML_REGEXP_SYMBOL:
2990 ret = xmlUCSIsCatS(codepoint);
2991 break;
2992 case XML_REGEXP_SYMBOL_MATH:
2993 ret = xmlUCSIsCatSm(codepoint);
2994 break;
2995 case XML_REGEXP_SYMBOL_CURRENCY:
2996 ret = xmlUCSIsCatSc(codepoint);
2997 break;
2998 case XML_REGEXP_SYMBOL_MODIFIER:
2999 ret = xmlUCSIsCatSk(codepoint);
3000 break;
3001 case XML_REGEXP_SYMBOL_OTHERS:
3002 ret = xmlUCSIsCatSo(codepoint);
3003 break;
3004 case XML_REGEXP_OTHER:
3005 ret = xmlUCSIsCatC(codepoint);
3006 break;
3007 case XML_REGEXP_OTHER_CONTROL:
3008 ret = xmlUCSIsCatCc(codepoint);
3009 break;
3010 case XML_REGEXP_OTHER_FORMAT:
3011 ret = xmlUCSIsCatCf(codepoint);
3012 break;
3013 case XML_REGEXP_OTHER_PRIVATE:
3014 ret = xmlUCSIsCatCo(codepoint);
3015 break;
3016 case XML_REGEXP_OTHER_NA:
3017 /* ret = xmlUCSIsCatCn(codepoint); */
3018 /* Seems it doesn't exist anymore in recent Unicode releases */
3019 ret = 0;
3020 break;
3021 case XML_REGEXP_BLOCK_NAME:
3022 ret = xmlUCSIsBlock(codepoint, (const char *) blockName);
3023 break;
3024 }
3025 if (neg)
3026 return(!ret);
3027 return(ret);
3028 }
3029
3030 static int
xmlRegCheckCharacter(xmlRegAtomPtr atom,int codepoint)3031 xmlRegCheckCharacter(xmlRegAtomPtr atom, int codepoint) {
3032 int i, ret = 0;
3033 xmlRegRangePtr range;
3034
3035 if ((atom == NULL) || (!IS_CHAR(codepoint)))
3036 return(-1);
3037
3038 switch (atom->type) {
3039 case XML_REGEXP_SUBREG:
3040 case XML_REGEXP_EPSILON:
3041 return(-1);
3042 case XML_REGEXP_CHARVAL:
3043 return(codepoint == atom->codepoint);
3044 case XML_REGEXP_RANGES: {
3045 int accept = 0;
3046
3047 for (i = 0;i < atom->nbRanges;i++) {
3048 range = atom->ranges[i];
3049 if (range->neg == 2) {
3050 ret = xmlRegCheckCharacterRange(range->type, codepoint,
3051 0, range->start, range->end,
3052 range->blockName);
3053 if (ret != 0)
3054 return(0); /* excluded char */
3055 } else if (range->neg) {
3056 ret = xmlRegCheckCharacterRange(range->type, codepoint,
3057 0, range->start, range->end,
3058 range->blockName);
3059 if (ret == 0)
3060 accept = 1;
3061 else
3062 return(0);
3063 } else {
3064 ret = xmlRegCheckCharacterRange(range->type, codepoint,
3065 0, range->start, range->end,
3066 range->blockName);
3067 if (ret != 0)
3068 accept = 1; /* might still be excluded */
3069 }
3070 }
3071 return(accept);
3072 }
3073 case XML_REGEXP_STRING:
3074 printf("TODO: XML_REGEXP_STRING\n");
3075 return(-1);
3076 case XML_REGEXP_ANYCHAR:
3077 case XML_REGEXP_ANYSPACE:
3078 case XML_REGEXP_NOTSPACE:
3079 case XML_REGEXP_INITNAME:
3080 case XML_REGEXP_NOTINITNAME:
3081 case XML_REGEXP_NAMECHAR:
3082 case XML_REGEXP_NOTNAMECHAR:
3083 case XML_REGEXP_DECIMAL:
3084 case XML_REGEXP_NOTDECIMAL:
3085 case XML_REGEXP_REALCHAR:
3086 case XML_REGEXP_NOTREALCHAR:
3087 case XML_REGEXP_LETTER:
3088 case XML_REGEXP_LETTER_UPPERCASE:
3089 case XML_REGEXP_LETTER_LOWERCASE:
3090 case XML_REGEXP_LETTER_TITLECASE:
3091 case XML_REGEXP_LETTER_MODIFIER:
3092 case XML_REGEXP_LETTER_OTHERS:
3093 case XML_REGEXP_MARK:
3094 case XML_REGEXP_MARK_NONSPACING:
3095 case XML_REGEXP_MARK_SPACECOMBINING:
3096 case XML_REGEXP_MARK_ENCLOSING:
3097 case XML_REGEXP_NUMBER:
3098 case XML_REGEXP_NUMBER_DECIMAL:
3099 case XML_REGEXP_NUMBER_LETTER:
3100 case XML_REGEXP_NUMBER_OTHERS:
3101 case XML_REGEXP_PUNCT:
3102 case XML_REGEXP_PUNCT_CONNECTOR:
3103 case XML_REGEXP_PUNCT_DASH:
3104 case XML_REGEXP_PUNCT_OPEN:
3105 case XML_REGEXP_PUNCT_CLOSE:
3106 case XML_REGEXP_PUNCT_INITQUOTE:
3107 case XML_REGEXP_PUNCT_FINQUOTE:
3108 case XML_REGEXP_PUNCT_OTHERS:
3109 case XML_REGEXP_SEPAR:
3110 case XML_REGEXP_SEPAR_SPACE:
3111 case XML_REGEXP_SEPAR_LINE:
3112 case XML_REGEXP_SEPAR_PARA:
3113 case XML_REGEXP_SYMBOL:
3114 case XML_REGEXP_SYMBOL_MATH:
3115 case XML_REGEXP_SYMBOL_CURRENCY:
3116 case XML_REGEXP_SYMBOL_MODIFIER:
3117 case XML_REGEXP_SYMBOL_OTHERS:
3118 case XML_REGEXP_OTHER:
3119 case XML_REGEXP_OTHER_CONTROL:
3120 case XML_REGEXP_OTHER_FORMAT:
3121 case XML_REGEXP_OTHER_PRIVATE:
3122 case XML_REGEXP_OTHER_NA:
3123 case XML_REGEXP_BLOCK_NAME:
3124 ret = xmlRegCheckCharacterRange(atom->type, codepoint, 0, 0, 0,
3125 (const xmlChar *)atom->valuep);
3126 if (atom->neg)
3127 ret = !ret;
3128 break;
3129 }
3130 return(ret);
3131 }
3132
3133 /************************************************************************
3134 * *
3135 * Saving and restoring state of an execution context *
3136 * *
3137 ************************************************************************/
3138
3139 #ifdef DEBUG_REGEXP_EXEC
3140 static void
xmlFARegDebugExec(xmlRegExecCtxtPtr exec)3141 xmlFARegDebugExec(xmlRegExecCtxtPtr exec) {
3142 printf("state: %d:%d:idx %d", exec->state->no, exec->transno, exec->index);
3143 if (exec->inputStack != NULL) {
3144 int i;
3145 printf(": ");
3146 for (i = 0;(i < 3) && (i < exec->inputStackNr);i++)
3147 printf("%s ", (const char *)
3148 exec->inputStack[exec->inputStackNr - (i + 1)].value);
3149 } else {
3150 printf(": %s", &(exec->inputString[exec->index]));
3151 }
3152 printf("\n");
3153 }
3154 #endif
3155
3156 static void
xmlFARegExecSave(xmlRegExecCtxtPtr exec)3157 xmlFARegExecSave(xmlRegExecCtxtPtr exec) {
3158 #ifdef DEBUG_REGEXP_EXEC
3159 printf("saving ");
3160 exec->transno++;
3161 xmlFARegDebugExec(exec);
3162 exec->transno--;
3163 #endif
3164 #ifdef MAX_PUSH
3165 if (exec->nbPush > MAX_PUSH) {
3166 return;
3167 }
3168 exec->nbPush++;
3169 #endif
3170
3171 if (exec->maxRollbacks == 0) {
3172 exec->maxRollbacks = 4;
3173 exec->rollbacks = (xmlRegExecRollback *) xmlMalloc(exec->maxRollbacks *
3174 sizeof(xmlRegExecRollback));
3175 if (exec->rollbacks == NULL) {
3176 xmlRegexpErrMemory(NULL, "saving regexp");
3177 exec->maxRollbacks = 0;
3178 return;
3179 }
3180 memset(exec->rollbacks, 0,
3181 exec->maxRollbacks * sizeof(xmlRegExecRollback));
3182 } else if (exec->nbRollbacks >= exec->maxRollbacks) {
3183 xmlRegExecRollback *tmp;
3184 int len = exec->maxRollbacks;
3185
3186 exec->maxRollbacks *= 2;
3187 tmp = (xmlRegExecRollback *) xmlRealloc(exec->rollbacks,
3188 exec->maxRollbacks * sizeof(xmlRegExecRollback));
3189 if (tmp == NULL) {
3190 xmlRegexpErrMemory(NULL, "saving regexp");
3191 exec->maxRollbacks /= 2;
3192 return;
3193 }
3194 exec->rollbacks = tmp;
3195 tmp = &exec->rollbacks[len];
3196 memset(tmp, 0, (exec->maxRollbacks - len) * sizeof(xmlRegExecRollback));
3197 }
3198 exec->rollbacks[exec->nbRollbacks].state = exec->state;
3199 exec->rollbacks[exec->nbRollbacks].index = exec->index;
3200 exec->rollbacks[exec->nbRollbacks].nextbranch = exec->transno + 1;
3201 if (exec->comp->nbCounters > 0) {
3202 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3203 exec->rollbacks[exec->nbRollbacks].counts = (int *)
3204 xmlMalloc(exec->comp->nbCounters * sizeof(int));
3205 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3206 xmlRegexpErrMemory(NULL, "saving regexp");
3207 exec->status = -5;
3208 return;
3209 }
3210 }
3211 memcpy(exec->rollbacks[exec->nbRollbacks].counts, exec->counts,
3212 exec->comp->nbCounters * sizeof(int));
3213 }
3214 exec->nbRollbacks++;
3215 }
3216
3217 static void
xmlFARegExecRollBack(xmlRegExecCtxtPtr exec)3218 xmlFARegExecRollBack(xmlRegExecCtxtPtr exec) {
3219 if (exec->nbRollbacks <= 0) {
3220 exec->status = -1;
3221 #ifdef DEBUG_REGEXP_EXEC
3222 printf("rollback failed on empty stack\n");
3223 #endif
3224 return;
3225 }
3226 exec->nbRollbacks--;
3227 exec->state = exec->rollbacks[exec->nbRollbacks].state;
3228 exec->index = exec->rollbacks[exec->nbRollbacks].index;
3229 exec->transno = exec->rollbacks[exec->nbRollbacks].nextbranch;
3230 if (exec->comp->nbCounters > 0) {
3231 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3232 fprintf(stderr, "exec save: allocation failed");
3233 exec->status = -6;
3234 return;
3235 }
3236 if (exec->counts) {
3237 memcpy(exec->counts, exec->rollbacks[exec->nbRollbacks].counts,
3238 exec->comp->nbCounters * sizeof(int));
3239 }
3240 }
3241
3242 #ifdef DEBUG_REGEXP_EXEC
3243 printf("restored ");
3244 xmlFARegDebugExec(exec);
3245 #endif
3246 }
3247
3248 /************************************************************************
3249 * *
3250 * Verifier, running an input against a compiled regexp *
3251 * *
3252 ************************************************************************/
3253
3254 static int
xmlFARegExec(xmlRegexpPtr comp,const xmlChar * content)3255 xmlFARegExec(xmlRegexpPtr comp, const xmlChar *content) {
3256 xmlRegExecCtxt execval;
3257 xmlRegExecCtxtPtr exec = &execval;
3258 int ret, codepoint = 0, len, deter;
3259
3260 exec->inputString = content;
3261 exec->index = 0;
3262 exec->nbPush = 0;
3263 exec->determinist = 1;
3264 exec->maxRollbacks = 0;
3265 exec->nbRollbacks = 0;
3266 exec->rollbacks = NULL;
3267 exec->status = 0;
3268 exec->comp = comp;
3269 exec->state = comp->states[0];
3270 exec->transno = 0;
3271 exec->transcount = 0;
3272 exec->inputStack = NULL;
3273 exec->inputStackMax = 0;
3274 if (comp->nbCounters > 0) {
3275 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int));
3276 if (exec->counts == NULL) {
3277 xmlRegexpErrMemory(NULL, "running regexp");
3278 return(-1);
3279 }
3280 memset(exec->counts, 0, comp->nbCounters * sizeof(int));
3281 } else
3282 exec->counts = NULL;
3283 while ((exec->status == 0) && (exec->state != NULL) &&
3284 ((exec->inputString[exec->index] != 0) ||
3285 ((exec->state != NULL) &&
3286 (exec->state->type != XML_REGEXP_FINAL_STATE)))) {
3287 xmlRegTransPtr trans;
3288 xmlRegAtomPtr atom;
3289
3290 /*
3291 * If end of input on non-terminal state, rollback, however we may
3292 * still have epsilon like transition for counted transitions
3293 * on counters, in that case don't break too early. Additionally,
3294 * if we are working on a range like "AB{0,2}", where B is not present,
3295 * we don't want to break.
3296 */
3297 len = 1;
3298 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL)) {
3299 /*
3300 * if there is a transition, we must check if
3301 * atom allows minOccurs of 0
3302 */
3303 if (exec->transno < exec->state->nbTrans) {
3304 trans = &exec->state->trans[exec->transno];
3305 if (trans->to >=0) {
3306 atom = trans->atom;
3307 if (!((atom->min == 0) && (atom->max > 0)))
3308 goto rollback;
3309 }
3310 } else
3311 goto rollback;
3312 }
3313
3314 exec->transcount = 0;
3315 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
3316 trans = &exec->state->trans[exec->transno];
3317 if (trans->to < 0)
3318 continue;
3319 atom = trans->atom;
3320 ret = 0;
3321 deter = 1;
3322 if (trans->count >= 0) {
3323 int count;
3324 xmlRegCounterPtr counter;
3325
3326 if (exec->counts == NULL) {
3327 exec->status = -1;
3328 goto error;
3329 }
3330 /*
3331 * A counted transition.
3332 */
3333
3334 count = exec->counts[trans->count];
3335 counter = &exec->comp->counters[trans->count];
3336 #ifdef DEBUG_REGEXP_EXEC
3337 printf("testing count %d: val %d, min %d, max %d\n",
3338 trans->count, count, counter->min, counter->max);
3339 #endif
3340 ret = ((count >= counter->min) && (count <= counter->max));
3341 if ((ret) && (counter->min != counter->max))
3342 deter = 0;
3343 } else if (atom == NULL) {
3344 fprintf(stderr, "epsilon transition left at runtime\n");
3345 exec->status = -2;
3346 break;
3347 } else if (exec->inputString[exec->index] != 0) {
3348 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
3349 ret = xmlRegCheckCharacter(atom, codepoint);
3350 if ((ret == 1) && (atom->min >= 0) && (atom->max > 0)) {
3351 xmlRegStatePtr to = comp->states[trans->to];
3352
3353 /*
3354 * this is a multiple input sequence
3355 * If there is a counter associated increment it now.
3356 * do not increment if the counter is already over the
3357 * maximum limit in which case get to next transition
3358 */
3359 if (trans->counter >= 0) {
3360 xmlRegCounterPtr counter;
3361
3362 if ((exec->counts == NULL) ||
3363 (exec->comp == NULL) ||
3364 (exec->comp->counters == NULL)) {
3365 exec->status = -1;
3366 goto error;
3367 }
3368 counter = &exec->comp->counters[trans->counter];
3369 if (exec->counts[trans->counter] >= counter->max)
3370 continue; /* for loop on transitions */
3371 }
3372 /* Save before incrementing */
3373 if (exec->state->nbTrans > exec->transno + 1) {
3374 xmlFARegExecSave(exec);
3375 }
3376 if (trans->counter >= 0) {
3377 #ifdef DEBUG_REGEXP_EXEC
3378 printf("Increasing count %d\n", trans->counter);
3379 #endif
3380 exec->counts[trans->counter]++;
3381 }
3382 exec->transcount = 1;
3383 do {
3384 /*
3385 * Try to progress as much as possible on the input
3386 */
3387 if (exec->transcount == atom->max) {
3388 break;
3389 }
3390 exec->index += len;
3391 /*
3392 * End of input: stop here
3393 */
3394 if (exec->inputString[exec->index] == 0) {
3395 exec->index -= len;
3396 break;
3397 }
3398 if (exec->transcount >= atom->min) {
3399 int transno = exec->transno;
3400 xmlRegStatePtr state = exec->state;
3401
3402 /*
3403 * The transition is acceptable save it
3404 */
3405 exec->transno = -1; /* trick */
3406 exec->state = to;
3407 xmlFARegExecSave(exec);
3408 exec->transno = transno;
3409 exec->state = state;
3410 }
3411 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
3412 len);
3413 ret = xmlRegCheckCharacter(atom, codepoint);
3414 exec->transcount++;
3415 } while (ret == 1);
3416 if (exec->transcount < atom->min)
3417 ret = 0;
3418
3419 /*
3420 * If the last check failed but one transition was found
3421 * possible, rollback
3422 */
3423 if (ret < 0)
3424 ret = 0;
3425 if (ret == 0) {
3426 goto rollback;
3427 }
3428 if (trans->counter >= 0) {
3429 if (exec->counts == NULL) {
3430 exec->status = -1;
3431 goto error;
3432 }
3433 #ifdef DEBUG_REGEXP_EXEC
3434 printf("Decreasing count %d\n", trans->counter);
3435 #endif
3436 exec->counts[trans->counter]--;
3437 }
3438 } else if ((ret == 0) && (atom->min == 0) && (atom->max > 0)) {
3439 /*
3440 * we don't match on the codepoint, but minOccurs of 0
3441 * says that's ok. Setting len to 0 inhibits stepping
3442 * over the codepoint.
3443 */
3444 exec->transcount = 1;
3445 len = 0;
3446 ret = 1;
3447 }
3448 } else if ((atom->min == 0) && (atom->max > 0)) {
3449 /* another spot to match when minOccurs is 0 */
3450 exec->transcount = 1;
3451 len = 0;
3452 ret = 1;
3453 }
3454 if (ret == 1) {
3455 if ((trans->nd == 1) ||
3456 ((trans->count >= 0) && (deter == 0) &&
3457 (exec->state->nbTrans > exec->transno + 1))) {
3458 #ifdef DEBUG_REGEXP_EXEC
3459 if (trans->nd == 1)
3460 printf("Saving on nd transition atom %d for %c at %d\n",
3461 trans->atom->no, codepoint, exec->index);
3462 else
3463 printf("Saving on counted transition count %d for %c at %d\n",
3464 trans->count, codepoint, exec->index);
3465 #endif
3466 xmlFARegExecSave(exec);
3467 }
3468 if (trans->counter >= 0) {
3469 xmlRegCounterPtr counter;
3470
3471 /* make sure we don't go over the counter maximum value */
3472 if ((exec->counts == NULL) ||
3473 (exec->comp == NULL) ||
3474 (exec->comp->counters == NULL)) {
3475 exec->status = -1;
3476 goto error;
3477 }
3478 counter = &exec->comp->counters[trans->counter];
3479 if (exec->counts[trans->counter] >= counter->max)
3480 continue; /* for loop on transitions */
3481 #ifdef DEBUG_REGEXP_EXEC
3482 printf("Increasing count %d\n", trans->counter);
3483 #endif
3484 exec->counts[trans->counter]++;
3485 }
3486 if ((trans->count >= 0) &&
3487 (trans->count < REGEXP_ALL_COUNTER)) {
3488 if (exec->counts == NULL) {
3489 exec->status = -1;
3490 goto error;
3491 }
3492 #ifdef DEBUG_REGEXP_EXEC
3493 printf("resetting count %d on transition\n",
3494 trans->count);
3495 #endif
3496 exec->counts[trans->count] = 0;
3497 }
3498 #ifdef DEBUG_REGEXP_EXEC
3499 printf("entering state %d\n", trans->to);
3500 #endif
3501 exec->state = comp->states[trans->to];
3502 exec->transno = 0;
3503 if (trans->atom != NULL) {
3504 exec->index += len;
3505 }
3506 goto progress;
3507 } else if (ret < 0) {
3508 exec->status = -4;
3509 break;
3510 }
3511 }
3512 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
3513 rollback:
3514 /*
3515 * Failed to find a way out
3516 */
3517 exec->determinist = 0;
3518 #ifdef DEBUG_REGEXP_EXEC
3519 printf("rollback from state %d on %d:%c\n", exec->state->no,
3520 codepoint,codepoint);
3521 #endif
3522 xmlFARegExecRollBack(exec);
3523 }
3524 progress:
3525 continue;
3526 }
3527 error:
3528 if (exec->rollbacks != NULL) {
3529 if (exec->counts != NULL) {
3530 int i;
3531
3532 for (i = 0;i < exec->maxRollbacks;i++)
3533 if (exec->rollbacks[i].counts != NULL)
3534 xmlFree(exec->rollbacks[i].counts);
3535 }
3536 xmlFree(exec->rollbacks);
3537 }
3538 if (exec->state == NULL)
3539 return(-1);
3540 if (exec->counts != NULL)
3541 xmlFree(exec->counts);
3542 if (exec->status == 0)
3543 return(1);
3544 if (exec->status == -1) {
3545 if (exec->nbPush > MAX_PUSH)
3546 return(-1);
3547 return(0);
3548 }
3549 return(exec->status);
3550 }
3551
3552 /************************************************************************
3553 * *
3554 * Progressive interface to the verifier one atom at a time *
3555 * *
3556 ************************************************************************/
3557 #ifdef DEBUG_ERR
3558 static void testerr(xmlRegExecCtxtPtr exec);
3559 #endif
3560
3561 /**
3562 * xmlRegNewExecCtxt:
3563 * @comp: a precompiled regular expression
3564 * @callback: a callback function used for handling progresses in the
3565 * automata matching phase
3566 * @data: the context data associated to the callback in this context
3567 *
3568 * Build a context used for progressive evaluation of a regexp.
3569 *
3570 * Returns the new context
3571 */
3572 xmlRegExecCtxtPtr
xmlRegNewExecCtxt(xmlRegexpPtr comp,xmlRegExecCallbacks callback,void * data)3573 xmlRegNewExecCtxt(xmlRegexpPtr comp, xmlRegExecCallbacks callback, void *data) {
3574 xmlRegExecCtxtPtr exec;
3575
3576 if (comp == NULL)
3577 return(NULL);
3578 if ((comp->compact == NULL) && (comp->states == NULL))
3579 return(NULL);
3580 exec = (xmlRegExecCtxtPtr) xmlMalloc(sizeof(xmlRegExecCtxt));
3581 if (exec == NULL) {
3582 xmlRegexpErrMemory(NULL, "creating execution context");
3583 return(NULL);
3584 }
3585 memset(exec, 0, sizeof(xmlRegExecCtxt));
3586 exec->inputString = NULL;
3587 exec->index = 0;
3588 exec->determinist = 1;
3589 exec->maxRollbacks = 0;
3590 exec->nbRollbacks = 0;
3591 exec->rollbacks = NULL;
3592 exec->status = 0;
3593 exec->comp = comp;
3594 if (comp->compact == NULL)
3595 exec->state = comp->states[0];
3596 exec->transno = 0;
3597 exec->transcount = 0;
3598 exec->callback = callback;
3599 exec->data = data;
3600 if (comp->nbCounters > 0) {
3601 /*
3602 * For error handling, exec->counts is allocated twice the size
3603 * the second half is used to store the data in case of rollback
3604 */
3605 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int)
3606 * 2);
3607 if (exec->counts == NULL) {
3608 xmlRegexpErrMemory(NULL, "creating execution context");
3609 xmlFree(exec);
3610 return(NULL);
3611 }
3612 memset(exec->counts, 0, comp->nbCounters * sizeof(int) * 2);
3613 exec->errCounts = &exec->counts[comp->nbCounters];
3614 } else {
3615 exec->counts = NULL;
3616 exec->errCounts = NULL;
3617 }
3618 exec->inputStackMax = 0;
3619 exec->inputStackNr = 0;
3620 exec->inputStack = NULL;
3621 exec->errStateNo = -1;
3622 exec->errString = NULL;
3623 exec->nbPush = 0;
3624 return(exec);
3625 }
3626
3627 /**
3628 * xmlRegFreeExecCtxt:
3629 * @exec: a regular expression evaluation context
3630 *
3631 * Free the structures associated to a regular expression evaluation context.
3632 */
3633 void
xmlRegFreeExecCtxt(xmlRegExecCtxtPtr exec)3634 xmlRegFreeExecCtxt(xmlRegExecCtxtPtr exec) {
3635 if (exec == NULL)
3636 return;
3637
3638 if (exec->rollbacks != NULL) {
3639 if (exec->counts != NULL) {
3640 int i;
3641
3642 for (i = 0;i < exec->maxRollbacks;i++)
3643 if (exec->rollbacks[i].counts != NULL)
3644 xmlFree(exec->rollbacks[i].counts);
3645 }
3646 xmlFree(exec->rollbacks);
3647 }
3648 if (exec->counts != NULL)
3649 xmlFree(exec->counts);
3650 if (exec->inputStack != NULL) {
3651 int i;
3652
3653 for (i = 0;i < exec->inputStackNr;i++) {
3654 if (exec->inputStack[i].value != NULL)
3655 xmlFree(exec->inputStack[i].value);
3656 }
3657 xmlFree(exec->inputStack);
3658 }
3659 if (exec->errString != NULL)
3660 xmlFree(exec->errString);
3661 xmlFree(exec);
3662 }
3663
3664 static void
xmlFARegExecSaveInputString(xmlRegExecCtxtPtr exec,const xmlChar * value,void * data)3665 xmlFARegExecSaveInputString(xmlRegExecCtxtPtr exec, const xmlChar *value,
3666 void *data) {
3667 #ifdef DEBUG_PUSH
3668 printf("saving value: %d:%s\n", exec->inputStackNr, value);
3669 #endif
3670 if (exec->inputStackMax == 0) {
3671 exec->inputStackMax = 4;
3672 exec->inputStack = (xmlRegInputTokenPtr)
3673 xmlMalloc(exec->inputStackMax * sizeof(xmlRegInputToken));
3674 if (exec->inputStack == NULL) {
3675 xmlRegexpErrMemory(NULL, "pushing input string");
3676 exec->inputStackMax = 0;
3677 return;
3678 }
3679 } else if (exec->inputStackNr + 1 >= exec->inputStackMax) {
3680 xmlRegInputTokenPtr tmp;
3681
3682 exec->inputStackMax *= 2;
3683 tmp = (xmlRegInputTokenPtr) xmlRealloc(exec->inputStack,
3684 exec->inputStackMax * sizeof(xmlRegInputToken));
3685 if (tmp == NULL) {
3686 xmlRegexpErrMemory(NULL, "pushing input string");
3687 exec->inputStackMax /= 2;
3688 return;
3689 }
3690 exec->inputStack = tmp;
3691 }
3692 exec->inputStack[exec->inputStackNr].value = xmlStrdup(value);
3693 exec->inputStack[exec->inputStackNr].data = data;
3694 exec->inputStackNr++;
3695 exec->inputStack[exec->inputStackNr].value = NULL;
3696 exec->inputStack[exec->inputStackNr].data = NULL;
3697 }
3698
3699 /**
3700 * xmlRegStrEqualWildcard:
3701 * @expStr: the string to be evaluated
3702 * @valStr: the validation string
3703 *
3704 * Checks if both strings are equal or have the same content. "*"
3705 * can be used as a wildcard in @valStr; "|" is used as a separator of
3706 * substrings in both @expStr and @valStr.
3707 *
3708 * Returns 1 if the comparison is satisfied and the number of substrings
3709 * is equal, 0 otherwise.
3710 */
3711
3712 static int
xmlRegStrEqualWildcard(const xmlChar * expStr,const xmlChar * valStr)3713 xmlRegStrEqualWildcard(const xmlChar *expStr, const xmlChar *valStr) {
3714 if (expStr == valStr) return(1);
3715 if (expStr == NULL) return(0);
3716 if (valStr == NULL) return(0);
3717 do {
3718 /*
3719 * Eval if we have a wildcard for the current item.
3720 */
3721 if (*expStr != *valStr) {
3722 /* if one of them starts with a wildcard make valStr be it */
3723 if (*valStr == '*') {
3724 const xmlChar *tmp;
3725
3726 tmp = valStr;
3727 valStr = expStr;
3728 expStr = tmp;
3729 }
3730 if ((*valStr != 0) && (*expStr != 0) && (*expStr++ == '*')) {
3731 do {
3732 if (*valStr == XML_REG_STRING_SEPARATOR)
3733 break;
3734 valStr++;
3735 } while (*valStr != 0);
3736 continue;
3737 } else
3738 return(0);
3739 }
3740 expStr++;
3741 valStr++;
3742 } while (*valStr != 0);
3743 if (*expStr != 0)
3744 return (0);
3745 else
3746 return (1);
3747 }
3748
3749 /**
3750 * xmlRegCompactPushString:
3751 * @exec: a regexp execution context
3752 * @comp: the precompiled exec with a compact table
3753 * @value: a string token input
3754 * @data: data associated to the token to reuse in callbacks
3755 *
3756 * Push one input token in the execution context
3757 *
3758 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
3759 * a negative value in case of error.
3760 */
3761 static int
xmlRegCompactPushString(xmlRegExecCtxtPtr exec,xmlRegexpPtr comp,const xmlChar * value,void * data)3762 xmlRegCompactPushString(xmlRegExecCtxtPtr exec,
3763 xmlRegexpPtr comp,
3764 const xmlChar *value,
3765 void *data) {
3766 int state = exec->index;
3767 int i, target;
3768
3769 if ((comp == NULL) || (comp->compact == NULL) || (comp->stringMap == NULL))
3770 return(-1);
3771
3772 if (value == NULL) {
3773 /*
3774 * are we at a final state ?
3775 */
3776 if (comp->compact[state * (comp->nbstrings + 1)] ==
3777 XML_REGEXP_FINAL_STATE)
3778 return(1);
3779 return(0);
3780 }
3781
3782 #ifdef DEBUG_PUSH
3783 printf("value pushed: %s\n", value);
3784 #endif
3785
3786 /*
3787 * Examine all outside transitions from current state
3788 */
3789 for (i = 0;i < comp->nbstrings;i++) {
3790 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
3791 if ((target > 0) && (target <= comp->nbstates)) {
3792 target--; /* to avoid 0 */
3793 if (xmlRegStrEqualWildcard(comp->stringMap[i], value)) {
3794 exec->index = target;
3795 if ((exec->callback != NULL) && (comp->transdata != NULL)) {
3796 exec->callback(exec->data, value,
3797 comp->transdata[state * comp->nbstrings + i], data);
3798 }
3799 #ifdef DEBUG_PUSH
3800 printf("entering state %d\n", target);
3801 #endif
3802 if (comp->compact[target * (comp->nbstrings + 1)] ==
3803 XML_REGEXP_SINK_STATE)
3804 goto error;
3805
3806 if (comp->compact[target * (comp->nbstrings + 1)] ==
3807 XML_REGEXP_FINAL_STATE)
3808 return(1);
3809 return(0);
3810 }
3811 }
3812 }
3813 /*
3814 * Failed to find an exit transition out from current state for the
3815 * current token
3816 */
3817 #ifdef DEBUG_PUSH
3818 printf("failed to find a transition for %s on state %d\n", value, state);
3819 #endif
3820 error:
3821 if (exec->errString != NULL)
3822 xmlFree(exec->errString);
3823 exec->errString = xmlStrdup(value);
3824 exec->errStateNo = state;
3825 exec->status = -1;
3826 #ifdef DEBUG_ERR
3827 testerr(exec);
3828 #endif
3829 return(-1);
3830 }
3831
3832 /**
3833 * xmlRegExecPushStringInternal:
3834 * @exec: a regexp execution context or NULL to indicate the end
3835 * @value: a string token input
3836 * @data: data associated to the token to reuse in callbacks
3837 * @compound: value was assembled from 2 strings
3838 *
3839 * Push one input token in the execution context
3840 *
3841 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
3842 * a negative value in case of error.
3843 */
3844 static int
xmlRegExecPushStringInternal(xmlRegExecCtxtPtr exec,const xmlChar * value,void * data,int compound)3845 xmlRegExecPushStringInternal(xmlRegExecCtxtPtr exec, const xmlChar *value,
3846 void *data, int compound) {
3847 xmlRegTransPtr trans;
3848 xmlRegAtomPtr atom;
3849 int ret;
3850 int final = 0;
3851 int progress = 1;
3852
3853 if (exec == NULL)
3854 return(-1);
3855 if (exec->comp == NULL)
3856 return(-1);
3857 if (exec->status != 0)
3858 return(exec->status);
3859
3860 if (exec->comp->compact != NULL)
3861 return(xmlRegCompactPushString(exec, exec->comp, value, data));
3862
3863 if (value == NULL) {
3864 if (exec->state->type == XML_REGEXP_FINAL_STATE)
3865 return(1);
3866 final = 1;
3867 }
3868
3869 #ifdef DEBUG_PUSH
3870 printf("value pushed: %s\n", value);
3871 #endif
3872 /*
3873 * If we have an active rollback stack push the new value there
3874 * and get back to where we were left
3875 */
3876 if ((value != NULL) && (exec->inputStackNr > 0)) {
3877 xmlFARegExecSaveInputString(exec, value, data);
3878 value = exec->inputStack[exec->index].value;
3879 data = exec->inputStack[exec->index].data;
3880 #ifdef DEBUG_PUSH
3881 printf("value loaded: %s\n", value);
3882 #endif
3883 }
3884
3885 while ((exec->status == 0) &&
3886 ((value != NULL) ||
3887 ((final == 1) &&
3888 (exec->state->type != XML_REGEXP_FINAL_STATE)))) {
3889
3890 /*
3891 * End of input on non-terminal state, rollback, however we may
3892 * still have epsilon like transition for counted transitions
3893 * on counters, in that case don't break too early.
3894 */
3895 if ((value == NULL) && (exec->counts == NULL))
3896 goto rollback;
3897
3898 exec->transcount = 0;
3899 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
3900 trans = &exec->state->trans[exec->transno];
3901 if (trans->to < 0)
3902 continue;
3903 atom = trans->atom;
3904 ret = 0;
3905 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
3906 int i;
3907 int count;
3908 xmlRegTransPtr t;
3909 xmlRegCounterPtr counter;
3910
3911 ret = 0;
3912
3913 #ifdef DEBUG_PUSH
3914 printf("testing all lax %d\n", trans->count);
3915 #endif
3916 /*
3917 * Check all counted transitions from the current state
3918 */
3919 if ((value == NULL) && (final)) {
3920 ret = 1;
3921 } else if (value != NULL) {
3922 for (i = 0;i < exec->state->nbTrans;i++) {
3923 t = &exec->state->trans[i];
3924 if ((t->counter < 0) || (t == trans))
3925 continue;
3926 counter = &exec->comp->counters[t->counter];
3927 count = exec->counts[t->counter];
3928 if ((count < counter->max) &&
3929 (t->atom != NULL) &&
3930 (xmlStrEqual(value, t->atom->valuep))) {
3931 ret = 0;
3932 break;
3933 }
3934 if ((count >= counter->min) &&
3935 (count < counter->max) &&
3936 (t->atom != NULL) &&
3937 (xmlStrEqual(value, t->atom->valuep))) {
3938 ret = 1;
3939 break;
3940 }
3941 }
3942 }
3943 } else if (trans->count == REGEXP_ALL_COUNTER) {
3944 int i;
3945 int count;
3946 xmlRegTransPtr t;
3947 xmlRegCounterPtr counter;
3948
3949 ret = 1;
3950
3951 #ifdef DEBUG_PUSH
3952 printf("testing all %d\n", trans->count);
3953 #endif
3954 /*
3955 * Check all counted transitions from the current state
3956 */
3957 for (i = 0;i < exec->state->nbTrans;i++) {
3958 t = &exec->state->trans[i];
3959 if ((t->counter < 0) || (t == trans))
3960 continue;
3961 counter = &exec->comp->counters[t->counter];
3962 count = exec->counts[t->counter];
3963 if ((count < counter->min) || (count > counter->max)) {
3964 ret = 0;
3965 break;
3966 }
3967 }
3968 } else if (trans->count >= 0) {
3969 int count;
3970 xmlRegCounterPtr counter;
3971
3972 /*
3973 * A counted transition.
3974 */
3975
3976 count = exec->counts[trans->count];
3977 counter = &exec->comp->counters[trans->count];
3978 #ifdef DEBUG_PUSH
3979 printf("testing count %d: val %d, min %d, max %d\n",
3980 trans->count, count, counter->min, counter->max);
3981 #endif
3982 ret = ((count >= counter->min) && (count <= counter->max));
3983 } else if (atom == NULL) {
3984 fprintf(stderr, "epsilon transition left at runtime\n");
3985 exec->status = -2;
3986 break;
3987 } else if (value != NULL) {
3988 ret = xmlRegStrEqualWildcard(atom->valuep, value);
3989 if (atom->neg) {
3990 ret = !ret;
3991 if (!compound)
3992 ret = 0;
3993 }
3994 if ((ret == 1) && (trans->counter >= 0)) {
3995 xmlRegCounterPtr counter;
3996 int count;
3997
3998 count = exec->counts[trans->counter];
3999 counter = &exec->comp->counters[trans->counter];
4000 if (count >= counter->max)
4001 ret = 0;
4002 }
4003
4004 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
4005 xmlRegStatePtr to = exec->comp->states[trans->to];
4006
4007 /*
4008 * this is a multiple input sequence
4009 */
4010 if (exec->state->nbTrans > exec->transno + 1) {
4011 if (exec->inputStackNr <= 0) {
4012 xmlFARegExecSaveInputString(exec, value, data);
4013 }
4014 xmlFARegExecSave(exec);
4015 }
4016 exec->transcount = 1;
4017 do {
4018 /*
4019 * Try to progress as much as possible on the input
4020 */
4021 if (exec->transcount == atom->max) {
4022 break;
4023 }
4024 exec->index++;
4025 value = exec->inputStack[exec->index].value;
4026 data = exec->inputStack[exec->index].data;
4027 #ifdef DEBUG_PUSH
4028 printf("value loaded: %s\n", value);
4029 #endif
4030
4031 /*
4032 * End of input: stop here
4033 */
4034 if (value == NULL) {
4035 exec->index --;
4036 break;
4037 }
4038 if (exec->transcount >= atom->min) {
4039 int transno = exec->transno;
4040 xmlRegStatePtr state = exec->state;
4041
4042 /*
4043 * The transition is acceptable save it
4044 */
4045 exec->transno = -1; /* trick */
4046 exec->state = to;
4047 if (exec->inputStackNr <= 0) {
4048 xmlFARegExecSaveInputString(exec, value, data);
4049 }
4050 xmlFARegExecSave(exec);
4051 exec->transno = transno;
4052 exec->state = state;
4053 }
4054 ret = xmlStrEqual(value, atom->valuep);
4055 exec->transcount++;
4056 } while (ret == 1);
4057 if (exec->transcount < atom->min)
4058 ret = 0;
4059
4060 /*
4061 * If the last check failed but one transition was found
4062 * possible, rollback
4063 */
4064 if (ret < 0)
4065 ret = 0;
4066 if (ret == 0) {
4067 goto rollback;
4068 }
4069 }
4070 }
4071 if (ret == 1) {
4072 if ((exec->callback != NULL) && (atom != NULL) &&
4073 (data != NULL)) {
4074 exec->callback(exec->data, atom->valuep,
4075 atom->data, data);
4076 }
4077 if (exec->state->nbTrans > exec->transno + 1) {
4078 if (exec->inputStackNr <= 0) {
4079 xmlFARegExecSaveInputString(exec, value, data);
4080 }
4081 xmlFARegExecSave(exec);
4082 }
4083 if (trans->counter >= 0) {
4084 #ifdef DEBUG_PUSH
4085 printf("Increasing count %d\n", trans->counter);
4086 #endif
4087 exec->counts[trans->counter]++;
4088 }
4089 if ((trans->count >= 0) &&
4090 (trans->count < REGEXP_ALL_COUNTER)) {
4091 #ifdef DEBUG_REGEXP_EXEC
4092 printf("resetting count %d on transition\n",
4093 trans->count);
4094 #endif
4095 exec->counts[trans->count] = 0;
4096 }
4097 #ifdef DEBUG_PUSH
4098 printf("entering state %d\n", trans->to);
4099 #endif
4100 if ((exec->comp->states[trans->to] != NULL) &&
4101 (exec->comp->states[trans->to]->type ==
4102 XML_REGEXP_SINK_STATE)) {
4103 /*
4104 * entering a sink state, save the current state as error
4105 * state.
4106 */
4107 if (exec->errString != NULL)
4108 xmlFree(exec->errString);
4109 exec->errString = xmlStrdup(value);
4110 exec->errState = exec->state;
4111 memcpy(exec->errCounts, exec->counts,
4112 exec->comp->nbCounters * sizeof(int));
4113 }
4114 exec->state = exec->comp->states[trans->to];
4115 exec->transno = 0;
4116 if (trans->atom != NULL) {
4117 if (exec->inputStack != NULL) {
4118 exec->index++;
4119 if (exec->index < exec->inputStackNr) {
4120 value = exec->inputStack[exec->index].value;
4121 data = exec->inputStack[exec->index].data;
4122 #ifdef DEBUG_PUSH
4123 printf("value loaded: %s\n", value);
4124 #endif
4125 } else {
4126 value = NULL;
4127 data = NULL;
4128 #ifdef DEBUG_PUSH
4129 printf("end of input\n");
4130 #endif
4131 }
4132 } else {
4133 value = NULL;
4134 data = NULL;
4135 #ifdef DEBUG_PUSH
4136 printf("end of input\n");
4137 #endif
4138 }
4139 }
4140 goto progress;
4141 } else if (ret < 0) {
4142 exec->status = -4;
4143 break;
4144 }
4145 }
4146 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
4147 rollback:
4148 /*
4149 * if we didn't yet rollback on the current input
4150 * store the current state as the error state.
4151 */
4152 if ((progress) && (exec->state != NULL) &&
4153 (exec->state->type != XML_REGEXP_SINK_STATE)) {
4154 progress = 0;
4155 if (exec->errString != NULL)
4156 xmlFree(exec->errString);
4157 exec->errString = xmlStrdup(value);
4158 exec->errState = exec->state;
4159 if (exec->comp->nbCounters)
4160 memcpy(exec->errCounts, exec->counts,
4161 exec->comp->nbCounters * sizeof(int));
4162 }
4163
4164 /*
4165 * Failed to find a way out
4166 */
4167 exec->determinist = 0;
4168 xmlFARegExecRollBack(exec);
4169 if ((exec->inputStack != NULL ) && (exec->status == 0)) {
4170 value = exec->inputStack[exec->index].value;
4171 data = exec->inputStack[exec->index].data;
4172 #ifdef DEBUG_PUSH
4173 printf("value loaded: %s\n", value);
4174 #endif
4175 }
4176 }
4177 continue;
4178 progress:
4179 progress = 1;
4180 continue;
4181 }
4182 if (exec->status == 0) {
4183 return(exec->state->type == XML_REGEXP_FINAL_STATE);
4184 }
4185 #ifdef DEBUG_ERR
4186 if (exec->status < 0) {
4187 testerr(exec);
4188 }
4189 #endif
4190 return(exec->status);
4191 }
4192
4193 /**
4194 * xmlRegExecPushString:
4195 * @exec: a regexp execution context or NULL to indicate the end
4196 * @value: a string token input
4197 * @data: data associated to the token to reuse in callbacks
4198 *
4199 * Push one input token in the execution context
4200 *
4201 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
4202 * a negative value in case of error.
4203 */
4204 int
xmlRegExecPushString(xmlRegExecCtxtPtr exec,const xmlChar * value,void * data)4205 xmlRegExecPushString(xmlRegExecCtxtPtr exec, const xmlChar *value,
4206 void *data) {
4207 return(xmlRegExecPushStringInternal(exec, value, data, 0));
4208 }
4209
4210 /**
4211 * xmlRegExecPushString2:
4212 * @exec: a regexp execution context or NULL to indicate the end
4213 * @value: the first string token input
4214 * @value2: the second string token input
4215 * @data: data associated to the token to reuse in callbacks
4216 *
4217 * Push one input token in the execution context
4218 *
4219 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
4220 * a negative value in case of error.
4221 */
4222 int
xmlRegExecPushString2(xmlRegExecCtxtPtr exec,const xmlChar * value,const xmlChar * value2,void * data)4223 xmlRegExecPushString2(xmlRegExecCtxtPtr exec, const xmlChar *value,
4224 const xmlChar *value2, void *data) {
4225 xmlChar buf[150];
4226 int lenn, lenp, ret;
4227 xmlChar *str;
4228
4229 if (exec == NULL)
4230 return(-1);
4231 if (exec->comp == NULL)
4232 return(-1);
4233 if (exec->status != 0)
4234 return(exec->status);
4235
4236 if (value2 == NULL)
4237 return(xmlRegExecPushString(exec, value, data));
4238
4239 lenn = strlen((char *) value2);
4240 lenp = strlen((char *) value);
4241
4242 if (150 < lenn + lenp + 2) {
4243 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
4244 if (str == NULL) {
4245 exec->status = -1;
4246 return(-1);
4247 }
4248 } else {
4249 str = buf;
4250 }
4251 memcpy(&str[0], value, lenp);
4252 str[lenp] = XML_REG_STRING_SEPARATOR;
4253 memcpy(&str[lenp + 1], value2, lenn);
4254 str[lenn + lenp + 1] = 0;
4255
4256 if (exec->comp->compact != NULL)
4257 ret = xmlRegCompactPushString(exec, exec->comp, str, data);
4258 else
4259 ret = xmlRegExecPushStringInternal(exec, str, data, 1);
4260
4261 if (str != buf)
4262 xmlFree(str);
4263 return(ret);
4264 }
4265
4266 /**
4267 * xmlRegExecGetValues:
4268 * @exec: a regexp execution context
4269 * @err: error extraction or normal one
4270 * @nbval: pointer to the number of accepted values IN/OUT
4271 * @nbneg: return number of negative transitions
4272 * @values: pointer to the array of acceptable values
4273 * @terminal: return value if this was a terminal state
4274 *
4275 * Extract information from the regexp execution, internal routine to
4276 * implement xmlRegExecNextValues() and xmlRegExecErrInfo()
4277 *
4278 * Returns: 0 in case of success or -1 in case of error.
4279 */
4280 static int
xmlRegExecGetValues(xmlRegExecCtxtPtr exec,int err,int * nbval,int * nbneg,xmlChar ** values,int * terminal)4281 xmlRegExecGetValues(xmlRegExecCtxtPtr exec, int err,
4282 int *nbval, int *nbneg,
4283 xmlChar **values, int *terminal) {
4284 int maxval;
4285 int nb = 0;
4286
4287 if ((exec == NULL) || (nbval == NULL) || (nbneg == NULL) ||
4288 (values == NULL) || (*nbval <= 0))
4289 return(-1);
4290
4291 maxval = *nbval;
4292 *nbval = 0;
4293 *nbneg = 0;
4294 if ((exec->comp != NULL) && (exec->comp->compact != NULL)) {
4295 xmlRegexpPtr comp;
4296 int target, i, state;
4297
4298 comp = exec->comp;
4299
4300 if (err) {
4301 if (exec->errStateNo == -1) return(-1);
4302 state = exec->errStateNo;
4303 } else {
4304 state = exec->index;
4305 }
4306 if (terminal != NULL) {
4307 if (comp->compact[state * (comp->nbstrings + 1)] ==
4308 XML_REGEXP_FINAL_STATE)
4309 *terminal = 1;
4310 else
4311 *terminal = 0;
4312 }
4313 for (i = 0;(i < comp->nbstrings) && (nb < maxval);i++) {
4314 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
4315 if ((target > 0) && (target <= comp->nbstates) &&
4316 (comp->compact[(target - 1) * (comp->nbstrings + 1)] !=
4317 XML_REGEXP_SINK_STATE)) {
4318 values[nb++] = comp->stringMap[i];
4319 (*nbval)++;
4320 }
4321 }
4322 for (i = 0;(i < comp->nbstrings) && (nb < maxval);i++) {
4323 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
4324 if ((target > 0) && (target <= comp->nbstates) &&
4325 (comp->compact[(target - 1) * (comp->nbstrings + 1)] ==
4326 XML_REGEXP_SINK_STATE)) {
4327 values[nb++] = comp->stringMap[i];
4328 (*nbneg)++;
4329 }
4330 }
4331 } else {
4332 int transno;
4333 xmlRegTransPtr trans;
4334 xmlRegAtomPtr atom;
4335 xmlRegStatePtr state;
4336
4337 if (terminal != NULL) {
4338 if (exec->state->type == XML_REGEXP_FINAL_STATE)
4339 *terminal = 1;
4340 else
4341 *terminal = 0;
4342 }
4343
4344 if (err) {
4345 if (exec->errState == NULL) return(-1);
4346 state = exec->errState;
4347 } else {
4348 if (exec->state == NULL) return(-1);
4349 state = exec->state;
4350 }
4351 for (transno = 0;
4352 (transno < state->nbTrans) && (nb < maxval);
4353 transno++) {
4354 trans = &state->trans[transno];
4355 if (trans->to < 0)
4356 continue;
4357 atom = trans->atom;
4358 if ((atom == NULL) || (atom->valuep == NULL))
4359 continue;
4360 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
4361 /* this should not be reached but ... */
4362 TODO;
4363 } else if (trans->count == REGEXP_ALL_COUNTER) {
4364 /* this should not be reached but ... */
4365 TODO;
4366 } else if (trans->counter >= 0) {
4367 xmlRegCounterPtr counter = NULL;
4368 int count;
4369
4370 if (err)
4371 count = exec->errCounts[trans->counter];
4372 else
4373 count = exec->counts[trans->counter];
4374 if (exec->comp != NULL)
4375 counter = &exec->comp->counters[trans->counter];
4376 if ((counter == NULL) || (count < counter->max)) {
4377 if (atom->neg)
4378 values[nb++] = (xmlChar *) atom->valuep2;
4379 else
4380 values[nb++] = (xmlChar *) atom->valuep;
4381 (*nbval)++;
4382 }
4383 } else {
4384 if ((exec->comp != NULL) && (exec->comp->states[trans->to] != NULL) &&
4385 (exec->comp->states[trans->to]->type !=
4386 XML_REGEXP_SINK_STATE)) {
4387 if (atom->neg)
4388 values[nb++] = (xmlChar *) atom->valuep2;
4389 else
4390 values[nb++] = (xmlChar *) atom->valuep;
4391 (*nbval)++;
4392 }
4393 }
4394 }
4395 for (transno = 0;
4396 (transno < state->nbTrans) && (nb < maxval);
4397 transno++) {
4398 trans = &state->trans[transno];
4399 if (trans->to < 0)
4400 continue;
4401 atom = trans->atom;
4402 if ((atom == NULL) || (atom->valuep == NULL))
4403 continue;
4404 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
4405 continue;
4406 } else if (trans->count == REGEXP_ALL_COUNTER) {
4407 continue;
4408 } else if (trans->counter >= 0) {
4409 continue;
4410 } else {
4411 if ((exec->comp->states[trans->to] != NULL) &&
4412 (exec->comp->states[trans->to]->type ==
4413 XML_REGEXP_SINK_STATE)) {
4414 if (atom->neg)
4415 values[nb++] = (xmlChar *) atom->valuep2;
4416 else
4417 values[nb++] = (xmlChar *) atom->valuep;
4418 (*nbneg)++;
4419 }
4420 }
4421 }
4422 }
4423 return(0);
4424 }
4425
4426 /**
4427 * xmlRegExecNextValues:
4428 * @exec: a regexp execution context
4429 * @nbval: pointer to the number of accepted values IN/OUT
4430 * @nbneg: return number of negative transitions
4431 * @values: pointer to the array of acceptable values
4432 * @terminal: return value if this was a terminal state
4433 *
4434 * Extract information from the regexp execution,
4435 * the parameter @values must point to an array of @nbval string pointers
4436 * on return nbval will contain the number of possible strings in that
4437 * state and the @values array will be updated with them. The string values
4438 * returned will be freed with the @exec context and don't need to be
4439 * deallocated.
4440 *
4441 * Returns: 0 in case of success or -1 in case of error.
4442 */
4443 int
xmlRegExecNextValues(xmlRegExecCtxtPtr exec,int * nbval,int * nbneg,xmlChar ** values,int * terminal)4444 xmlRegExecNextValues(xmlRegExecCtxtPtr exec, int *nbval, int *nbneg,
4445 xmlChar **values, int *terminal) {
4446 return(xmlRegExecGetValues(exec, 0, nbval, nbneg, values, terminal));
4447 }
4448
4449 /**
4450 * xmlRegExecErrInfo:
4451 * @exec: a regexp execution context generating an error
4452 * @string: return value for the error string
4453 * @nbval: pointer to the number of accepted values IN/OUT
4454 * @nbneg: return number of negative transitions
4455 * @values: pointer to the array of acceptable values
4456 * @terminal: return value if this was a terminal state
4457 *
4458 * Extract error information from the regexp execution, the parameter
4459 * @string will be updated with the value pushed and not accepted,
4460 * the parameter @values must point to an array of @nbval string pointers
4461 * on return nbval will contain the number of possible strings in that
4462 * state and the @values array will be updated with them. The string values
4463 * returned will be freed with the @exec context and don't need to be
4464 * deallocated.
4465 *
4466 * Returns: 0 in case of success or -1 in case of error.
4467 */
4468 int
xmlRegExecErrInfo(xmlRegExecCtxtPtr exec,const xmlChar ** string,int * nbval,int * nbneg,xmlChar ** values,int * terminal)4469 xmlRegExecErrInfo(xmlRegExecCtxtPtr exec, const xmlChar **string,
4470 int *nbval, int *nbneg, xmlChar **values, int *terminal) {
4471 if (exec == NULL)
4472 return(-1);
4473 if (string != NULL) {
4474 if (exec->status != 0)
4475 *string = exec->errString;
4476 else
4477 *string = NULL;
4478 }
4479 return(xmlRegExecGetValues(exec, 1, nbval, nbneg, values, terminal));
4480 }
4481
4482 #ifdef DEBUG_ERR
testerr(xmlRegExecCtxtPtr exec)4483 static void testerr(xmlRegExecCtxtPtr exec) {
4484 const xmlChar *string;
4485 xmlChar *values[5];
4486 int nb = 5;
4487 int nbneg;
4488 int terminal;
4489 xmlRegExecErrInfo(exec, &string, &nb, &nbneg, &values[0], &terminal);
4490 }
4491 #endif
4492
4493 #if 0
4494 static int
4495 xmlRegExecPushChar(xmlRegExecCtxtPtr exec, int UCS) {
4496 xmlRegTransPtr trans;
4497 xmlRegAtomPtr atom;
4498 int ret;
4499 int codepoint, len;
4500
4501 if (exec == NULL)
4502 return(-1);
4503 if (exec->status != 0)
4504 return(exec->status);
4505
4506 while ((exec->status == 0) &&
4507 ((exec->inputString[exec->index] != 0) ||
4508 (exec->state->type != XML_REGEXP_FINAL_STATE))) {
4509
4510 /*
4511 * End of input on non-terminal state, rollback, however we may
4512 * still have epsilon like transition for counted transitions
4513 * on counters, in that case don't break too early.
4514 */
4515 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL))
4516 goto rollback;
4517
4518 exec->transcount = 0;
4519 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
4520 trans = &exec->state->trans[exec->transno];
4521 if (trans->to < 0)
4522 continue;
4523 atom = trans->atom;
4524 ret = 0;
4525 if (trans->count >= 0) {
4526 int count;
4527 xmlRegCounterPtr counter;
4528
4529 /*
4530 * A counted transition.
4531 */
4532
4533 count = exec->counts[trans->count];
4534 counter = &exec->comp->counters[trans->count];
4535 #ifdef DEBUG_REGEXP_EXEC
4536 printf("testing count %d: val %d, min %d, max %d\n",
4537 trans->count, count, counter->min, counter->max);
4538 #endif
4539 ret = ((count >= counter->min) && (count <= counter->max));
4540 } else if (atom == NULL) {
4541 fprintf(stderr, "epsilon transition left at runtime\n");
4542 exec->status = -2;
4543 break;
4544 } else if (exec->inputString[exec->index] != 0) {
4545 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
4546 ret = xmlRegCheckCharacter(atom, codepoint);
4547 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
4548 xmlRegStatePtr to = exec->comp->states[trans->to];
4549
4550 /*
4551 * this is a multiple input sequence
4552 */
4553 if (exec->state->nbTrans > exec->transno + 1) {
4554 xmlFARegExecSave(exec);
4555 }
4556 exec->transcount = 1;
4557 do {
4558 /*
4559 * Try to progress as much as possible on the input
4560 */
4561 if (exec->transcount == atom->max) {
4562 break;
4563 }
4564 exec->index += len;
4565 /*
4566 * End of input: stop here
4567 */
4568 if (exec->inputString[exec->index] == 0) {
4569 exec->index -= len;
4570 break;
4571 }
4572 if (exec->transcount >= atom->min) {
4573 int transno = exec->transno;
4574 xmlRegStatePtr state = exec->state;
4575
4576 /*
4577 * The transition is acceptable save it
4578 */
4579 exec->transno = -1; /* trick */
4580 exec->state = to;
4581 xmlFARegExecSave(exec);
4582 exec->transno = transno;
4583 exec->state = state;
4584 }
4585 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
4586 len);
4587 ret = xmlRegCheckCharacter(atom, codepoint);
4588 exec->transcount++;
4589 } while (ret == 1);
4590 if (exec->transcount < atom->min)
4591 ret = 0;
4592
4593 /*
4594 * If the last check failed but one transition was found
4595 * possible, rollback
4596 */
4597 if (ret < 0)
4598 ret = 0;
4599 if (ret == 0) {
4600 goto rollback;
4601 }
4602 }
4603 }
4604 if (ret == 1) {
4605 if (exec->state->nbTrans > exec->transno + 1) {
4606 xmlFARegExecSave(exec);
4607 }
4608 /*
4609 * restart count for expressions like this ((abc){2})*
4610 */
4611 if (trans->count >= 0) {
4612 #ifdef DEBUG_REGEXP_EXEC
4613 printf("Reset count %d\n", trans->count);
4614 #endif
4615 exec->counts[trans->count] = 0;
4616 }
4617 if (trans->counter >= 0) {
4618 #ifdef DEBUG_REGEXP_EXEC
4619 printf("Increasing count %d\n", trans->counter);
4620 #endif
4621 exec->counts[trans->counter]++;
4622 }
4623 #ifdef DEBUG_REGEXP_EXEC
4624 printf("entering state %d\n", trans->to);
4625 #endif
4626 exec->state = exec->comp->states[trans->to];
4627 exec->transno = 0;
4628 if (trans->atom != NULL) {
4629 exec->index += len;
4630 }
4631 goto progress;
4632 } else if (ret < 0) {
4633 exec->status = -4;
4634 break;
4635 }
4636 }
4637 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
4638 rollback:
4639 /*
4640 * Failed to find a way out
4641 */
4642 exec->determinist = 0;
4643 xmlFARegExecRollBack(exec);
4644 }
4645 progress:
4646 continue;
4647 }
4648 }
4649 #endif
4650 /************************************************************************
4651 * *
4652 * Parser for the Schemas Datatype Regular Expressions *
4653 * http://www.w3.org/TR/2001/REC-xmlschema-2-20010502/#regexs *
4654 * *
4655 ************************************************************************/
4656
4657 /**
4658 * xmlFAIsChar:
4659 * @ctxt: a regexp parser context
4660 *
4661 * [10] Char ::= [^.\?*+()|#x5B#x5D]
4662 */
4663 static int
xmlFAIsChar(xmlRegParserCtxtPtr ctxt)4664 xmlFAIsChar(xmlRegParserCtxtPtr ctxt) {
4665 int cur;
4666 int len;
4667
4668 cur = CUR_SCHAR(ctxt->cur, len);
4669 if ((cur == '.') || (cur == '\\') || (cur == '?') ||
4670 (cur == '*') || (cur == '+') || (cur == '(') ||
4671 (cur == ')') || (cur == '|') || (cur == 0x5B) ||
4672 (cur == 0x5D) || (cur == 0))
4673 return(-1);
4674 return(cur);
4675 }
4676
4677 /**
4678 * xmlFAParseCharProp:
4679 * @ctxt: a regexp parser context
4680 *
4681 * [27] charProp ::= IsCategory | IsBlock
4682 * [28] IsCategory ::= Letters | Marks | Numbers | Punctuation |
4683 * Separators | Symbols | Others
4684 * [29] Letters ::= 'L' [ultmo]?
4685 * [30] Marks ::= 'M' [nce]?
4686 * [31] Numbers ::= 'N' [dlo]?
4687 * [32] Punctuation ::= 'P' [cdseifo]?
4688 * [33] Separators ::= 'Z' [slp]?
4689 * [34] Symbols ::= 'S' [mcko]?
4690 * [35] Others ::= 'C' [cfon]?
4691 * [36] IsBlock ::= 'Is' [a-zA-Z0-9#x2D]+
4692 */
4693 static void
xmlFAParseCharProp(xmlRegParserCtxtPtr ctxt)4694 xmlFAParseCharProp(xmlRegParserCtxtPtr ctxt) {
4695 int cur;
4696 xmlRegAtomType type = (xmlRegAtomType) 0;
4697 xmlChar *blockName = NULL;
4698
4699 cur = CUR;
4700 if (cur == 'L') {
4701 NEXT;
4702 cur = CUR;
4703 if (cur == 'u') {
4704 NEXT;
4705 type = XML_REGEXP_LETTER_UPPERCASE;
4706 } else if (cur == 'l') {
4707 NEXT;
4708 type = XML_REGEXP_LETTER_LOWERCASE;
4709 } else if (cur == 't') {
4710 NEXT;
4711 type = XML_REGEXP_LETTER_TITLECASE;
4712 } else if (cur == 'm') {
4713 NEXT;
4714 type = XML_REGEXP_LETTER_MODIFIER;
4715 } else if (cur == 'o') {
4716 NEXT;
4717 type = XML_REGEXP_LETTER_OTHERS;
4718 } else {
4719 type = XML_REGEXP_LETTER;
4720 }
4721 } else if (cur == 'M') {
4722 NEXT;
4723 cur = CUR;
4724 if (cur == 'n') {
4725 NEXT;
4726 /* nonspacing */
4727 type = XML_REGEXP_MARK_NONSPACING;
4728 } else if (cur == 'c') {
4729 NEXT;
4730 /* spacing combining */
4731 type = XML_REGEXP_MARK_SPACECOMBINING;
4732 } else if (cur == 'e') {
4733 NEXT;
4734 /* enclosing */
4735 type = XML_REGEXP_MARK_ENCLOSING;
4736 } else {
4737 /* all marks */
4738 type = XML_REGEXP_MARK;
4739 }
4740 } else if (cur == 'N') {
4741 NEXT;
4742 cur = CUR;
4743 if (cur == 'd') {
4744 NEXT;
4745 /* digital */
4746 type = XML_REGEXP_NUMBER_DECIMAL;
4747 } else if (cur == 'l') {
4748 NEXT;
4749 /* letter */
4750 type = XML_REGEXP_NUMBER_LETTER;
4751 } else if (cur == 'o') {
4752 NEXT;
4753 /* other */
4754 type = XML_REGEXP_NUMBER_OTHERS;
4755 } else {
4756 /* all numbers */
4757 type = XML_REGEXP_NUMBER;
4758 }
4759 } else if (cur == 'P') {
4760 NEXT;
4761 cur = CUR;
4762 if (cur == 'c') {
4763 NEXT;
4764 /* connector */
4765 type = XML_REGEXP_PUNCT_CONNECTOR;
4766 } else if (cur == 'd') {
4767 NEXT;
4768 /* dash */
4769 type = XML_REGEXP_PUNCT_DASH;
4770 } else if (cur == 's') {
4771 NEXT;
4772 /* open */
4773 type = XML_REGEXP_PUNCT_OPEN;
4774 } else if (cur == 'e') {
4775 NEXT;
4776 /* close */
4777 type = XML_REGEXP_PUNCT_CLOSE;
4778 } else if (cur == 'i') {
4779 NEXT;
4780 /* initial quote */
4781 type = XML_REGEXP_PUNCT_INITQUOTE;
4782 } else if (cur == 'f') {
4783 NEXT;
4784 /* final quote */
4785 type = XML_REGEXP_PUNCT_FINQUOTE;
4786 } else if (cur == 'o') {
4787 NEXT;
4788 /* other */
4789 type = XML_REGEXP_PUNCT_OTHERS;
4790 } else {
4791 /* all punctuation */
4792 type = XML_REGEXP_PUNCT;
4793 }
4794 } else if (cur == 'Z') {
4795 NEXT;
4796 cur = CUR;
4797 if (cur == 's') {
4798 NEXT;
4799 /* space */
4800 type = XML_REGEXP_SEPAR_SPACE;
4801 } else if (cur == 'l') {
4802 NEXT;
4803 /* line */
4804 type = XML_REGEXP_SEPAR_LINE;
4805 } else if (cur == 'p') {
4806 NEXT;
4807 /* paragraph */
4808 type = XML_REGEXP_SEPAR_PARA;
4809 } else {
4810 /* all separators */
4811 type = XML_REGEXP_SEPAR;
4812 }
4813 } else if (cur == 'S') {
4814 NEXT;
4815 cur = CUR;
4816 if (cur == 'm') {
4817 NEXT;
4818 type = XML_REGEXP_SYMBOL_MATH;
4819 /* math */
4820 } else if (cur == 'c') {
4821 NEXT;
4822 type = XML_REGEXP_SYMBOL_CURRENCY;
4823 /* currency */
4824 } else if (cur == 'k') {
4825 NEXT;
4826 type = XML_REGEXP_SYMBOL_MODIFIER;
4827 /* modifiers */
4828 } else if (cur == 'o') {
4829 NEXT;
4830 type = XML_REGEXP_SYMBOL_OTHERS;
4831 /* other */
4832 } else {
4833 /* all symbols */
4834 type = XML_REGEXP_SYMBOL;
4835 }
4836 } else if (cur == 'C') {
4837 NEXT;
4838 cur = CUR;
4839 if (cur == 'c') {
4840 NEXT;
4841 /* control */
4842 type = XML_REGEXP_OTHER_CONTROL;
4843 } else if (cur == 'f') {
4844 NEXT;
4845 /* format */
4846 type = XML_REGEXP_OTHER_FORMAT;
4847 } else if (cur == 'o') {
4848 NEXT;
4849 /* private use */
4850 type = XML_REGEXP_OTHER_PRIVATE;
4851 } else if (cur == 'n') {
4852 NEXT;
4853 /* not assigned */
4854 type = XML_REGEXP_OTHER_NA;
4855 } else {
4856 /* all others */
4857 type = XML_REGEXP_OTHER;
4858 }
4859 } else if (cur == 'I') {
4860 const xmlChar *start;
4861 NEXT;
4862 cur = CUR;
4863 if (cur != 's') {
4864 ERROR("IsXXXX expected");
4865 return;
4866 }
4867 NEXT;
4868 start = ctxt->cur;
4869 cur = CUR;
4870 if (((cur >= 'a') && (cur <= 'z')) ||
4871 ((cur >= 'A') && (cur <= 'Z')) ||
4872 ((cur >= '0') && (cur <= '9')) ||
4873 (cur == 0x2D)) {
4874 NEXT;
4875 cur = CUR;
4876 while (((cur >= 'a') && (cur <= 'z')) ||
4877 ((cur >= 'A') && (cur <= 'Z')) ||
4878 ((cur >= '0') && (cur <= '9')) ||
4879 (cur == 0x2D)) {
4880 NEXT;
4881 cur = CUR;
4882 }
4883 }
4884 type = XML_REGEXP_BLOCK_NAME;
4885 blockName = xmlStrndup(start, ctxt->cur - start);
4886 } else {
4887 ERROR("Unknown char property");
4888 return;
4889 }
4890 if (ctxt->atom == NULL) {
4891 ctxt->atom = xmlRegNewAtom(ctxt, type);
4892 if (ctxt->atom == NULL) {
4893 xmlFree(blockName);
4894 return;
4895 }
4896 ctxt->atom->valuep = blockName;
4897 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4898 if (xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4899 type, 0, 0, blockName) == NULL) {
4900 xmlFree(blockName);
4901 }
4902 }
4903 }
4904
parse_escaped_codeunit(xmlRegParserCtxtPtr ctxt)4905 static int parse_escaped_codeunit(xmlRegParserCtxtPtr ctxt)
4906 {
4907 int val = 0, i, cur;
4908 for (i = 0; i < 4; i++) {
4909 NEXT;
4910 val *= 16;
4911 cur = CUR;
4912 if (cur >= '0' && cur <= '9') {
4913 val += cur - '0';
4914 } else if (cur >= 'A' && cur <= 'F') {
4915 val += cur - 'A' + 10;
4916 } else if (cur >= 'a' && cur <= 'f') {
4917 val += cur - 'a' + 10;
4918 } else {
4919 ERROR("Expecting hex digit");
4920 return -1;
4921 }
4922 }
4923 return val;
4924 }
4925
parse_escaped_codepoint(xmlRegParserCtxtPtr ctxt)4926 static int parse_escaped_codepoint(xmlRegParserCtxtPtr ctxt)
4927 {
4928 int val = parse_escaped_codeunit(ctxt);
4929 if (0xD800 <= val && val <= 0xDBFF) {
4930 NEXT;
4931 if (CUR == '\\') {
4932 NEXT;
4933 if (CUR == 'u') {
4934 int low = parse_escaped_codeunit(ctxt);
4935 if (0xDC00 <= low && low <= 0xDFFF) {
4936 return (val - 0xD800) * 0x400 + (low - 0xDC00) + 0x10000;
4937 }
4938 }
4939 }
4940 ERROR("Invalid low surrogate pair code unit");
4941 val = -1;
4942 }
4943 return val;
4944 }
4945
4946 /**
4947 * xmlFAParseCharClassEsc:
4948 * @ctxt: a regexp parser context
4949 *
4950 * [23] charClassEsc ::= ( SingleCharEsc | MultiCharEsc | catEsc | complEsc )
4951 * [24] SingleCharEsc ::= '\' [nrt\|.?*+(){}#x2D#x5B#x5D#x5E]
4952 * [25] catEsc ::= '\p{' charProp '}'
4953 * [26] complEsc ::= '\P{' charProp '}'
4954 * [37] MultiCharEsc ::= '.' | ('\' [sSiIcCdDwW])
4955 */
4956 static void
xmlFAParseCharClassEsc(xmlRegParserCtxtPtr ctxt)4957 xmlFAParseCharClassEsc(xmlRegParserCtxtPtr ctxt) {
4958 int cur;
4959
4960 if (CUR == '.') {
4961 if (ctxt->atom == NULL) {
4962 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_ANYCHAR);
4963 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4964 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4965 XML_REGEXP_ANYCHAR, 0, 0, NULL);
4966 }
4967 NEXT;
4968 return;
4969 }
4970 if (CUR != '\\') {
4971 ERROR("Escaped sequence: expecting \\");
4972 return;
4973 }
4974 NEXT;
4975 cur = CUR;
4976 if (cur == 'p') {
4977 NEXT;
4978 if (CUR != '{') {
4979 ERROR("Expecting '{'");
4980 return;
4981 }
4982 NEXT;
4983 xmlFAParseCharProp(ctxt);
4984 if (CUR != '}') {
4985 ERROR("Expecting '}'");
4986 return;
4987 }
4988 NEXT;
4989 } else if (cur == 'P') {
4990 NEXT;
4991 if (CUR != '{') {
4992 ERROR("Expecting '{'");
4993 return;
4994 }
4995 NEXT;
4996 xmlFAParseCharProp(ctxt);
4997 if (ctxt->atom != NULL)
4998 ctxt->atom->neg = 1;
4999 if (CUR != '}') {
5000 ERROR("Expecting '}'");
5001 return;
5002 }
5003 NEXT;
5004 } else if ((cur == 'n') || (cur == 'r') || (cur == 't') || (cur == '\\') ||
5005 (cur == '|') || (cur == '.') || (cur == '?') || (cur == '*') ||
5006 (cur == '+') || (cur == '(') || (cur == ')') || (cur == '{') ||
5007 (cur == '}') || (cur == 0x2D) || (cur == 0x5B) || (cur == 0x5D) ||
5008 (cur == 0x5E) ||
5009
5010 /* Non-standard escape sequences:
5011 * Java 1.8|.NET Core 3.1|MSXML 6 */
5012 (cur == '!') || /* + | + | + */
5013 (cur == '"') || /* + | + | + */
5014 (cur == '#') || /* + | + | + */
5015 (cur == '$') || /* + | + | + */
5016 (cur == '%') || /* + | + | + */
5017 (cur == ',') || /* + | + | + */
5018 (cur == '/') || /* + | + | + */
5019 (cur == ':') || /* + | + | + */
5020 (cur == ';') || /* + | + | + */
5021 (cur == '=') || /* + | + | + */
5022 (cur == '>') || /* | + | + */
5023 (cur == '@') || /* + | + | + */
5024 (cur == '`') || /* + | + | + */
5025 (cur == '~') || /* + | + | + */
5026 (cur == 'u')) { /* | + | + */
5027 if (ctxt->atom == NULL) {
5028 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
5029 if (ctxt->atom != NULL) {
5030 switch (cur) {
5031 case 'n':
5032 ctxt->atom->codepoint = '\n';
5033 break;
5034 case 'r':
5035 ctxt->atom->codepoint = '\r';
5036 break;
5037 case 't':
5038 ctxt->atom->codepoint = '\t';
5039 break;
5040 case 'u':
5041 cur = parse_escaped_codepoint(ctxt);
5042 if (cur < 0) {
5043 return;
5044 }
5045 ctxt->atom->codepoint = cur;
5046 break;
5047 default:
5048 ctxt->atom->codepoint = cur;
5049 }
5050 }
5051 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
5052 switch (cur) {
5053 case 'n':
5054 cur = '\n';
5055 break;
5056 case 'r':
5057 cur = '\r';
5058 break;
5059 case 't':
5060 cur = '\t';
5061 break;
5062 }
5063 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5064 XML_REGEXP_CHARVAL, cur, cur, NULL);
5065 }
5066 NEXT;
5067 } else if ((cur == 's') || (cur == 'S') || (cur == 'i') || (cur == 'I') ||
5068 (cur == 'c') || (cur == 'C') || (cur == 'd') || (cur == 'D') ||
5069 (cur == 'w') || (cur == 'W')) {
5070 xmlRegAtomType type = XML_REGEXP_ANYSPACE;
5071
5072 switch (cur) {
5073 case 's':
5074 type = XML_REGEXP_ANYSPACE;
5075 break;
5076 case 'S':
5077 type = XML_REGEXP_NOTSPACE;
5078 break;
5079 case 'i':
5080 type = XML_REGEXP_INITNAME;
5081 break;
5082 case 'I':
5083 type = XML_REGEXP_NOTINITNAME;
5084 break;
5085 case 'c':
5086 type = XML_REGEXP_NAMECHAR;
5087 break;
5088 case 'C':
5089 type = XML_REGEXP_NOTNAMECHAR;
5090 break;
5091 case 'd':
5092 type = XML_REGEXP_DECIMAL;
5093 break;
5094 case 'D':
5095 type = XML_REGEXP_NOTDECIMAL;
5096 break;
5097 case 'w':
5098 type = XML_REGEXP_REALCHAR;
5099 break;
5100 case 'W':
5101 type = XML_REGEXP_NOTREALCHAR;
5102 break;
5103 }
5104 NEXT;
5105 if (ctxt->atom == NULL) {
5106 ctxt->atom = xmlRegNewAtom(ctxt, type);
5107 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
5108 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5109 type, 0, 0, NULL);
5110 }
5111 } else {
5112 ERROR("Wrong escape sequence, misuse of character '\\'");
5113 }
5114 }
5115
5116 /**
5117 * xmlFAParseCharRange:
5118 * @ctxt: a regexp parser context
5119 *
5120 * [17] charRange ::= seRange | XmlCharRef | XmlCharIncDash
5121 * [18] seRange ::= charOrEsc '-' charOrEsc
5122 * [20] charOrEsc ::= XmlChar | SingleCharEsc
5123 * [21] XmlChar ::= [^\#x2D#x5B#x5D]
5124 * [22] XmlCharIncDash ::= [^\#x5B#x5D]
5125 */
5126 static void
xmlFAParseCharRange(xmlRegParserCtxtPtr ctxt)5127 xmlFAParseCharRange(xmlRegParserCtxtPtr ctxt) {
5128 int cur, len;
5129 int start = -1;
5130 int end = -1;
5131
5132 if (CUR == '\0') {
5133 ERROR("Expecting ']'");
5134 return;
5135 }
5136
5137 cur = CUR;
5138 if (cur == '\\') {
5139 NEXT;
5140 cur = CUR;
5141 switch (cur) {
5142 case 'n': start = 0xA; break;
5143 case 'r': start = 0xD; break;
5144 case 't': start = 0x9; break;
5145 case '\\': case '|': case '.': case '-': case '^': case '?':
5146 case '*': case '+': case '{': case '}': case '(': case ')':
5147 case '[': case ']':
5148 start = cur; break;
5149 default:
5150 ERROR("Invalid escape value");
5151 return;
5152 }
5153 end = start;
5154 len = 1;
5155 } else if ((cur != 0x5B) && (cur != 0x5D)) {
5156 end = start = CUR_SCHAR(ctxt->cur, len);
5157 } else {
5158 ERROR("Expecting a char range");
5159 return;
5160 }
5161 /*
5162 * Since we are "inside" a range, we can assume ctxt->cur is past
5163 * the start of ctxt->string, and PREV should be safe
5164 */
5165 if ((start == '-') && (NXT(1) != ']') && (PREV != '[') && (PREV != '^')) {
5166 NEXTL(len);
5167 return;
5168 }
5169 NEXTL(len);
5170 cur = CUR;
5171 if ((cur != '-') || (NXT(1) == '[') || (NXT(1) == ']')) {
5172 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5173 XML_REGEXP_CHARVAL, start, end, NULL);
5174 return;
5175 }
5176 NEXT;
5177 cur = CUR;
5178 if (cur == '\\') {
5179 NEXT;
5180 cur = CUR;
5181 switch (cur) {
5182 case 'n': end = 0xA; break;
5183 case 'r': end = 0xD; break;
5184 case 't': end = 0x9; break;
5185 case '\\': case '|': case '.': case '-': case '^': case '?':
5186 case '*': case '+': case '{': case '}': case '(': case ')':
5187 case '[': case ']':
5188 end = cur; break;
5189 default:
5190 ERROR("Invalid escape value");
5191 return;
5192 }
5193 len = 1;
5194 } else if ((cur != '\0') && (cur != 0x5B) && (cur != 0x5D)) {
5195 end = CUR_SCHAR(ctxt->cur, len);
5196 } else {
5197 ERROR("Expecting the end of a char range");
5198 return;
5199 }
5200
5201 /* TODO check that the values are acceptable character ranges for XML */
5202 if (end < start) {
5203 ERROR("End of range is before start of range");
5204 } else {
5205 NEXTL(len);
5206 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5207 XML_REGEXP_CHARVAL, start, end, NULL);
5208 }
5209 return;
5210 }
5211
5212 /**
5213 * xmlFAParsePosCharGroup:
5214 * @ctxt: a regexp parser context
5215 *
5216 * [14] posCharGroup ::= ( charRange | charClassEsc )+
5217 */
5218 static void
xmlFAParsePosCharGroup(xmlRegParserCtxtPtr ctxt)5219 xmlFAParsePosCharGroup(xmlRegParserCtxtPtr ctxt) {
5220 do {
5221 if (CUR == '\\') {
5222 xmlFAParseCharClassEsc(ctxt);
5223 } else {
5224 xmlFAParseCharRange(ctxt);
5225 }
5226 } while ((CUR != ']') && (CUR != '-') &&
5227 (CUR != 0) && (ctxt->error == 0));
5228 }
5229
5230 /**
5231 * xmlFAParseCharGroup:
5232 * @ctxt: a regexp parser context
5233 *
5234 * [13] charGroup ::= posCharGroup | negCharGroup | charClassSub
5235 * [15] negCharGroup ::= '^' posCharGroup
5236 * [16] charClassSub ::= ( posCharGroup | negCharGroup ) '-' charClassExpr
5237 * [12] charClassExpr ::= '[' charGroup ']'
5238 */
5239 static void
xmlFAParseCharGroup(xmlRegParserCtxtPtr ctxt)5240 xmlFAParseCharGroup(xmlRegParserCtxtPtr ctxt) {
5241 int neg = ctxt->neg;
5242
5243 if (CUR == '^') {
5244 NEXT;
5245 ctxt->neg = !ctxt->neg;
5246 xmlFAParsePosCharGroup(ctxt);
5247 ctxt->neg = neg;
5248 }
5249 while ((CUR != ']') && (ctxt->error == 0)) {
5250 if ((CUR == '-') && (NXT(1) == '[')) {
5251 NEXT; /* eat the '-' */
5252 NEXT; /* eat the '[' */
5253 ctxt->neg = 2;
5254 xmlFAParseCharGroup(ctxt);
5255 ctxt->neg = neg;
5256 if (CUR == ']') {
5257 NEXT;
5258 } else {
5259 ERROR("charClassExpr: ']' expected");
5260 }
5261 break;
5262 } else {
5263 xmlFAParsePosCharGroup(ctxt);
5264 }
5265 }
5266 }
5267
5268 /**
5269 * xmlFAParseCharClass:
5270 * @ctxt: a regexp parser context
5271 *
5272 * [11] charClass ::= charClassEsc | charClassExpr
5273 * [12] charClassExpr ::= '[' charGroup ']'
5274 */
5275 static void
xmlFAParseCharClass(xmlRegParserCtxtPtr ctxt)5276 xmlFAParseCharClass(xmlRegParserCtxtPtr ctxt) {
5277 if (CUR == '[') {
5278 NEXT;
5279 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_RANGES);
5280 if (ctxt->atom == NULL)
5281 return;
5282 xmlFAParseCharGroup(ctxt);
5283 if (CUR == ']') {
5284 NEXT;
5285 } else {
5286 ERROR("xmlFAParseCharClass: ']' expected");
5287 }
5288 } else {
5289 xmlFAParseCharClassEsc(ctxt);
5290 }
5291 }
5292
5293 /**
5294 * xmlFAParseQuantExact:
5295 * @ctxt: a regexp parser context
5296 *
5297 * [8] QuantExact ::= [0-9]+
5298 *
5299 * Returns 0 if success or -1 in case of error
5300 */
5301 static int
xmlFAParseQuantExact(xmlRegParserCtxtPtr ctxt)5302 xmlFAParseQuantExact(xmlRegParserCtxtPtr ctxt) {
5303 int ret = 0;
5304 int ok = 0;
5305 int overflow = 0;
5306
5307 while ((CUR >= '0') && (CUR <= '9')) {
5308 if (ret > INT_MAX / 10) {
5309 overflow = 1;
5310 } else {
5311 int digit = CUR - '0';
5312
5313 ret *= 10;
5314 if (ret > INT_MAX - digit)
5315 overflow = 1;
5316 else
5317 ret += digit;
5318 }
5319 ok = 1;
5320 NEXT;
5321 }
5322 if ((ok != 1) || (overflow == 1)) {
5323 return(-1);
5324 }
5325 return(ret);
5326 }
5327
5328 /**
5329 * xmlFAParseQuantifier:
5330 * @ctxt: a regexp parser context
5331 *
5332 * [4] quantifier ::= [?*+] | ( '{' quantity '}' )
5333 * [5] quantity ::= quantRange | quantMin | QuantExact
5334 * [6] quantRange ::= QuantExact ',' QuantExact
5335 * [7] quantMin ::= QuantExact ','
5336 * [8] QuantExact ::= [0-9]+
5337 */
5338 static int
xmlFAParseQuantifier(xmlRegParserCtxtPtr ctxt)5339 xmlFAParseQuantifier(xmlRegParserCtxtPtr ctxt) {
5340 int cur;
5341
5342 cur = CUR;
5343 if ((cur == '?') || (cur == '*') || (cur == '+')) {
5344 if (ctxt->atom != NULL) {
5345 if (cur == '?')
5346 ctxt->atom->quant = XML_REGEXP_QUANT_OPT;
5347 else if (cur == '*')
5348 ctxt->atom->quant = XML_REGEXP_QUANT_MULT;
5349 else if (cur == '+')
5350 ctxt->atom->quant = XML_REGEXP_QUANT_PLUS;
5351 }
5352 NEXT;
5353 return(1);
5354 }
5355 if (cur == '{') {
5356 int min = 0, max = 0;
5357
5358 NEXT;
5359 cur = xmlFAParseQuantExact(ctxt);
5360 if (cur >= 0)
5361 min = cur;
5362 else {
5363 ERROR("Improper quantifier");
5364 }
5365 if (CUR == ',') {
5366 NEXT;
5367 if (CUR == '}')
5368 max = INT_MAX;
5369 else {
5370 cur = xmlFAParseQuantExact(ctxt);
5371 if (cur >= 0)
5372 max = cur;
5373 else {
5374 ERROR("Improper quantifier");
5375 }
5376 }
5377 }
5378 if (CUR == '}') {
5379 NEXT;
5380 } else {
5381 ERROR("Unterminated quantifier");
5382 }
5383 if (max == 0)
5384 max = min;
5385 if (ctxt->atom != NULL) {
5386 ctxt->atom->quant = XML_REGEXP_QUANT_RANGE;
5387 ctxt->atom->min = min;
5388 ctxt->atom->max = max;
5389 }
5390 return(1);
5391 }
5392 return(0);
5393 }
5394
5395 /**
5396 * xmlFAParseAtom:
5397 * @ctxt: a regexp parser context
5398 *
5399 * [9] atom ::= Char | charClass | ( '(' regExp ')' )
5400 */
5401 static int
xmlFAParseAtom(xmlRegParserCtxtPtr ctxt)5402 xmlFAParseAtom(xmlRegParserCtxtPtr ctxt) {
5403 int codepoint, len;
5404
5405 codepoint = xmlFAIsChar(ctxt);
5406 if (codepoint > 0) {
5407 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
5408 if (ctxt->atom == NULL)
5409 return(-1);
5410 codepoint = CUR_SCHAR(ctxt->cur, len);
5411 ctxt->atom->codepoint = codepoint;
5412 NEXTL(len);
5413 return(1);
5414 } else if (CUR == '|') {
5415 return(0);
5416 } else if (CUR == 0) {
5417 return(0);
5418 } else if (CUR == ')') {
5419 return(0);
5420 } else if (CUR == '(') {
5421 xmlRegStatePtr start, oldend, start0;
5422
5423 NEXT;
5424 if (ctxt->depth >= 50) {
5425 ERROR("xmlFAParseAtom: maximum nesting depth exceeded");
5426 return(-1);
5427 }
5428 /*
5429 * this extra Epsilon transition is needed if we count with 0 allowed
5430 * unfortunately this can't be known at that point
5431 */
5432 xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, NULL);
5433 start0 = ctxt->state;
5434 xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, NULL);
5435 start = ctxt->state;
5436 oldend = ctxt->end;
5437 ctxt->end = NULL;
5438 ctxt->atom = NULL;
5439 ctxt->depth++;
5440 xmlFAParseRegExp(ctxt, 0);
5441 ctxt->depth--;
5442 if (CUR == ')') {
5443 NEXT;
5444 } else {
5445 ERROR("xmlFAParseAtom: expecting ')'");
5446 }
5447 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_SUBREG);
5448 if (ctxt->atom == NULL)
5449 return(-1);
5450 ctxt->atom->start = start;
5451 ctxt->atom->start0 = start0;
5452 ctxt->atom->stop = ctxt->state;
5453 ctxt->end = oldend;
5454 return(1);
5455 } else if ((CUR == '[') || (CUR == '\\') || (CUR == '.')) {
5456 xmlFAParseCharClass(ctxt);
5457 return(1);
5458 }
5459 return(0);
5460 }
5461
5462 /**
5463 * xmlFAParsePiece:
5464 * @ctxt: a regexp parser context
5465 *
5466 * [3] piece ::= atom quantifier?
5467 */
5468 static int
xmlFAParsePiece(xmlRegParserCtxtPtr ctxt)5469 xmlFAParsePiece(xmlRegParserCtxtPtr ctxt) {
5470 int ret;
5471
5472 ctxt->atom = NULL;
5473 ret = xmlFAParseAtom(ctxt);
5474 if (ret == 0)
5475 return(0);
5476 if (ctxt->atom == NULL) {
5477 ERROR("internal: no atom generated");
5478 }
5479 xmlFAParseQuantifier(ctxt);
5480 return(1);
5481 }
5482
5483 /**
5484 * xmlFAParseBranch:
5485 * @ctxt: a regexp parser context
5486 * @to: optional target to the end of the branch
5487 *
5488 * @to is used to optimize by removing duplicate path in automata
5489 * in expressions like (a|b)(c|d)
5490 *
5491 * [2] branch ::= piece*
5492 */
5493 static int
xmlFAParseBranch(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr to)5494 xmlFAParseBranch(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr to) {
5495 xmlRegStatePtr previous;
5496 int ret;
5497
5498 previous = ctxt->state;
5499 ret = xmlFAParsePiece(ctxt);
5500 if (ret == 0) {
5501 /* Empty branch */
5502 xmlFAGenerateEpsilonTransition(ctxt, previous, to);
5503 } else {
5504 if (xmlFAGenerateTransitions(ctxt, previous,
5505 (CUR=='|' || CUR==')' || CUR==0) ? to : NULL,
5506 ctxt->atom) < 0) {
5507 xmlRegFreeAtom(ctxt->atom);
5508 ctxt->atom = NULL;
5509 return(-1);
5510 }
5511 previous = ctxt->state;
5512 ctxt->atom = NULL;
5513 }
5514 while ((ret != 0) && (ctxt->error == 0)) {
5515 ret = xmlFAParsePiece(ctxt);
5516 if (ret != 0) {
5517 if (xmlFAGenerateTransitions(ctxt, previous,
5518 (CUR=='|' || CUR==')' || CUR==0) ? to : NULL,
5519 ctxt->atom) < 0) {
5520 xmlRegFreeAtom(ctxt->atom);
5521 ctxt->atom = NULL;
5522 return(-1);
5523 }
5524 previous = ctxt->state;
5525 ctxt->atom = NULL;
5526 }
5527 }
5528 return(0);
5529 }
5530
5531 /**
5532 * xmlFAParseRegExp:
5533 * @ctxt: a regexp parser context
5534 * @top: is this the top-level expression ?
5535 *
5536 * [1] regExp ::= branch ( '|' branch )*
5537 */
5538 static void
xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt,int top)5539 xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top) {
5540 xmlRegStatePtr start, end;
5541
5542 /* if not top start should have been generated by an epsilon trans */
5543 start = ctxt->state;
5544 ctxt->end = NULL;
5545 xmlFAParseBranch(ctxt, NULL);
5546 if (top) {
5547 #ifdef DEBUG_REGEXP_GRAPH
5548 printf("State %d is final\n", ctxt->state->no);
5549 #endif
5550 ctxt->state->type = XML_REGEXP_FINAL_STATE;
5551 }
5552 if (CUR != '|') {
5553 ctxt->end = ctxt->state;
5554 return;
5555 }
5556 end = ctxt->state;
5557 while ((CUR == '|') && (ctxt->error == 0)) {
5558 NEXT;
5559 ctxt->state = start;
5560 ctxt->end = NULL;
5561 xmlFAParseBranch(ctxt, end);
5562 }
5563 if (!top) {
5564 ctxt->state = end;
5565 ctxt->end = end;
5566 }
5567 }
5568
5569 /************************************************************************
5570 * *
5571 * The basic API *
5572 * *
5573 ************************************************************************/
5574
5575 /**
5576 * xmlRegexpPrint:
5577 * @output: the file for the output debug
5578 * @regexp: the compiled regexp
5579 *
5580 * Print the content of the compiled regular expression
5581 */
5582 void
xmlRegexpPrint(FILE * output,xmlRegexpPtr regexp)5583 xmlRegexpPrint(FILE *output, xmlRegexpPtr regexp) {
5584 int i;
5585
5586 if (output == NULL)
5587 return;
5588 fprintf(output, " regexp: ");
5589 if (regexp == NULL) {
5590 fprintf(output, "NULL\n");
5591 return;
5592 }
5593 fprintf(output, "'%s' ", regexp->string);
5594 fprintf(output, "\n");
5595 fprintf(output, "%d atoms:\n", regexp->nbAtoms);
5596 for (i = 0;i < regexp->nbAtoms; i++) {
5597 fprintf(output, " %02d ", i);
5598 xmlRegPrintAtom(output, regexp->atoms[i]);
5599 }
5600 fprintf(output, "%d states:", regexp->nbStates);
5601 fprintf(output, "\n");
5602 for (i = 0;i < regexp->nbStates; i++) {
5603 xmlRegPrintState(output, regexp->states[i]);
5604 }
5605 fprintf(output, "%d counters:\n", regexp->nbCounters);
5606 for (i = 0;i < regexp->nbCounters; i++) {
5607 fprintf(output, " %d: min %d max %d\n", i, regexp->counters[i].min,
5608 regexp->counters[i].max);
5609 }
5610 }
5611
5612 /**
5613 * xmlRegexpCompile:
5614 * @regexp: a regular expression string
5615 *
5616 * Parses a regular expression conforming to XML Schemas Part 2 Datatype
5617 * Appendix F and builds an automata suitable for testing strings against
5618 * that regular expression
5619 *
5620 * Returns the compiled expression or NULL in case of error
5621 */
5622 xmlRegexpPtr
xmlRegexpCompile(const xmlChar * regexp)5623 xmlRegexpCompile(const xmlChar *regexp) {
5624 xmlRegexpPtr ret = NULL;
5625 xmlRegParserCtxtPtr ctxt;
5626
5627 ctxt = xmlRegNewParserCtxt(regexp);
5628 if (ctxt == NULL)
5629 return(NULL);
5630
5631 /* initialize the parser */
5632 ctxt->state = xmlRegStatePush(ctxt);
5633 if (ctxt->state == NULL)
5634 goto error;
5635 ctxt->start = ctxt->state;
5636 ctxt->end = NULL;
5637
5638 /* parse the expression building an automata */
5639 xmlFAParseRegExp(ctxt, 1);
5640 if (CUR != 0) {
5641 ERROR("xmlFAParseRegExp: extra characters");
5642 }
5643 if (ctxt->error != 0)
5644 goto error;
5645 ctxt->end = ctxt->state;
5646 ctxt->start->type = XML_REGEXP_START_STATE;
5647 ctxt->end->type = XML_REGEXP_FINAL_STATE;
5648
5649 /* remove the Epsilon except for counted transitions */
5650 xmlFAEliminateEpsilonTransitions(ctxt);
5651
5652
5653 if (ctxt->error != 0)
5654 goto error;
5655 ret = xmlRegEpxFromParse(ctxt);
5656
5657 error:
5658 xmlRegFreeParserCtxt(ctxt);
5659 return(ret);
5660 }
5661
5662 /**
5663 * xmlRegexpExec:
5664 * @comp: the compiled regular expression
5665 * @content: the value to check against the regular expression
5666 *
5667 * Check if the regular expression generates the value
5668 *
5669 * Returns 1 if it matches, 0 if not and a negative value in case of error
5670 */
5671 int
xmlRegexpExec(xmlRegexpPtr comp,const xmlChar * content)5672 xmlRegexpExec(xmlRegexpPtr comp, const xmlChar *content) {
5673 if ((comp == NULL) || (content == NULL))
5674 return(-1);
5675 return(xmlFARegExec(comp, content));
5676 }
5677
5678 /**
5679 * xmlRegexpIsDeterminist:
5680 * @comp: the compiled regular expression
5681 *
5682 * Check if the regular expression is determinist
5683 *
5684 * Returns 1 if it yes, 0 if not and a negative value in case of error
5685 */
5686 int
xmlRegexpIsDeterminist(xmlRegexpPtr comp)5687 xmlRegexpIsDeterminist(xmlRegexpPtr comp) {
5688 xmlAutomataPtr am;
5689 int ret;
5690
5691 if (comp == NULL)
5692 return(-1);
5693 if (comp->determinist != -1)
5694 return(comp->determinist);
5695
5696 am = xmlNewAutomata();
5697 if (am == NULL)
5698 return(-1);
5699 if (am->states != NULL) {
5700 int i;
5701
5702 for (i = 0;i < am->nbStates;i++)
5703 xmlRegFreeState(am->states[i]);
5704 xmlFree(am->states);
5705 }
5706 am->nbAtoms = comp->nbAtoms;
5707 am->atoms = comp->atoms;
5708 am->nbStates = comp->nbStates;
5709 am->states = comp->states;
5710 am->determinist = -1;
5711 am->flags = comp->flags;
5712 ret = xmlFAComputesDeterminism(am);
5713 am->atoms = NULL;
5714 am->states = NULL;
5715 xmlFreeAutomata(am);
5716 comp->determinist = ret;
5717 return(ret);
5718 }
5719
5720 /**
5721 * xmlRegFreeRegexp:
5722 * @regexp: the regexp
5723 *
5724 * Free a regexp
5725 */
5726 void
xmlRegFreeRegexp(xmlRegexpPtr regexp)5727 xmlRegFreeRegexp(xmlRegexpPtr regexp) {
5728 int i;
5729 if (regexp == NULL)
5730 return;
5731
5732 if (regexp->string != NULL)
5733 xmlFree(regexp->string);
5734 if (regexp->states != NULL) {
5735 for (i = 0;i < regexp->nbStates;i++)
5736 xmlRegFreeState(regexp->states[i]);
5737 xmlFree(regexp->states);
5738 }
5739 if (regexp->atoms != NULL) {
5740 for (i = 0;i < regexp->nbAtoms;i++)
5741 xmlRegFreeAtom(regexp->atoms[i]);
5742 xmlFree(regexp->atoms);
5743 }
5744 if (regexp->counters != NULL)
5745 xmlFree(regexp->counters);
5746 if (regexp->compact != NULL)
5747 xmlFree(regexp->compact);
5748 if (regexp->transdata != NULL)
5749 xmlFree(regexp->transdata);
5750 if (regexp->stringMap != NULL) {
5751 for (i = 0; i < regexp->nbstrings;i++)
5752 xmlFree(regexp->stringMap[i]);
5753 xmlFree(regexp->stringMap);
5754 }
5755
5756 xmlFree(regexp);
5757 }
5758
5759 #ifdef LIBXML_AUTOMATA_ENABLED
5760 /************************************************************************
5761 * *
5762 * The Automata interface *
5763 * *
5764 ************************************************************************/
5765
5766 /**
5767 * xmlNewAutomata:
5768 *
5769 * Create a new automata
5770 *
5771 * Returns the new object or NULL in case of failure
5772 */
5773 xmlAutomataPtr
xmlNewAutomata(void)5774 xmlNewAutomata(void) {
5775 xmlAutomataPtr ctxt;
5776
5777 ctxt = xmlRegNewParserCtxt(NULL);
5778 if (ctxt == NULL)
5779 return(NULL);
5780
5781 /* initialize the parser */
5782 ctxt->state = xmlRegStatePush(ctxt);
5783 if (ctxt->state == NULL) {
5784 xmlFreeAutomata(ctxt);
5785 return(NULL);
5786 }
5787 ctxt->start = ctxt->state;
5788 ctxt->end = NULL;
5789
5790 ctxt->start->type = XML_REGEXP_START_STATE;
5791 ctxt->flags = 0;
5792
5793 return(ctxt);
5794 }
5795
5796 /**
5797 * xmlFreeAutomata:
5798 * @am: an automata
5799 *
5800 * Free an automata
5801 */
5802 void
xmlFreeAutomata(xmlAutomataPtr am)5803 xmlFreeAutomata(xmlAutomataPtr am) {
5804 if (am == NULL)
5805 return;
5806 xmlRegFreeParserCtxt(am);
5807 }
5808
5809 /**
5810 * xmlAutomataSetFlags:
5811 * @am: an automata
5812 * @flags: a set of internal flags
5813 *
5814 * Set some flags on the automata
5815 */
5816 void
xmlAutomataSetFlags(xmlAutomataPtr am,int flags)5817 xmlAutomataSetFlags(xmlAutomataPtr am, int flags) {
5818 if (am == NULL)
5819 return;
5820 am->flags |= flags;
5821 }
5822
5823 /**
5824 * xmlAutomataGetInitState:
5825 * @am: an automata
5826 *
5827 * Initial state lookup
5828 *
5829 * Returns the initial state of the automata
5830 */
5831 xmlAutomataStatePtr
xmlAutomataGetInitState(xmlAutomataPtr am)5832 xmlAutomataGetInitState(xmlAutomataPtr am) {
5833 if (am == NULL)
5834 return(NULL);
5835 return(am->start);
5836 }
5837
5838 /**
5839 * xmlAutomataSetFinalState:
5840 * @am: an automata
5841 * @state: a state in this automata
5842 *
5843 * Makes that state a final state
5844 *
5845 * Returns 0 or -1 in case of error
5846 */
5847 int
xmlAutomataSetFinalState(xmlAutomataPtr am,xmlAutomataStatePtr state)5848 xmlAutomataSetFinalState(xmlAutomataPtr am, xmlAutomataStatePtr state) {
5849 if ((am == NULL) || (state == NULL))
5850 return(-1);
5851 state->type = XML_REGEXP_FINAL_STATE;
5852 return(0);
5853 }
5854
5855 /**
5856 * xmlAutomataNewTransition:
5857 * @am: an automata
5858 * @from: the starting point of the transition
5859 * @to: the target point of the transition or NULL
5860 * @token: the input string associated to that transition
5861 * @data: data passed to the callback function if the transition is activated
5862 *
5863 * If @to is NULL, this creates first a new target state in the automata
5864 * and then adds a transition from the @from state to the target state
5865 * activated by the value of @token
5866 *
5867 * Returns the target state or NULL in case of error
5868 */
5869 xmlAutomataStatePtr
xmlAutomataNewTransition(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,void * data)5870 xmlAutomataNewTransition(xmlAutomataPtr am, xmlAutomataStatePtr from,
5871 xmlAutomataStatePtr to, const xmlChar *token,
5872 void *data) {
5873 xmlRegAtomPtr atom;
5874
5875 if ((am == NULL) || (from == NULL) || (token == NULL))
5876 return(NULL);
5877 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5878 if (atom == NULL)
5879 return(NULL);
5880 atom->data = data;
5881 atom->valuep = xmlStrdup(token);
5882
5883 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5884 xmlRegFreeAtom(atom);
5885 return(NULL);
5886 }
5887 if (to == NULL)
5888 return(am->state);
5889 return(to);
5890 }
5891
5892 /**
5893 * xmlAutomataNewTransition2:
5894 * @am: an automata
5895 * @from: the starting point of the transition
5896 * @to: the target point of the transition or NULL
5897 * @token: the first input string associated to that transition
5898 * @token2: the second input string associated to that transition
5899 * @data: data passed to the callback function if the transition is activated
5900 *
5901 * If @to is NULL, this creates first a new target state in the automata
5902 * and then adds a transition from the @from state to the target state
5903 * activated by the value of @token
5904 *
5905 * Returns the target state or NULL in case of error
5906 */
5907 xmlAutomataStatePtr
xmlAutomataNewTransition2(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,const xmlChar * token2,void * data)5908 xmlAutomataNewTransition2(xmlAutomataPtr am, xmlAutomataStatePtr from,
5909 xmlAutomataStatePtr to, const xmlChar *token,
5910 const xmlChar *token2, void *data) {
5911 xmlRegAtomPtr atom;
5912
5913 if ((am == NULL) || (from == NULL) || (token == NULL))
5914 return(NULL);
5915 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5916 if (atom == NULL)
5917 return(NULL);
5918 atom->data = data;
5919 if ((token2 == NULL) || (*token2 == 0)) {
5920 atom->valuep = xmlStrdup(token);
5921 } else {
5922 int lenn, lenp;
5923 xmlChar *str;
5924
5925 lenn = strlen((char *) token2);
5926 lenp = strlen((char *) token);
5927
5928 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5929 if (str == NULL) {
5930 xmlRegFreeAtom(atom);
5931 return(NULL);
5932 }
5933 memcpy(&str[0], token, lenp);
5934 str[lenp] = '|';
5935 memcpy(&str[lenp + 1], token2, lenn);
5936 str[lenn + lenp + 1] = 0;
5937
5938 atom->valuep = str;
5939 }
5940
5941 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5942 xmlRegFreeAtom(atom);
5943 return(NULL);
5944 }
5945 if (to == NULL)
5946 return(am->state);
5947 return(to);
5948 }
5949
5950 /**
5951 * xmlAutomataNewNegTrans:
5952 * @am: an automata
5953 * @from: the starting point of the transition
5954 * @to: the target point of the transition or NULL
5955 * @token: the first input string associated to that transition
5956 * @token2: the second input string associated to that transition
5957 * @data: data passed to the callback function if the transition is activated
5958 *
5959 * If @to is NULL, this creates first a new target state in the automata
5960 * and then adds a transition from the @from state to the target state
5961 * activated by any value except (@token,@token2)
5962 * Note that if @token2 is not NULL, then (X, NULL) won't match to follow
5963 # the semantic of XSD ##other
5964 *
5965 * Returns the target state or NULL in case of error
5966 */
5967 xmlAutomataStatePtr
xmlAutomataNewNegTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,const xmlChar * token2,void * data)5968 xmlAutomataNewNegTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
5969 xmlAutomataStatePtr to, const xmlChar *token,
5970 const xmlChar *token2, void *data) {
5971 xmlRegAtomPtr atom;
5972 xmlChar err_msg[200];
5973
5974 if ((am == NULL) || (from == NULL) || (token == NULL))
5975 return(NULL);
5976 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5977 if (atom == NULL)
5978 return(NULL);
5979 atom->data = data;
5980 atom->neg = 1;
5981 if ((token2 == NULL) || (*token2 == 0)) {
5982 atom->valuep = xmlStrdup(token);
5983 } else {
5984 int lenn, lenp;
5985 xmlChar *str;
5986
5987 lenn = strlen((char *) token2);
5988 lenp = strlen((char *) token);
5989
5990 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5991 if (str == NULL) {
5992 xmlRegFreeAtom(atom);
5993 return(NULL);
5994 }
5995 memcpy(&str[0], token, lenp);
5996 str[lenp] = '|';
5997 memcpy(&str[lenp + 1], token2, lenn);
5998 str[lenn + lenp + 1] = 0;
5999
6000 atom->valuep = str;
6001 }
6002 snprintf((char *) err_msg, 199, "not %s", (const char *) atom->valuep);
6003 err_msg[199] = 0;
6004 atom->valuep2 = xmlStrdup(err_msg);
6005
6006 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
6007 xmlRegFreeAtom(atom);
6008 return(NULL);
6009 }
6010 am->negs++;
6011 if (to == NULL)
6012 return(am->state);
6013 return(to);
6014 }
6015
6016 /**
6017 * xmlAutomataNewCountTrans2:
6018 * @am: an automata
6019 * @from: the starting point of the transition
6020 * @to: the target point of the transition or NULL
6021 * @token: the input string associated to that transition
6022 * @token2: the second input string associated to that transition
6023 * @min: the minimum successive occurrences of token
6024 * @max: the maximum successive occurrences of token
6025 * @data: data associated to the transition
6026 *
6027 * If @to is NULL, this creates first a new target state in the automata
6028 * and then adds a transition from the @from state to the target state
6029 * activated by a succession of input of value @token and @token2 and
6030 * whose number is between @min and @max
6031 *
6032 * Returns the target state or NULL in case of error
6033 */
6034 xmlAutomataStatePtr
xmlAutomataNewCountTrans2(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,const xmlChar * token2,int min,int max,void * data)6035 xmlAutomataNewCountTrans2(xmlAutomataPtr am, xmlAutomataStatePtr from,
6036 xmlAutomataStatePtr to, const xmlChar *token,
6037 const xmlChar *token2,
6038 int min, int max, void *data) {
6039 xmlRegAtomPtr atom;
6040 int counter;
6041
6042 if ((am == NULL) || (from == NULL) || (token == NULL))
6043 return(NULL);
6044 if (min < 0)
6045 return(NULL);
6046 if ((max < min) || (max < 1))
6047 return(NULL);
6048 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6049 if (atom == NULL)
6050 return(NULL);
6051 if ((token2 == NULL) || (*token2 == 0)) {
6052 atom->valuep = xmlStrdup(token);
6053 if (atom->valuep == NULL)
6054 goto error;
6055 } else {
6056 int lenn, lenp;
6057 xmlChar *str;
6058
6059 lenn = strlen((char *) token2);
6060 lenp = strlen((char *) token);
6061
6062 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
6063 if (str == NULL)
6064 goto error;
6065 memcpy(&str[0], token, lenp);
6066 str[lenp] = '|';
6067 memcpy(&str[lenp + 1], token2, lenn);
6068 str[lenn + lenp + 1] = 0;
6069
6070 atom->valuep = str;
6071 }
6072 atom->data = data;
6073 if (min == 0)
6074 atom->min = 1;
6075 else
6076 atom->min = min;
6077 atom->max = max;
6078
6079 /*
6080 * associate a counter to the transition.
6081 */
6082 counter = xmlRegGetCounter(am);
6083 if (counter < 0)
6084 goto error;
6085 am->counters[counter].min = min;
6086 am->counters[counter].max = max;
6087
6088 /* xmlFAGenerateTransitions(am, from, to, atom); */
6089 if (to == NULL) {
6090 to = xmlRegStatePush(am);
6091 if (to == NULL)
6092 goto error;
6093 }
6094 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6095 if (xmlRegAtomPush(am, atom) < 0)
6096 goto error;
6097 am->state = to;
6098
6099 if (to == NULL)
6100 to = am->state;
6101 if (to == NULL)
6102 return(NULL);
6103 if (min == 0)
6104 xmlFAGenerateEpsilonTransition(am, from, to);
6105 return(to);
6106
6107 error:
6108 xmlRegFreeAtom(atom);
6109 return(NULL);
6110 }
6111
6112 /**
6113 * xmlAutomataNewCountTrans:
6114 * @am: an automata
6115 * @from: the starting point of the transition
6116 * @to: the target point of the transition or NULL
6117 * @token: the input string associated to that transition
6118 * @min: the minimum successive occurrences of token
6119 * @max: the maximum successive occurrences of token
6120 * @data: data associated to the transition
6121 *
6122 * If @to is NULL, this creates first a new target state in the automata
6123 * and then adds a transition from the @from state to the target state
6124 * activated by a succession of input of value @token and whose number
6125 * is between @min and @max
6126 *
6127 * Returns the target state or NULL in case of error
6128 */
6129 xmlAutomataStatePtr
xmlAutomataNewCountTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,int min,int max,void * data)6130 xmlAutomataNewCountTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6131 xmlAutomataStatePtr to, const xmlChar *token,
6132 int min, int max, void *data) {
6133 xmlRegAtomPtr atom;
6134 int counter;
6135
6136 if ((am == NULL) || (from == NULL) || (token == NULL))
6137 return(NULL);
6138 if (min < 0)
6139 return(NULL);
6140 if ((max < min) || (max < 1))
6141 return(NULL);
6142 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6143 if (atom == NULL)
6144 return(NULL);
6145 atom->valuep = xmlStrdup(token);
6146 if (atom->valuep == NULL)
6147 goto error;
6148 atom->data = data;
6149 if (min == 0)
6150 atom->min = 1;
6151 else
6152 atom->min = min;
6153 atom->max = max;
6154
6155 /*
6156 * associate a counter to the transition.
6157 */
6158 counter = xmlRegGetCounter(am);
6159 if (counter < 0)
6160 goto error;
6161 am->counters[counter].min = min;
6162 am->counters[counter].max = max;
6163
6164 /* xmlFAGenerateTransitions(am, from, to, atom); */
6165 if (to == NULL) {
6166 to = xmlRegStatePush(am);
6167 if (to == NULL)
6168 goto error;
6169 }
6170 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6171 if (xmlRegAtomPush(am, atom) < 0)
6172 goto error;
6173 am->state = to;
6174
6175 if (to == NULL)
6176 to = am->state;
6177 if (to == NULL)
6178 return(NULL);
6179 if (min == 0)
6180 xmlFAGenerateEpsilonTransition(am, from, to);
6181 return(to);
6182
6183 error:
6184 xmlRegFreeAtom(atom);
6185 return(NULL);
6186 }
6187
6188 /**
6189 * xmlAutomataNewOnceTrans2:
6190 * @am: an automata
6191 * @from: the starting point of the transition
6192 * @to: the target point of the transition or NULL
6193 * @token: the input string associated to that transition
6194 * @token2: the second input string associated to that transition
6195 * @min: the minimum successive occurrences of token
6196 * @max: the maximum successive occurrences of token
6197 * @data: data associated to the transition
6198 *
6199 * If @to is NULL, this creates first a new target state in the automata
6200 * and then adds a transition from the @from state to the target state
6201 * activated by a succession of input of value @token and @token2 and whose
6202 * number is between @min and @max, moreover that transition can only be
6203 * crossed once.
6204 *
6205 * Returns the target state or NULL in case of error
6206 */
6207 xmlAutomataStatePtr
xmlAutomataNewOnceTrans2(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,const xmlChar * token2,int min,int max,void * data)6208 xmlAutomataNewOnceTrans2(xmlAutomataPtr am, xmlAutomataStatePtr from,
6209 xmlAutomataStatePtr to, const xmlChar *token,
6210 const xmlChar *token2,
6211 int min, int max, void *data) {
6212 xmlRegAtomPtr atom;
6213 int counter;
6214
6215 if ((am == NULL) || (from == NULL) || (token == NULL))
6216 return(NULL);
6217 if (min < 1)
6218 return(NULL);
6219 if (max < min)
6220 return(NULL);
6221 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6222 if (atom == NULL)
6223 return(NULL);
6224 if ((token2 == NULL) || (*token2 == 0)) {
6225 atom->valuep = xmlStrdup(token);
6226 if (atom->valuep == NULL)
6227 goto error;
6228 } else {
6229 int lenn, lenp;
6230 xmlChar *str;
6231
6232 lenn = strlen((char *) token2);
6233 lenp = strlen((char *) token);
6234
6235 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
6236 if (str == NULL)
6237 goto error;
6238 memcpy(&str[0], token, lenp);
6239 str[lenp] = '|';
6240 memcpy(&str[lenp + 1], token2, lenn);
6241 str[lenn + lenp + 1] = 0;
6242
6243 atom->valuep = str;
6244 }
6245 atom->data = data;
6246 atom->quant = XML_REGEXP_QUANT_ONCEONLY;
6247 atom->min = min;
6248 atom->max = max;
6249 /*
6250 * associate a counter to the transition.
6251 */
6252 counter = xmlRegGetCounter(am);
6253 if (counter < 0)
6254 goto error;
6255 am->counters[counter].min = 1;
6256 am->counters[counter].max = 1;
6257
6258 /* xmlFAGenerateTransitions(am, from, to, atom); */
6259 if (to == NULL) {
6260 to = xmlRegStatePush(am);
6261 if (to == NULL)
6262 goto error;
6263 }
6264 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6265 if (xmlRegAtomPush(am, atom) < 0)
6266 goto error;
6267 am->state = to;
6268 return(to);
6269
6270 error:
6271 xmlRegFreeAtom(atom);
6272 return(NULL);
6273 }
6274
6275
6276
6277 /**
6278 * xmlAutomataNewOnceTrans:
6279 * @am: an automata
6280 * @from: the starting point of the transition
6281 * @to: the target point of the transition or NULL
6282 * @token: the input string associated to that transition
6283 * @min: the minimum successive occurrences of token
6284 * @max: the maximum successive occurrences of token
6285 * @data: data associated to the transition
6286 *
6287 * If @to is NULL, this creates first a new target state in the automata
6288 * and then adds a transition from the @from state to the target state
6289 * activated by a succession of input of value @token and whose number
6290 * is between @min and @max, moreover that transition can only be crossed
6291 * once.
6292 *
6293 * Returns the target state or NULL in case of error
6294 */
6295 xmlAutomataStatePtr
xmlAutomataNewOnceTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,int min,int max,void * data)6296 xmlAutomataNewOnceTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6297 xmlAutomataStatePtr to, const xmlChar *token,
6298 int min, int max, void *data) {
6299 xmlRegAtomPtr atom;
6300 int counter;
6301
6302 if ((am == NULL) || (from == NULL) || (token == NULL))
6303 return(NULL);
6304 if (min < 1)
6305 return(NULL);
6306 if (max < min)
6307 return(NULL);
6308 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6309 if (atom == NULL)
6310 return(NULL);
6311 atom->valuep = xmlStrdup(token);
6312 atom->data = data;
6313 atom->quant = XML_REGEXP_QUANT_ONCEONLY;
6314 atom->min = min;
6315 atom->max = max;
6316 /*
6317 * associate a counter to the transition.
6318 */
6319 counter = xmlRegGetCounter(am);
6320 if (counter < 0)
6321 goto error;
6322 am->counters[counter].min = 1;
6323 am->counters[counter].max = 1;
6324
6325 /* xmlFAGenerateTransitions(am, from, to, atom); */
6326 if (to == NULL) {
6327 to = xmlRegStatePush(am);
6328 if (to == NULL)
6329 goto error;
6330 }
6331 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6332 if (xmlRegAtomPush(am, atom) < 0)
6333 goto error;
6334 am->state = to;
6335 return(to);
6336
6337 error:
6338 xmlRegFreeAtom(atom);
6339 return(NULL);
6340 }
6341
6342 /**
6343 * xmlAutomataNewState:
6344 * @am: an automata
6345 *
6346 * Create a new disconnected state in the automata
6347 *
6348 * Returns the new state or NULL in case of error
6349 */
6350 xmlAutomataStatePtr
xmlAutomataNewState(xmlAutomataPtr am)6351 xmlAutomataNewState(xmlAutomataPtr am) {
6352 if (am == NULL)
6353 return(NULL);
6354 return(xmlRegStatePush(am));
6355 }
6356
6357 /**
6358 * xmlAutomataNewEpsilon:
6359 * @am: an automata
6360 * @from: the starting point of the transition
6361 * @to: the target point of the transition or NULL
6362 *
6363 * If @to is NULL, this creates first a new target state in the automata
6364 * and then adds an epsilon transition from the @from state to the
6365 * target state
6366 *
6367 * Returns the target state or NULL in case of error
6368 */
6369 xmlAutomataStatePtr
xmlAutomataNewEpsilon(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to)6370 xmlAutomataNewEpsilon(xmlAutomataPtr am, xmlAutomataStatePtr from,
6371 xmlAutomataStatePtr to) {
6372 if ((am == NULL) || (from == NULL))
6373 return(NULL);
6374 xmlFAGenerateEpsilonTransition(am, from, to);
6375 if (to == NULL)
6376 return(am->state);
6377 return(to);
6378 }
6379
6380 /**
6381 * xmlAutomataNewAllTrans:
6382 * @am: an automata
6383 * @from: the starting point of the transition
6384 * @to: the target point of the transition or NULL
6385 * @lax: allow to transition if not all all transitions have been activated
6386 *
6387 * If @to is NULL, this creates first a new target state in the automata
6388 * and then adds a an ALL transition from the @from state to the
6389 * target state. That transition is an epsilon transition allowed only when
6390 * all transitions from the @from node have been activated.
6391 *
6392 * Returns the target state or NULL in case of error
6393 */
6394 xmlAutomataStatePtr
xmlAutomataNewAllTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,int lax)6395 xmlAutomataNewAllTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6396 xmlAutomataStatePtr to, int lax) {
6397 if ((am == NULL) || (from == NULL))
6398 return(NULL);
6399 xmlFAGenerateAllTransition(am, from, to, lax);
6400 if (to == NULL)
6401 return(am->state);
6402 return(to);
6403 }
6404
6405 /**
6406 * xmlAutomataNewCounter:
6407 * @am: an automata
6408 * @min: the minimal value on the counter
6409 * @max: the maximal value on the counter
6410 *
6411 * Create a new counter
6412 *
6413 * Returns the counter number or -1 in case of error
6414 */
6415 int
xmlAutomataNewCounter(xmlAutomataPtr am,int min,int max)6416 xmlAutomataNewCounter(xmlAutomataPtr am, int min, int max) {
6417 int ret;
6418
6419 if (am == NULL)
6420 return(-1);
6421
6422 ret = xmlRegGetCounter(am);
6423 if (ret < 0)
6424 return(-1);
6425 am->counters[ret].min = min;
6426 am->counters[ret].max = max;
6427 return(ret);
6428 }
6429
6430 /**
6431 * xmlAutomataNewCountedTrans:
6432 * @am: an automata
6433 * @from: the starting point of the transition
6434 * @to: the target point of the transition or NULL
6435 * @counter: the counter associated to that transition
6436 *
6437 * If @to is NULL, this creates first a new target state in the automata
6438 * and then adds an epsilon transition from the @from state to the target state
6439 * which will increment the counter provided
6440 *
6441 * Returns the target state or NULL in case of error
6442 */
6443 xmlAutomataStatePtr
xmlAutomataNewCountedTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,int counter)6444 xmlAutomataNewCountedTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6445 xmlAutomataStatePtr to, int counter) {
6446 if ((am == NULL) || (from == NULL) || (counter < 0))
6447 return(NULL);
6448 xmlFAGenerateCountedEpsilonTransition(am, from, to, counter);
6449 if (to == NULL)
6450 return(am->state);
6451 return(to);
6452 }
6453
6454 /**
6455 * xmlAutomataNewCounterTrans:
6456 * @am: an automata
6457 * @from: the starting point of the transition
6458 * @to: the target point of the transition or NULL
6459 * @counter: the counter associated to that transition
6460 *
6461 * If @to is NULL, this creates first a new target state in the automata
6462 * and then adds an epsilon transition from the @from state to the target state
6463 * which will be allowed only if the counter is within the right range.
6464 *
6465 * Returns the target state or NULL in case of error
6466 */
6467 xmlAutomataStatePtr
xmlAutomataNewCounterTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,int counter)6468 xmlAutomataNewCounterTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6469 xmlAutomataStatePtr to, int counter) {
6470 if ((am == NULL) || (from == NULL) || (counter < 0))
6471 return(NULL);
6472 xmlFAGenerateCountedTransition(am, from, to, counter);
6473 if (to == NULL)
6474 return(am->state);
6475 return(to);
6476 }
6477
6478 /**
6479 * xmlAutomataCompile:
6480 * @am: an automata
6481 *
6482 * Compile the automata into a Reg Exp ready for being executed.
6483 * The automata should be free after this point.
6484 *
6485 * Returns the compiled regexp or NULL in case of error
6486 */
6487 xmlRegexpPtr
xmlAutomataCompile(xmlAutomataPtr am)6488 xmlAutomataCompile(xmlAutomataPtr am) {
6489 xmlRegexpPtr ret;
6490
6491 if ((am == NULL) || (am->error != 0)) return(NULL);
6492 xmlFAEliminateEpsilonTransitions(am);
6493 /* xmlFAComputesDeterminism(am); */
6494 ret = xmlRegEpxFromParse(am);
6495
6496 return(ret);
6497 }
6498
6499 /**
6500 * xmlAutomataIsDeterminist:
6501 * @am: an automata
6502 *
6503 * Checks if an automata is determinist.
6504 *
6505 * Returns 1 if true, 0 if not, and -1 in case of error
6506 */
6507 int
xmlAutomataIsDeterminist(xmlAutomataPtr am)6508 xmlAutomataIsDeterminist(xmlAutomataPtr am) {
6509 int ret;
6510
6511 if (am == NULL)
6512 return(-1);
6513
6514 ret = xmlFAComputesDeterminism(am);
6515 return(ret);
6516 }
6517 #endif /* LIBXML_AUTOMATA_ENABLED */
6518
6519 #ifdef LIBXML_EXPR_ENABLED
6520 /************************************************************************
6521 * *
6522 * Formal Expression handling code *
6523 * *
6524 ************************************************************************/
6525 /************************************************************************
6526 * *
6527 * Expression handling context *
6528 * *
6529 ************************************************************************/
6530
6531 struct _xmlExpCtxt {
6532 xmlDictPtr dict;
6533 xmlExpNodePtr *table;
6534 int size;
6535 int nbElems;
6536 int nb_nodes;
6537 int maxNodes;
6538 const char *expr;
6539 const char *cur;
6540 int nb_cons;
6541 int tabSize;
6542 };
6543
6544 /**
6545 * xmlExpNewCtxt:
6546 * @maxNodes: the maximum number of nodes
6547 * @dict: optional dictionary to use internally
6548 *
6549 * Creates a new context for manipulating expressions
6550 *
6551 * Returns the context or NULL in case of error
6552 */
6553 xmlExpCtxtPtr
xmlExpNewCtxt(int maxNodes,xmlDictPtr dict)6554 xmlExpNewCtxt(int maxNodes, xmlDictPtr dict) {
6555 xmlExpCtxtPtr ret;
6556 int size = 256;
6557
6558 if (maxNodes <= 4096)
6559 maxNodes = 4096;
6560
6561 ret = (xmlExpCtxtPtr) xmlMalloc(sizeof(xmlExpCtxt));
6562 if (ret == NULL)
6563 return(NULL);
6564 memset(ret, 0, sizeof(xmlExpCtxt));
6565 ret->size = size;
6566 ret->nbElems = 0;
6567 ret->maxNodes = maxNodes;
6568 ret->table = xmlMalloc(size * sizeof(xmlExpNodePtr));
6569 if (ret->table == NULL) {
6570 xmlFree(ret);
6571 return(NULL);
6572 }
6573 memset(ret->table, 0, size * sizeof(xmlExpNodePtr));
6574 if (dict == NULL) {
6575 ret->dict = xmlDictCreate();
6576 if (ret->dict == NULL) {
6577 xmlFree(ret->table);
6578 xmlFree(ret);
6579 return(NULL);
6580 }
6581 } else {
6582 ret->dict = dict;
6583 xmlDictReference(ret->dict);
6584 }
6585 return(ret);
6586 }
6587
6588 /**
6589 * xmlExpFreeCtxt:
6590 * @ctxt: an expression context
6591 *
6592 * Free an expression context
6593 */
6594 void
xmlExpFreeCtxt(xmlExpCtxtPtr ctxt)6595 xmlExpFreeCtxt(xmlExpCtxtPtr ctxt) {
6596 if (ctxt == NULL)
6597 return;
6598 xmlDictFree(ctxt->dict);
6599 if (ctxt->table != NULL)
6600 xmlFree(ctxt->table);
6601 xmlFree(ctxt);
6602 }
6603
6604 /************************************************************************
6605 * *
6606 * Structure associated to an expression node *
6607 * *
6608 ************************************************************************/
6609 #define MAX_NODES 10000
6610
6611 /* #define DEBUG_DERIV */
6612
6613 /*
6614 * TODO:
6615 * - Wildcards
6616 * - public API for creation
6617 *
6618 * Started
6619 * - regression testing
6620 *
6621 * Done
6622 * - split into module and test tool
6623 * - memleaks
6624 */
6625
6626 typedef enum {
6627 XML_EXP_NILABLE = (1 << 0)
6628 } xmlExpNodeInfo;
6629
6630 #define IS_NILLABLE(node) ((node)->info & XML_EXP_NILABLE)
6631
6632 struct _xmlExpNode {
6633 unsigned char type;/* xmlExpNodeType */
6634 unsigned char info;/* OR of xmlExpNodeInfo */
6635 unsigned short key; /* the hash key */
6636 unsigned int ref; /* The number of references */
6637 int c_max; /* the maximum length it can consume */
6638 xmlExpNodePtr exp_left;
6639 xmlExpNodePtr next;/* the next node in the hash table or free list */
6640 union {
6641 struct {
6642 int f_min;
6643 int f_max;
6644 } count;
6645 struct {
6646 xmlExpNodePtr f_right;
6647 } children;
6648 const xmlChar *f_str;
6649 } field;
6650 };
6651
6652 #define exp_min field.count.f_min
6653 #define exp_max field.count.f_max
6654 /* #define exp_left field.children.f_left */
6655 #define exp_right field.children.f_right
6656 #define exp_str field.f_str
6657
6658 static xmlExpNodePtr xmlExpNewNode(xmlExpCtxtPtr ctxt, xmlExpNodeType type);
6659 static xmlExpNode forbiddenExpNode = {
6660 XML_EXP_FORBID, 0, 0, 0, 0, NULL, NULL, {{ 0, 0}}
6661 };
6662 xmlExpNodePtr forbiddenExp = &forbiddenExpNode;
6663 static xmlExpNode emptyExpNode = {
6664 XML_EXP_EMPTY, 1, 0, 0, 0, NULL, NULL, {{ 0, 0}}
6665 };
6666 xmlExpNodePtr emptyExp = &emptyExpNode;
6667
6668 /************************************************************************
6669 * *
6670 * The custom hash table for unicity and canonicalization *
6671 * of sub-expressions pointers *
6672 * *
6673 ************************************************************************/
6674 /*
6675 * xmlExpHashNameComputeKey:
6676 * Calculate the hash key for a token
6677 */
6678 static unsigned short
xmlExpHashNameComputeKey(const xmlChar * name)6679 xmlExpHashNameComputeKey(const xmlChar *name) {
6680 unsigned short value = 0L;
6681 char ch;
6682
6683 if (name != NULL) {
6684 value += 30 * (*name);
6685 while ((ch = *name++) != 0) {
6686 value = value ^ ((value << 5) + (value >> 3) + (unsigned long)ch);
6687 }
6688 }
6689 return (value);
6690 }
6691
6692 /*
6693 * xmlExpHashComputeKey:
6694 * Calculate the hash key for a compound expression
6695 */
6696 static unsigned short
xmlExpHashComputeKey(xmlExpNodeType type,xmlExpNodePtr left,xmlExpNodePtr right)6697 xmlExpHashComputeKey(xmlExpNodeType type, xmlExpNodePtr left,
6698 xmlExpNodePtr right) {
6699 unsigned long value;
6700 unsigned short ret;
6701
6702 switch (type) {
6703 case XML_EXP_SEQ:
6704 value = left->key;
6705 value += right->key;
6706 value *= 3;
6707 ret = (unsigned short) value;
6708 break;
6709 case XML_EXP_OR:
6710 value = left->key;
6711 value += right->key;
6712 value *= 7;
6713 ret = (unsigned short) value;
6714 break;
6715 case XML_EXP_COUNT:
6716 value = left->key;
6717 value += right->key;
6718 ret = (unsigned short) value;
6719 break;
6720 default:
6721 ret = 0;
6722 }
6723 return(ret);
6724 }
6725
6726
6727 static xmlExpNodePtr
xmlExpNewNode(xmlExpCtxtPtr ctxt,xmlExpNodeType type)6728 xmlExpNewNode(xmlExpCtxtPtr ctxt, xmlExpNodeType type) {
6729 xmlExpNodePtr ret;
6730
6731 if (ctxt->nb_nodes >= MAX_NODES)
6732 return(NULL);
6733 ret = (xmlExpNodePtr) xmlMalloc(sizeof(xmlExpNode));
6734 if (ret == NULL)
6735 return(NULL);
6736 memset(ret, 0, sizeof(xmlExpNode));
6737 ret->type = type;
6738 ret->next = NULL;
6739 ctxt->nb_nodes++;
6740 ctxt->nb_cons++;
6741 return(ret);
6742 }
6743
6744 /**
6745 * xmlExpHashGetEntry:
6746 * @table: the hash table
6747 *
6748 * Get the unique entry from the hash table. The entry is created if
6749 * needed. @left and @right are consumed, i.e. their ref count will
6750 * be decremented by the operation.
6751 *
6752 * Returns the pointer or NULL in case of error
6753 */
6754 static xmlExpNodePtr
xmlExpHashGetEntry(xmlExpCtxtPtr ctxt,xmlExpNodeType type,xmlExpNodePtr left,xmlExpNodePtr right,const xmlChar * name,int min,int max)6755 xmlExpHashGetEntry(xmlExpCtxtPtr ctxt, xmlExpNodeType type,
6756 xmlExpNodePtr left, xmlExpNodePtr right,
6757 const xmlChar *name, int min, int max) {
6758 unsigned short kbase, key;
6759 xmlExpNodePtr entry;
6760 xmlExpNodePtr insert;
6761
6762 if (ctxt == NULL)
6763 return(NULL);
6764
6765 /*
6766 * Check for duplicate and insertion location.
6767 */
6768 if (type == XML_EXP_ATOM) {
6769 kbase = xmlExpHashNameComputeKey(name);
6770 } else if (type == XML_EXP_COUNT) {
6771 /* COUNT reduction rule 1 */
6772 /* a{1} -> a */
6773 if (min == max) {
6774 if (min == 1) {
6775 return(left);
6776 }
6777 if (min == 0) {
6778 xmlExpFree(ctxt, left);
6779 return(emptyExp);
6780 }
6781 }
6782 if (min < 0) {
6783 xmlExpFree(ctxt, left);
6784 return(forbiddenExp);
6785 }
6786 if (max == -1)
6787 kbase = min + 79;
6788 else
6789 kbase = max - min;
6790 kbase += left->key;
6791 } else if (type == XML_EXP_OR) {
6792 /* Forbid reduction rules */
6793 if (left->type == XML_EXP_FORBID) {
6794 xmlExpFree(ctxt, left);
6795 return(right);
6796 }
6797 if (right->type == XML_EXP_FORBID) {
6798 xmlExpFree(ctxt, right);
6799 return(left);
6800 }
6801
6802 /* OR reduction rule 1 */
6803 /* a | a reduced to a */
6804 if (left == right) {
6805 xmlExpFree(ctxt, right);
6806 return(left);
6807 }
6808 /* OR canonicalization rule 1 */
6809 /* linearize (a | b) | c into a | (b | c) */
6810 if ((left->type == XML_EXP_OR) && (right->type != XML_EXP_OR)) {
6811 xmlExpNodePtr tmp = left;
6812 left = right;
6813 right = tmp;
6814 }
6815 /* OR reduction rule 2 */
6816 /* a | (a | b) and b | (a | b) are reduced to a | b */
6817 if (right->type == XML_EXP_OR) {
6818 if ((left == right->exp_left) ||
6819 (left == right->exp_right)) {
6820 xmlExpFree(ctxt, left);
6821 return(right);
6822 }
6823 }
6824 /* OR canonicalization rule 2 */
6825 /* linearize (a | b) | c into a | (b | c) */
6826 if (left->type == XML_EXP_OR) {
6827 xmlExpNodePtr tmp;
6828
6829 /* OR canonicalization rule 2 */
6830 if ((left->exp_right->type != XML_EXP_OR) &&
6831 (left->exp_right->key < left->exp_left->key)) {
6832 tmp = left->exp_right;
6833 left->exp_right = left->exp_left;
6834 left->exp_left = tmp;
6835 }
6836 left->exp_right->ref++;
6837 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left->exp_right, right,
6838 NULL, 0, 0);
6839 left->exp_left->ref++;
6840 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left->exp_left, tmp,
6841 NULL, 0, 0);
6842
6843 xmlExpFree(ctxt, left);
6844 return(tmp);
6845 }
6846 if (right->type == XML_EXP_OR) {
6847 /* Ordering in the tree */
6848 /* C | (A | B) -> A | (B | C) */
6849 if (left->key > right->exp_right->key) {
6850 xmlExpNodePtr tmp;
6851 right->exp_right->ref++;
6852 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_right,
6853 left, NULL, 0, 0);
6854 right->exp_left->ref++;
6855 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_left,
6856 tmp, NULL, 0, 0);
6857 xmlExpFree(ctxt, right);
6858 return(tmp);
6859 }
6860 /* Ordering in the tree */
6861 /* B | (A | C) -> A | (B | C) */
6862 if (left->key > right->exp_left->key) {
6863 xmlExpNodePtr tmp;
6864 right->exp_right->ref++;
6865 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left,
6866 right->exp_right, NULL, 0, 0);
6867 right->exp_left->ref++;
6868 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_left,
6869 tmp, NULL, 0, 0);
6870 xmlExpFree(ctxt, right);
6871 return(tmp);
6872 }
6873 }
6874 /* we know both types are != XML_EXP_OR here */
6875 else if (left->key > right->key) {
6876 xmlExpNodePtr tmp = left;
6877 left = right;
6878 right = tmp;
6879 }
6880 kbase = xmlExpHashComputeKey(type, left, right);
6881 } else if (type == XML_EXP_SEQ) {
6882 /* Forbid reduction rules */
6883 if (left->type == XML_EXP_FORBID) {
6884 xmlExpFree(ctxt, right);
6885 return(left);
6886 }
6887 if (right->type == XML_EXP_FORBID) {
6888 xmlExpFree(ctxt, left);
6889 return(right);
6890 }
6891 /* Empty reduction rules */
6892 if (right->type == XML_EXP_EMPTY) {
6893 return(left);
6894 }
6895 if (left->type == XML_EXP_EMPTY) {
6896 return(right);
6897 }
6898 kbase = xmlExpHashComputeKey(type, left, right);
6899 } else
6900 return(NULL);
6901
6902 key = kbase % ctxt->size;
6903 if (ctxt->table[key] != NULL) {
6904 for (insert = ctxt->table[key]; insert != NULL;
6905 insert = insert->next) {
6906 if ((insert->key == kbase) &&
6907 (insert->type == type)) {
6908 if (type == XML_EXP_ATOM) {
6909 if (name == insert->exp_str) {
6910 insert->ref++;
6911 return(insert);
6912 }
6913 } else if (type == XML_EXP_COUNT) {
6914 if ((insert->exp_min == min) && (insert->exp_max == max) &&
6915 (insert->exp_left == left)) {
6916 insert->ref++;
6917 left->ref--;
6918 return(insert);
6919 }
6920 } else if ((insert->exp_left == left) &&
6921 (insert->exp_right == right)) {
6922 insert->ref++;
6923 left->ref--;
6924 right->ref--;
6925 return(insert);
6926 }
6927 }
6928 }
6929 }
6930
6931 entry = xmlExpNewNode(ctxt, type);
6932 if (entry == NULL)
6933 return(NULL);
6934 entry->key = kbase;
6935 if (type == XML_EXP_ATOM) {
6936 entry->exp_str = name;
6937 entry->c_max = 1;
6938 } else if (type == XML_EXP_COUNT) {
6939 entry->exp_min = min;
6940 entry->exp_max = max;
6941 entry->exp_left = left;
6942 if ((min == 0) || (IS_NILLABLE(left)))
6943 entry->info |= XML_EXP_NILABLE;
6944 if (max < 0)
6945 entry->c_max = -1;
6946 else
6947 entry->c_max = max * entry->exp_left->c_max;
6948 } else {
6949 entry->exp_left = left;
6950 entry->exp_right = right;
6951 if (type == XML_EXP_OR) {
6952 if ((IS_NILLABLE(left)) || (IS_NILLABLE(right)))
6953 entry->info |= XML_EXP_NILABLE;
6954 if ((entry->exp_left->c_max == -1) ||
6955 (entry->exp_right->c_max == -1))
6956 entry->c_max = -1;
6957 else if (entry->exp_left->c_max > entry->exp_right->c_max)
6958 entry->c_max = entry->exp_left->c_max;
6959 else
6960 entry->c_max = entry->exp_right->c_max;
6961 } else {
6962 if ((IS_NILLABLE(left)) && (IS_NILLABLE(right)))
6963 entry->info |= XML_EXP_NILABLE;
6964 if ((entry->exp_left->c_max == -1) ||
6965 (entry->exp_right->c_max == -1))
6966 entry->c_max = -1;
6967 else
6968 entry->c_max = entry->exp_left->c_max + entry->exp_right->c_max;
6969 }
6970 }
6971 entry->ref = 1;
6972 if (ctxt->table[key] != NULL)
6973 entry->next = ctxt->table[key];
6974
6975 ctxt->table[key] = entry;
6976 ctxt->nbElems++;
6977
6978 return(entry);
6979 }
6980
6981 /**
6982 * xmlExpFree:
6983 * @ctxt: the expression context
6984 * @exp: the expression
6985 *
6986 * Dereference the expression
6987 */
6988 void
xmlExpFree(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp)6989 xmlExpFree(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp) {
6990 if ((exp == NULL) || (exp == forbiddenExp) || (exp == emptyExp))
6991 return;
6992 exp->ref--;
6993 if (exp->ref == 0) {
6994 unsigned short key;
6995
6996 /* Unlink it first from the hash table */
6997 key = exp->key % ctxt->size;
6998 if (ctxt->table[key] == exp) {
6999 ctxt->table[key] = exp->next;
7000 } else {
7001 xmlExpNodePtr tmp;
7002
7003 tmp = ctxt->table[key];
7004 while (tmp != NULL) {
7005 if (tmp->next == exp) {
7006 tmp->next = exp->next;
7007 break;
7008 }
7009 tmp = tmp->next;
7010 }
7011 }
7012
7013 if ((exp->type == XML_EXP_SEQ) || (exp->type == XML_EXP_OR)) {
7014 xmlExpFree(ctxt, exp->exp_left);
7015 xmlExpFree(ctxt, exp->exp_right);
7016 } else if (exp->type == XML_EXP_COUNT) {
7017 xmlExpFree(ctxt, exp->exp_left);
7018 }
7019 xmlFree(exp);
7020 ctxt->nb_nodes--;
7021 }
7022 }
7023
7024 /**
7025 * xmlExpRef:
7026 * @exp: the expression
7027 *
7028 * Increase the reference count of the expression
7029 */
7030 void
xmlExpRef(xmlExpNodePtr exp)7031 xmlExpRef(xmlExpNodePtr exp) {
7032 if (exp != NULL)
7033 exp->ref++;
7034 }
7035
7036 /**
7037 * xmlExpNewAtom:
7038 * @ctxt: the expression context
7039 * @name: the atom name
7040 * @len: the atom name length in byte (or -1);
7041 *
7042 * Get the atom associated to this name from that context
7043 *
7044 * Returns the node or NULL in case of error
7045 */
7046 xmlExpNodePtr
xmlExpNewAtom(xmlExpCtxtPtr ctxt,const xmlChar * name,int len)7047 xmlExpNewAtom(xmlExpCtxtPtr ctxt, const xmlChar *name, int len) {
7048 if ((ctxt == NULL) || (name == NULL))
7049 return(NULL);
7050 name = xmlDictLookup(ctxt->dict, name, len);
7051 if (name == NULL)
7052 return(NULL);
7053 return(xmlExpHashGetEntry(ctxt, XML_EXP_ATOM, NULL, NULL, name, 0, 0));
7054 }
7055
7056 /**
7057 * xmlExpNewOr:
7058 * @ctxt: the expression context
7059 * @left: left expression
7060 * @right: right expression
7061 *
7062 * Get the atom associated to the choice @left | @right
7063 * Note that @left and @right are consumed in the operation, to keep
7064 * an handle on them use xmlExpRef() and use xmlExpFree() to release them,
7065 * this is true even in case of failure (unless ctxt == NULL).
7066 *
7067 * Returns the node or NULL in case of error
7068 */
7069 xmlExpNodePtr
xmlExpNewOr(xmlExpCtxtPtr ctxt,xmlExpNodePtr left,xmlExpNodePtr right)7070 xmlExpNewOr(xmlExpCtxtPtr ctxt, xmlExpNodePtr left, xmlExpNodePtr right) {
7071 if (ctxt == NULL)
7072 return(NULL);
7073 if ((left == NULL) || (right == NULL)) {
7074 xmlExpFree(ctxt, left);
7075 xmlExpFree(ctxt, right);
7076 return(NULL);
7077 }
7078 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, left, right, NULL, 0, 0));
7079 }
7080
7081 /**
7082 * xmlExpNewSeq:
7083 * @ctxt: the expression context
7084 * @left: left expression
7085 * @right: right expression
7086 *
7087 * Get the atom associated to the sequence @left , @right
7088 * Note that @left and @right are consumed in the operation, to keep
7089 * an handle on them use xmlExpRef() and use xmlExpFree() to release them,
7090 * this is true even in case of failure (unless ctxt == NULL).
7091 *
7092 * Returns the node or NULL in case of error
7093 */
7094 xmlExpNodePtr
xmlExpNewSeq(xmlExpCtxtPtr ctxt,xmlExpNodePtr left,xmlExpNodePtr right)7095 xmlExpNewSeq(xmlExpCtxtPtr ctxt, xmlExpNodePtr left, xmlExpNodePtr right) {
7096 if (ctxt == NULL)
7097 return(NULL);
7098 if ((left == NULL) || (right == NULL)) {
7099 xmlExpFree(ctxt, left);
7100 xmlExpFree(ctxt, right);
7101 return(NULL);
7102 }
7103 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, left, right, NULL, 0, 0));
7104 }
7105
7106 /**
7107 * xmlExpNewRange:
7108 * @ctxt: the expression context
7109 * @subset: the expression to be repeated
7110 * @min: the lower bound for the repetition
7111 * @max: the upper bound for the repetition, -1 means infinite
7112 *
7113 * Get the atom associated to the range (@subset){@min, @max}
7114 * Note that @subset is consumed in the operation, to keep
7115 * an handle on it use xmlExpRef() and use xmlExpFree() to release it,
7116 * this is true even in case of failure (unless ctxt == NULL).
7117 *
7118 * Returns the node or NULL in case of error
7119 */
7120 xmlExpNodePtr
xmlExpNewRange(xmlExpCtxtPtr ctxt,xmlExpNodePtr subset,int min,int max)7121 xmlExpNewRange(xmlExpCtxtPtr ctxt, xmlExpNodePtr subset, int min, int max) {
7122 if (ctxt == NULL)
7123 return(NULL);
7124 if ((subset == NULL) || (min < 0) || (max < -1) ||
7125 ((max >= 0) && (min > max))) {
7126 xmlExpFree(ctxt, subset);
7127 return(NULL);
7128 }
7129 return(xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, subset,
7130 NULL, NULL, min, max));
7131 }
7132
7133 /************************************************************************
7134 * *
7135 * Public API for operations on expressions *
7136 * *
7137 ************************************************************************/
7138
7139 static int
xmlExpGetLanguageInt(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar ** list,int len,int nb)7140 xmlExpGetLanguageInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7141 const xmlChar**list, int len, int nb) {
7142 int tmp, tmp2;
7143 tail:
7144 switch (exp->type) {
7145 case XML_EXP_EMPTY:
7146 return(0);
7147 case XML_EXP_ATOM:
7148 for (tmp = 0;tmp < nb;tmp++)
7149 if (list[tmp] == exp->exp_str)
7150 return(0);
7151 if (nb >= len)
7152 return(-2);
7153 list[nb] = exp->exp_str;
7154 return(1);
7155 case XML_EXP_COUNT:
7156 exp = exp->exp_left;
7157 goto tail;
7158 case XML_EXP_SEQ:
7159 case XML_EXP_OR:
7160 tmp = xmlExpGetLanguageInt(ctxt, exp->exp_left, list, len, nb);
7161 if (tmp < 0)
7162 return(tmp);
7163 tmp2 = xmlExpGetLanguageInt(ctxt, exp->exp_right, list, len,
7164 nb + tmp);
7165 if (tmp2 < 0)
7166 return(tmp2);
7167 return(tmp + tmp2);
7168 }
7169 return(-1);
7170 }
7171
7172 /**
7173 * xmlExpGetLanguage:
7174 * @ctxt: the expression context
7175 * @exp: the expression
7176 * @langList: where to store the tokens
7177 * @len: the allocated length of @list
7178 *
7179 * Find all the strings used in @exp and store them in @list
7180 *
7181 * Returns the number of unique strings found, -1 in case of errors and
7182 * -2 if there is more than @len strings
7183 */
7184 int
xmlExpGetLanguage(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar ** langList,int len)7185 xmlExpGetLanguage(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7186 const xmlChar**langList, int len) {
7187 if ((ctxt == NULL) || (exp == NULL) || (langList == NULL) || (len <= 0))
7188 return(-1);
7189 return(xmlExpGetLanguageInt(ctxt, exp, langList, len, 0));
7190 }
7191
7192 static int
xmlExpGetStartInt(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar ** list,int len,int nb)7193 xmlExpGetStartInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7194 const xmlChar**list, int len, int nb) {
7195 int tmp, tmp2;
7196 tail:
7197 switch (exp->type) {
7198 case XML_EXP_FORBID:
7199 return(0);
7200 case XML_EXP_EMPTY:
7201 return(0);
7202 case XML_EXP_ATOM:
7203 for (tmp = 0;tmp < nb;tmp++)
7204 if (list[tmp] == exp->exp_str)
7205 return(0);
7206 if (nb >= len)
7207 return(-2);
7208 list[nb] = exp->exp_str;
7209 return(1);
7210 case XML_EXP_COUNT:
7211 exp = exp->exp_left;
7212 goto tail;
7213 case XML_EXP_SEQ:
7214 tmp = xmlExpGetStartInt(ctxt, exp->exp_left, list, len, nb);
7215 if (tmp < 0)
7216 return(tmp);
7217 if (IS_NILLABLE(exp->exp_left)) {
7218 tmp2 = xmlExpGetStartInt(ctxt, exp->exp_right, list, len,
7219 nb + tmp);
7220 if (tmp2 < 0)
7221 return(tmp2);
7222 tmp += tmp2;
7223 }
7224 return(tmp);
7225 case XML_EXP_OR:
7226 tmp = xmlExpGetStartInt(ctxt, exp->exp_left, list, len, nb);
7227 if (tmp < 0)
7228 return(tmp);
7229 tmp2 = xmlExpGetStartInt(ctxt, exp->exp_right, list, len,
7230 nb + tmp);
7231 if (tmp2 < 0)
7232 return(tmp2);
7233 return(tmp + tmp2);
7234 }
7235 return(-1);
7236 }
7237
7238 /**
7239 * xmlExpGetStart:
7240 * @ctxt: the expression context
7241 * @exp: the expression
7242 * @tokList: where to store the tokens
7243 * @len: the allocated length of @list
7244 *
7245 * Find all the strings that appears at the start of the languages
7246 * accepted by @exp and store them in @list. E.g. for (a, b) | c
7247 * it will return the list [a, c]
7248 *
7249 * Returns the number of unique strings found, -1 in case of errors and
7250 * -2 if there is more than @len strings
7251 */
7252 int
xmlExpGetStart(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar ** tokList,int len)7253 xmlExpGetStart(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7254 const xmlChar**tokList, int len) {
7255 if ((ctxt == NULL) || (exp == NULL) || (tokList == NULL) || (len <= 0))
7256 return(-1);
7257 return(xmlExpGetStartInt(ctxt, exp, tokList, len, 0));
7258 }
7259
7260 /**
7261 * xmlExpIsNillable:
7262 * @exp: the expression
7263 *
7264 * Finds if the expression is nillable, i.e. if it accepts the empty sequence
7265 *
7266 * Returns 1 if nillable, 0 if not and -1 in case of error
7267 */
7268 int
xmlExpIsNillable(xmlExpNodePtr exp)7269 xmlExpIsNillable(xmlExpNodePtr exp) {
7270 if (exp == NULL)
7271 return(-1);
7272 return(IS_NILLABLE(exp) != 0);
7273 }
7274
7275 static xmlExpNodePtr
xmlExpStringDeriveInt(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar * str)7276 xmlExpStringDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, const xmlChar *str)
7277 {
7278 xmlExpNodePtr ret;
7279
7280 switch (exp->type) {
7281 case XML_EXP_EMPTY:
7282 return(forbiddenExp);
7283 case XML_EXP_FORBID:
7284 return(forbiddenExp);
7285 case XML_EXP_ATOM:
7286 if (exp->exp_str == str) {
7287 #ifdef DEBUG_DERIV
7288 printf("deriv atom: equal => Empty\n");
7289 #endif
7290 ret = emptyExp;
7291 } else {
7292 #ifdef DEBUG_DERIV
7293 printf("deriv atom: mismatch => forbid\n");
7294 #endif
7295 /* TODO wildcards here */
7296 ret = forbiddenExp;
7297 }
7298 return(ret);
7299 case XML_EXP_OR: {
7300 xmlExpNodePtr tmp;
7301
7302 #ifdef DEBUG_DERIV
7303 printf("deriv or: => or(derivs)\n");
7304 #endif
7305 tmp = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7306 if (tmp == NULL) {
7307 return(NULL);
7308 }
7309 ret = xmlExpStringDeriveInt(ctxt, exp->exp_right, str);
7310 if (ret == NULL) {
7311 xmlExpFree(ctxt, tmp);
7312 return(NULL);
7313 }
7314 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, tmp, ret,
7315 NULL, 0, 0);
7316 return(ret);
7317 }
7318 case XML_EXP_SEQ:
7319 #ifdef DEBUG_DERIV
7320 printf("deriv seq: starting with left\n");
7321 #endif
7322 ret = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7323 if (ret == NULL) {
7324 return(NULL);
7325 } else if (ret == forbiddenExp) {
7326 if (IS_NILLABLE(exp->exp_left)) {
7327 #ifdef DEBUG_DERIV
7328 printf("deriv seq: left failed but nillable\n");
7329 #endif
7330 ret = xmlExpStringDeriveInt(ctxt, exp->exp_right, str);
7331 }
7332 } else {
7333 #ifdef DEBUG_DERIV
7334 printf("deriv seq: left match => sequence\n");
7335 #endif
7336 exp->exp_right->ref++;
7337 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, exp->exp_right,
7338 NULL, 0, 0);
7339 }
7340 return(ret);
7341 case XML_EXP_COUNT: {
7342 int min, max;
7343 xmlExpNodePtr tmp;
7344
7345 if (exp->exp_max == 0)
7346 return(forbiddenExp);
7347 ret = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7348 if (ret == NULL)
7349 return(NULL);
7350 if (ret == forbiddenExp) {
7351 #ifdef DEBUG_DERIV
7352 printf("deriv count: pattern mismatch => forbid\n");
7353 #endif
7354 return(ret);
7355 }
7356 if (exp->exp_max == 1)
7357 return(ret);
7358 if (exp->exp_max < 0) /* unbounded */
7359 max = -1;
7360 else
7361 max = exp->exp_max - 1;
7362 if (exp->exp_min > 0)
7363 min = exp->exp_min - 1;
7364 else
7365 min = 0;
7366 exp->exp_left->ref++;
7367 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left, NULL,
7368 NULL, min, max);
7369 if (ret == emptyExp) {
7370 #ifdef DEBUG_DERIV
7371 printf("deriv count: match to empty => new count\n");
7372 #endif
7373 return(tmp);
7374 }
7375 #ifdef DEBUG_DERIV
7376 printf("deriv count: match => sequence with new count\n");
7377 #endif
7378 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, tmp,
7379 NULL, 0, 0));
7380 }
7381 }
7382 return(NULL);
7383 }
7384
7385 /**
7386 * xmlExpStringDerive:
7387 * @ctxt: the expression context
7388 * @exp: the expression
7389 * @str: the string
7390 * @len: the string len in bytes if available
7391 *
7392 * Do one step of Brzozowski derivation of the expression @exp with
7393 * respect to the input string
7394 *
7395 * Returns the resulting expression or NULL in case of internal error
7396 */
7397 xmlExpNodePtr
xmlExpStringDerive(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar * str,int len)7398 xmlExpStringDerive(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7399 const xmlChar *str, int len) {
7400 const xmlChar *input;
7401
7402 if ((exp == NULL) || (ctxt == NULL) || (str == NULL)) {
7403 return(NULL);
7404 }
7405 /*
7406 * check the string is in the dictionary, if yes use an interned
7407 * copy, otherwise we know it's not an acceptable input
7408 */
7409 input = xmlDictExists(ctxt->dict, str, len);
7410 if (input == NULL) {
7411 return(forbiddenExp);
7412 }
7413 return(xmlExpStringDeriveInt(ctxt, exp, input));
7414 }
7415
7416 static int
xmlExpCheckCard(xmlExpNodePtr exp,xmlExpNodePtr sub)7417 xmlExpCheckCard(xmlExpNodePtr exp, xmlExpNodePtr sub) {
7418 int ret = 1;
7419
7420 if (sub->c_max == -1) {
7421 if (exp->c_max != -1)
7422 ret = 0;
7423 } else if ((exp->c_max >= 0) && (exp->c_max < sub->c_max)) {
7424 ret = 0;
7425 }
7426 #if 0
7427 if ((IS_NILLABLE(sub)) && (!IS_NILLABLE(exp)))
7428 ret = 0;
7429 #endif
7430 return(ret);
7431 }
7432
7433 static xmlExpNodePtr xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7434 xmlExpNodePtr sub);
7435 /**
7436 * xmlExpDivide:
7437 * @ctxt: the expressions context
7438 * @exp: the englobing expression
7439 * @sub: the subexpression
7440 * @mult: the multiple expression
7441 * @remain: the remain from the derivation of the multiple
7442 *
7443 * Check if exp is a multiple of sub, i.e. if there is a finite number n
7444 * so that sub{n} subsume exp
7445 *
7446 * Returns the multiple value if successful, 0 if it is not a multiple
7447 * and -1 in case of internal error.
7448 */
7449
7450 static int
xmlExpDivide(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,xmlExpNodePtr sub,xmlExpNodePtr * mult,xmlExpNodePtr * remain)7451 xmlExpDivide(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub,
7452 xmlExpNodePtr *mult, xmlExpNodePtr *remain) {
7453 int i;
7454 xmlExpNodePtr tmp, tmp2;
7455
7456 if (mult != NULL) *mult = NULL;
7457 if (remain != NULL) *remain = NULL;
7458 if (exp->c_max == -1) return(0);
7459 if (IS_NILLABLE(exp) && (!IS_NILLABLE(sub))) return(0);
7460
7461 for (i = 1;i <= exp->c_max;i++) {
7462 sub->ref++;
7463 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT,
7464 sub, NULL, NULL, i, i);
7465 if (tmp == NULL) {
7466 return(-1);
7467 }
7468 if (!xmlExpCheckCard(tmp, exp)) {
7469 xmlExpFree(ctxt, tmp);
7470 continue;
7471 }
7472 tmp2 = xmlExpExpDeriveInt(ctxt, tmp, exp);
7473 if (tmp2 == NULL) {
7474 xmlExpFree(ctxt, tmp);
7475 return(-1);
7476 }
7477 if ((tmp2 != forbiddenExp) && (IS_NILLABLE(tmp2))) {
7478 if (remain != NULL)
7479 *remain = tmp2;
7480 else
7481 xmlExpFree(ctxt, tmp2);
7482 if (mult != NULL)
7483 *mult = tmp;
7484 else
7485 xmlExpFree(ctxt, tmp);
7486 #ifdef DEBUG_DERIV
7487 printf("Divide succeeded %d\n", i);
7488 #endif
7489 return(i);
7490 }
7491 xmlExpFree(ctxt, tmp);
7492 xmlExpFree(ctxt, tmp2);
7493 }
7494 #ifdef DEBUG_DERIV
7495 printf("Divide failed\n");
7496 #endif
7497 return(0);
7498 }
7499
7500 /**
7501 * xmlExpExpDeriveInt:
7502 * @ctxt: the expressions context
7503 * @exp: the englobing expression
7504 * @sub: the subexpression
7505 *
7506 * Try to do a step of Brzozowski derivation but at a higher level
7507 * the input being a subexpression.
7508 *
7509 * Returns the resulting expression or NULL in case of internal error
7510 */
7511 static xmlExpNodePtr
xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,xmlExpNodePtr sub)7512 xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7513 xmlExpNodePtr ret, tmp, tmp2, tmp3;
7514 const xmlChar **tab;
7515 int len, i;
7516
7517 /*
7518 * In case of equality and if the expression can only consume a finite
7519 * amount, then the derivation is empty
7520 */
7521 if ((exp == sub) && (exp->c_max >= 0)) {
7522 #ifdef DEBUG_DERIV
7523 printf("Equal(exp, sub) and finite -> Empty\n");
7524 #endif
7525 return(emptyExp);
7526 }
7527 /*
7528 * decompose sub sequence first
7529 */
7530 if (sub->type == XML_EXP_EMPTY) {
7531 #ifdef DEBUG_DERIV
7532 printf("Empty(sub) -> Empty\n");
7533 #endif
7534 exp->ref++;
7535 return(exp);
7536 }
7537 if (sub->type == XML_EXP_SEQ) {
7538 #ifdef DEBUG_DERIV
7539 printf("Seq(sub) -> decompose\n");
7540 #endif
7541 tmp = xmlExpExpDeriveInt(ctxt, exp, sub->exp_left);
7542 if (tmp == NULL)
7543 return(NULL);
7544 if (tmp == forbiddenExp)
7545 return(tmp);
7546 ret = xmlExpExpDeriveInt(ctxt, tmp, sub->exp_right);
7547 xmlExpFree(ctxt, tmp);
7548 return(ret);
7549 }
7550 if (sub->type == XML_EXP_OR) {
7551 #ifdef DEBUG_DERIV
7552 printf("Or(sub) -> decompose\n");
7553 #endif
7554 tmp = xmlExpExpDeriveInt(ctxt, exp, sub->exp_left);
7555 if (tmp == forbiddenExp)
7556 return(tmp);
7557 if (tmp == NULL)
7558 return(NULL);
7559 ret = xmlExpExpDeriveInt(ctxt, exp, sub->exp_right);
7560 if ((ret == NULL) || (ret == forbiddenExp)) {
7561 xmlExpFree(ctxt, tmp);
7562 return(ret);
7563 }
7564 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, tmp, ret, NULL, 0, 0));
7565 }
7566 if (!xmlExpCheckCard(exp, sub)) {
7567 #ifdef DEBUG_DERIV
7568 printf("CheckCard(exp, sub) failed -> Forbid\n");
7569 #endif
7570 return(forbiddenExp);
7571 }
7572 switch (exp->type) {
7573 case XML_EXP_EMPTY:
7574 if (sub == emptyExp)
7575 return(emptyExp);
7576 #ifdef DEBUG_DERIV
7577 printf("Empty(exp) -> Forbid\n");
7578 #endif
7579 return(forbiddenExp);
7580 case XML_EXP_FORBID:
7581 #ifdef DEBUG_DERIV
7582 printf("Forbid(exp) -> Forbid\n");
7583 #endif
7584 return(forbiddenExp);
7585 case XML_EXP_ATOM:
7586 if (sub->type == XML_EXP_ATOM) {
7587 /* TODO: handle wildcards */
7588 if (exp->exp_str == sub->exp_str) {
7589 #ifdef DEBUG_DERIV
7590 printf("Atom match -> Empty\n");
7591 #endif
7592 return(emptyExp);
7593 }
7594 #ifdef DEBUG_DERIV
7595 printf("Atom mismatch -> Forbid\n");
7596 #endif
7597 return(forbiddenExp);
7598 }
7599 if ((sub->type == XML_EXP_COUNT) &&
7600 (sub->exp_max == 1) &&
7601 (sub->exp_left->type == XML_EXP_ATOM)) {
7602 /* TODO: handle wildcards */
7603 if (exp->exp_str == sub->exp_left->exp_str) {
7604 #ifdef DEBUG_DERIV
7605 printf("Atom match -> Empty\n");
7606 #endif
7607 return(emptyExp);
7608 }
7609 #ifdef DEBUG_DERIV
7610 printf("Atom mismatch -> Forbid\n");
7611 #endif
7612 return(forbiddenExp);
7613 }
7614 #ifdef DEBUG_DERIV
7615 printf("Complex exp vs Atom -> Forbid\n");
7616 #endif
7617 return(forbiddenExp);
7618 case XML_EXP_SEQ:
7619 /* try to get the sequence consumed only if possible */
7620 if (xmlExpCheckCard(exp->exp_left, sub)) {
7621 /* See if the sequence can be consumed directly */
7622 #ifdef DEBUG_DERIV
7623 printf("Seq trying left only\n");
7624 #endif
7625 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7626 if ((ret != forbiddenExp) && (ret != NULL)) {
7627 #ifdef DEBUG_DERIV
7628 printf("Seq trying left only worked\n");
7629 #endif
7630 /*
7631 * TODO: assumption here that we are determinist
7632 * i.e. we won't get to a nillable exp left
7633 * subset which could be matched by the right
7634 * part too.
7635 * e.g.: (a | b)+,(a | c) and 'a+,a'
7636 */
7637 exp->exp_right->ref++;
7638 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret,
7639 exp->exp_right, NULL, 0, 0));
7640 }
7641 #ifdef DEBUG_DERIV
7642 } else {
7643 printf("Seq: left too short\n");
7644 #endif
7645 }
7646 /* Try instead to decompose */
7647 if (sub->type == XML_EXP_COUNT) {
7648 int min, max;
7649
7650 #ifdef DEBUG_DERIV
7651 printf("Seq: sub is a count\n");
7652 #endif
7653 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub->exp_left);
7654 if (ret == NULL)
7655 return(NULL);
7656 if (ret != forbiddenExp) {
7657 #ifdef DEBUG_DERIV
7658 printf("Seq , Count match on left\n");
7659 #endif
7660 if (sub->exp_max < 0)
7661 max = -1;
7662 else
7663 max = sub->exp_max -1;
7664 if (sub->exp_min > 0)
7665 min = sub->exp_min -1;
7666 else
7667 min = 0;
7668 exp->exp_right->ref++;
7669 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret,
7670 exp->exp_right, NULL, 0, 0);
7671 if (tmp == NULL)
7672 return(NULL);
7673
7674 sub->exp_left->ref++;
7675 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT,
7676 sub->exp_left, NULL, NULL, min, max);
7677 if (tmp2 == NULL) {
7678 xmlExpFree(ctxt, tmp);
7679 return(NULL);
7680 }
7681 ret = xmlExpExpDeriveInt(ctxt, tmp, tmp2);
7682 xmlExpFree(ctxt, tmp);
7683 xmlExpFree(ctxt, tmp2);
7684 return(ret);
7685 }
7686 }
7687 /* we made no progress on structured operations */
7688 break;
7689 case XML_EXP_OR:
7690 #ifdef DEBUG_DERIV
7691 printf("Or , trying both side\n");
7692 #endif
7693 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7694 if (ret == NULL)
7695 return(NULL);
7696 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_right, sub);
7697 if (tmp == NULL) {
7698 xmlExpFree(ctxt, ret);
7699 return(NULL);
7700 }
7701 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, tmp, NULL, 0, 0));
7702 case XML_EXP_COUNT: {
7703 int min, max;
7704
7705 if (sub->type == XML_EXP_COUNT) {
7706 /*
7707 * Try to see if the loop is completely subsumed
7708 */
7709 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub->exp_left);
7710 if (tmp == NULL)
7711 return(NULL);
7712 if (tmp == forbiddenExp) {
7713 int mult;
7714
7715 #ifdef DEBUG_DERIV
7716 printf("Count, Count inner don't subsume\n");
7717 #endif
7718 mult = xmlExpDivide(ctxt, sub->exp_left, exp->exp_left,
7719 NULL, &tmp);
7720 if (mult <= 0) {
7721 #ifdef DEBUG_DERIV
7722 printf("Count, Count not multiple => forbidden\n");
7723 #endif
7724 return(forbiddenExp);
7725 }
7726 if (sub->exp_max == -1) {
7727 max = -1;
7728 if (exp->exp_max == -1) {
7729 if (exp->exp_min <= sub->exp_min * mult)
7730 min = 0;
7731 else
7732 min = exp->exp_min - sub->exp_min * mult;
7733 } else {
7734 #ifdef DEBUG_DERIV
7735 printf("Count, Count finite can't subsume infinite\n");
7736 #endif
7737 xmlExpFree(ctxt, tmp);
7738 return(forbiddenExp);
7739 }
7740 } else {
7741 if (exp->exp_max == -1) {
7742 #ifdef DEBUG_DERIV
7743 printf("Infinite loop consume mult finite loop\n");
7744 #endif
7745 if (exp->exp_min > sub->exp_min * mult) {
7746 max = -1;
7747 min = exp->exp_min - sub->exp_min * mult;
7748 } else {
7749 max = -1;
7750 min = 0;
7751 }
7752 } else {
7753 if (exp->exp_max < sub->exp_max * mult) {
7754 #ifdef DEBUG_DERIV
7755 printf("loops max mult mismatch => forbidden\n");
7756 #endif
7757 xmlExpFree(ctxt, tmp);
7758 return(forbiddenExp);
7759 }
7760 if (sub->exp_max * mult > exp->exp_min)
7761 min = 0;
7762 else
7763 min = exp->exp_min - sub->exp_max * mult;
7764 max = exp->exp_max - sub->exp_max * mult;
7765 }
7766 }
7767 } else if (!IS_NILLABLE(tmp)) {
7768 /*
7769 * TODO: loop here to try to grow if working on finite
7770 * blocks.
7771 */
7772 #ifdef DEBUG_DERIV
7773 printf("Count, Count remain not nillable => forbidden\n");
7774 #endif
7775 xmlExpFree(ctxt, tmp);
7776 return(forbiddenExp);
7777 } else if (sub->exp_max == -1) {
7778 if (exp->exp_max == -1) {
7779 if (exp->exp_min <= sub->exp_min) {
7780 #ifdef DEBUG_DERIV
7781 printf("Infinite loops Okay => COUNT(0,Inf)\n");
7782 #endif
7783 max = -1;
7784 min = 0;
7785 } else {
7786 #ifdef DEBUG_DERIV
7787 printf("Infinite loops min => Count(X,Inf)\n");
7788 #endif
7789 max = -1;
7790 min = exp->exp_min - sub->exp_min;
7791 }
7792 } else if (exp->exp_min > sub->exp_min) {
7793 #ifdef DEBUG_DERIV
7794 printf("loops min mismatch 1 => forbidden ???\n");
7795 #endif
7796 xmlExpFree(ctxt, tmp);
7797 return(forbiddenExp);
7798 } else {
7799 max = -1;
7800 min = 0;
7801 }
7802 } else {
7803 if (exp->exp_max == -1) {
7804 #ifdef DEBUG_DERIV
7805 printf("Infinite loop consume finite loop\n");
7806 #endif
7807 if (exp->exp_min > sub->exp_min) {
7808 max = -1;
7809 min = exp->exp_min - sub->exp_min;
7810 } else {
7811 max = -1;
7812 min = 0;
7813 }
7814 } else {
7815 if (exp->exp_max < sub->exp_max) {
7816 #ifdef DEBUG_DERIV
7817 printf("loops max mismatch => forbidden\n");
7818 #endif
7819 xmlExpFree(ctxt, tmp);
7820 return(forbiddenExp);
7821 }
7822 if (sub->exp_max > exp->exp_min)
7823 min = 0;
7824 else
7825 min = exp->exp_min - sub->exp_max;
7826 max = exp->exp_max - sub->exp_max;
7827 }
7828 }
7829 #ifdef DEBUG_DERIV
7830 printf("loops match => SEQ(COUNT())\n");
7831 #endif
7832 exp->exp_left->ref++;
7833 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left,
7834 NULL, NULL, min, max);
7835 if (tmp2 == NULL) {
7836 return(NULL);
7837 }
7838 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, tmp, tmp2,
7839 NULL, 0, 0);
7840 return(ret);
7841 }
7842 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7843 if (tmp == NULL)
7844 return(NULL);
7845 if (tmp == forbiddenExp) {
7846 #ifdef DEBUG_DERIV
7847 printf("loop mismatch => forbidden\n");
7848 #endif
7849 return(forbiddenExp);
7850 }
7851 if (exp->exp_min > 0)
7852 min = exp->exp_min - 1;
7853 else
7854 min = 0;
7855 if (exp->exp_max < 0)
7856 max = -1;
7857 else
7858 max = exp->exp_max - 1;
7859
7860 #ifdef DEBUG_DERIV
7861 printf("loop match => SEQ(COUNT())\n");
7862 #endif
7863 exp->exp_left->ref++;
7864 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left,
7865 NULL, NULL, min, max);
7866 if (tmp2 == NULL)
7867 return(NULL);
7868 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, tmp, tmp2,
7869 NULL, 0, 0);
7870 return(ret);
7871 }
7872 }
7873
7874 #ifdef DEBUG_DERIV
7875 printf("Fallback to derivative\n");
7876 #endif
7877 if (IS_NILLABLE(sub)) {
7878 if (!(IS_NILLABLE(exp)))
7879 return(forbiddenExp);
7880 else
7881 ret = emptyExp;
7882 } else
7883 ret = NULL;
7884 /*
7885 * here the structured derivation made no progress so
7886 * we use the default token based derivation to force one more step
7887 */
7888 if (ctxt->tabSize == 0)
7889 ctxt->tabSize = 40;
7890
7891 tab = (const xmlChar **) xmlMalloc(ctxt->tabSize *
7892 sizeof(const xmlChar *));
7893 if (tab == NULL) {
7894 return(NULL);
7895 }
7896
7897 /*
7898 * collect all the strings accepted by the subexpression on input
7899 */
7900 len = xmlExpGetStartInt(ctxt, sub, tab, ctxt->tabSize, 0);
7901 while (len < 0) {
7902 const xmlChar **temp;
7903 temp = (const xmlChar **) xmlRealloc((xmlChar **) tab, ctxt->tabSize * 2 *
7904 sizeof(const xmlChar *));
7905 if (temp == NULL) {
7906 xmlFree((xmlChar **) tab);
7907 return(NULL);
7908 }
7909 tab = temp;
7910 ctxt->tabSize *= 2;
7911 len = xmlExpGetStartInt(ctxt, sub, tab, ctxt->tabSize, 0);
7912 }
7913 for (i = 0;i < len;i++) {
7914 tmp = xmlExpStringDeriveInt(ctxt, exp, tab[i]);
7915 if ((tmp == NULL) || (tmp == forbiddenExp)) {
7916 xmlExpFree(ctxt, ret);
7917 xmlFree((xmlChar **) tab);
7918 return(tmp);
7919 }
7920 tmp2 = xmlExpStringDeriveInt(ctxt, sub, tab[i]);
7921 if ((tmp2 == NULL) || (tmp2 == forbiddenExp)) {
7922 xmlExpFree(ctxt, tmp);
7923 xmlExpFree(ctxt, ret);
7924 xmlFree((xmlChar **) tab);
7925 return(tmp);
7926 }
7927 tmp3 = xmlExpExpDeriveInt(ctxt, tmp, tmp2);
7928 xmlExpFree(ctxt, tmp);
7929 xmlExpFree(ctxt, tmp2);
7930
7931 if ((tmp3 == NULL) || (tmp3 == forbiddenExp)) {
7932 xmlExpFree(ctxt, ret);
7933 xmlFree((xmlChar **) tab);
7934 return(tmp3);
7935 }
7936
7937 if (ret == NULL)
7938 ret = tmp3;
7939 else {
7940 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, tmp3, NULL, 0, 0);
7941 if (ret == NULL) {
7942 xmlFree((xmlChar **) tab);
7943 return(NULL);
7944 }
7945 }
7946 }
7947 xmlFree((xmlChar **) tab);
7948 return(ret);
7949 }
7950
7951 /**
7952 * xmlExpExpDerive:
7953 * @ctxt: the expressions context
7954 * @exp: the englobing expression
7955 * @sub: the subexpression
7956 *
7957 * Evaluates the expression resulting from @exp consuming a sub expression @sub
7958 * Based on algebraic derivation and sometimes direct Brzozowski derivation
7959 * it usually takes less than linear time and can handle expressions generating
7960 * infinite languages.
7961 *
7962 * Returns the resulting expression or NULL in case of internal error, the
7963 * result must be freed
7964 */
7965 xmlExpNodePtr
xmlExpExpDerive(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,xmlExpNodePtr sub)7966 xmlExpExpDerive(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7967 if ((exp == NULL) || (ctxt == NULL) || (sub == NULL))
7968 return(NULL);
7969
7970 /*
7971 * O(1) speedups
7972 */
7973 if (IS_NILLABLE(sub) && (!IS_NILLABLE(exp))) {
7974 #ifdef DEBUG_DERIV
7975 printf("Sub nillable and not exp : can't subsume\n");
7976 #endif
7977 return(forbiddenExp);
7978 }
7979 if (xmlExpCheckCard(exp, sub) == 0) {
7980 #ifdef DEBUG_DERIV
7981 printf("sub generate longer sequences than exp : can't subsume\n");
7982 #endif
7983 return(forbiddenExp);
7984 }
7985 return(xmlExpExpDeriveInt(ctxt, exp, sub));
7986 }
7987
7988 /**
7989 * xmlExpSubsume:
7990 * @ctxt: the expressions context
7991 * @exp: the englobing expression
7992 * @sub: the subexpression
7993 *
7994 * Check whether @exp accepts all the languages accepted by @sub
7995 * the input being a subexpression.
7996 *
7997 * Returns 1 if true 0 if false and -1 in case of failure.
7998 */
7999 int
xmlExpSubsume(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,xmlExpNodePtr sub)8000 xmlExpSubsume(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
8001 xmlExpNodePtr tmp;
8002
8003 if ((exp == NULL) || (ctxt == NULL) || (sub == NULL))
8004 return(-1);
8005
8006 /*
8007 * TODO: speedup by checking the language of sub is a subset of the
8008 * language of exp
8009 */
8010 /*
8011 * O(1) speedups
8012 */
8013 if (IS_NILLABLE(sub) && (!IS_NILLABLE(exp))) {
8014 #ifdef DEBUG_DERIV
8015 printf("Sub nillable and not exp : can't subsume\n");
8016 #endif
8017 return(0);
8018 }
8019 if (xmlExpCheckCard(exp, sub) == 0) {
8020 #ifdef DEBUG_DERIV
8021 printf("sub generate longer sequences than exp : can't subsume\n");
8022 #endif
8023 return(0);
8024 }
8025 tmp = xmlExpExpDeriveInt(ctxt, exp, sub);
8026 #ifdef DEBUG_DERIV
8027 printf("Result derivation :\n");
8028 PRINT_EXP(tmp);
8029 #endif
8030 if (tmp == NULL)
8031 return(-1);
8032 if (tmp == forbiddenExp)
8033 return(0);
8034 if (tmp == emptyExp)
8035 return(1);
8036 if ((tmp != NULL) && (IS_NILLABLE(tmp))) {
8037 xmlExpFree(ctxt, tmp);
8038 return(1);
8039 }
8040 xmlExpFree(ctxt, tmp);
8041 return(0);
8042 }
8043
8044 /************************************************************************
8045 * *
8046 * Parsing expression *
8047 * *
8048 ************************************************************************/
8049
8050 static xmlExpNodePtr xmlExpParseExpr(xmlExpCtxtPtr ctxt);
8051
8052 #undef CUR
8053 #define CUR (*ctxt->cur)
8054 #undef NEXT
8055 #define NEXT ctxt->cur++;
8056 #undef IS_BLANK
8057 #define IS_BLANK(c) ((c == ' ') || (c == '\n') || (c == '\r') || (c == '\t'))
8058 #define SKIP_BLANKS while (IS_BLANK(*ctxt->cur)) ctxt->cur++;
8059
8060 static int
xmlExpParseNumber(xmlExpCtxtPtr ctxt)8061 xmlExpParseNumber(xmlExpCtxtPtr ctxt) {
8062 int ret = 0;
8063
8064 SKIP_BLANKS
8065 if (CUR == '*') {
8066 NEXT
8067 return(-1);
8068 }
8069 if ((CUR < '0') || (CUR > '9'))
8070 return(-1);
8071 while ((CUR >= '0') && (CUR <= '9')) {
8072 ret = ret * 10 + (CUR - '0');
8073 NEXT
8074 }
8075 return(ret);
8076 }
8077
8078 static xmlExpNodePtr
xmlExpParseOr(xmlExpCtxtPtr ctxt)8079 xmlExpParseOr(xmlExpCtxtPtr ctxt) {
8080 const char *base;
8081 xmlExpNodePtr ret;
8082 const xmlChar *val;
8083
8084 SKIP_BLANKS
8085 base = ctxt->cur;
8086 if (*ctxt->cur == '(') {
8087 NEXT
8088 ret = xmlExpParseExpr(ctxt);
8089 SKIP_BLANKS
8090 if (*ctxt->cur != ')') {
8091 fprintf(stderr, "unbalanced '(' : %s\n", base);
8092 xmlExpFree(ctxt, ret);
8093 return(NULL);
8094 }
8095 NEXT;
8096 SKIP_BLANKS
8097 goto parse_quantifier;
8098 }
8099 while ((CUR != 0) && (!(IS_BLANK(CUR))) && (CUR != '(') &&
8100 (CUR != ')') && (CUR != '|') && (CUR != ',') && (CUR != '{') &&
8101 (CUR != '*') && (CUR != '+') && (CUR != '?') && (CUR != '}'))
8102 NEXT;
8103 val = xmlDictLookup(ctxt->dict, BAD_CAST base, ctxt->cur - base);
8104 if (val == NULL)
8105 return(NULL);
8106 ret = xmlExpHashGetEntry(ctxt, XML_EXP_ATOM, NULL, NULL, val, 0, 0);
8107 if (ret == NULL)
8108 return(NULL);
8109 SKIP_BLANKS
8110 parse_quantifier:
8111 if (CUR == '{') {
8112 int min, max;
8113
8114 NEXT
8115 min = xmlExpParseNumber(ctxt);
8116 if (min < 0) {
8117 xmlExpFree(ctxt, ret);
8118 return(NULL);
8119 }
8120 SKIP_BLANKS
8121 if (CUR == ',') {
8122 NEXT
8123 max = xmlExpParseNumber(ctxt);
8124 SKIP_BLANKS
8125 } else
8126 max = min;
8127 if (CUR != '}') {
8128 xmlExpFree(ctxt, ret);
8129 return(NULL);
8130 }
8131 NEXT
8132 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
8133 min, max);
8134 SKIP_BLANKS
8135 } else if (CUR == '?') {
8136 NEXT
8137 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
8138 0, 1);
8139 SKIP_BLANKS
8140 } else if (CUR == '+') {
8141 NEXT
8142 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
8143 1, -1);
8144 SKIP_BLANKS
8145 } else if (CUR == '*') {
8146 NEXT
8147 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
8148 0, -1);
8149 SKIP_BLANKS
8150 }
8151 return(ret);
8152 }
8153
8154
8155 static xmlExpNodePtr
xmlExpParseSeq(xmlExpCtxtPtr ctxt)8156 xmlExpParseSeq(xmlExpCtxtPtr ctxt) {
8157 xmlExpNodePtr ret, right;
8158
8159 ret = xmlExpParseOr(ctxt);
8160 SKIP_BLANKS
8161 while (CUR == '|') {
8162 NEXT
8163 right = xmlExpParseOr(ctxt);
8164 if (right == NULL) {
8165 xmlExpFree(ctxt, ret);
8166 return(NULL);
8167 }
8168 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, right, NULL, 0, 0);
8169 if (ret == NULL)
8170 return(NULL);
8171 }
8172 return(ret);
8173 }
8174
8175 static xmlExpNodePtr
xmlExpParseExpr(xmlExpCtxtPtr ctxt)8176 xmlExpParseExpr(xmlExpCtxtPtr ctxt) {
8177 xmlExpNodePtr ret, right;
8178
8179 ret = xmlExpParseSeq(ctxt);
8180 SKIP_BLANKS
8181 while (CUR == ',') {
8182 NEXT
8183 right = xmlExpParseSeq(ctxt);
8184 if (right == NULL) {
8185 xmlExpFree(ctxt, ret);
8186 return(NULL);
8187 }
8188 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, right, NULL, 0, 0);
8189 if (ret == NULL)
8190 return(NULL);
8191 }
8192 return(ret);
8193 }
8194
8195 /**
8196 * xmlExpParse:
8197 * @ctxt: the expressions context
8198 * @expr: the 0 terminated string
8199 *
8200 * Minimal parser for regexps, it understand the following constructs
8201 * - string terminals
8202 * - choice operator |
8203 * - sequence operator ,
8204 * - subexpressions (...)
8205 * - usual cardinality operators + * and ?
8206 * - finite sequences { min, max }
8207 * - infinite sequences { min, * }
8208 * There is minimal checkings made especially no checking on strings values
8209 *
8210 * Returns a new expression or NULL in case of failure
8211 */
8212 xmlExpNodePtr
xmlExpParse(xmlExpCtxtPtr ctxt,const char * expr)8213 xmlExpParse(xmlExpCtxtPtr ctxt, const char *expr) {
8214 xmlExpNodePtr ret;
8215
8216 ctxt->expr = expr;
8217 ctxt->cur = expr;
8218
8219 ret = xmlExpParseExpr(ctxt);
8220 SKIP_BLANKS
8221 if (*ctxt->cur != 0) {
8222 xmlExpFree(ctxt, ret);
8223 return(NULL);
8224 }
8225 return(ret);
8226 }
8227
8228 static void
xmlExpDumpInt(xmlBufferPtr buf,xmlExpNodePtr expr,int glob)8229 xmlExpDumpInt(xmlBufferPtr buf, xmlExpNodePtr expr, int glob) {
8230 xmlExpNodePtr c;
8231
8232 if (expr == NULL) return;
8233 if (glob) xmlBufferWriteChar(buf, "(");
8234 switch (expr->type) {
8235 case XML_EXP_EMPTY:
8236 xmlBufferWriteChar(buf, "empty");
8237 break;
8238 case XML_EXP_FORBID:
8239 xmlBufferWriteChar(buf, "forbidden");
8240 break;
8241 case XML_EXP_ATOM:
8242 xmlBufferWriteCHAR(buf, expr->exp_str);
8243 break;
8244 case XML_EXP_SEQ:
8245 c = expr->exp_left;
8246 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8247 xmlExpDumpInt(buf, c, 1);
8248 else
8249 xmlExpDumpInt(buf, c, 0);
8250 xmlBufferWriteChar(buf, " , ");
8251 c = expr->exp_right;
8252 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8253 xmlExpDumpInt(buf, c, 1);
8254 else
8255 xmlExpDumpInt(buf, c, 0);
8256 break;
8257 case XML_EXP_OR:
8258 c = expr->exp_left;
8259 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8260 xmlExpDumpInt(buf, c, 1);
8261 else
8262 xmlExpDumpInt(buf, c, 0);
8263 xmlBufferWriteChar(buf, " | ");
8264 c = expr->exp_right;
8265 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8266 xmlExpDumpInt(buf, c, 1);
8267 else
8268 xmlExpDumpInt(buf, c, 0);
8269 break;
8270 case XML_EXP_COUNT: {
8271 char rep[40];
8272
8273 c = expr->exp_left;
8274 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8275 xmlExpDumpInt(buf, c, 1);
8276 else
8277 xmlExpDumpInt(buf, c, 0);
8278 if ((expr->exp_min == 0) && (expr->exp_max == 1)) {
8279 rep[0] = '?';
8280 rep[1] = 0;
8281 } else if ((expr->exp_min == 0) && (expr->exp_max == -1)) {
8282 rep[0] = '*';
8283 rep[1] = 0;
8284 } else if ((expr->exp_min == 1) && (expr->exp_max == -1)) {
8285 rep[0] = '+';
8286 rep[1] = 0;
8287 } else if (expr->exp_max == expr->exp_min) {
8288 snprintf(rep, 39, "{%d}", expr->exp_min);
8289 } else if (expr->exp_max < 0) {
8290 snprintf(rep, 39, "{%d,inf}", expr->exp_min);
8291 } else {
8292 snprintf(rep, 39, "{%d,%d}", expr->exp_min, expr->exp_max);
8293 }
8294 rep[39] = 0;
8295 xmlBufferWriteChar(buf, rep);
8296 break;
8297 }
8298 default:
8299 fprintf(stderr, "Error in tree\n");
8300 }
8301 if (glob)
8302 xmlBufferWriteChar(buf, ")");
8303 }
8304 /**
8305 * xmlExpDump:
8306 * @buf: a buffer to receive the output
8307 * @expr: the compiled expression
8308 *
8309 * Serialize the expression as compiled to the buffer
8310 */
8311 void
xmlExpDump(xmlBufferPtr buf,xmlExpNodePtr expr)8312 xmlExpDump(xmlBufferPtr buf, xmlExpNodePtr expr) {
8313 if ((buf == NULL) || (expr == NULL))
8314 return;
8315 xmlExpDumpInt(buf, expr, 0);
8316 }
8317
8318 /**
8319 * xmlExpMaxToken:
8320 * @expr: a compiled expression
8321 *
8322 * Indicate the maximum number of input a expression can accept
8323 *
8324 * Returns the maximum length or -1 in case of error
8325 */
8326 int
xmlExpMaxToken(xmlExpNodePtr expr)8327 xmlExpMaxToken(xmlExpNodePtr expr) {
8328 if (expr == NULL)
8329 return(-1);
8330 return(expr->c_max);
8331 }
8332
8333 /**
8334 * xmlExpCtxtNbNodes:
8335 * @ctxt: an expression context
8336 *
8337 * Debugging facility provides the number of allocated nodes at a that point
8338 *
8339 * Returns the number of nodes in use or -1 in case of error
8340 */
8341 int
xmlExpCtxtNbNodes(xmlExpCtxtPtr ctxt)8342 xmlExpCtxtNbNodes(xmlExpCtxtPtr ctxt) {
8343 if (ctxt == NULL)
8344 return(-1);
8345 return(ctxt->nb_nodes);
8346 }
8347
8348 /**
8349 * xmlExpCtxtNbCons:
8350 * @ctxt: an expression context
8351 *
8352 * Debugging facility provides the number of allocated nodes over lifetime
8353 *
8354 * Returns the number of nodes ever allocated or -1 in case of error
8355 */
8356 int
xmlExpCtxtNbCons(xmlExpCtxtPtr ctxt)8357 xmlExpCtxtNbCons(xmlExpCtxtPtr ctxt) {
8358 if (ctxt == NULL)
8359 return(-1);
8360 return(ctxt->nb_cons);
8361 }
8362
8363 #endif /* LIBXML_EXPR_ENABLED */
8364
8365 #endif /* LIBXML_REGEXP_ENABLED */
8366