VirtualBox

source: vbox/trunk/src/libs/libxml2-2.9.14/xmlregexp.c@ 100906

Last change on this file since 100906 was 95312, checked in by vboxsync, 2 years ago

libs/{curl,libxml2}: OSE export fixes, bugref:8515
  • Property svn:eol-style set to native
File size: 217.8 KB
Line 
1/*
2 * regexp.c: generic and extensible Regular Expression engine
3 *
4 * Basically designed with the purpose of compiling regexps for
5 * the variety of validation/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 <[email protected]>
15 */
16
17#define IN_LIBXML
18#include "libxml.h"
19
20#ifdef LIBXML_REGEXP_ENABLED
21
22/* #define DEBUG_ERR */
23
24#include <stdio.h>
25#include <string.h>
26#ifdef HAVE_LIMITS_H
27#include <limits.h>
28#endif
29#ifdef HAVE_STDINT_H
30#include <stdint.h>
31#endif
32
33#include <libxml/tree.h>
34#include <libxml/parserInternals.h>
35#include <libxml/xmlregexp.h>
36#include <libxml/xmlautomata.h>
37#include <libxml/xmlunicode.h>
38
39#ifndef INT_MAX
40#define INT_MAX 123456789 /* easy to flag and big enough for our needs */
41#endif
42#ifndef SIZE_MAX
43#define SIZE_MAX ((size_t) -1)
44#endif
45
46/* #define DEBUG_REGEXP_GRAPH */
47/* #define DEBUG_REGEXP_EXEC */
48/* #define DEBUG_PUSH */
49/* #define DEBUG_COMPACTION */
50
51#define MAX_PUSH 10000000
52
53#ifdef ERROR
54#undef ERROR
55#endif
56#define ERROR(str) \
57 ctxt->error = XML_REGEXP_COMPILE_ERROR; \
58 xmlRegexpErrCompile(ctxt, str);
59#define NEXT ctxt->cur++
60#define CUR (*(ctxt->cur))
61#define NXT(index) (ctxt->cur[index])
62
63#define CUR_SCHAR(s, l) xmlStringCurrentChar(NULL, s, &l)
64#define NEXTL(l) ctxt->cur += l;
65#define XML_REG_STRING_SEPARATOR '|'
66/*
67 * Need PREV to check on a '-' within a Character Group. May only be used
68 * when it's guaranteed that cur is not at the beginning of ctxt->string!
69 */
70#define PREV (ctxt->cur[-1])
71
72/**
73 * TODO:
74 *
75 * macro to flag unimplemented blocks
76 */
77#define TODO \
78 xmlGenericError(xmlGenericErrorContext, \
79 "Unimplemented block at %s:%d\n", \
80 __FILE__, __LINE__);
81
82/************************************************************************
83 * *
84 * Datatypes and structures *
85 * *
86 ************************************************************************/
87
88/*
89 * Note: the order of the enums below is significant, do not shuffle
90 */
91typedef enum {
92 XML_REGEXP_EPSILON = 1,
93 XML_REGEXP_CHARVAL,
94 XML_REGEXP_RANGES,
95 XML_REGEXP_SUBREG, /* used for () sub regexps */
96 XML_REGEXP_STRING,
97 XML_REGEXP_ANYCHAR, /* . */
98 XML_REGEXP_ANYSPACE, /* \s */
99 XML_REGEXP_NOTSPACE, /* \S */
100 XML_REGEXP_INITNAME, /* \l */
101 XML_REGEXP_NOTINITNAME, /* \L */
102 XML_REGEXP_NAMECHAR, /* \c */
103 XML_REGEXP_NOTNAMECHAR, /* \C */
104 XML_REGEXP_DECIMAL, /* \d */
105 XML_REGEXP_NOTDECIMAL, /* \D */
106 XML_REGEXP_REALCHAR, /* \w */
107 XML_REGEXP_NOTREALCHAR, /* \W */
108 XML_REGEXP_LETTER = 100,
109 XML_REGEXP_LETTER_UPPERCASE,
110 XML_REGEXP_LETTER_LOWERCASE,
111 XML_REGEXP_LETTER_TITLECASE,
112 XML_REGEXP_LETTER_MODIFIER,
113 XML_REGEXP_LETTER_OTHERS,
114 XML_REGEXP_MARK,
115 XML_REGEXP_MARK_NONSPACING,
116 XML_REGEXP_MARK_SPACECOMBINING,
117 XML_REGEXP_MARK_ENCLOSING,
118 XML_REGEXP_NUMBER,
119 XML_REGEXP_NUMBER_DECIMAL,
120 XML_REGEXP_NUMBER_LETTER,
121 XML_REGEXP_NUMBER_OTHERS,
122 XML_REGEXP_PUNCT,
123 XML_REGEXP_PUNCT_CONNECTOR,
124 XML_REGEXP_PUNCT_DASH,
125 XML_REGEXP_PUNCT_OPEN,
126 XML_REGEXP_PUNCT_CLOSE,
127 XML_REGEXP_PUNCT_INITQUOTE,
128 XML_REGEXP_PUNCT_FINQUOTE,
129 XML_REGEXP_PUNCT_OTHERS,
130 XML_REGEXP_SEPAR,
131 XML_REGEXP_SEPAR_SPACE,
132 XML_REGEXP_SEPAR_LINE,
133 XML_REGEXP_SEPAR_PARA,
134 XML_REGEXP_SYMBOL,
135 XML_REGEXP_SYMBOL_MATH,
136 XML_REGEXP_SYMBOL_CURRENCY,
137 XML_REGEXP_SYMBOL_MODIFIER,
138 XML_REGEXP_SYMBOL_OTHERS,
139 XML_REGEXP_OTHER,
140 XML_REGEXP_OTHER_CONTROL,
141 XML_REGEXP_OTHER_FORMAT,
142 XML_REGEXP_OTHER_PRIVATE,
143 XML_REGEXP_OTHER_NA,
144 XML_REGEXP_BLOCK_NAME
145} xmlRegAtomType;
146
147typedef enum {
148 XML_REGEXP_QUANT_EPSILON = 1,
149 XML_REGEXP_QUANT_ONCE,
150 XML_REGEXP_QUANT_OPT,
151 XML_REGEXP_QUANT_MULT,
152 XML_REGEXP_QUANT_PLUS,
153 XML_REGEXP_QUANT_ONCEONLY,
154 XML_REGEXP_QUANT_ALL,
155 XML_REGEXP_QUANT_RANGE
156} xmlRegQuantType;
157
158typedef enum {
159 XML_REGEXP_START_STATE = 1,
160 XML_REGEXP_FINAL_STATE,
161 XML_REGEXP_TRANS_STATE,
162 XML_REGEXP_SINK_STATE,
163 XML_REGEXP_UNREACH_STATE
164} xmlRegStateType;
165
166typedef enum {
167 XML_REGEXP_MARK_NORMAL = 0,
168 XML_REGEXP_MARK_START,
169 XML_REGEXP_MARK_VISITED
170} xmlRegMarkedType;
171
172typedef struct _xmlRegRange xmlRegRange;
173typedef xmlRegRange *xmlRegRangePtr;
174
175struct _xmlRegRange {
176 int neg; /* 0 normal, 1 not, 2 exclude */
177 xmlRegAtomType type;
178 int start;
179 int end;
180 xmlChar *blockName;
181};
182
183typedef struct _xmlRegAtom xmlRegAtom;
184typedef xmlRegAtom *xmlRegAtomPtr;
185
186typedef struct _xmlAutomataState xmlRegState;
187typedef xmlRegState *xmlRegStatePtr;
188
189struct _xmlRegAtom {
190 int no;
191 xmlRegAtomType type;
192 xmlRegQuantType quant;
193 int min;
194 int max;
195
196 void *valuep;
197 void *valuep2;
198 int neg;
199 int codepoint;
200 xmlRegStatePtr start;
201 xmlRegStatePtr start0;
202 xmlRegStatePtr stop;
203 int maxRanges;
204 int nbRanges;
205 xmlRegRangePtr *ranges;
206 void *data;
207};
208
209typedef struct _xmlRegCounter xmlRegCounter;
210typedef xmlRegCounter *xmlRegCounterPtr;
211
212struct _xmlRegCounter {
213 int min;
214 int max;
215};
216
217typedef struct _xmlRegTrans xmlRegTrans;
218typedef xmlRegTrans *xmlRegTransPtr;
219
220struct _xmlRegTrans {
221 xmlRegAtomPtr atom;
222 int to;
223 int counter;
224 int count;
225 int nd;
226};
227
228struct _xmlAutomataState {
229 xmlRegStateType type;
230 xmlRegMarkedType mark;
231 xmlRegMarkedType markd;
232 xmlRegMarkedType reached;
233 int no;
234 int maxTrans;
235 int nbTrans;
236 xmlRegTrans *trans;
237 /* knowing states pointing to us can speed things up */
238 int maxTransTo;
239 int nbTransTo;
240 int *transTo;
241};
242
243typedef struct _xmlAutomata xmlRegParserCtxt;
244typedef xmlRegParserCtxt *xmlRegParserCtxtPtr;
245
246#define AM_AUTOMATA_RNG 1
247
248struct _xmlAutomata {
249 xmlChar *string;
250 xmlChar *cur;
251
252 int error;
253 int neg;
254
255 xmlRegStatePtr start;
256 xmlRegStatePtr end;
257 xmlRegStatePtr state;
258
259 xmlRegAtomPtr atom;
260
261 int maxAtoms;
262 int nbAtoms;
263 xmlRegAtomPtr *atoms;
264
265 int maxStates;
266 int nbStates;
267 xmlRegStatePtr *states;
268
269 int maxCounters;
270 int nbCounters;
271 xmlRegCounter *counters;
272
273 int determinist;
274 int negs;
275 int flags;
276
277 int depth;
278};
279
280struct _xmlRegexp {
281 xmlChar *string;
282 int nbStates;
283 xmlRegStatePtr *states;
284 int nbAtoms;
285 xmlRegAtomPtr *atoms;
286 int nbCounters;
287 xmlRegCounter *counters;
288 int determinist;
289 int flags;
290 /*
291 * That's the compact form for determinists automatas
292 */
293 int nbstates;
294 int *compact;
295 void **transdata;
296 int nbstrings;
297 xmlChar **stringMap;
298};
299
300typedef struct _xmlRegExecRollback xmlRegExecRollback;
301typedef xmlRegExecRollback *xmlRegExecRollbackPtr;
302
303struct _xmlRegExecRollback {
304 xmlRegStatePtr state;/* the current state */
305 int index; /* the index in the input stack */
306 int nextbranch; /* the next transition to explore in that state */
307 int *counts; /* save the automata state if it has some */
308};
309
310typedef struct _xmlRegInputToken xmlRegInputToken;
311typedef xmlRegInputToken *xmlRegInputTokenPtr;
312
313struct _xmlRegInputToken {
314 xmlChar *value;
315 void *data;
316};
317
318struct _xmlRegExecCtxt {
319 int status; /* execution status != 0 indicate an error */
320 int determinist; /* did we find an indeterministic behaviour */
321 xmlRegexpPtr comp; /* the compiled regexp */
322 xmlRegExecCallbacks callback;
323 void *data;
324
325 xmlRegStatePtr state;/* the current state */
326 int transno; /* the current transition on that state */
327 int transcount; /* the number of chars in char counted transitions */
328
329 /*
330 * A stack of rollback states
331 */
332 int maxRollbacks;
333 int nbRollbacks;
334 xmlRegExecRollback *rollbacks;
335
336 /*
337 * The state of the automata if any
338 */
339 int *counts;
340
341 /*
342 * The input stack
343 */
344 int inputStackMax;
345 int inputStackNr;
346 int index;
347 int *charStack;
348 const xmlChar *inputString; /* when operating on characters */
349 xmlRegInputTokenPtr inputStack;/* when operating on strings */
350
351 /*
352 * error handling
353 */
354 int errStateNo; /* the error state number */
355 xmlRegStatePtr errState; /* the error state */
356 xmlChar *errString; /* the string raising the error */
357 int *errCounts; /* counters at the error state */
358 int nbPush;
359};
360
361#define REGEXP_ALL_COUNTER 0x123456
362#define REGEXP_ALL_LAX_COUNTER 0x123457
363
364static void xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top);
365static void xmlRegFreeState(xmlRegStatePtr state);
366static void xmlRegFreeAtom(xmlRegAtomPtr atom);
367static int xmlRegStrEqualWildcard(const xmlChar *expStr, const xmlChar *valStr);
368static int xmlRegCheckCharacter(xmlRegAtomPtr atom, int codepoint);
369static int xmlRegCheckCharacterRange(xmlRegAtomType type, int codepoint,
370 int neg, int start, int end, const xmlChar *blockName);
371
372void xmlAutomataSetFlags(xmlAutomataPtr am, int flags);
373
374/************************************************************************
375 * *
376 * Regexp memory error handler *
377 * *
378 ************************************************************************/
379/**
380 * xmlRegexpErrMemory:
381 * @extra: extra information
382 *
383 * Handle an out of memory condition
384 */
385static void
386xmlRegexpErrMemory(xmlRegParserCtxtPtr ctxt, const char *extra)
387{
388 const char *regexp = NULL;
389 if (ctxt != NULL) {
390 regexp = (const char *) ctxt->string;
391 ctxt->error = XML_ERR_NO_MEMORY;
392 }
393 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_REGEXP,
394 XML_ERR_NO_MEMORY, XML_ERR_FATAL, NULL, 0, extra,
395 regexp, NULL, 0, 0,
396 "Memory allocation failed : %s\n", extra);
397}
398
399/**
400 * xmlRegexpErrCompile:
401 * @extra: extra information
402 *
403 * Handle a compilation failure
404 */
405static void
406xmlRegexpErrCompile(xmlRegParserCtxtPtr ctxt, const char *extra)
407{
408 const char *regexp = NULL;
409 int idx = 0;
410
411 if (ctxt != NULL) {
412 regexp = (const char *) ctxt->string;
413 idx = ctxt->cur - ctxt->string;
414 ctxt->error = XML_REGEXP_COMPILE_ERROR;
415 }
416 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_REGEXP,
417 XML_REGEXP_COMPILE_ERROR, XML_ERR_FATAL, NULL, 0, extra,
418 regexp, NULL, idx, 0,
419 "failed to compile: %s\n", extra);
420}
421
422/************************************************************************
423 * *
424 * Allocation/Deallocation *
425 * *
426 ************************************************************************/
427
428static int xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt);
429
430/**
431 * xmlRegCalloc2:
432 * @dim1: size of first dimension
433 * @dim2: size of second dimension
434 * @elemSize: size of element
435 *
436 * Allocate a two-dimensional array and set all elements to zero.
437 *
438 * Returns the new array or NULL in case of error.
439 */
440static void*
441xmlRegCalloc2(size_t dim1, size_t dim2, size_t elemSize) {
442 size_t totalSize;
443 void *ret;
444
445 /* Check for overflow */
446 if (dim1 > SIZE_MAX / dim2 / elemSize)
447 return (NULL);
448 totalSize = dim1 * dim2 * elemSize;
449 ret = xmlMalloc(totalSize);
450 if (ret != NULL)
451 memset(ret, 0, totalSize);
452 return (ret);
453}
454
455/**
456 * xmlRegEpxFromParse:
457 * @ctxt: the parser context used to build it
458 *
459 * Allocate a new regexp and fill it with the result from the parser
460 *
461 * Returns the new regexp or NULL in case of error
462 */
463static xmlRegexpPtr
464xmlRegEpxFromParse(xmlRegParserCtxtPtr ctxt) {
465 xmlRegexpPtr ret;
466
467 ret = (xmlRegexpPtr) xmlMalloc(sizeof(xmlRegexp));
468 if (ret == NULL) {
469 xmlRegexpErrMemory(ctxt, "compiling regexp");
470 return(NULL);
471 }
472 memset(ret, 0, sizeof(xmlRegexp));
473 ret->string = ctxt->string;
474 ret->nbStates = ctxt->nbStates;
475 ret->states = ctxt->states;
476 ret->nbAtoms = ctxt->nbAtoms;
477 ret->atoms = ctxt->atoms;
478 ret->nbCounters = ctxt->nbCounters;
479 ret->counters = ctxt->counters;
480 ret->determinist = ctxt->determinist;
481 ret->flags = ctxt->flags;
482 if (ret->determinist == -1) {
483 xmlRegexpIsDeterminist(ret);
484 }
485
486 if ((ret->determinist != 0) &&
487 (ret->nbCounters == 0) &&
488 (ctxt->negs == 0) &&
489 (ret->atoms != NULL) &&
490 (ret->atoms[0] != NULL) &&
491 (ret->atoms[0]->type == XML_REGEXP_STRING)) {
492 int i, j, nbstates = 0, nbatoms = 0;
493 int *stateRemap;
494 int *stringRemap;
495 int *transitions;
496 void **transdata;
497 xmlChar **stringMap;
498 xmlChar *value;
499
500 /*
501 * Switch to a compact representation
502 * 1/ counting the effective number of states left
503 * 2/ counting the unique number of atoms, and check that
504 * they are all of the string type
505 * 3/ build a table state x atom for the transitions
506 */
507
508 stateRemap = xmlMalloc(ret->nbStates * sizeof(int));
509 if (stateRemap == NULL) {
510 xmlRegexpErrMemory(ctxt, "compiling regexp");
511 xmlFree(ret);
512 return(NULL);
513 }
514 for (i = 0;i < ret->nbStates;i++) {
515 if (ret->states[i] != NULL) {
516 stateRemap[i] = nbstates;
517 nbstates++;
518 } else {
519 stateRemap[i] = -1;
520 }
521 }
522#ifdef DEBUG_COMPACTION
523 printf("Final: %d states\n", nbstates);
524#endif
525 stringMap = xmlMalloc(ret->nbAtoms * sizeof(char *));
526 if (stringMap == NULL) {
527 xmlRegexpErrMemory(ctxt, "compiling regexp");
528 xmlFree(stateRemap);
529 xmlFree(ret);
530 return(NULL);
531 }
532 stringRemap = xmlMalloc(ret->nbAtoms * sizeof(int));
533 if (stringRemap == NULL) {
534 xmlRegexpErrMemory(ctxt, "compiling regexp");
535 xmlFree(stringMap);
536 xmlFree(stateRemap);
537 xmlFree(ret);
538 return(NULL);
539 }
540 for (i = 0;i < ret->nbAtoms;i++) {
541 if ((ret->atoms[i]->type == XML_REGEXP_STRING) &&
542 (ret->atoms[i]->quant == XML_REGEXP_QUANT_ONCE)) {
543 value = ret->atoms[i]->valuep;
544 for (j = 0;j < nbatoms;j++) {
545 if (xmlStrEqual(stringMap[j], value)) {
546 stringRemap[i] = j;
547 break;
548 }
549 }
550 if (j >= nbatoms) {
551 stringRemap[i] = nbatoms;
552 stringMap[nbatoms] = xmlStrdup(value);
553 if (stringMap[nbatoms] == NULL) {
554 for (i = 0;i < nbatoms;i++)
555 xmlFree(stringMap[i]);
556 xmlFree(stringRemap);
557 xmlFree(stringMap);
558 xmlFree(stateRemap);
559 xmlFree(ret);
560 return(NULL);
561 }
562 nbatoms++;
563 }
564 } else {
565 xmlFree(stateRemap);
566 xmlFree(stringRemap);
567 for (i = 0;i < nbatoms;i++)
568 xmlFree(stringMap[i]);
569 xmlFree(stringMap);
570 xmlFree(ret);
571 return(NULL);
572 }
573 }
574#ifdef DEBUG_COMPACTION
575 printf("Final: %d atoms\n", nbatoms);
576#endif
577 transitions = (int *) xmlRegCalloc2(nbstates + 1, nbatoms + 1,
578 sizeof(int));
579 if (transitions == NULL) {
580 xmlFree(stateRemap);
581 xmlFree(stringRemap);
582 for (i = 0;i < nbatoms;i++)
583 xmlFree(stringMap[i]);
584 xmlFree(stringMap);
585 xmlFree(ret);
586 return(NULL);
587 }
588
589 /*
590 * Allocate the transition table. The first entry for each
591 * state corresponds to the state type.
592 */
593 transdata = NULL;
594
595 for (i = 0;i < ret->nbStates;i++) {
596 int stateno, atomno, targetno, prev;
597 xmlRegStatePtr state;
598 xmlRegTransPtr trans;
599
600 stateno = stateRemap[i];
601 if (stateno == -1)
602 continue;
603 state = ret->states[i];
604
605 transitions[stateno * (nbatoms + 1)] = state->type;
606
607 for (j = 0;j < state->nbTrans;j++) {
608 trans = &(state->trans[j]);
609 if ((trans->to == -1) || (trans->atom == NULL))
610 continue;
611 atomno = stringRemap[trans->atom->no];
612 if ((trans->atom->data != NULL) && (transdata == NULL)) {
613 transdata = (void **) xmlRegCalloc2(nbstates, nbatoms,
614 sizeof(void *));
615 if (transdata == NULL) {
616 xmlRegexpErrMemory(ctxt, "compiling regexp");
617 break;
618 }
619 }
620 targetno = stateRemap[trans->to];
621 /*
622 * if the same atom can generate transitions to 2 different
623 * states then it means the automata is not deterministic and
624 * the compact form can't be used !
625 */
626 prev = transitions[stateno * (nbatoms + 1) + atomno + 1];
627 if (prev != 0) {
628 if (prev != targetno + 1) {
629 ret->determinist = 0;
630#ifdef DEBUG_COMPACTION
631 printf("Indet: state %d trans %d, atom %d to %d : %d to %d\n",
632 i, j, trans->atom->no, trans->to, atomno, targetno);
633 printf(" previous to is %d\n", prev);
634#endif
635 if (transdata != NULL)
636 xmlFree(transdata);
637 xmlFree(transitions);
638 xmlFree(stateRemap);
639 xmlFree(stringRemap);
640 for (i = 0;i < nbatoms;i++)
641 xmlFree(stringMap[i]);
642 xmlFree(stringMap);
643 goto not_determ;
644 }
645 } else {
646#if 0
647 printf("State %d trans %d: atom %d to %d : %d to %d\n",
648 i, j, trans->atom->no, trans->to, atomno, targetno);
649#endif
650 transitions[stateno * (nbatoms + 1) + atomno + 1] =
651 targetno + 1; /* to avoid 0 */
652 if (transdata != NULL)
653 transdata[stateno * nbatoms + atomno] =
654 trans->atom->data;
655 }
656 }
657 }
658 ret->determinist = 1;
659#ifdef DEBUG_COMPACTION
660 /*
661 * Debug
662 */
663 for (i = 0;i < nbstates;i++) {
664 for (j = 0;j < nbatoms + 1;j++) {
665 printf("%02d ", transitions[i * (nbatoms + 1) + j]);
666 }
667 printf("\n");
668 }
669 printf("\n");
670#endif
671 /*
672 * Cleanup of the old data
673 */
674 if (ret->states != NULL) {
675 for (i = 0;i < ret->nbStates;i++)
676 xmlRegFreeState(ret->states[i]);
677 xmlFree(ret->states);
678 }
679 ret->states = NULL;
680 ret->nbStates = 0;
681 if (ret->atoms != NULL) {
682 for (i = 0;i < ret->nbAtoms;i++)
683 xmlRegFreeAtom(ret->atoms[i]);
684 xmlFree(ret->atoms);
685 }
686 ret->atoms = NULL;
687 ret->nbAtoms = 0;
688
689 ret->compact = transitions;
690 ret->transdata = transdata;
691 ret->stringMap = stringMap;
692 ret->nbstrings = nbatoms;
693 ret->nbstates = nbstates;
694 xmlFree(stateRemap);
695 xmlFree(stringRemap);
696 }
697not_determ:
698 ctxt->string = NULL;
699 ctxt->nbStates = 0;
700 ctxt->states = NULL;
701 ctxt->nbAtoms = 0;
702 ctxt->atoms = NULL;
703 ctxt->nbCounters = 0;
704 ctxt->counters = NULL;
705 return(ret);
706}
707
708/**
709 * xmlRegNewParserCtxt:
710 * @string: the string to parse
711 *
712 * Allocate a new regexp parser context
713 *
714 * Returns the new context or NULL in case of error
715 */
716static xmlRegParserCtxtPtr
717xmlRegNewParserCtxt(const xmlChar *string) {
718 xmlRegParserCtxtPtr ret;
719
720 ret = (xmlRegParserCtxtPtr) xmlMalloc(sizeof(xmlRegParserCtxt));
721 if (ret == NULL)
722 return(NULL);
723 memset(ret, 0, sizeof(xmlRegParserCtxt));
724 if (string != NULL)
725 ret->string = xmlStrdup(string);
726 ret->cur = ret->string;
727 ret->neg = 0;
728 ret->negs = 0;
729 ret->error = 0;
730 ret->determinist = -1;
731 return(ret);
732}
733
734/**
735 * xmlRegNewRange:
736 * @ctxt: the regexp parser context
737 * @neg: is that negative
738 * @type: the type of range
739 * @start: the start codepoint
740 * @end: the end codepoint
741 *
742 * Allocate a new regexp range
743 *
744 * Returns the new range or NULL in case of error
745 */
746static xmlRegRangePtr
747xmlRegNewRange(xmlRegParserCtxtPtr ctxt,
748 int neg, xmlRegAtomType type, int start, int end) {
749 xmlRegRangePtr ret;
750
751 ret = (xmlRegRangePtr) xmlMalloc(sizeof(xmlRegRange));
752 if (ret == NULL) {
753 xmlRegexpErrMemory(ctxt, "allocating range");
754 return(NULL);
755 }
756 ret->neg = neg;
757 ret->type = type;
758 ret->start = start;
759 ret->end = end;
760 return(ret);
761}
762
763/**
764 * xmlRegFreeRange:
765 * @range: the regexp range
766 *
767 * Free a regexp range
768 */
769static void
770xmlRegFreeRange(xmlRegRangePtr range) {
771 if (range == NULL)
772 return;
773
774 if (range->blockName != NULL)
775 xmlFree(range->blockName);
776 xmlFree(range);
777}
778
779/**
780 * xmlRegCopyRange:
781 * @range: the regexp range
782 *
783 * Copy a regexp range
784 *
785 * Returns the new copy or NULL in case of error.
786 */
787static xmlRegRangePtr
788xmlRegCopyRange(xmlRegParserCtxtPtr ctxt, xmlRegRangePtr range) {
789 xmlRegRangePtr ret;
790
791 if (range == NULL)
792 return(NULL);
793
794 ret = xmlRegNewRange(ctxt, range->neg, range->type, range->start,
795 range->end);
796 if (ret == NULL)
797 return(NULL);
798 if (range->blockName != NULL) {
799 ret->blockName = xmlStrdup(range->blockName);
800 if (ret->blockName == NULL) {
801 xmlRegexpErrMemory(ctxt, "allocating range");
802 xmlRegFreeRange(ret);
803 return(NULL);
804 }
805 }
806 return(ret);
807}
808
809/**
810 * xmlRegNewAtom:
811 * @ctxt: the regexp parser context
812 * @type: the type of atom
813 *
814 * Allocate a new atom
815 *
816 * Returns the new atom or NULL in case of error
817 */
818static xmlRegAtomPtr
819xmlRegNewAtom(xmlRegParserCtxtPtr ctxt, xmlRegAtomType type) {
820 xmlRegAtomPtr ret;
821
822 ret = (xmlRegAtomPtr) xmlMalloc(sizeof(xmlRegAtom));
823 if (ret == NULL) {
824 xmlRegexpErrMemory(ctxt, "allocating atom");
825 return(NULL);
826 }
827 memset(ret, 0, sizeof(xmlRegAtom));
828 ret->type = type;
829 ret->quant = XML_REGEXP_QUANT_ONCE;
830 ret->min = 0;
831 ret->max = 0;
832 return(ret);
833}
834
835/**
836 * xmlRegFreeAtom:
837 * @atom: the regexp atom
838 *
839 * Free a regexp atom
840 */
841static void
842xmlRegFreeAtom(xmlRegAtomPtr atom) {
843 int i;
844
845 if (atom == NULL)
846 return;
847
848 for (i = 0;i < atom->nbRanges;i++)
849 xmlRegFreeRange(atom->ranges[i]);
850 if (atom->ranges != NULL)
851 xmlFree(atom->ranges);
852 if ((atom->type == XML_REGEXP_STRING) && (atom->valuep != NULL))
853 xmlFree(atom->valuep);
854 if ((atom->type == XML_REGEXP_STRING) && (atom->valuep2 != NULL))
855 xmlFree(atom->valuep2);
856 if ((atom->type == XML_REGEXP_BLOCK_NAME) && (atom->valuep != NULL))
857 xmlFree(atom->valuep);
858 xmlFree(atom);
859}
860
861/**
862 * xmlRegCopyAtom:
863 * @ctxt: the regexp parser context
864 * @atom: the original atom
865 *
866 * Allocate a new regexp range
867 *
868 * Returns the new atom or NULL in case of error
869 */
870static xmlRegAtomPtr
871xmlRegCopyAtom(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom) {
872 xmlRegAtomPtr ret;
873
874 ret = (xmlRegAtomPtr) xmlMalloc(sizeof(xmlRegAtom));
875 if (ret == NULL) {
876 xmlRegexpErrMemory(ctxt, "copying atom");
877 return(NULL);
878 }
879 memset(ret, 0, sizeof(xmlRegAtom));
880 ret->type = atom->type;
881 ret->quant = atom->quant;
882 ret->min = atom->min;
883 ret->max = atom->max;
884 if (atom->nbRanges > 0) {
885 int i;
886
887 ret->ranges = (xmlRegRangePtr *) xmlMalloc(sizeof(xmlRegRangePtr) *
888 atom->nbRanges);
889 if (ret->ranges == NULL) {
890 xmlRegexpErrMemory(ctxt, "copying atom");
891 goto error;
892 }
893 for (i = 0;i < atom->nbRanges;i++) {
894 ret->ranges[i] = xmlRegCopyRange(ctxt, atom->ranges[i]);
895 if (ret->ranges[i] == NULL)
896 goto error;
897 ret->nbRanges = i + 1;
898 }
899 }
900 return(ret);
901
902error:
903 xmlRegFreeAtom(ret);
904 return(NULL);
905}
906
907static xmlRegStatePtr
908xmlRegNewState(xmlRegParserCtxtPtr ctxt) {
909 xmlRegStatePtr ret;
910
911 ret = (xmlRegStatePtr) xmlMalloc(sizeof(xmlRegState));
912 if (ret == NULL) {
913 xmlRegexpErrMemory(ctxt, "allocating state");
914 return(NULL);
915 }
916 memset(ret, 0, sizeof(xmlRegState));
917 ret->type = XML_REGEXP_TRANS_STATE;
918 ret->mark = XML_REGEXP_MARK_NORMAL;
919 return(ret);
920}
921
922/**
923 * xmlRegFreeState:
924 * @state: the regexp state
925 *
926 * Free a regexp state
927 */
928static void
929xmlRegFreeState(xmlRegStatePtr state) {
930 if (state == NULL)
931 return;
932
933 if (state->trans != NULL)
934 xmlFree(state->trans);
935 if (state->transTo != NULL)
936 xmlFree(state->transTo);
937 xmlFree(state);
938}
939
940/**
941 * xmlRegFreeParserCtxt:
942 * @ctxt: the regexp parser context
943 *
944 * Free a regexp parser context
945 */
946static void
947xmlRegFreeParserCtxt(xmlRegParserCtxtPtr ctxt) {
948 int i;
949 if (ctxt == NULL)
950 return;
951
952 if (ctxt->string != NULL)
953 xmlFree(ctxt->string);
954 if (ctxt->states != NULL) {
955 for (i = 0;i < ctxt->nbStates;i++)
956 xmlRegFreeState(ctxt->states[i]);
957 xmlFree(ctxt->states);
958 }
959 if (ctxt->atoms != NULL) {
960 for (i = 0;i < ctxt->nbAtoms;i++)
961 xmlRegFreeAtom(ctxt->atoms[i]);
962 xmlFree(ctxt->atoms);
963 }
964 if (ctxt->counters != NULL)
965 xmlFree(ctxt->counters);
966 xmlFree(ctxt);
967}
968
969/************************************************************************
970 * *
971 * Display of Data structures *
972 * *
973 ************************************************************************/
974
975static void
976xmlRegPrintAtomType(FILE *output, xmlRegAtomType type) {
977 switch (type) {
978 case XML_REGEXP_EPSILON:
979 fprintf(output, "epsilon "); break;
980 case XML_REGEXP_CHARVAL:
981 fprintf(output, "charval "); break;
982 case XML_REGEXP_RANGES:
983 fprintf(output, "ranges "); break;
984 case XML_REGEXP_SUBREG:
985 fprintf(output, "subexpr "); break;
986 case XML_REGEXP_STRING:
987 fprintf(output, "string "); break;
988 case XML_REGEXP_ANYCHAR:
989 fprintf(output, "anychar "); break;
990 case XML_REGEXP_ANYSPACE:
991 fprintf(output, "anyspace "); break;
992 case XML_REGEXP_NOTSPACE:
993 fprintf(output, "notspace "); break;
994 case XML_REGEXP_INITNAME:
995 fprintf(output, "initname "); break;
996 case XML_REGEXP_NOTINITNAME:
997 fprintf(output, "notinitname "); break;
998 case XML_REGEXP_NAMECHAR:
999 fprintf(output, "namechar "); break;
1000 case XML_REGEXP_NOTNAMECHAR:
1001 fprintf(output, "notnamechar "); break;
1002 case XML_REGEXP_DECIMAL:
1003 fprintf(output, "decimal "); break;
1004 case XML_REGEXP_NOTDECIMAL:
1005 fprintf(output, "notdecimal "); break;
1006 case XML_REGEXP_REALCHAR:
1007 fprintf(output, "realchar "); break;
1008 case XML_REGEXP_NOTREALCHAR:
1009 fprintf(output, "notrealchar "); break;
1010 case XML_REGEXP_LETTER:
1011 fprintf(output, "LETTER "); break;
1012 case XML_REGEXP_LETTER_UPPERCASE:
1013 fprintf(output, "LETTER_UPPERCASE "); break;
1014 case XML_REGEXP_LETTER_LOWERCASE:
1015 fprintf(output, "LETTER_LOWERCASE "); break;
1016 case XML_REGEXP_LETTER_TITLECASE:
1017 fprintf(output, "LETTER_TITLECASE "); break;
1018 case XML_REGEXP_LETTER_MODIFIER:
1019 fprintf(output, "LETTER_MODIFIER "); break;
1020 case XML_REGEXP_LETTER_OTHERS:
1021 fprintf(output, "LETTER_OTHERS "); break;
1022 case XML_REGEXP_MARK:
1023 fprintf(output, "MARK "); break;
1024 case XML_REGEXP_MARK_NONSPACING:
1025 fprintf(output, "MARK_NONSPACING "); break;
1026 case XML_REGEXP_MARK_SPACECOMBINING:
1027 fprintf(output, "MARK_SPACECOMBINING "); break;
1028 case XML_REGEXP_MARK_ENCLOSING:
1029 fprintf(output, "MARK_ENCLOSING "); break;
1030 case XML_REGEXP_NUMBER:
1031 fprintf(output, "NUMBER "); break;
1032 case XML_REGEXP_NUMBER_DECIMAL:
1033 fprintf(output, "NUMBER_DECIMAL "); break;
1034 case XML_REGEXP_NUMBER_LETTER:
1035 fprintf(output, "NUMBER_LETTER "); break;
1036 case XML_REGEXP_NUMBER_OTHERS:
1037 fprintf(output, "NUMBER_OTHERS "); break;
1038 case XML_REGEXP_PUNCT:
1039 fprintf(output, "PUNCT "); break;
1040 case XML_REGEXP_PUNCT_CONNECTOR:
1041 fprintf(output, "PUNCT_CONNECTOR "); break;
1042 case XML_REGEXP_PUNCT_DASH:
1043 fprintf(output, "PUNCT_DASH "); break;
1044 case XML_REGEXP_PUNCT_OPEN:
1045 fprintf(output, "PUNCT_OPEN "); break;
1046 case XML_REGEXP_PUNCT_CLOSE:
1047 fprintf(output, "PUNCT_CLOSE "); break;
1048 case XML_REGEXP_PUNCT_INITQUOTE:
1049 fprintf(output, "PUNCT_INITQUOTE "); break;
1050 case XML_REGEXP_PUNCT_FINQUOTE:
1051 fprintf(output, "PUNCT_FINQUOTE "); break;
1052 case XML_REGEXP_PUNCT_OTHERS:
1053 fprintf(output, "PUNCT_OTHERS "); break;
1054 case XML_REGEXP_SEPAR:
1055 fprintf(output, "SEPAR "); break;
1056 case XML_REGEXP_SEPAR_SPACE:
1057 fprintf(output, "SEPAR_SPACE "); break;
1058 case XML_REGEXP_SEPAR_LINE:
1059 fprintf(output, "SEPAR_LINE "); break;
1060 case XML_REGEXP_SEPAR_PARA:
1061 fprintf(output, "SEPAR_PARA "); break;
1062 case XML_REGEXP_SYMBOL:
1063 fprintf(output, "SYMBOL "); break;
1064 case XML_REGEXP_SYMBOL_MATH:
1065 fprintf(output, "SYMBOL_MATH "); break;
1066 case XML_REGEXP_SYMBOL_CURRENCY:
1067 fprintf(output, "SYMBOL_CURRENCY "); break;
1068 case XML_REGEXP_SYMBOL_MODIFIER:
1069 fprintf(output, "SYMBOL_MODIFIER "); break;
1070 case XML_REGEXP_SYMBOL_OTHERS:
1071 fprintf(output, "SYMBOL_OTHERS "); break;
1072 case XML_REGEXP_OTHER:
1073 fprintf(output, "OTHER "); break;
1074 case XML_REGEXP_OTHER_CONTROL:
1075 fprintf(output, "OTHER_CONTROL "); break;
1076 case XML_REGEXP_OTHER_FORMAT:
1077 fprintf(output, "OTHER_FORMAT "); break;
1078 case XML_REGEXP_OTHER_PRIVATE:
1079 fprintf(output, "OTHER_PRIVATE "); break;
1080 case XML_REGEXP_OTHER_NA:
1081 fprintf(output, "OTHER_NA "); break;
1082 case XML_REGEXP_BLOCK_NAME:
1083 fprintf(output, "BLOCK "); break;
1084 }
1085}
1086
1087static void
1088xmlRegPrintQuantType(FILE *output, xmlRegQuantType type) {
1089 switch (type) {
1090 case XML_REGEXP_QUANT_EPSILON:
1091 fprintf(output, "epsilon "); break;
1092 case XML_REGEXP_QUANT_ONCE:
1093 fprintf(output, "once "); break;
1094 case XML_REGEXP_QUANT_OPT:
1095 fprintf(output, "? "); break;
1096 case XML_REGEXP_QUANT_MULT:
1097 fprintf(output, "* "); break;
1098 case XML_REGEXP_QUANT_PLUS:
1099 fprintf(output, "+ "); break;
1100 case XML_REGEXP_QUANT_RANGE:
1101 fprintf(output, "range "); break;
1102 case XML_REGEXP_QUANT_ONCEONLY:
1103 fprintf(output, "onceonly "); break;
1104 case XML_REGEXP_QUANT_ALL:
1105 fprintf(output, "all "); break;
1106 }
1107}
1108static void
1109xmlRegPrintRange(FILE *output, xmlRegRangePtr range) {
1110 fprintf(output, " range: ");
1111 if (range->neg)
1112 fprintf(output, "negative ");
1113 xmlRegPrintAtomType(output, range->type);
1114 fprintf(output, "%c - %c\n", range->start, range->end);
1115}
1116
1117static void
1118xmlRegPrintAtom(FILE *output, xmlRegAtomPtr atom) {
1119 fprintf(output, " atom: ");
1120 if (atom == NULL) {
1121 fprintf(output, "NULL\n");
1122 return;
1123 }
1124 if (atom->neg)
1125 fprintf(output, "not ");
1126 xmlRegPrintAtomType(output, atom->type);
1127 xmlRegPrintQuantType(output, atom->quant);
1128 if (atom->quant == XML_REGEXP_QUANT_RANGE)
1129 fprintf(output, "%d-%d ", atom->min, atom->max);
1130 if (atom->type == XML_REGEXP_STRING)
1131 fprintf(output, "'%s' ", (char *) atom->valuep);
1132 if (atom->type == XML_REGEXP_CHARVAL)
1133 fprintf(output, "char %c\n", atom->codepoint);
1134 else if (atom->type == XML_REGEXP_RANGES) {
1135 int i;
1136 fprintf(output, "%d entries\n", atom->nbRanges);
1137 for (i = 0; i < atom->nbRanges;i++)
1138 xmlRegPrintRange(output, atom->ranges[i]);
1139 } else if (atom->type == XML_REGEXP_SUBREG) {
1140 fprintf(output, "start %d end %d\n", atom->start->no, atom->stop->no);
1141 } else {
1142 fprintf(output, "\n");
1143 }
1144}
1145
1146static void
1147xmlRegPrintTrans(FILE *output, xmlRegTransPtr trans) {
1148 fprintf(output, " trans: ");
1149 if (trans == NULL) {
1150 fprintf(output, "NULL\n");
1151 return;
1152 }
1153 if (trans->to < 0) {
1154 fprintf(output, "removed\n");
1155 return;
1156 }
1157 if (trans->nd != 0) {
1158 if (trans->nd == 2)
1159 fprintf(output, "last not determinist, ");
1160 else
1161 fprintf(output, "not determinist, ");
1162 }
1163 if (trans->counter >= 0) {
1164 fprintf(output, "counted %d, ", trans->counter);
1165 }
1166 if (trans->count == REGEXP_ALL_COUNTER) {
1167 fprintf(output, "all transition, ");
1168 } else if (trans->count >= 0) {
1169 fprintf(output, "count based %d, ", trans->count);
1170 }
1171 if (trans->atom == NULL) {
1172 fprintf(output, "epsilon to %d\n", trans->to);
1173 return;
1174 }
1175 if (trans->atom->type == XML_REGEXP_CHARVAL)
1176 fprintf(output, "char %c ", trans->atom->codepoint);
1177 fprintf(output, "atom %d, to %d\n", trans->atom->no, trans->to);
1178}
1179
1180static void
1181xmlRegPrintState(FILE *output, xmlRegStatePtr state) {
1182 int i;
1183
1184 fprintf(output, " state: ");
1185 if (state == NULL) {
1186 fprintf(output, "NULL\n");
1187 return;
1188 }
1189 if (state->type == XML_REGEXP_START_STATE)
1190 fprintf(output, "START ");
1191 if (state->type == XML_REGEXP_FINAL_STATE)
1192 fprintf(output, "FINAL ");
1193
1194 fprintf(output, "%d, %d transitions:\n", state->no, state->nbTrans);
1195 for (i = 0;i < state->nbTrans; i++) {
1196 xmlRegPrintTrans(output, &(state->trans[i]));
1197 }
1198}
1199
1200#ifdef DEBUG_REGEXP_GRAPH
1201static void
1202xmlRegPrintCtxt(FILE *output, xmlRegParserCtxtPtr ctxt) {
1203 int i;
1204
1205 fprintf(output, " ctxt: ");
1206 if (ctxt == NULL) {
1207 fprintf(output, "NULL\n");
1208 return;
1209 }
1210 fprintf(output, "'%s' ", ctxt->string);
1211 if (ctxt->error)
1212 fprintf(output, "error ");
1213 if (ctxt->neg)
1214 fprintf(output, "neg ");
1215 fprintf(output, "\n");
1216 fprintf(output, "%d atoms:\n", ctxt->nbAtoms);
1217 for (i = 0;i < ctxt->nbAtoms; i++) {
1218 fprintf(output, " %02d ", i);
1219 xmlRegPrintAtom(output, ctxt->atoms[i]);
1220 }
1221 if (ctxt->atom != NULL) {
1222 fprintf(output, "current atom:\n");
1223 xmlRegPrintAtom(output, ctxt->atom);
1224 }
1225 fprintf(output, "%d states:", ctxt->nbStates);
1226 if (ctxt->start != NULL)
1227 fprintf(output, " start: %d", ctxt->start->no);
1228 if (ctxt->end != NULL)
1229 fprintf(output, " end: %d", ctxt->end->no);
1230 fprintf(output, "\n");
1231 for (i = 0;i < ctxt->nbStates; i++) {
1232 xmlRegPrintState(output, ctxt->states[i]);
1233 }
1234 fprintf(output, "%d counters:\n", ctxt->nbCounters);
1235 for (i = 0;i < ctxt->nbCounters; i++) {
1236 fprintf(output, " %d: min %d max %d\n", i, ctxt->counters[i].min,
1237 ctxt->counters[i].max);
1238 }
1239}
1240#endif
1241
1242/************************************************************************
1243 * *
1244 * Finite Automata structures manipulations *
1245 * *
1246 ************************************************************************/
1247
1248static void
1249xmlRegAtomAddRange(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom,
1250 int neg, xmlRegAtomType type, int start, int end,
1251 xmlChar *blockName) {
1252 xmlRegRangePtr range;
1253
1254 if (atom == NULL) {
1255 ERROR("add range: atom is NULL");
1256 return;
1257 }
1258 if (atom->type != XML_REGEXP_RANGES) {
1259 ERROR("add range: atom is not ranges");
1260 return;
1261 }
1262 if (atom->maxRanges == 0) {
1263 atom->maxRanges = 4;
1264 atom->ranges = (xmlRegRangePtr *) xmlMalloc(atom->maxRanges *
1265 sizeof(xmlRegRangePtr));
1266 if (atom->ranges == NULL) {
1267 xmlRegexpErrMemory(ctxt, "adding ranges");
1268 atom->maxRanges = 0;
1269 return;
1270 }
1271 } else if (atom->nbRanges >= atom->maxRanges) {
1272 xmlRegRangePtr *tmp;
1273 atom->maxRanges *= 2;
1274 tmp = (xmlRegRangePtr *) xmlRealloc(atom->ranges, atom->maxRanges *
1275 sizeof(xmlRegRangePtr));
1276 if (tmp == NULL) {
1277 xmlRegexpErrMemory(ctxt, "adding ranges");
1278 atom->maxRanges /= 2;
1279 return;
1280 }
1281 atom->ranges = tmp;
1282 }
1283 range = xmlRegNewRange(ctxt, neg, type, start, end);
1284 if (range == NULL)
1285 return;
1286 range->blockName = blockName;
1287 atom->ranges[atom->nbRanges++] = range;
1288
1289}
1290
1291static int
1292xmlRegGetCounter(xmlRegParserCtxtPtr ctxt) {
1293 if (ctxt->maxCounters == 0) {
1294 ctxt->maxCounters = 4;
1295 ctxt->counters = (xmlRegCounter *) xmlMalloc(ctxt->maxCounters *
1296 sizeof(xmlRegCounter));
1297 if (ctxt->counters == NULL) {
1298 xmlRegexpErrMemory(ctxt, "allocating counter");
1299 ctxt->maxCounters = 0;
1300 return(-1);
1301 }
1302 } else if (ctxt->nbCounters >= ctxt->maxCounters) {
1303 xmlRegCounter *tmp;
1304 ctxt->maxCounters *= 2;
1305 tmp = (xmlRegCounter *) xmlRealloc(ctxt->counters, ctxt->maxCounters *
1306 sizeof(xmlRegCounter));
1307 if (tmp == NULL) {
1308 xmlRegexpErrMemory(ctxt, "allocating counter");
1309 ctxt->maxCounters /= 2;
1310 return(-1);
1311 }
1312 ctxt->counters = tmp;
1313 }
1314 ctxt->counters[ctxt->nbCounters].min = -1;
1315 ctxt->counters[ctxt->nbCounters].max = -1;
1316 return(ctxt->nbCounters++);
1317}
1318
1319static int
1320xmlRegAtomPush(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom) {
1321 if (atom == NULL) {
1322 ERROR("atom push: atom is NULL");
1323 return(-1);
1324 }
1325 if (ctxt->maxAtoms == 0) {
1326 ctxt->maxAtoms = 4;
1327 ctxt->atoms = (xmlRegAtomPtr *) xmlMalloc(ctxt->maxAtoms *
1328 sizeof(xmlRegAtomPtr));
1329 if (ctxt->atoms == NULL) {
1330 xmlRegexpErrMemory(ctxt, "pushing atom");
1331 ctxt->maxAtoms = 0;
1332 return(-1);
1333 }
1334 } else if (ctxt->nbAtoms >= ctxt->maxAtoms) {
1335 xmlRegAtomPtr *tmp;
1336 ctxt->maxAtoms *= 2;
1337 tmp = (xmlRegAtomPtr *) xmlRealloc(ctxt->atoms, ctxt->maxAtoms *
1338 sizeof(xmlRegAtomPtr));
1339 if (tmp == NULL) {
1340 xmlRegexpErrMemory(ctxt, "allocating counter");
1341 ctxt->maxAtoms /= 2;
1342 return(-1);
1343 }
1344 ctxt->atoms = tmp;
1345 }
1346 atom->no = ctxt->nbAtoms;
1347 ctxt->atoms[ctxt->nbAtoms++] = atom;
1348 return(0);
1349}
1350
1351static void
1352xmlRegStateAddTransTo(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr target,
1353 int from) {
1354 if (target->maxTransTo == 0) {
1355 target->maxTransTo = 8;
1356 target->transTo = (int *) xmlMalloc(target->maxTransTo *
1357 sizeof(int));
1358 if (target->transTo == NULL) {
1359 xmlRegexpErrMemory(ctxt, "adding transition");
1360 target->maxTransTo = 0;
1361 return;
1362 }
1363 } else if (target->nbTransTo >= target->maxTransTo) {
1364 int *tmp;
1365 target->maxTransTo *= 2;
1366 tmp = (int *) xmlRealloc(target->transTo, target->maxTransTo *
1367 sizeof(int));
1368 if (tmp == NULL) {
1369 xmlRegexpErrMemory(ctxt, "adding transition");
1370 target->maxTransTo /= 2;
1371 return;
1372 }
1373 target->transTo = tmp;
1374 }
1375 target->transTo[target->nbTransTo] = from;
1376 target->nbTransTo++;
1377}
1378
1379static void
1380xmlRegStateAddTrans(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
1381 xmlRegAtomPtr atom, xmlRegStatePtr target,
1382 int counter, int count) {
1383
1384 int nrtrans;
1385
1386 if (state == NULL) {
1387 ERROR("add state: state is NULL");
1388 return;
1389 }
1390 if (target == NULL) {
1391 ERROR("add state: target is NULL");
1392 return;
1393 }
1394 /*
1395 * Other routines follow the philosophy 'When in doubt, add a transition'
1396 * so we check here whether such a transition is already present and, if
1397 * so, silently ignore this request.
1398 */
1399
1400 for (nrtrans = state->nbTrans - 1; nrtrans >= 0; nrtrans--) {
1401 xmlRegTransPtr trans = &(state->trans[nrtrans]);
1402 if ((trans->atom == atom) &&
1403 (trans->to == target->no) &&
1404 (trans->counter == counter) &&
1405 (trans->count == count)) {
1406#ifdef DEBUG_REGEXP_GRAPH
1407 printf("Ignoring duplicate transition from %d to %d\n",
1408 state->no, target->no);
1409#endif
1410 return;
1411 }
1412 }
1413
1414 if (state->maxTrans == 0) {
1415 state->maxTrans = 8;
1416 state->trans = (xmlRegTrans *) xmlMalloc(state->maxTrans *
1417 sizeof(xmlRegTrans));
1418 if (state->trans == NULL) {
1419 xmlRegexpErrMemory(ctxt, "adding transition");
1420 state->maxTrans = 0;
1421 return;
1422 }
1423 } else if (state->nbTrans >= state->maxTrans) {
1424 xmlRegTrans *tmp;
1425 state->maxTrans *= 2;
1426 tmp = (xmlRegTrans *) xmlRealloc(state->trans, state->maxTrans *
1427 sizeof(xmlRegTrans));
1428 if (tmp == NULL) {
1429 xmlRegexpErrMemory(ctxt, "adding transition");
1430 state->maxTrans /= 2;
1431 return;
1432 }
1433 state->trans = tmp;
1434 }
1435#ifdef DEBUG_REGEXP_GRAPH
1436 printf("Add trans from %d to %d ", state->no, target->no);
1437 if (count == REGEXP_ALL_COUNTER)
1438 printf("all transition\n");
1439 else if (count >= 0)
1440 printf("count based %d\n", count);
1441 else if (counter >= 0)
1442 printf("counted %d\n", counter);
1443 else if (atom == NULL)
1444 printf("epsilon transition\n");
1445 else if (atom != NULL)
1446 xmlRegPrintAtom(stdout, atom);
1447#endif
1448
1449 state->trans[state->nbTrans].atom = atom;
1450 state->trans[state->nbTrans].to = target->no;
1451 state->trans[state->nbTrans].counter = counter;
1452 state->trans[state->nbTrans].count = count;
1453 state->trans[state->nbTrans].nd = 0;
1454 state->nbTrans++;
1455 xmlRegStateAddTransTo(ctxt, target, state->no);
1456}
1457
1458static int
1459xmlRegStatePush(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state) {
1460 if (state == NULL) return(-1);
1461 if (ctxt->maxStates == 0) {
1462 ctxt->maxStates = 4;
1463 ctxt->states = (xmlRegStatePtr *) xmlMalloc(ctxt->maxStates *
1464 sizeof(xmlRegStatePtr));
1465 if (ctxt->states == NULL) {
1466 xmlRegexpErrMemory(ctxt, "adding state");
1467 ctxt->maxStates = 0;
1468 return(-1);
1469 }
1470 } else if (ctxt->nbStates >= ctxt->maxStates) {
1471 xmlRegStatePtr *tmp;
1472 ctxt->maxStates *= 2;
1473 tmp = (xmlRegStatePtr *) xmlRealloc(ctxt->states, ctxt->maxStates *
1474 sizeof(xmlRegStatePtr));
1475 if (tmp == NULL) {
1476 xmlRegexpErrMemory(ctxt, "adding state");
1477 ctxt->maxStates /= 2;
1478 return(-1);
1479 }
1480 ctxt->states = tmp;
1481 }
1482 state->no = ctxt->nbStates;
1483 ctxt->states[ctxt->nbStates++] = state;
1484 return(0);
1485}
1486
1487/**
1488 * xmlFAGenerateAllTransition:
1489 * @ctxt: a regexp parser context
1490 * @from: the from state
1491 * @to: the target state or NULL for building a new one
1492 * @lax:
1493 *
1494 */
1495static void
1496xmlFAGenerateAllTransition(xmlRegParserCtxtPtr ctxt,
1497 xmlRegStatePtr from, xmlRegStatePtr to,
1498 int lax) {
1499 if (to == NULL) {
1500 to = xmlRegNewState(ctxt);
1501 xmlRegStatePush(ctxt, to);
1502 ctxt->state = to;
1503 }
1504 if (lax)
1505 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_LAX_COUNTER);
1506 else
1507 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_COUNTER);
1508}
1509
1510/**
1511 * xmlFAGenerateEpsilonTransition:
1512 * @ctxt: a regexp parser context
1513 * @from: the from state
1514 * @to: the target state or NULL for building a new one
1515 *
1516 */
1517static void
1518xmlFAGenerateEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1519 xmlRegStatePtr from, xmlRegStatePtr to) {
1520 if (to == NULL) {
1521 to = xmlRegNewState(ctxt);
1522 xmlRegStatePush(ctxt, to);
1523 ctxt->state = to;
1524 }
1525 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, -1);
1526}
1527
1528/**
1529 * xmlFAGenerateCountedEpsilonTransition:
1530 * @ctxt: a regexp parser context
1531 * @from: the from state
1532 * @to: the target state or NULL for building a new one
1533 * counter: the counter for that transition
1534 *
1535 */
1536static void
1537xmlFAGenerateCountedEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1538 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1539 if (to == NULL) {
1540 to = xmlRegNewState(ctxt);
1541 xmlRegStatePush(ctxt, to);
1542 ctxt->state = to;
1543 }
1544 xmlRegStateAddTrans(ctxt, from, NULL, to, counter, -1);
1545}
1546
1547/**
1548 * xmlFAGenerateCountedTransition:
1549 * @ctxt: a regexp parser context
1550 * @from: the from state
1551 * @to: the target state or NULL for building a new one
1552 * counter: the counter for that transition
1553 *
1554 */
1555static void
1556xmlFAGenerateCountedTransition(xmlRegParserCtxtPtr ctxt,
1557 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1558 if (to == NULL) {
1559 to = xmlRegNewState(ctxt);
1560 xmlRegStatePush(ctxt, to);
1561 ctxt->state = to;
1562 }
1563 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, counter);
1564}
1565
1566/**
1567 * xmlFAGenerateTransitions:
1568 * @ctxt: a regexp parser context
1569 * @from: the from state
1570 * @to: the target state or NULL for building a new one
1571 * @atom: the atom generating the transition
1572 *
1573 * Returns 0 if success and -1 in case of error.
1574 */
1575static int
1576xmlFAGenerateTransitions(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr from,
1577 xmlRegStatePtr to, xmlRegAtomPtr atom) {
1578 xmlRegStatePtr end;
1579 int nullable = 0;
1580
1581 if (atom == NULL) {
1582 ERROR("generate transition: atom == NULL");
1583 return(-1);
1584 }
1585 if (atom->type == XML_REGEXP_SUBREG) {
1586 /*
1587 * this is a subexpression handling one should not need to
1588 * create a new node except for XML_REGEXP_QUANT_RANGE.
1589 */
1590 if (xmlRegAtomPush(ctxt, atom) < 0) {
1591 return(-1);
1592 }
1593 if ((to != NULL) && (atom->stop != to) &&
1594 (atom->quant != XML_REGEXP_QUANT_RANGE)) {
1595 /*
1596 * Generate an epsilon transition to link to the target
1597 */
1598 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, to);
1599#ifdef DV
1600 } else if ((to == NULL) && (atom->quant != XML_REGEXP_QUANT_RANGE) &&
1601 (atom->quant != XML_REGEXP_QUANT_ONCE)) {
1602 to = xmlRegNewState(ctxt);
1603 xmlRegStatePush(ctxt, to);
1604 ctxt->state = to;
1605 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, to);
1606#endif
1607 }
1608 switch (atom->quant) {
1609 case XML_REGEXP_QUANT_OPT:
1610 atom->quant = XML_REGEXP_QUANT_ONCE;
1611 /*
1612 * transition done to the state after end of atom.
1613 * 1. set transition from atom start to new state
1614 * 2. set transition from atom end to this state.
1615 */
1616 if (to == NULL) {
1617 xmlFAGenerateEpsilonTransition(ctxt, atom->start, 0);
1618 xmlFAGenerateEpsilonTransition(ctxt, atom->stop,
1619 ctxt->state);
1620 } else {
1621 xmlFAGenerateEpsilonTransition(ctxt, atom->start, to);
1622 }
1623 break;
1624 case XML_REGEXP_QUANT_MULT:
1625 atom->quant = XML_REGEXP_QUANT_ONCE;
1626 xmlFAGenerateEpsilonTransition(ctxt, atom->start, atom->stop);
1627 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1628 break;
1629 case XML_REGEXP_QUANT_PLUS:
1630 atom->quant = XML_REGEXP_QUANT_ONCE;
1631 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1632 break;
1633 case XML_REGEXP_QUANT_RANGE: {
1634 int counter;
1635 xmlRegStatePtr inter, newstate;
1636
1637 /*
1638 * create the final state now if needed
1639 */
1640 if (to != NULL) {
1641 newstate = to;
1642 } else {
1643 newstate = xmlRegNewState(ctxt);
1644 xmlRegStatePush(ctxt, newstate);
1645 }
1646
1647 /*
1648 * The principle here is to use counted transition
1649 * to avoid explosion in the number of states in the
1650 * graph. This is clearly more complex but should not
1651 * be exploitable at runtime.
1652 */
1653 if ((atom->min == 0) && (atom->start0 == NULL)) {
1654 xmlRegAtomPtr copy;
1655 /*
1656 * duplicate a transition based on atom to count next
1657 * occurrences after 1. We cannot loop to atom->start
1658 * directly because we need an epsilon transition to
1659 * newstate.
1660 */
1661 /* ???? For some reason it seems we never reach that
1662 case, I suppose this got optimized out before when
1663 building the automata */
1664 copy = xmlRegCopyAtom(ctxt, atom);
1665 if (copy == NULL)
1666 return(-1);
1667 copy->quant = XML_REGEXP_QUANT_ONCE;
1668 copy->min = 0;
1669 copy->max = 0;
1670
1671 if (xmlFAGenerateTransitions(ctxt, atom->start, NULL, copy)
1672 < 0)
1673 return(-1);
1674 inter = ctxt->state;
1675 counter = xmlRegGetCounter(ctxt);
1676 ctxt->counters[counter].min = atom->min - 1;
1677 ctxt->counters[counter].max = atom->max - 1;
1678 /* count the number of times we see it again */
1679 xmlFAGenerateCountedEpsilonTransition(ctxt, inter,
1680 atom->stop, counter);
1681 /* allow a way out based on the count */
1682 xmlFAGenerateCountedTransition(ctxt, inter,
1683 newstate, counter);
1684 /* and also allow a direct exit for 0 */
1685 xmlFAGenerateEpsilonTransition(ctxt, atom->start,
1686 newstate);
1687 } else {
1688 /*
1689 * either we need the atom at least once or there
1690 * is an atom->start0 allowing to easily plug the
1691 * epsilon transition.
1692 */
1693 counter = xmlRegGetCounter(ctxt);
1694 ctxt->counters[counter].min = atom->min - 1;
1695 ctxt->counters[counter].max = atom->max - 1;
1696 /* allow a way out based on the count */
1697 xmlFAGenerateCountedTransition(ctxt, atom->stop,
1698 newstate, counter);
1699 /* count the number of times we see it again */
1700 xmlFAGenerateCountedEpsilonTransition(ctxt, atom->stop,
1701 atom->start, counter);
1702 /* and if needed allow a direct exit for 0 */
1703 if (atom->min == 0)
1704 xmlFAGenerateEpsilonTransition(ctxt, atom->start0,
1705 newstate);
1706
1707 }
1708 atom->min = 0;
1709 atom->max = 0;
1710 atom->quant = XML_REGEXP_QUANT_ONCE;
1711 ctxt->state = newstate;
1712 }
1713 default:
1714 break;
1715 }
1716 return(0);
1717 }
1718 if ((atom->min == 0) && (atom->max == 0) &&
1719 (atom->quant == XML_REGEXP_QUANT_RANGE)) {
1720 /*
1721 * we can discard the atom and generate an epsilon transition instead
1722 */
1723 if (to == NULL) {
1724 to = xmlRegNewState(ctxt);
1725 if (to != NULL)
1726 xmlRegStatePush(ctxt, to);
1727 else {
1728 return(-1);
1729 }
1730 }
1731 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1732 ctxt->state = to;
1733 xmlRegFreeAtom(atom);
1734 return(0);
1735 }
1736 if (to == NULL) {
1737 to = xmlRegNewState(ctxt);
1738 if (to != NULL)
1739 xmlRegStatePush(ctxt, to);
1740 else {
1741 return(-1);
1742 }
1743 }
1744 end = to;
1745 if ((atom->quant == XML_REGEXP_QUANT_MULT) ||
1746 (atom->quant == XML_REGEXP_QUANT_PLUS)) {
1747 /*
1748 * Do not pollute the target state by adding transitions from
1749 * it as it is likely to be the shared target of multiple branches.
1750 * So isolate with an epsilon transition.
1751 */
1752 xmlRegStatePtr tmp;
1753
1754 tmp = xmlRegNewState(ctxt);
1755 if (tmp != NULL)
1756 xmlRegStatePush(ctxt, tmp);
1757 else {
1758 return(-1);
1759 }
1760 xmlFAGenerateEpsilonTransition(ctxt, tmp, to);
1761 to = tmp;
1762 }
1763 if (xmlRegAtomPush(ctxt, atom) < 0) {
1764 return(-1);
1765 }
1766 if ((atom->quant == XML_REGEXP_QUANT_RANGE) &&
1767 (atom->min == 0) && (atom->max > 0)) {
1768 nullable = 1;
1769 atom->min = 1;
1770 if (atom->max == 1)
1771 atom->quant = XML_REGEXP_QUANT_OPT;
1772 }
1773 xmlRegStateAddTrans(ctxt, from, atom, to, -1, -1);
1774 ctxt->state = end;
1775 switch (atom->quant) {
1776 case XML_REGEXP_QUANT_OPT:
1777 atom->quant = XML_REGEXP_QUANT_ONCE;
1778 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1779 break;
1780 case XML_REGEXP_QUANT_MULT:
1781 atom->quant = XML_REGEXP_QUANT_ONCE;
1782 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1783 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1784 break;
1785 case XML_REGEXP_QUANT_PLUS:
1786 atom->quant = XML_REGEXP_QUANT_ONCE;
1787 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1788 break;
1789 case XML_REGEXP_QUANT_RANGE:
1790 if (nullable)
1791 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1792 break;
1793 default:
1794 break;
1795 }
1796 return(0);
1797}
1798
1799/**
1800 * xmlFAReduceEpsilonTransitions:
1801 * @ctxt: a regexp parser context
1802 * @fromnr: the from state
1803 * @tonr: the to state
1804 * @counter: should that transition be associated to a counted
1805 *
1806 */
1807static void
1808xmlFAReduceEpsilonTransitions(xmlRegParserCtxtPtr ctxt, int fromnr,
1809 int tonr, int counter) {
1810 int transnr;
1811 xmlRegStatePtr from;
1812 xmlRegStatePtr to;
1813
1814#ifdef DEBUG_REGEXP_GRAPH
1815 printf("xmlFAReduceEpsilonTransitions(%d, %d)\n", fromnr, tonr);
1816#endif
1817 from = ctxt->states[fromnr];
1818 if (from == NULL)
1819 return;
1820 to = ctxt->states[tonr];
1821 if (to == NULL)
1822 return;
1823 if ((to->mark == XML_REGEXP_MARK_START) ||
1824 (to->mark == XML_REGEXP_MARK_VISITED))
1825 return;
1826
1827 to->mark = XML_REGEXP_MARK_VISITED;
1828 if (to->type == XML_REGEXP_FINAL_STATE) {
1829#ifdef DEBUG_REGEXP_GRAPH
1830 printf("State %d is final, so %d becomes final\n", tonr, fromnr);
1831#endif
1832 from->type = XML_REGEXP_FINAL_STATE;
1833 }
1834 for (transnr = 0;transnr < to->nbTrans;transnr++) {
1835 if (to->trans[transnr].to < 0)
1836 continue;
1837 if (to->trans[transnr].atom == NULL) {
1838 /*
1839 * Don't remove counted transitions
1840 * Don't loop either
1841 */
1842 if (to->trans[transnr].to != fromnr) {
1843 if (to->trans[transnr].count >= 0) {
1844 int newto = to->trans[transnr].to;
1845
1846 xmlRegStateAddTrans(ctxt, from, NULL,
1847 ctxt->states[newto],
1848 -1, to->trans[transnr].count);
1849 } else {
1850#ifdef DEBUG_REGEXP_GRAPH
1851 printf("Found epsilon trans %d from %d to %d\n",
1852 transnr, tonr, to->trans[transnr].to);
1853#endif
1854 if (to->trans[transnr].counter >= 0) {
1855 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1856 to->trans[transnr].to,
1857 to->trans[transnr].counter);
1858 } else {
1859 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1860 to->trans[transnr].to,
1861 counter);
1862 }
1863 }
1864 }
1865 } else {
1866 int newto = to->trans[transnr].to;
1867
1868 if (to->trans[transnr].counter >= 0) {
1869 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1870 ctxt->states[newto],
1871 to->trans[transnr].counter, -1);
1872 } else {
1873 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1874 ctxt->states[newto], counter, -1);
1875 }
1876 }
1877 }
1878 to->mark = XML_REGEXP_MARK_NORMAL;
1879}
1880
1881/**
1882 * xmlFAEliminateSimpleEpsilonTransitions:
1883 * @ctxt: a regexp parser context
1884 *
1885 * Eliminating general epsilon transitions can get costly in the general
1886 * algorithm due to the large amount of generated new transitions and
1887 * associated comparisons. However for simple epsilon transition used just
1888 * to separate building blocks when generating the automata this can be
1889 * reduced to state elimination:
1890 * - if there exists an epsilon from X to Y
1891 * - if there is no other transition from X
1892 * then X and Y are semantically equivalent and X can be eliminated
1893 * If X is the start state then make Y the start state, else replace the
1894 * target of all transitions to X by transitions to Y.
1895 *
1896 * If X is a final state, skip it.
1897 * Otherwise it would be necessary to manipulate counters for this case when
1898 * eliminating state 2:
1899 * State 1 has a transition with an atom to state 2.
1900 * State 2 is final and has an epsilon transition to state 1.
1901 */
1902static void
1903xmlFAEliminateSimpleEpsilonTransitions(xmlRegParserCtxtPtr ctxt) {
1904 int statenr, i, j, newto;
1905 xmlRegStatePtr state, tmp;
1906
1907 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1908 state = ctxt->states[statenr];
1909 if (state == NULL)
1910 continue;
1911 if (state->nbTrans != 1)
1912 continue;
1913 if (state->type == XML_REGEXP_UNREACH_STATE ||
1914 state->type == XML_REGEXP_FINAL_STATE)
1915 continue;
1916 /* is the only transition out a basic transition */
1917 if ((state->trans[0].atom == NULL) &&
1918 (state->trans[0].to >= 0) &&
1919 (state->trans[0].to != statenr) &&
1920 (state->trans[0].counter < 0) &&
1921 (state->trans[0].count < 0)) {
1922 newto = state->trans[0].to;
1923
1924 if (state->type == XML_REGEXP_START_STATE) {
1925#ifdef DEBUG_REGEXP_GRAPH
1926 printf("Found simple epsilon trans from start %d to %d\n",
1927 statenr, newto);
1928#endif
1929 } else {
1930#ifdef DEBUG_REGEXP_GRAPH
1931 printf("Found simple epsilon trans from %d to %d\n",
1932 statenr, newto);
1933#endif
1934 for (i = 0;i < state->nbTransTo;i++) {
1935 tmp = ctxt->states[state->transTo[i]];
1936 for (j = 0;j < tmp->nbTrans;j++) {
1937 if (tmp->trans[j].to == statenr) {
1938#ifdef DEBUG_REGEXP_GRAPH
1939 printf("Changed transition %d on %d to go to %d\n",
1940 j, tmp->no, newto);
1941#endif
1942 tmp->trans[j].to = -1;
1943 xmlRegStateAddTrans(ctxt, tmp, tmp->trans[j].atom,
1944 ctxt->states[newto],
1945 tmp->trans[j].counter,
1946 tmp->trans[j].count);
1947 }
1948 }
1949 }
1950 if (state->type == XML_REGEXP_FINAL_STATE)
1951 ctxt->states[newto]->type = XML_REGEXP_FINAL_STATE;
1952 /* eliminate the transition completely */
1953 state->nbTrans = 0;
1954
1955 state->type = XML_REGEXP_UNREACH_STATE;
1956
1957 }
1958
1959 }
1960 }
1961}
1962/**
1963 * xmlFAEliminateEpsilonTransitions:
1964 * @ctxt: a regexp parser context
1965 *
1966 */
1967static void
1968xmlFAEliminateEpsilonTransitions(xmlRegParserCtxtPtr ctxt) {
1969 int statenr, transnr;
1970 xmlRegStatePtr state;
1971 int has_epsilon;
1972
1973 if (ctxt->states == NULL) return;
1974
1975 /*
1976 * Eliminate simple epsilon transition and the associated unreachable
1977 * states.
1978 */
1979 xmlFAEliminateSimpleEpsilonTransitions(ctxt);
1980 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1981 state = ctxt->states[statenr];
1982 if ((state != NULL) && (state->type == XML_REGEXP_UNREACH_STATE)) {
1983#ifdef DEBUG_REGEXP_GRAPH
1984 printf("Removed unreachable state %d\n", statenr);
1985#endif
1986 xmlRegFreeState(state);
1987 ctxt->states[statenr] = NULL;
1988 }
1989 }
1990
1991 has_epsilon = 0;
1992
1993 /*
1994 * Build the completed transitions bypassing the epsilons
1995 * Use a marking algorithm to avoid loops
1996 * Mark sink states too.
1997 * Process from the latest states backward to the start when
1998 * there is long cascading epsilon chains this minimize the
1999 * recursions and transition compares when adding the new ones
2000 */
2001 for (statenr = ctxt->nbStates - 1;statenr >= 0;statenr--) {
2002 state = ctxt->states[statenr];
2003 if (state == NULL)
2004 continue;
2005 if ((state->nbTrans == 0) &&
2006 (state->type != XML_REGEXP_FINAL_STATE)) {
2007 state->type = XML_REGEXP_SINK_STATE;
2008 }
2009 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2010 if ((state->trans[transnr].atom == NULL) &&
2011 (state->trans[transnr].to >= 0)) {
2012 if (state->trans[transnr].to == statenr) {
2013 state->trans[transnr].to = -1;
2014#ifdef DEBUG_REGEXP_GRAPH
2015 printf("Removed loopback epsilon trans %d on %d\n",
2016 transnr, statenr);
2017#endif
2018 } else if (state->trans[transnr].count < 0) {
2019 int newto = state->trans[transnr].to;
2020
2021#ifdef DEBUG_REGEXP_GRAPH
2022 printf("Found epsilon trans %d from %d to %d\n",
2023 transnr, statenr, newto);
2024#endif
2025 has_epsilon = 1;
2026 state->trans[transnr].to = -2;
2027 state->mark = XML_REGEXP_MARK_START;
2028 xmlFAReduceEpsilonTransitions(ctxt, statenr,
2029 newto, state->trans[transnr].counter);
2030 state->mark = XML_REGEXP_MARK_NORMAL;
2031#ifdef DEBUG_REGEXP_GRAPH
2032 } else {
2033 printf("Found counted transition %d on %d\n",
2034 transnr, statenr);
2035#endif
2036 }
2037 }
2038 }
2039 }
2040 /*
2041 * Eliminate the epsilon transitions
2042 */
2043 if (has_epsilon) {
2044 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2045 state = ctxt->states[statenr];
2046 if (state == NULL)
2047 continue;
2048 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2049 xmlRegTransPtr trans = &(state->trans[transnr]);
2050 if ((trans->atom == NULL) &&
2051 (trans->count < 0) &&
2052 (trans->to >= 0)) {
2053 trans->to = -1;
2054 }
2055 }
2056 }
2057 }
2058
2059 /*
2060 * Use this pass to detect unreachable states too
2061 */
2062 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2063 state = ctxt->states[statenr];
2064 if (state != NULL)
2065 state->reached = XML_REGEXP_MARK_NORMAL;
2066 }
2067 state = ctxt->states[0];
2068 if (state != NULL)
2069 state->reached = XML_REGEXP_MARK_START;
2070 while (state != NULL) {
2071 xmlRegStatePtr target = NULL;
2072 state->reached = XML_REGEXP_MARK_VISITED;
2073 /*
2074 * Mark all states reachable from the current reachable state
2075 */
2076 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2077 if ((state->trans[transnr].to >= 0) &&
2078 ((state->trans[transnr].atom != NULL) ||
2079 (state->trans[transnr].count >= 0))) {
2080 int newto = state->trans[transnr].to;
2081
2082 if (ctxt->states[newto] == NULL)
2083 continue;
2084 if (ctxt->states[newto]->reached == XML_REGEXP_MARK_NORMAL) {
2085 ctxt->states[newto]->reached = XML_REGEXP_MARK_START;
2086 target = ctxt->states[newto];
2087 }
2088 }
2089 }
2090
2091 /*
2092 * find the next accessible state not explored
2093 */
2094 if (target == NULL) {
2095 for (statenr = 1;statenr < ctxt->nbStates;statenr++) {
2096 state = ctxt->states[statenr];
2097 if ((state != NULL) && (state->reached ==
2098 XML_REGEXP_MARK_START)) {
2099 target = state;
2100 break;
2101 }
2102 }
2103 }
2104 state = target;
2105 }
2106 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2107 state = ctxt->states[statenr];
2108 if ((state != NULL) && (state->reached == XML_REGEXP_MARK_NORMAL)) {
2109#ifdef DEBUG_REGEXP_GRAPH
2110 printf("Removed unreachable state %d\n", statenr);
2111#endif
2112 xmlRegFreeState(state);
2113 ctxt->states[statenr] = NULL;
2114 }
2115 }
2116
2117}
2118
2119static int
2120xmlFACompareRanges(xmlRegRangePtr range1, xmlRegRangePtr range2) {
2121 int ret = 0;
2122
2123 if ((range1->type == XML_REGEXP_RANGES) ||
2124 (range2->type == XML_REGEXP_RANGES) ||
2125 (range2->type == XML_REGEXP_SUBREG) ||
2126 (range1->type == XML_REGEXP_SUBREG) ||
2127 (range1->type == XML_REGEXP_STRING) ||
2128 (range2->type == XML_REGEXP_STRING))
2129 return(-1);
2130
2131 /* put them in order */
2132 if (range1->type > range2->type) {
2133 xmlRegRangePtr tmp;
2134
2135 tmp = range1;
2136 range1 = range2;
2137 range2 = tmp;
2138 }
2139 if ((range1->type == XML_REGEXP_ANYCHAR) ||
2140 (range2->type == XML_REGEXP_ANYCHAR)) {
2141 ret = 1;
2142 } else if ((range1->type == XML_REGEXP_EPSILON) ||
2143 (range2->type == XML_REGEXP_EPSILON)) {
2144 return(0);
2145 } else if (range1->type == range2->type) {
2146 if (range1->type != XML_REGEXP_CHARVAL)
2147 ret = 1;
2148 else if ((range1->end < range2->start) ||
2149 (range2->end < range1->start))
2150 ret = 0;
2151 else
2152 ret = 1;
2153 } else if (range1->type == XML_REGEXP_CHARVAL) {
2154 int codepoint;
2155 int neg = 0;
2156
2157 /*
2158 * just check all codepoints in the range for acceptance,
2159 * this is usually way cheaper since done only once at
2160 * compilation than testing over and over at runtime or
2161 * pushing too many states when evaluating.
2162 */
2163 if (((range1->neg == 0) && (range2->neg != 0)) ||
2164 ((range1->neg != 0) && (range2->neg == 0)))
2165 neg = 1;
2166
2167 for (codepoint = range1->start;codepoint <= range1->end ;codepoint++) {
2168 ret = xmlRegCheckCharacterRange(range2->type, codepoint,
2169 0, range2->start, range2->end,
2170 range2->blockName);
2171 if (ret < 0)
2172 return(-1);
2173 if (((neg == 1) && (ret == 0)) ||
2174 ((neg == 0) && (ret == 1)))
2175 return(1);
2176 }
2177 return(0);
2178 } else if ((range1->type == XML_REGEXP_BLOCK_NAME) ||
2179 (range2->type == XML_REGEXP_BLOCK_NAME)) {
2180 if (range1->type == range2->type) {
2181 ret = xmlStrEqual(range1->blockName, range2->blockName);
2182 } else {
2183 /*
2184 * comparing a block range with anything else is way
2185 * too costly, and maintaining the table is like too much
2186 * memory too, so let's force the automata to save state
2187 * here.
2188 */
2189 return(1);
2190 }
2191 } else if ((range1->type < XML_REGEXP_LETTER) ||
2192 (range2->type < XML_REGEXP_LETTER)) {
2193 if ((range1->type == XML_REGEXP_ANYSPACE) &&
2194 (range2->type == XML_REGEXP_NOTSPACE))
2195 ret = 0;
2196 else if ((range1->type == XML_REGEXP_INITNAME) &&
2197 (range2->type == XML_REGEXP_NOTINITNAME))
2198 ret = 0;
2199 else if ((range1->type == XML_REGEXP_NAMECHAR) &&
2200 (range2->type == XML_REGEXP_NOTNAMECHAR))
2201 ret = 0;
2202 else if ((range1->type == XML_REGEXP_DECIMAL) &&
2203 (range2->type == XML_REGEXP_NOTDECIMAL))
2204 ret = 0;
2205 else if ((range1->type == XML_REGEXP_REALCHAR) &&
2206 (range2->type == XML_REGEXP_NOTREALCHAR))
2207 ret = 0;
2208 else {
2209 /* same thing to limit complexity */
2210 return(1);
2211 }
2212 } else {
2213 ret = 0;
2214 /* range1->type < range2->type here */
2215 switch (range1->type) {
2216 case XML_REGEXP_LETTER:
2217 /* all disjoint except in the subgroups */
2218 if ((range2->type == XML_REGEXP_LETTER_UPPERCASE) ||
2219 (range2->type == XML_REGEXP_LETTER_LOWERCASE) ||
2220 (range2->type == XML_REGEXP_LETTER_TITLECASE) ||
2221 (range2->type == XML_REGEXP_LETTER_MODIFIER) ||
2222 (range2->type == XML_REGEXP_LETTER_OTHERS))
2223 ret = 1;
2224 break;
2225 case XML_REGEXP_MARK:
2226 if ((range2->type == XML_REGEXP_MARK_NONSPACING) ||
2227 (range2->type == XML_REGEXP_MARK_SPACECOMBINING) ||
2228 (range2->type == XML_REGEXP_MARK_ENCLOSING))
2229 ret = 1;
2230 break;
2231 case XML_REGEXP_NUMBER:
2232 if ((range2->type == XML_REGEXP_NUMBER_DECIMAL) ||
2233 (range2->type == XML_REGEXP_NUMBER_LETTER) ||
2234 (range2->type == XML_REGEXP_NUMBER_OTHERS))
2235 ret = 1;
2236 break;
2237 case XML_REGEXP_PUNCT:
2238 if ((range2->type == XML_REGEXP_PUNCT_CONNECTOR) ||
2239 (range2->type == XML_REGEXP_PUNCT_DASH) ||
2240 (range2->type == XML_REGEXP_PUNCT_OPEN) ||
2241 (range2->type == XML_REGEXP_PUNCT_CLOSE) ||
2242 (range2->type == XML_REGEXP_PUNCT_INITQUOTE) ||
2243 (range2->type == XML_REGEXP_PUNCT_FINQUOTE) ||
2244 (range2->type == XML_REGEXP_PUNCT_OTHERS))
2245 ret = 1;
2246 break;
2247 case XML_REGEXP_SEPAR:
2248 if ((range2->type == XML_REGEXP_SEPAR_SPACE) ||
2249 (range2->type == XML_REGEXP_SEPAR_LINE) ||
2250 (range2->type == XML_REGEXP_SEPAR_PARA))
2251 ret = 1;
2252 break;
2253 case XML_REGEXP_SYMBOL:
2254 if ((range2->type == XML_REGEXP_SYMBOL_MATH) ||
2255 (range2->type == XML_REGEXP_SYMBOL_CURRENCY) ||
2256 (range2->type == XML_REGEXP_SYMBOL_MODIFIER) ||
2257 (range2->type == XML_REGEXP_SYMBOL_OTHERS))
2258 ret = 1;
2259 break;
2260 case XML_REGEXP_OTHER:
2261 if ((range2->type == XML_REGEXP_OTHER_CONTROL) ||
2262 (range2->type == XML_REGEXP_OTHER_FORMAT) ||
2263 (range2->type == XML_REGEXP_OTHER_PRIVATE))
2264 ret = 1;
2265 break;
2266 default:
2267 if ((range2->type >= XML_REGEXP_LETTER) &&
2268 (range2->type < XML_REGEXP_BLOCK_NAME))
2269 ret = 0;
2270 else {
2271 /* safety net ! */
2272 return(1);
2273 }
2274 }
2275 }
2276 if (((range1->neg == 0) && (range2->neg != 0)) ||
2277 ((range1->neg != 0) && (range2->neg == 0)))
2278 ret = !ret;
2279 return(ret);
2280}
2281
2282/**
2283 * xmlFACompareAtomTypes:
2284 * @type1: an atom type
2285 * @type2: an atom type
2286 *
2287 * Compares two atoms type to check whether they intersect in some ways,
2288 * this is used by xmlFACompareAtoms only
2289 *
2290 * Returns 1 if they may intersect and 0 otherwise
2291 */
2292static int
2293xmlFACompareAtomTypes(xmlRegAtomType type1, xmlRegAtomType type2) {
2294 if ((type1 == XML_REGEXP_EPSILON) ||
2295 (type1 == XML_REGEXP_CHARVAL) ||
2296 (type1 == XML_REGEXP_RANGES) ||
2297 (type1 == XML_REGEXP_SUBREG) ||
2298 (type1 == XML_REGEXP_STRING) ||
2299 (type1 == XML_REGEXP_ANYCHAR))
2300 return(1);
2301 if ((type2 == XML_REGEXP_EPSILON) ||
2302 (type2 == XML_REGEXP_CHARVAL) ||
2303 (type2 == XML_REGEXP_RANGES) ||
2304 (type2 == XML_REGEXP_SUBREG) ||
2305 (type2 == XML_REGEXP_STRING) ||
2306 (type2 == XML_REGEXP_ANYCHAR))
2307 return(1);
2308
2309 if (type1 == type2) return(1);
2310
2311 /* simplify subsequent compares by making sure type1 < type2 */
2312 if (type1 > type2) {
2313 xmlRegAtomType tmp = type1;
2314 type1 = type2;
2315 type2 = tmp;
2316 }
2317 switch (type1) {
2318 case XML_REGEXP_ANYSPACE: /* \s */
2319 /* can't be a letter, number, mark, punctuation, symbol */
2320 if ((type2 == XML_REGEXP_NOTSPACE) ||
2321 ((type2 >= XML_REGEXP_LETTER) &&
2322 (type2 <= XML_REGEXP_LETTER_OTHERS)) ||
2323 ((type2 >= XML_REGEXP_NUMBER) &&
2324 (type2 <= XML_REGEXP_NUMBER_OTHERS)) ||
2325 ((type2 >= XML_REGEXP_MARK) &&
2326 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2327 ((type2 >= XML_REGEXP_PUNCT) &&
2328 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2329 ((type2 >= XML_REGEXP_SYMBOL) &&
2330 (type2 <= XML_REGEXP_SYMBOL_OTHERS))
2331 ) return(0);
2332 break;
2333 case XML_REGEXP_NOTSPACE: /* \S */
2334 break;
2335 case XML_REGEXP_INITNAME: /* \l */
2336 /* can't be a number, mark, separator, punctuation, symbol or other */
2337 if ((type2 == XML_REGEXP_NOTINITNAME) ||
2338 ((type2 >= XML_REGEXP_NUMBER) &&
2339 (type2 <= XML_REGEXP_NUMBER_OTHERS)) ||
2340 ((type2 >= XML_REGEXP_MARK) &&
2341 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2342 ((type2 >= XML_REGEXP_SEPAR) &&
2343 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2344 ((type2 >= XML_REGEXP_PUNCT) &&
2345 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2346 ((type2 >= XML_REGEXP_SYMBOL) &&
2347 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2348 ((type2 >= XML_REGEXP_OTHER) &&
2349 (type2 <= XML_REGEXP_OTHER_NA))
2350 ) return(0);
2351 break;
2352 case XML_REGEXP_NOTINITNAME: /* \L */
2353 break;
2354 case XML_REGEXP_NAMECHAR: /* \c */
2355 /* can't be a mark, separator, punctuation, symbol or other */
2356 if ((type2 == XML_REGEXP_NOTNAMECHAR) ||
2357 ((type2 >= XML_REGEXP_MARK) &&
2358 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2359 ((type2 >= XML_REGEXP_PUNCT) &&
2360 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2361 ((type2 >= XML_REGEXP_SEPAR) &&
2362 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2363 ((type2 >= XML_REGEXP_SYMBOL) &&
2364 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2365 ((type2 >= XML_REGEXP_OTHER) &&
2366 (type2 <= XML_REGEXP_OTHER_NA))
2367 ) return(0);
2368 break;
2369 case XML_REGEXP_NOTNAMECHAR: /* \C */
2370 break;
2371 case XML_REGEXP_DECIMAL: /* \d */
2372 /* can't be a letter, mark, separator, punctuation, symbol or other */
2373 if ((type2 == XML_REGEXP_NOTDECIMAL) ||
2374 (type2 == XML_REGEXP_REALCHAR) ||
2375 ((type2 >= XML_REGEXP_LETTER) &&
2376 (type2 <= XML_REGEXP_LETTER_OTHERS)) ||
2377 ((type2 >= XML_REGEXP_MARK) &&
2378 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2379 ((type2 >= XML_REGEXP_PUNCT) &&
2380 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2381 ((type2 >= XML_REGEXP_SEPAR) &&
2382 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2383 ((type2 >= XML_REGEXP_SYMBOL) &&
2384 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2385 ((type2 >= XML_REGEXP_OTHER) &&
2386 (type2 <= XML_REGEXP_OTHER_NA))
2387 )return(0);
2388 break;
2389 case XML_REGEXP_NOTDECIMAL: /* \D */
2390 break;
2391 case XML_REGEXP_REALCHAR: /* \w */
2392 /* can't be a mark, separator, punctuation, symbol or other */
2393 if ((type2 == XML_REGEXP_NOTDECIMAL) ||
2394 ((type2 >= XML_REGEXP_MARK) &&
2395 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2396 ((type2 >= XML_REGEXP_PUNCT) &&
2397 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2398 ((type2 >= XML_REGEXP_SEPAR) &&
2399 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2400 ((type2 >= XML_REGEXP_SYMBOL) &&
2401 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2402 ((type2 >= XML_REGEXP_OTHER) &&
2403 (type2 <= XML_REGEXP_OTHER_NA))
2404 )return(0);
2405 break;
2406 case XML_REGEXP_NOTREALCHAR: /* \W */
2407 break;
2408 /*
2409 * at that point we know both type 1 and type2 are from
2410 * character categories are ordered and are different,
2411 * it becomes simple because this is a partition
2412 */
2413 case XML_REGEXP_LETTER:
2414 if (type2 <= XML_REGEXP_LETTER_OTHERS)
2415 return(1);
2416 return(0);
2417 case XML_REGEXP_LETTER_UPPERCASE:
2418 case XML_REGEXP_LETTER_LOWERCASE:
2419 case XML_REGEXP_LETTER_TITLECASE:
2420 case XML_REGEXP_LETTER_MODIFIER:
2421 case XML_REGEXP_LETTER_OTHERS:
2422 return(0);
2423 case XML_REGEXP_MARK:
2424 if (type2 <= XML_REGEXP_MARK_ENCLOSING)
2425 return(1);
2426 return(0);
2427 case XML_REGEXP_MARK_NONSPACING:
2428 case XML_REGEXP_MARK_SPACECOMBINING:
2429 case XML_REGEXP_MARK_ENCLOSING:
2430 return(0);
2431 case XML_REGEXP_NUMBER:
2432 if (type2 <= XML_REGEXP_NUMBER_OTHERS)
2433 return(1);
2434 return(0);
2435 case XML_REGEXP_NUMBER_DECIMAL:
2436 case XML_REGEXP_NUMBER_LETTER:
2437 case XML_REGEXP_NUMBER_OTHERS:
2438 return(0);
2439 case XML_REGEXP_PUNCT:
2440 if (type2 <= XML_REGEXP_PUNCT_OTHERS)
2441 return(1);
2442 return(0);
2443 case XML_REGEXP_PUNCT_CONNECTOR:
2444 case XML_REGEXP_PUNCT_DASH:
2445 case XML_REGEXP_PUNCT_OPEN:
2446 case XML_REGEXP_PUNCT_CLOSE:
2447 case XML_REGEXP_PUNCT_INITQUOTE:
2448 case XML_REGEXP_PUNCT_FINQUOTE:
2449 case XML_REGEXP_PUNCT_OTHERS:
2450 return(0);
2451 case XML_REGEXP_SEPAR:
2452 if (type2 <= XML_REGEXP_SEPAR_PARA)
2453 return(1);
2454 return(0);
2455 case XML_REGEXP_SEPAR_SPACE:
2456 case XML_REGEXP_SEPAR_LINE:
2457 case XML_REGEXP_SEPAR_PARA:
2458 return(0);
2459 case XML_REGEXP_SYMBOL:
2460 if (type2 <= XML_REGEXP_SYMBOL_OTHERS)
2461 return(1);
2462 return(0);
2463 case XML_REGEXP_SYMBOL_MATH:
2464 case XML_REGEXP_SYMBOL_CURRENCY:
2465 case XML_REGEXP_SYMBOL_MODIFIER:
2466 case XML_REGEXP_SYMBOL_OTHERS:
2467 return(0);
2468 case XML_REGEXP_OTHER:
2469 if (type2 <= XML_REGEXP_OTHER_NA)
2470 return(1);
2471 return(0);
2472 case XML_REGEXP_OTHER_CONTROL:
2473 case XML_REGEXP_OTHER_FORMAT:
2474 case XML_REGEXP_OTHER_PRIVATE:
2475 case XML_REGEXP_OTHER_NA:
2476 return(0);
2477 default:
2478 break;
2479 }
2480 return(1);
2481}
2482
2483/**
2484 * xmlFAEqualAtoms:
2485 * @atom1: an atom
2486 * @atom2: an atom
2487 * @deep: if not set only compare string pointers
2488 *
2489 * Compares two atoms to check whether they are the same exactly
2490 * this is used to remove equivalent transitions
2491 *
2492 * Returns 1 if same and 0 otherwise
2493 */
2494static int
2495xmlFAEqualAtoms(xmlRegAtomPtr atom1, xmlRegAtomPtr atom2, int deep) {
2496 int ret = 0;
2497
2498 if (atom1 == atom2)
2499 return(1);
2500 if ((atom1 == NULL) || (atom2 == NULL))
2501 return(0);
2502
2503 if (atom1->type != atom2->type)
2504 return(0);
2505 switch (atom1->type) {
2506 case XML_REGEXP_EPSILON:
2507 ret = 0;
2508 break;
2509 case XML_REGEXP_STRING:
2510 if (!deep)
2511 ret = (atom1->valuep == atom2->valuep);
2512 else
2513 ret = xmlStrEqual((xmlChar *)atom1->valuep,
2514 (xmlChar *)atom2->valuep);
2515 break;
2516 case XML_REGEXP_CHARVAL:
2517 ret = (atom1->codepoint == atom2->codepoint);
2518 break;
2519 case XML_REGEXP_RANGES:
2520 /* too hard to do in the general case */
2521 ret = 0;
2522 default:
2523 break;
2524 }
2525 return(ret);
2526}
2527
2528/**
2529 * xmlFACompareAtoms:
2530 * @atom1: an atom
2531 * @atom2: an atom
2532 * @deep: if not set only compare string pointers
2533 *
2534 * Compares two atoms to check whether they intersect in some ways,
2535 * this is used by xmlFAComputesDeterminism and xmlFARecurseDeterminism only
2536 *
2537 * Returns 1 if yes and 0 otherwise
2538 */
2539static int
2540xmlFACompareAtoms(xmlRegAtomPtr atom1, xmlRegAtomPtr atom2, int deep) {
2541 int ret = 1;
2542
2543 if (atom1 == atom2)
2544 return(1);
2545 if ((atom1 == NULL) || (atom2 == NULL))
2546 return(0);
2547
2548 if ((atom1->type == XML_REGEXP_ANYCHAR) ||
2549 (atom2->type == XML_REGEXP_ANYCHAR))
2550 return(1);
2551
2552 if (atom1->type > atom2->type) {
2553 xmlRegAtomPtr tmp;
2554 tmp = atom1;
2555 atom1 = atom2;
2556 atom2 = tmp;
2557 }
2558 if (atom1->type != atom2->type) {
2559 ret = xmlFACompareAtomTypes(atom1->type, atom2->type);
2560 /* if they can't intersect at the type level break now */
2561 if (ret == 0)
2562 return(0);
2563 }
2564 switch (atom1->type) {
2565 case XML_REGEXP_STRING:
2566 if (!deep)
2567 ret = (atom1->valuep != atom2->valuep);
2568 else {
2569 xmlChar *val1 = (xmlChar *)atom1->valuep;
2570 xmlChar *val2 = (xmlChar *)atom2->valuep;
2571 int compound1 = (xmlStrchr(val1, '|') != NULL);
2572 int compound2 = (xmlStrchr(val2, '|') != NULL);
2573
2574 /* Ignore negative match flag for ##other namespaces */
2575 if (compound1 != compound2)
2576 return(0);
2577
2578 ret = xmlRegStrEqualWildcard(val1, val2);
2579 }
2580 break;
2581 case XML_REGEXP_EPSILON:
2582 goto not_determinist;
2583 case XML_REGEXP_CHARVAL:
2584 if (atom2->type == XML_REGEXP_CHARVAL) {
2585 ret = (atom1->codepoint == atom2->codepoint);
2586 } else {
2587 ret = xmlRegCheckCharacter(atom2, atom1->codepoint);
2588 if (ret < 0)
2589 ret = 1;
2590 }
2591 break;
2592 case XML_REGEXP_RANGES:
2593 if (atom2->type == XML_REGEXP_RANGES) {
2594 int i, j, res;
2595 xmlRegRangePtr r1, r2;
2596
2597 /*
2598 * need to check that none of the ranges eventually matches
2599 */
2600 for (i = 0;i < atom1->nbRanges;i++) {
2601 for (j = 0;j < atom2->nbRanges;j++) {
2602 r1 = atom1->ranges[i];
2603 r2 = atom2->ranges[j];
2604 res = xmlFACompareRanges(r1, r2);
2605 if (res == 1) {
2606 ret = 1;
2607 goto done;
2608 }
2609 }
2610 }
2611 ret = 0;
2612 }
2613 break;
2614 default:
2615 goto not_determinist;
2616 }
2617done:
2618 if (atom1->neg != atom2->neg) {
2619 ret = !ret;
2620 }
2621 if (ret == 0)
2622 return(0);
2623not_determinist:
2624 return(1);
2625}
2626
2627/**
2628 * xmlFARecurseDeterminism:
2629 * @ctxt: a regexp parser context
2630 *
2631 * Check whether the associated regexp is determinist,
2632 * should be called after xmlFAEliminateEpsilonTransitions()
2633 *
2634 */
2635static int
2636xmlFARecurseDeterminism(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
2637 int to, xmlRegAtomPtr atom) {
2638 int ret = 1;
2639 int res;
2640 int transnr, nbTrans;
2641 xmlRegTransPtr t1;
2642 int deep = 1;
2643
2644 if (state == NULL)
2645 return(ret);
2646 if (state->markd == XML_REGEXP_MARK_VISITED)
2647 return(ret);
2648
2649 if (ctxt->flags & AM_AUTOMATA_RNG)
2650 deep = 0;
2651
2652 /*
2653 * don't recurse on transitions potentially added in the course of
2654 * the elimination.
2655 */
2656 nbTrans = state->nbTrans;
2657 for (transnr = 0;transnr < nbTrans;transnr++) {
2658 t1 = &(state->trans[transnr]);
2659 /*
2660 * check transitions conflicting with the one looked at
2661 */
2662 if (t1->atom == NULL) {
2663 if (t1->to < 0)
2664 continue;
2665 state->markd = XML_REGEXP_MARK_VISITED;
2666 res = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
2667 to, atom);
2668 if (res == 0) {
2669 ret = 0;
2670 /* t1->nd = 1; */
2671 }
2672 continue;
2673 }
2674 if (t1->to != to)
2675 continue;
2676 if (xmlFACompareAtoms(t1->atom, atom, deep)) {
2677 ret = 0;
2678 /* mark the transition as non-deterministic */
2679 t1->nd = 1;
2680 }
2681 }
2682 return(ret);
2683}
2684
2685/**
2686 * xmlFAFinishRecurseDeterminism:
2687 * @ctxt: a regexp parser context
2688 *
2689 * Reset flags after checking determinism.
2690 */
2691static void
2692xmlFAFinishRecurseDeterminism(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state) {
2693 int transnr, nbTrans;
2694
2695 if (state == NULL)
2696 return;
2697 if (state->markd != XML_REGEXP_MARK_VISITED)
2698 return;
2699 state->markd = 0;
2700
2701 nbTrans = state->nbTrans;
2702 for (transnr = 0; transnr < nbTrans; transnr++) {
2703 xmlRegTransPtr t1 = &state->trans[transnr];
2704 if ((t1->atom == NULL) && (t1->to >= 0))
2705 xmlFAFinishRecurseDeterminism(ctxt, ctxt->states[t1->to]);
2706 }
2707}
2708
2709/**
2710 * xmlFAComputesDeterminism:
2711 * @ctxt: a regexp parser context
2712 *
2713 * Check whether the associated regexp is determinist,
2714 * should be called after xmlFAEliminateEpsilonTransitions()
2715 *
2716 */
2717static int
2718xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt) {
2719 int statenr, transnr;
2720 xmlRegStatePtr state;
2721 xmlRegTransPtr t1, t2, last;
2722 int i;
2723 int ret = 1;
2724 int deep = 1;
2725
2726#ifdef DEBUG_REGEXP_GRAPH
2727 printf("xmlFAComputesDeterminism\n");
2728 xmlRegPrintCtxt(stdout, ctxt);
2729#endif
2730 if (ctxt->determinist != -1)
2731 return(ctxt->determinist);
2732
2733 if (ctxt->flags & AM_AUTOMATA_RNG)
2734 deep = 0;
2735
2736 /*
2737 * First cleanup the automata removing cancelled transitions
2738 */
2739 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2740 state = ctxt->states[statenr];
2741 if (state == NULL)
2742 continue;
2743 if (state->nbTrans < 2)
2744 continue;
2745 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2746 t1 = &(state->trans[transnr]);
2747 /*
2748 * Determinism checks in case of counted or all transitions
2749 * will have to be handled separately
2750 */
2751 if (t1->atom == NULL) {
2752 /* t1->nd = 1; */
2753 continue;
2754 }
2755 if (t1->to == -1) /* eliminated */
2756 continue;
2757 for (i = 0;i < transnr;i++) {
2758 t2 = &(state->trans[i]);
2759 if (t2->to == -1) /* eliminated */
2760 continue;
2761 if (t2->atom != NULL) {
2762 if (t1->to == t2->to) {
2763 /*
2764 * Here we use deep because we want to keep the
2765 * transitions which indicate a conflict
2766 */
2767 if (xmlFAEqualAtoms(t1->atom, t2->atom, deep) &&
2768 (t1->counter == t2->counter) &&
2769 (t1->count == t2->count))
2770 t2->to = -1; /* eliminated */
2771 }
2772 }
2773 }
2774 }
2775 }
2776
2777 /*
2778 * Check for all states that there aren't 2 transitions
2779 * with the same atom and a different target.
2780 */
2781 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2782 state = ctxt->states[statenr];
2783 if (state == NULL)
2784 continue;
2785 if (state->nbTrans < 2)
2786 continue;
2787 last = NULL;
2788 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2789 t1 = &(state->trans[transnr]);
2790 /*
2791 * Determinism checks in case of counted or all transitions
2792 * will have to be handled separately
2793 */
2794 if (t1->atom == NULL) {
2795 continue;
2796 }
2797 if (t1->to == -1) /* eliminated */
2798 continue;
2799 for (i = 0;i < transnr;i++) {
2800 t2 = &(state->trans[i]);
2801 if (t2->to == -1) /* eliminated */
2802 continue;
2803 if (t2->atom != NULL) {
2804 /*
2805 * But here we don't use deep because we want to
2806 * find transitions which indicate a conflict
2807 */
2808 if (xmlFACompareAtoms(t1->atom, t2->atom, 1)) {
2809 ret = 0;
2810 /* mark the transitions as non-deterministic ones */
2811 t1->nd = 1;
2812 t2->nd = 1;
2813 last = t1;
2814 }
2815 } else if (t1->to != -1) {
2816 /*
2817 * do the closure in case of remaining specific
2818 * epsilon transitions like choices or all
2819 */
2820 ret = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
2821 t2->to, t2->atom);
2822 xmlFAFinishRecurseDeterminism(ctxt, ctxt->states[t1->to]);
2823 /* don't shortcut the computation so all non deterministic
2824 transition get marked down
2825 if (ret == 0)
2826 return(0);
2827 */
2828 if (ret == 0) {
2829 t1->nd = 1;
2830 /* t2->nd = 1; */
2831 last = t1;
2832 }
2833 }
2834 }
2835 /* don't shortcut the computation so all non deterministic
2836 transition get marked down
2837 if (ret == 0)
2838 break; */
2839 }
2840
2841 /*
2842 * mark specifically the last non-deterministic transition
2843 * from a state since there is no need to set-up rollback
2844 * from it
2845 */
2846 if (last != NULL) {
2847 last->nd = 2;
2848 }
2849
2850 /* don't shortcut the computation so all non deterministic
2851 transition get marked down
2852 if (ret == 0)
2853 break; */
2854 }
2855
2856 ctxt->determinist = ret;
2857 return(ret);
2858}
2859
2860/************************************************************************
2861 * *
2862 * Routines to check input against transition atoms *
2863 * *
2864 ************************************************************************/
2865
2866static int
2867xmlRegCheckCharacterRange(xmlRegAtomType type, int codepoint, int neg,
2868 int start, int end, const xmlChar *blockName) {
2869 int ret = 0;
2870
2871 switch (type) {
2872 case XML_REGEXP_STRING:
2873 case XML_REGEXP_SUBREG:
2874 case XML_REGEXP_RANGES:
2875 case XML_REGEXP_EPSILON:
2876 return(-1);
2877 case XML_REGEXP_ANYCHAR:
2878 ret = ((codepoint != '\n') && (codepoint != '\r'));
2879 break;
2880 case XML_REGEXP_CHARVAL:
2881 ret = ((codepoint >= start) && (codepoint <= end));
2882 break;
2883 case XML_REGEXP_NOTSPACE:
2884 neg = !neg;
2885 /* Falls through. */
2886 case XML_REGEXP_ANYSPACE:
2887 ret = ((codepoint == '\n') || (codepoint == '\r') ||
2888 (codepoint == '\t') || (codepoint == ' '));
2889 break;
2890 case XML_REGEXP_NOTINITNAME:
2891 neg = !neg;
2892 /* Falls through. */
2893 case XML_REGEXP_INITNAME:
2894 ret = (IS_LETTER(codepoint) ||
2895 (codepoint == '_') || (codepoint == ':'));
2896 break;
2897 case XML_REGEXP_NOTNAMECHAR:
2898 neg = !neg;
2899 /* Falls through. */
2900 case XML_REGEXP_NAMECHAR:
2901 ret = (IS_LETTER(codepoint) || IS_DIGIT(codepoint) ||
2902 (codepoint == '.') || (codepoint == '-') ||
2903 (codepoint == '_') || (codepoint == ':') ||
2904 IS_COMBINING(codepoint) || IS_EXTENDER(codepoint));
2905 break;
2906 case XML_REGEXP_NOTDECIMAL:
2907 neg = !neg;
2908 /* Falls through. */
2909 case XML_REGEXP_DECIMAL:
2910 ret = xmlUCSIsCatNd(codepoint);
2911 break;
2912 case XML_REGEXP_REALCHAR:
2913 neg = !neg;
2914 /* Falls through. */
2915 case XML_REGEXP_NOTREALCHAR:
2916 ret = xmlUCSIsCatP(codepoint);
2917 if (ret == 0)
2918 ret = xmlUCSIsCatZ(codepoint);
2919 if (ret == 0)
2920 ret = xmlUCSIsCatC(codepoint);
2921 break;
2922 case XML_REGEXP_LETTER:
2923 ret = xmlUCSIsCatL(codepoint);
2924 break;
2925 case XML_REGEXP_LETTER_UPPERCASE:
2926 ret = xmlUCSIsCatLu(codepoint);
2927 break;
2928 case XML_REGEXP_LETTER_LOWERCASE:
2929 ret = xmlUCSIsCatLl(codepoint);
2930 break;
2931 case XML_REGEXP_LETTER_TITLECASE:
2932 ret = xmlUCSIsCatLt(codepoint);
2933 break;
2934 case XML_REGEXP_LETTER_MODIFIER:
2935 ret = xmlUCSIsCatLm(codepoint);
2936 break;
2937 case XML_REGEXP_LETTER_OTHERS:
2938 ret = xmlUCSIsCatLo(codepoint);
2939 break;
2940 case XML_REGEXP_MARK:
2941 ret = xmlUCSIsCatM(codepoint);
2942 break;
2943 case XML_REGEXP_MARK_NONSPACING:
2944 ret = xmlUCSIsCatMn(codepoint);
2945 break;
2946 case XML_REGEXP_MARK_SPACECOMBINING:
2947 ret = xmlUCSIsCatMc(codepoint);
2948 break;
2949 case XML_REGEXP_MARK_ENCLOSING:
2950 ret = xmlUCSIsCatMe(codepoint);
2951 break;
2952 case XML_REGEXP_NUMBER:
2953 ret = xmlUCSIsCatN(codepoint);
2954 break;
2955 case XML_REGEXP_NUMBER_DECIMAL:
2956 ret = xmlUCSIsCatNd(codepoint);
2957 break;
2958 case XML_REGEXP_NUMBER_LETTER:
2959 ret = xmlUCSIsCatNl(codepoint);
2960 break;
2961 case XML_REGEXP_NUMBER_OTHERS:
2962 ret = xmlUCSIsCatNo(codepoint);
2963 break;
2964 case XML_REGEXP_PUNCT:
2965 ret = xmlUCSIsCatP(codepoint);
2966 break;
2967 case XML_REGEXP_PUNCT_CONNECTOR:
2968 ret = xmlUCSIsCatPc(codepoint);
2969 break;
2970 case XML_REGEXP_PUNCT_DASH:
2971 ret = xmlUCSIsCatPd(codepoint);
2972 break;
2973 case XML_REGEXP_PUNCT_OPEN:
2974 ret = xmlUCSIsCatPs(codepoint);
2975 break;
2976 case XML_REGEXP_PUNCT_CLOSE:
2977 ret = xmlUCSIsCatPe(codepoint);
2978 break;
2979 case XML_REGEXP_PUNCT_INITQUOTE:
2980 ret = xmlUCSIsCatPi(codepoint);
2981 break;
2982 case XML_REGEXP_PUNCT_FINQUOTE:
2983 ret = xmlUCSIsCatPf(codepoint);
2984 break;
2985 case XML_REGEXP_PUNCT_OTHERS:
2986 ret = xmlUCSIsCatPo(codepoint);
2987 break;
2988 case XML_REGEXP_SEPAR:
2989 ret = xmlUCSIsCatZ(codepoint);
2990 break;
2991 case XML_REGEXP_SEPAR_SPACE:
2992 ret = xmlUCSIsCatZs(codepoint);
2993 break;
2994 case XML_REGEXP_SEPAR_LINE:
2995 ret = xmlUCSIsCatZl(codepoint);
2996 break;
2997 case XML_REGEXP_SEPAR_PARA:
2998 ret = xmlUCSIsCatZp(codepoint);
2999 break;
3000 case XML_REGEXP_SYMBOL:
3001 ret = xmlUCSIsCatS(codepoint);
3002 break;
3003 case XML_REGEXP_SYMBOL_MATH:
3004 ret = xmlUCSIsCatSm(codepoint);
3005 break;
3006 case XML_REGEXP_SYMBOL_CURRENCY:
3007 ret = xmlUCSIsCatSc(codepoint);
3008 break;
3009 case XML_REGEXP_SYMBOL_MODIFIER:
3010 ret = xmlUCSIsCatSk(codepoint);
3011 break;
3012 case XML_REGEXP_SYMBOL_OTHERS:
3013 ret = xmlUCSIsCatSo(codepoint);
3014 break;
3015 case XML_REGEXP_OTHER:
3016 ret = xmlUCSIsCatC(codepoint);
3017 break;
3018 case XML_REGEXP_OTHER_CONTROL:
3019 ret = xmlUCSIsCatCc(codepoint);
3020 break;
3021 case XML_REGEXP_OTHER_FORMAT:
3022 ret = xmlUCSIsCatCf(codepoint);
3023 break;
3024 case XML_REGEXP_OTHER_PRIVATE:
3025 ret = xmlUCSIsCatCo(codepoint);
3026 break;
3027 case XML_REGEXP_OTHER_NA:
3028 /* ret = xmlUCSIsCatCn(codepoint); */
3029 /* Seems it doesn't exist anymore in recent Unicode releases */
3030 ret = 0;
3031 break;
3032 case XML_REGEXP_BLOCK_NAME:
3033 ret = xmlUCSIsBlock(codepoint, (const char *) blockName);
3034 break;
3035 }
3036 if (neg)
3037 return(!ret);
3038 return(ret);
3039}
3040
3041static int
3042xmlRegCheckCharacter(xmlRegAtomPtr atom, int codepoint) {
3043 int i, ret = 0;
3044 xmlRegRangePtr range;
3045
3046 if ((atom == NULL) || (!IS_CHAR(codepoint)))
3047 return(-1);
3048
3049 switch (atom->type) {
3050 case XML_REGEXP_SUBREG:
3051 case XML_REGEXP_EPSILON:
3052 return(-1);
3053 case XML_REGEXP_CHARVAL:
3054 return(codepoint == atom->codepoint);
3055 case XML_REGEXP_RANGES: {
3056 int accept = 0;
3057
3058 for (i = 0;i < atom->nbRanges;i++) {
3059 range = atom->ranges[i];
3060 if (range->neg == 2) {
3061 ret = xmlRegCheckCharacterRange(range->type, codepoint,
3062 0, range->start, range->end,
3063 range->blockName);
3064 if (ret != 0)
3065 return(0); /* excluded char */
3066 } else if (range->neg) {
3067 ret = xmlRegCheckCharacterRange(range->type, codepoint,
3068 0, range->start, range->end,
3069 range->blockName);
3070 if (ret == 0)
3071 accept = 1;
3072 else
3073 return(0);
3074 } else {
3075 ret = xmlRegCheckCharacterRange(range->type, codepoint,
3076 0, range->start, range->end,
3077 range->blockName);
3078 if (ret != 0)
3079 accept = 1; /* might still be excluded */
3080 }
3081 }
3082 return(accept);
3083 }
3084 case XML_REGEXP_STRING:
3085 printf("TODO: XML_REGEXP_STRING\n");
3086 return(-1);
3087 case XML_REGEXP_ANYCHAR:
3088 case XML_REGEXP_ANYSPACE:
3089 case XML_REGEXP_NOTSPACE:
3090 case XML_REGEXP_INITNAME:
3091 case XML_REGEXP_NOTINITNAME:
3092 case XML_REGEXP_NAMECHAR:
3093 case XML_REGEXP_NOTNAMECHAR:
3094 case XML_REGEXP_DECIMAL:
3095 case XML_REGEXP_NOTDECIMAL:
3096 case XML_REGEXP_REALCHAR:
3097 case XML_REGEXP_NOTREALCHAR:
3098 case XML_REGEXP_LETTER:
3099 case XML_REGEXP_LETTER_UPPERCASE:
3100 case XML_REGEXP_LETTER_LOWERCASE:
3101 case XML_REGEXP_LETTER_TITLECASE:
3102 case XML_REGEXP_LETTER_MODIFIER:
3103 case XML_REGEXP_LETTER_OTHERS:
3104 case XML_REGEXP_MARK:
3105 case XML_REGEXP_MARK_NONSPACING:
3106 case XML_REGEXP_MARK_SPACECOMBINING:
3107 case XML_REGEXP_MARK_ENCLOSING:
3108 case XML_REGEXP_NUMBER:
3109 case XML_REGEXP_NUMBER_DECIMAL:
3110 case XML_REGEXP_NUMBER_LETTER:
3111 case XML_REGEXP_NUMBER_OTHERS:
3112 case XML_REGEXP_PUNCT:
3113 case XML_REGEXP_PUNCT_CONNECTOR:
3114 case XML_REGEXP_PUNCT_DASH:
3115 case XML_REGEXP_PUNCT_OPEN:
3116 case XML_REGEXP_PUNCT_CLOSE:
3117 case XML_REGEXP_PUNCT_INITQUOTE:
3118 case XML_REGEXP_PUNCT_FINQUOTE:
3119 case XML_REGEXP_PUNCT_OTHERS:
3120 case XML_REGEXP_SEPAR:
3121 case XML_REGEXP_SEPAR_SPACE:
3122 case XML_REGEXP_SEPAR_LINE:
3123 case XML_REGEXP_SEPAR_PARA:
3124 case XML_REGEXP_SYMBOL:
3125 case XML_REGEXP_SYMBOL_MATH:
3126 case XML_REGEXP_SYMBOL_CURRENCY:
3127 case XML_REGEXP_SYMBOL_MODIFIER:
3128 case XML_REGEXP_SYMBOL_OTHERS:
3129 case XML_REGEXP_OTHER:
3130 case XML_REGEXP_OTHER_CONTROL:
3131 case XML_REGEXP_OTHER_FORMAT:
3132 case XML_REGEXP_OTHER_PRIVATE:
3133 case XML_REGEXP_OTHER_NA:
3134 case XML_REGEXP_BLOCK_NAME:
3135 ret = xmlRegCheckCharacterRange(atom->type, codepoint, 0, 0, 0,
3136 (const xmlChar *)atom->valuep);
3137 if (atom->neg)
3138 ret = !ret;
3139 break;
3140 }
3141 return(ret);
3142}
3143
3144/************************************************************************
3145 * *
3146 * Saving and restoring state of an execution context *
3147 * *
3148 ************************************************************************/
3149
3150#ifdef DEBUG_REGEXP_EXEC
3151static void
3152xmlFARegDebugExec(xmlRegExecCtxtPtr exec) {
3153 printf("state: %d:%d:idx %d", exec->state->no, exec->transno, exec->index);
3154 if (exec->inputStack != NULL) {
3155 int i;
3156 printf(": ");
3157 for (i = 0;(i < 3) && (i < exec->inputStackNr);i++)
3158 printf("%s ", (const char *)
3159 exec->inputStack[exec->inputStackNr - (i + 1)].value);
3160 } else {
3161 printf(": %s", &(exec->inputString[exec->index]));
3162 }
3163 printf("\n");
3164}
3165#endif
3166
3167static void
3168xmlFARegExecSave(xmlRegExecCtxtPtr exec) {
3169#ifdef DEBUG_REGEXP_EXEC
3170 printf("saving ");
3171 exec->transno++;
3172 xmlFARegDebugExec(exec);
3173 exec->transno--;
3174#endif
3175#ifdef MAX_PUSH
3176 if (exec->nbPush > MAX_PUSH) {
3177 return;
3178 }
3179 exec->nbPush++;
3180#endif
3181
3182 if (exec->maxRollbacks == 0) {
3183 exec->maxRollbacks = 4;
3184 exec->rollbacks = (xmlRegExecRollback *) xmlMalloc(exec->maxRollbacks *
3185 sizeof(xmlRegExecRollback));
3186 if (exec->rollbacks == NULL) {
3187 xmlRegexpErrMemory(NULL, "saving regexp");
3188 exec->maxRollbacks = 0;
3189 return;
3190 }
3191 memset(exec->rollbacks, 0,
3192 exec->maxRollbacks * sizeof(xmlRegExecRollback));
3193 } else if (exec->nbRollbacks >= exec->maxRollbacks) {
3194 xmlRegExecRollback *tmp;
3195 int len = exec->maxRollbacks;
3196
3197 exec->maxRollbacks *= 2;
3198 tmp = (xmlRegExecRollback *) xmlRealloc(exec->rollbacks,
3199 exec->maxRollbacks * sizeof(xmlRegExecRollback));
3200 if (tmp == NULL) {
3201 xmlRegexpErrMemory(NULL, "saving regexp");
3202 exec->maxRollbacks /= 2;
3203 return;
3204 }
3205 exec->rollbacks = tmp;
3206 tmp = &exec->rollbacks[len];
3207 memset(tmp, 0, (exec->maxRollbacks - len) * sizeof(xmlRegExecRollback));
3208 }
3209 exec->rollbacks[exec->nbRollbacks].state = exec->state;
3210 exec->rollbacks[exec->nbRollbacks].index = exec->index;
3211 exec->rollbacks[exec->nbRollbacks].nextbranch = exec->transno + 1;
3212 if (exec->comp->nbCounters > 0) {
3213 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3214 exec->rollbacks[exec->nbRollbacks].counts = (int *)
3215 xmlMalloc(exec->comp->nbCounters * sizeof(int));
3216 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3217 xmlRegexpErrMemory(NULL, "saving regexp");
3218 exec->status = -5;
3219 return;
3220 }
3221 }
3222 memcpy(exec->rollbacks[exec->nbRollbacks].counts, exec->counts,
3223 exec->comp->nbCounters * sizeof(int));
3224 }
3225 exec->nbRollbacks++;
3226}
3227
3228static void
3229xmlFARegExecRollBack(xmlRegExecCtxtPtr exec) {
3230 if (exec->nbRollbacks <= 0) {
3231 exec->status = -1;
3232#ifdef DEBUG_REGEXP_EXEC
3233 printf("rollback failed on empty stack\n");
3234#endif
3235 return;
3236 }
3237 exec->nbRollbacks--;
3238 exec->state = exec->rollbacks[exec->nbRollbacks].state;
3239 exec->index = exec->rollbacks[exec->nbRollbacks].index;
3240 exec->transno = exec->rollbacks[exec->nbRollbacks].nextbranch;
3241 if (exec->comp->nbCounters > 0) {
3242 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3243 fprintf(stderr, "exec save: allocation failed");
3244 exec->status = -6;
3245 return;
3246 }
3247 if (exec->counts) {
3248 memcpy(exec->counts, exec->rollbacks[exec->nbRollbacks].counts,
3249 exec->comp->nbCounters * sizeof(int));
3250 }
3251 }
3252
3253#ifdef DEBUG_REGEXP_EXEC
3254 printf("restored ");
3255 xmlFARegDebugExec(exec);
3256#endif
3257}
3258
3259/************************************************************************
3260 * *
3261 * Verifier, running an input against a compiled regexp *
3262 * *
3263 ************************************************************************/
3264
3265static int
3266xmlFARegExec(xmlRegexpPtr comp, const xmlChar *content) {
3267 xmlRegExecCtxt execval;
3268 xmlRegExecCtxtPtr exec = &execval;
3269 int ret, codepoint = 0, len, deter;
3270
3271 exec->inputString = content;
3272 exec->index = 0;
3273 exec->nbPush = 0;
3274 exec->determinist = 1;
3275 exec->maxRollbacks = 0;
3276 exec->nbRollbacks = 0;
3277 exec->rollbacks = NULL;
3278 exec->status = 0;
3279 exec->comp = comp;
3280 exec->state = comp->states[0];
3281 exec->transno = 0;
3282 exec->transcount = 0;
3283 exec->inputStack = NULL;
3284 exec->inputStackMax = 0;
3285 if (comp->nbCounters > 0) {
3286 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int));
3287 if (exec->counts == NULL) {
3288 xmlRegexpErrMemory(NULL, "running regexp");
3289 return(-1);
3290 }
3291 memset(exec->counts, 0, comp->nbCounters * sizeof(int));
3292 } else
3293 exec->counts = NULL;
3294 while ((exec->status == 0) && (exec->state != NULL) &&
3295 ((exec->inputString[exec->index] != 0) ||
3296 ((exec->state != NULL) &&
3297 (exec->state->type != XML_REGEXP_FINAL_STATE)))) {
3298 xmlRegTransPtr trans;
3299 xmlRegAtomPtr atom;
3300
3301 /*
3302 * If end of input on non-terminal state, rollback, however we may
3303 * still have epsilon like transition for counted transitions
3304 * on counters, in that case don't break too early. Additionally,
3305 * if we are working on a range like "AB{0,2}", where B is not present,
3306 * we don't want to break.
3307 */
3308 len = 1;
3309 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL)) {
3310 /*
3311 * if there is a transition, we must check if
3312 * atom allows minOccurs of 0
3313 */
3314 if (exec->transno < exec->state->nbTrans) {
3315 trans = &exec->state->trans[exec->transno];
3316 if (trans->to >=0) {
3317 atom = trans->atom;
3318 if (!((atom->min == 0) && (atom->max > 0)))
3319 goto rollback;
3320 }
3321 } else
3322 goto rollback;
3323 }
3324
3325 exec->transcount = 0;
3326 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
3327 trans = &exec->state->trans[exec->transno];
3328 if (trans->to < 0)
3329 continue;
3330 atom = trans->atom;
3331 ret = 0;
3332 deter = 1;
3333 if (trans->count >= 0) {
3334 int count;
3335 xmlRegCounterPtr counter;
3336
3337 if (exec->counts == NULL) {
3338 exec->status = -1;
3339 goto error;
3340 }
3341 /*
3342 * A counted transition.
3343 */
3344
3345 count = exec->counts[trans->count];
3346 counter = &exec->comp->counters[trans->count];
3347#ifdef DEBUG_REGEXP_EXEC
3348 printf("testing count %d: val %d, min %d, max %d\n",
3349 trans->count, count, counter->min, counter->max);
3350#endif
3351 ret = ((count >= counter->min) && (count <= counter->max));
3352 if ((ret) && (counter->min != counter->max))
3353 deter = 0;
3354 } else if (atom == NULL) {
3355 fprintf(stderr, "epsilon transition left at runtime\n");
3356 exec->status = -2;
3357 break;
3358 } else if (exec->inputString[exec->index] != 0) {
3359 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
3360 ret = xmlRegCheckCharacter(atom, codepoint);
3361 if ((ret == 1) && (atom->min >= 0) && (atom->max > 0)) {
3362 xmlRegStatePtr to = comp->states[trans->to];
3363
3364 /*
3365 * this is a multiple input sequence
3366 * If there is a counter associated increment it now.
3367 * do not increment if the counter is already over the
3368 * maximum limit in which case get to next transition
3369 */
3370 if (trans->counter >= 0) {
3371 xmlRegCounterPtr counter;
3372
3373 if ((exec->counts == NULL) ||
3374 (exec->comp == NULL) ||
3375 (exec->comp->counters == NULL)) {
3376 exec->status = -1;
3377 goto error;
3378 }
3379 counter = &exec->comp->counters[trans->counter];
3380 if (exec->counts[trans->counter] >= counter->max)
3381 continue; /* for loop on transitions */
3382 }
3383 /* Save before incrementing */
3384 if (exec->state->nbTrans > exec->transno + 1) {
3385 xmlFARegExecSave(exec);
3386 }
3387 if (trans->counter >= 0) {
3388#ifdef DEBUG_REGEXP_EXEC
3389 printf("Increasing count %d\n", trans->counter);
3390#endif
3391 exec->counts[trans->counter]++;
3392 }
3393 exec->transcount = 1;
3394 do {
3395 /*
3396 * Try to progress as much as possible on the input
3397 */
3398 if (exec->transcount == atom->max) {
3399 break;
3400 }
3401 exec->index += len;
3402 /*
3403 * End of input: stop here
3404 */
3405 if (exec->inputString[exec->index] == 0) {
3406 exec->index -= len;
3407 break;
3408 }
3409 if (exec->transcount >= atom->min) {
3410 int transno = exec->transno;
3411 xmlRegStatePtr state = exec->state;
3412
3413 /*
3414 * The transition is acceptable save it
3415 */
3416 exec->transno = -1; /* trick */
3417 exec->state = to;
3418 xmlFARegExecSave(exec);
3419 exec->transno = transno;
3420 exec->state = state;
3421 }
3422 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
3423 len);
3424 ret = xmlRegCheckCharacter(atom, codepoint);
3425 exec->transcount++;
3426 } while (ret == 1);
3427 if (exec->transcount < atom->min)
3428 ret = 0;
3429
3430 /*
3431 * If the last check failed but one transition was found
3432 * possible, rollback
3433 */
3434 if (ret < 0)
3435 ret = 0;
3436 if (ret == 0) {
3437 goto rollback;
3438 }
3439 if (trans->counter >= 0) {
3440 if (exec->counts == NULL) {
3441 exec->status = -1;
3442 goto error;
3443 }
3444#ifdef DEBUG_REGEXP_EXEC
3445 printf("Decreasing count %d\n", trans->counter);
3446#endif
3447 exec->counts[trans->counter]--;
3448 }
3449 } else if ((ret == 0) && (atom->min == 0) && (atom->max > 0)) {
3450 /*
3451 * we don't match on the codepoint, but minOccurs of 0
3452 * says that's ok. Setting len to 0 inhibits stepping
3453 * over the codepoint.
3454 */
3455 exec->transcount = 1;
3456 len = 0;
3457 ret = 1;
3458 }
3459 } else if ((atom->min == 0) && (atom->max > 0)) {
3460 /* another spot to match when minOccurs is 0 */
3461 exec->transcount = 1;
3462 len = 0;
3463 ret = 1;
3464 }
3465 if (ret == 1) {
3466 if ((trans->nd == 1) ||
3467 ((trans->count >= 0) && (deter == 0) &&
3468 (exec->state->nbTrans > exec->transno + 1))) {
3469#ifdef DEBUG_REGEXP_EXEC
3470 if (trans->nd == 1)
3471 printf("Saving on nd transition atom %d for %c at %d\n",
3472 trans->atom->no, codepoint, exec->index);
3473 else
3474 printf("Saving on counted transition count %d for %c at %d\n",
3475 trans->count, codepoint, exec->index);
3476#endif
3477 xmlFARegExecSave(exec);
3478 }
3479 if (trans->counter >= 0) {
3480 xmlRegCounterPtr counter;
3481
3482 /* make sure we don't go over the counter maximum value */
3483 if ((exec->counts == NULL) ||
3484 (exec->comp == NULL) ||
3485 (exec->comp->counters == NULL)) {
3486 exec->status = -1;
3487 goto error;
3488 }
3489 counter = &exec->comp->counters[trans->counter];
3490 if (exec->counts[trans->counter] >= counter->max)
3491 continue; /* for loop on transitions */
3492#ifdef DEBUG_REGEXP_EXEC
3493 printf("Increasing count %d\n", trans->counter);
3494#endif
3495 exec->counts[trans->counter]++;
3496 }
3497 if ((trans->count >= 0) &&
3498 (trans->count < REGEXP_ALL_COUNTER)) {
3499 if (exec->counts == NULL) {
3500 exec->status = -1;
3501 goto error;
3502 }
3503#ifdef DEBUG_REGEXP_EXEC
3504 printf("resetting count %d on transition\n",
3505 trans->count);
3506#endif
3507 exec->counts[trans->count] = 0;
3508 }
3509#ifdef DEBUG_REGEXP_EXEC
3510 printf("entering state %d\n", trans->to);
3511#endif
3512 exec->state = comp->states[trans->to];
3513 exec->transno = 0;
3514 if (trans->atom != NULL) {
3515 exec->index += len;
3516 }
3517 goto progress;
3518 } else if (ret < 0) {
3519 exec->status = -4;
3520 break;
3521 }
3522 }
3523 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
3524rollback:
3525 /*
3526 * Failed to find a way out
3527 */
3528 exec->determinist = 0;
3529#ifdef DEBUG_REGEXP_EXEC
3530 printf("rollback from state %d on %d:%c\n", exec->state->no,
3531 codepoint,codepoint);
3532#endif
3533 xmlFARegExecRollBack(exec);
3534 }
3535progress:
3536 continue;
3537 }
3538error:
3539 if (exec->rollbacks != NULL) {
3540 if (exec->counts != NULL) {
3541 int i;
3542
3543 for (i = 0;i < exec->maxRollbacks;i++)
3544 if (exec->rollbacks[i].counts != NULL)
3545 xmlFree(exec->rollbacks[i].counts);
3546 }
3547 xmlFree(exec->rollbacks);
3548 }
3549 if (exec->state == NULL)
3550 return(-1);
3551 if (exec->counts != NULL)
3552 xmlFree(exec->counts);
3553 if (exec->status == 0)
3554 return(1);
3555 if (exec->status == -1) {
3556 if (exec->nbPush > MAX_PUSH)
3557 return(-1);
3558 return(0);
3559 }
3560 return(exec->status);
3561}
3562
3563/************************************************************************
3564 * *
3565 * Progressive interface to the verifier one atom at a time *
3566 * *
3567 ************************************************************************/
3568#ifdef DEBUG_ERR
3569static void testerr(xmlRegExecCtxtPtr exec);
3570#endif
3571
3572/**
3573 * xmlRegNewExecCtxt:
3574 * @comp: a precompiled regular expression
3575 * @callback: a callback function used for handling progresses in the
3576 * automata matching phase
3577 * @data: the context data associated to the callback in this context
3578 *
3579 * Build a context used for progressive evaluation of a regexp.
3580 *
3581 * Returns the new context
3582 */
3583xmlRegExecCtxtPtr
3584xmlRegNewExecCtxt(xmlRegexpPtr comp, xmlRegExecCallbacks callback, void *data) {
3585 xmlRegExecCtxtPtr exec;
3586
3587 if (comp == NULL)
3588 return(NULL);
3589 if ((comp->compact == NULL) && (comp->states == NULL))
3590 return(NULL);
3591 exec = (xmlRegExecCtxtPtr) xmlMalloc(sizeof(xmlRegExecCtxt));
3592 if (exec == NULL) {
3593 xmlRegexpErrMemory(NULL, "creating execution context");
3594 return(NULL);
3595 }
3596 memset(exec, 0, sizeof(xmlRegExecCtxt));
3597 exec->inputString = NULL;
3598 exec->index = 0;
3599 exec->determinist = 1;
3600 exec->maxRollbacks = 0;
3601 exec->nbRollbacks = 0;
3602 exec->rollbacks = NULL;
3603 exec->status = 0;
3604 exec->comp = comp;
3605 if (comp->compact == NULL)
3606 exec->state = comp->states[0];
3607 exec->transno = 0;
3608 exec->transcount = 0;
3609 exec->callback = callback;
3610 exec->data = data;
3611 if (comp->nbCounters > 0) {
3612 /*
3613 * For error handling, exec->counts is allocated twice the size
3614 * the second half is used to store the data in case of rollback
3615 */
3616 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int)
3617 * 2);
3618 if (exec->counts == NULL) {
3619 xmlRegexpErrMemory(NULL, "creating execution context");
3620 xmlFree(exec);
3621 return(NULL);
3622 }
3623 memset(exec->counts, 0, comp->nbCounters * sizeof(int) * 2);
3624 exec->errCounts = &exec->counts[comp->nbCounters];
3625 } else {
3626 exec->counts = NULL;
3627 exec->errCounts = NULL;
3628 }
3629 exec->inputStackMax = 0;
3630 exec->inputStackNr = 0;
3631 exec->inputStack = NULL;
3632 exec->errStateNo = -1;
3633 exec->errString = NULL;
3634 exec->nbPush = 0;
3635 return(exec);
3636}
3637
3638/**
3639 * xmlRegFreeExecCtxt:
3640 * @exec: a regular expression evaluation context
3641 *
3642 * Free the structures associated to a regular expression evaluation context.
3643 */
3644void
3645xmlRegFreeExecCtxt(xmlRegExecCtxtPtr exec) {
3646 if (exec == NULL)
3647 return;
3648
3649 if (exec->rollbacks != NULL) {
3650 if (exec->counts != NULL) {
3651 int i;
3652
3653 for (i = 0;i < exec->maxRollbacks;i++)
3654 if (exec->rollbacks[i].counts != NULL)
3655 xmlFree(exec->rollbacks[i].counts);
3656 }
3657 xmlFree(exec->rollbacks);
3658 }
3659 if (exec->counts != NULL)
3660 xmlFree(exec->counts);
3661 if (exec->inputStack != NULL) {
3662 int i;
3663
3664 for (i = 0;i < exec->inputStackNr;i++) {
3665 if (exec->inputStack[i].value != NULL)
3666 xmlFree(exec->inputStack[i].value);
3667 }
3668 xmlFree(exec->inputStack);
3669 }
3670 if (exec->errString != NULL)
3671 xmlFree(exec->errString);
3672 xmlFree(exec);
3673}
3674
3675static void
3676xmlFARegExecSaveInputString(xmlRegExecCtxtPtr exec, const xmlChar *value,
3677 void *data) {
3678#ifdef DEBUG_PUSH
3679 printf("saving value: %d:%s\n", exec->inputStackNr, value);
3680#endif
3681 if (exec->inputStackMax == 0) {
3682 exec->inputStackMax = 4;
3683 exec->inputStack = (xmlRegInputTokenPtr)
3684 xmlMalloc(exec->inputStackMax * sizeof(xmlRegInputToken));
3685 if (exec->inputStack == NULL) {
3686 xmlRegexpErrMemory(NULL, "pushing input string");
3687 exec->inputStackMax = 0;
3688 return;
3689 }
3690 } else if (exec->inputStackNr + 1 >= exec->inputStackMax) {
3691 xmlRegInputTokenPtr tmp;
3692
3693 exec->inputStackMax *= 2;
3694 tmp = (xmlRegInputTokenPtr) xmlRealloc(exec->inputStack,
3695 exec->inputStackMax * sizeof(xmlRegInputToken));
3696 if (tmp == NULL) {
3697 xmlRegexpErrMemory(NULL, "pushing input string");
3698 exec->inputStackMax /= 2;
3699 return;
3700 }
3701 exec->inputStack = tmp;
3702 }
3703 exec->inputStack[exec->inputStackNr].value = xmlStrdup(value);
3704 exec->inputStack[exec->inputStackNr].data = data;
3705 exec->inputStackNr++;
3706 exec->inputStack[exec->inputStackNr].value = NULL;
3707 exec->inputStack[exec->inputStackNr].data = NULL;
3708}
3709
3710/**
3711 * xmlRegStrEqualWildcard:
3712 * @expStr: the string to be evaluated
3713 * @valStr: the validation string
3714 *
3715 * Checks if both strings are equal or have the same content. "*"
3716 * can be used as a wildcard in @valStr; "|" is used as a separator of
3717 * substrings in both @expStr and @valStr.
3718 *
3719 * Returns 1 if the comparison is satisfied and the number of substrings
3720 * is equal, 0 otherwise.
3721 */
3722
3723static int
3724xmlRegStrEqualWildcard(const xmlChar *expStr, const xmlChar *valStr) {
3725 if (expStr == valStr) return(1);
3726 if (expStr == NULL) return(0);
3727 if (valStr == NULL) return(0);
3728 do {
3729 /*
3730 * Eval if we have a wildcard for the current item.
3731 */
3732 if (*expStr != *valStr) {
3733 /* if one of them starts with a wildcard make valStr be it */
3734 if (*valStr == '*') {
3735 const xmlChar *tmp;
3736
3737 tmp = valStr;
3738 valStr = expStr;
3739 expStr = tmp;
3740 }
3741 if ((*valStr != 0) && (*expStr != 0) && (*expStr++ == '*')) {
3742 do {
3743 if (*valStr == XML_REG_STRING_SEPARATOR)
3744 break;
3745 valStr++;
3746 } while (*valStr != 0);
3747 continue;
3748 } else
3749 return(0);
3750 }
3751 expStr++;
3752 valStr++;
3753 } while (*valStr != 0);
3754 if (*expStr != 0)
3755 return (0);
3756 else
3757 return (1);
3758}
3759
3760/**
3761 * xmlRegCompactPushString:
3762 * @exec: a regexp execution context
3763 * @comp: the precompiled exec with a compact table
3764 * @value: a string token input
3765 * @data: data associated to the token to reuse in callbacks
3766 *
3767 * Push one input token in the execution context
3768 *
3769 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
3770 * a negative value in case of error.
3771 */
3772static int
3773xmlRegCompactPushString(xmlRegExecCtxtPtr exec,
3774 xmlRegexpPtr comp,
3775 const xmlChar *value,
3776 void *data) {
3777 int state = exec->index;
3778 int i, target;
3779
3780 if ((comp == NULL) || (comp->compact == NULL) || (comp->stringMap == NULL))
3781 return(-1);
3782
3783 if (value == NULL) {
3784 /*
3785 * are we at a final state ?
3786 */
3787 if (comp->compact[state * (comp->nbstrings + 1)] ==
3788 XML_REGEXP_FINAL_STATE)
3789 return(1);
3790 return(0);
3791 }
3792
3793#ifdef DEBUG_PUSH
3794 printf("value pushed: %s\n", value);
3795#endif
3796
3797 /*
3798 * Examine all outside transitions from current state
3799 */
3800 for (i = 0;i < comp->nbstrings;i++) {
3801 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
3802 if ((target > 0) && (target <= comp->nbstates)) {
3803 target--; /* to avoid 0 */
3804 if (xmlRegStrEqualWildcard(comp->stringMap[i], value)) {
3805 exec->index = target;
3806 if ((exec->callback != NULL) && (comp->transdata != NULL)) {
3807 exec->callback(exec->data, value,
3808 comp->transdata[state * comp->nbstrings + i], data);
3809 }
3810#ifdef DEBUG_PUSH
3811 printf("entering state %d\n", target);
3812#endif
3813 if (comp->compact[target * (comp->nbstrings + 1)] ==
3814 XML_REGEXP_SINK_STATE)
3815 goto error;
3816
3817 if (comp->compact[target * (comp->nbstrings + 1)] ==
3818 XML_REGEXP_FINAL_STATE)
3819 return(1);
3820 return(0);
3821 }
3822 }
3823 }
3824 /*
3825 * Failed to find an exit transition out from current state for the
3826 * current token
3827 */
3828#ifdef DEBUG_PUSH
3829 printf("failed to find a transition for %s on state %d\n", value, state);
3830#endif
3831error:
3832 if (exec->errString != NULL)
3833 xmlFree(exec->errString);
3834 exec->errString = xmlStrdup(value);
3835 exec->errStateNo = state;
3836 exec->status = -1;
3837#ifdef DEBUG_ERR
3838 testerr(exec);
3839#endif
3840 return(-1);
3841}
3842
3843/**
3844 * xmlRegExecPushStringInternal:
3845 * @exec: a regexp execution context or NULL to indicate the end
3846 * @value: a string token input
3847 * @data: data associated to the token to reuse in callbacks
3848 * @compound: value was assembled from 2 strings
3849 *
3850 * Push one input token in the execution context
3851 *
3852 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
3853 * a negative value in case of error.
3854 */
3855static int
3856xmlRegExecPushStringInternal(xmlRegExecCtxtPtr exec, const xmlChar *value,
3857 void *data, int compound) {
3858 xmlRegTransPtr trans;
3859 xmlRegAtomPtr atom;
3860 int ret;
3861 int final = 0;
3862 int progress = 1;
3863
3864 if (exec == NULL)
3865 return(-1);
3866 if (exec->comp == NULL)
3867 return(-1);
3868 if (exec->status != 0)
3869 return(exec->status);
3870
3871 if (exec->comp->compact != NULL)
3872 return(xmlRegCompactPushString(exec, exec->comp, value, data));
3873
3874 if (value == NULL) {
3875 if (exec->state->type == XML_REGEXP_FINAL_STATE)
3876 return(1);
3877 final = 1;
3878 }
3879
3880#ifdef DEBUG_PUSH
3881 printf("value pushed: %s\n", value);
3882#endif
3883 /*
3884 * If we have an active rollback stack push the new value there
3885 * and get back to where we were left
3886 */
3887 if ((value != NULL) && (exec->inputStackNr > 0)) {
3888 xmlFARegExecSaveInputString(exec, value, data);
3889 value = exec->inputStack[exec->index].value;
3890 data = exec->inputStack[exec->index].data;
3891#ifdef DEBUG_PUSH
3892 printf("value loaded: %s\n", value);
3893#endif
3894 }
3895
3896 while ((exec->status == 0) &&
3897 ((value != NULL) ||
3898 ((final == 1) &&
3899 (exec->state->type != XML_REGEXP_FINAL_STATE)))) {
3900
3901 /*
3902 * End of input on non-terminal state, rollback, however we may
3903 * still have epsilon like transition for counted transitions
3904 * on counters, in that case don't break too early.
3905 */
3906 if ((value == NULL) && (exec->counts == NULL))
3907 goto rollback;
3908
3909 exec->transcount = 0;
3910 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
3911 trans = &exec->state->trans[exec->transno];
3912 if (trans->to < 0)
3913 continue;
3914 atom = trans->atom;
3915 ret = 0;
3916 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
3917 int i;
3918 int count;
3919 xmlRegTransPtr t;
3920 xmlRegCounterPtr counter;
3921
3922 ret = 0;
3923
3924#ifdef DEBUG_PUSH
3925 printf("testing all lax %d\n", trans->count);
3926#endif
3927 /*
3928 * Check all counted transitions from the current state
3929 */
3930 if ((value == NULL) && (final)) {
3931 ret = 1;
3932 } else if (value != NULL) {
3933 for (i = 0;i < exec->state->nbTrans;i++) {
3934 t = &exec->state->trans[i];
3935 if ((t->counter < 0) || (t == trans))
3936 continue;
3937 counter = &exec->comp->counters[t->counter];
3938 count = exec->counts[t->counter];
3939 if ((count < counter->max) &&
3940 (t->atom != NULL) &&
3941 (xmlStrEqual(value, t->atom->valuep))) {
3942 ret = 0;
3943 break;
3944 }
3945 if ((count >= counter->min) &&
3946 (count < counter->max) &&
3947 (t->atom != NULL) &&
3948 (xmlStrEqual(value, t->atom->valuep))) {
3949 ret = 1;
3950 break;
3951 }
3952 }
3953 }
3954 } else if (trans->count == REGEXP_ALL_COUNTER) {
3955 int i;
3956 int count;
3957 xmlRegTransPtr t;
3958 xmlRegCounterPtr counter;
3959
3960 ret = 1;
3961
3962#ifdef DEBUG_PUSH
3963 printf("testing all %d\n", trans->count);
3964#endif
3965 /*
3966 * Check all counted transitions from the current state
3967 */
3968 for (i = 0;i < exec->state->nbTrans;i++) {
3969 t = &exec->state->trans[i];
3970 if ((t->counter < 0) || (t == trans))
3971 continue;
3972 counter = &exec->comp->counters[t->counter];
3973 count = exec->counts[t->counter];
3974 if ((count < counter->min) || (count > counter->max)) {
3975 ret = 0;
3976 break;
3977 }
3978 }
3979 } else if (trans->count >= 0) {
3980 int count;
3981 xmlRegCounterPtr counter;
3982
3983 /*
3984 * A counted transition.
3985 */
3986
3987 count = exec->counts[trans->count];
3988 counter = &exec->comp->counters[trans->count];
3989#ifdef DEBUG_PUSH
3990 printf("testing count %d: val %d, min %d, max %d\n",
3991 trans->count, count, counter->min, counter->max);
3992#endif
3993 ret = ((count >= counter->min) && (count <= counter->max));
3994 } else if (atom == NULL) {
3995 fprintf(stderr, "epsilon transition left at runtime\n");
3996 exec->status = -2;
3997 break;
3998 } else if (value != NULL) {
3999 ret = xmlRegStrEqualWildcard(atom->valuep, value);
4000 if (atom->neg) {
4001 ret = !ret;
4002 if (!compound)
4003 ret = 0;
4004 }
4005 if ((ret == 1) && (trans->counter >= 0)) {
4006 xmlRegCounterPtr counter;
4007 int count;
4008
4009 count = exec->counts[trans->counter];
4010 counter = &exec->comp->counters[trans->counter];
4011 if (count >= counter->max)
4012 ret = 0;
4013 }
4014
4015 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
4016 xmlRegStatePtr to = exec->comp->states[trans->to];
4017
4018 /*
4019 * this is a multiple input sequence
4020 */
4021 if (exec->state->nbTrans > exec->transno + 1) {
4022 if (exec->inputStackNr <= 0) {
4023 xmlFARegExecSaveInputString(exec, value, data);
4024 }
4025 xmlFARegExecSave(exec);
4026 }
4027 exec->transcount = 1;
4028 do {
4029 /*
4030 * Try to progress as much as possible on the input
4031 */
4032 if (exec->transcount == atom->max) {
4033 break;
4034 }
4035 exec->index++;
4036 value = exec->inputStack[exec->index].value;
4037 data = exec->inputStack[exec->index].data;
4038#ifdef DEBUG_PUSH
4039 printf("value loaded: %s\n", value);
4040#endif
4041
4042 /*
4043 * End of input: stop here
4044 */
4045 if (value == NULL) {
4046 exec->index --;
4047 break;
4048 }
4049 if (exec->transcount >= atom->min) {
4050 int transno = exec->transno;
4051 xmlRegStatePtr state = exec->state;
4052
4053 /*
4054 * The transition is acceptable save it
4055 */
4056 exec->transno = -1; /* trick */
4057 exec->state = to;
4058 if (exec->inputStackNr <= 0) {
4059 xmlFARegExecSaveInputString(exec, value, data);
4060 }
4061 xmlFARegExecSave(exec);
4062 exec->transno = transno;
4063 exec->state = state;
4064 }
4065 ret = xmlStrEqual(value, atom->valuep);
4066 exec->transcount++;
4067 } while (ret == 1);
4068 if (exec->transcount < atom->min)
4069 ret = 0;
4070
4071 /*
4072 * If the last check failed but one transition was found
4073 * possible, rollback
4074 */
4075 if (ret < 0)
4076 ret = 0;
4077 if (ret == 0) {
4078 goto rollback;
4079 }
4080 }
4081 }
4082 if (ret == 1) {
4083 if ((exec->callback != NULL) && (atom != NULL) &&
4084 (data != NULL)) {
4085 exec->callback(exec->data, atom->valuep,
4086 atom->data, data);
4087 }
4088 if (exec->state->nbTrans > exec->transno + 1) {
4089 if (exec->inputStackNr <= 0) {
4090 xmlFARegExecSaveInputString(exec, value, data);
4091 }
4092 xmlFARegExecSave(exec);
4093 }
4094 if (trans->counter >= 0) {
4095#ifdef DEBUG_PUSH
4096 printf("Increasing count %d\n", trans->counter);
4097#endif
4098 exec->counts[trans->counter]++;
4099 }
4100 if ((trans->count >= 0) &&
4101 (trans->count < REGEXP_ALL_COUNTER)) {
4102#ifdef DEBUG_REGEXP_EXEC
4103 printf("resetting count %d on transition\n",
4104 trans->count);
4105#endif
4106 exec->counts[trans->count] = 0;
4107 }
4108#ifdef DEBUG_PUSH
4109 printf("entering state %d\n", trans->to);
4110#endif
4111 if ((exec->comp->states[trans->to] != NULL) &&
4112 (exec->comp->states[trans->to]->type ==
4113 XML_REGEXP_SINK_STATE)) {
4114 /*
4115 * entering a sink state, save the current state as error
4116 * state.
4117 */
4118 if (exec->errString != NULL)
4119 xmlFree(exec->errString);
4120 exec->errString = xmlStrdup(value);
4121 exec->errState = exec->state;
4122 memcpy(exec->errCounts, exec->counts,
4123 exec->comp->nbCounters * sizeof(int));
4124 }
4125 exec->state = exec->comp->states[trans->to];
4126 exec->transno = 0;
4127 if (trans->atom != NULL) {
4128 if (exec->inputStack != NULL) {
4129 exec->index++;
4130 if (exec->index < exec->inputStackNr) {
4131 value = exec->inputStack[exec->index].value;
4132 data = exec->inputStack[exec->index].data;
4133#ifdef DEBUG_PUSH
4134 printf("value loaded: %s\n", value);
4135#endif
4136 } else {
4137 value = NULL;
4138 data = NULL;
4139#ifdef DEBUG_PUSH
4140 printf("end of input\n");
4141#endif
4142 }
4143 } else {
4144 value = NULL;
4145 data = NULL;
4146#ifdef DEBUG_PUSH
4147 printf("end of input\n");
4148#endif
4149 }
4150 }
4151 goto progress;
4152 } else if (ret < 0) {
4153 exec->status = -4;
4154 break;
4155 }
4156 }
4157 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
4158rollback:
4159 /*
4160 * if we didn't yet rollback on the current input
4161 * store the current state as the error state.
4162 */
4163 if ((progress) && (exec->state != NULL) &&
4164 (exec->state->type != XML_REGEXP_SINK_STATE)) {
4165 progress = 0;
4166 if (exec->errString != NULL)
4167 xmlFree(exec->errString);
4168 exec->errString = xmlStrdup(value);
4169 exec->errState = exec->state;
4170 if (exec->comp->nbCounters)
4171 memcpy(exec->errCounts, exec->counts,
4172 exec->comp->nbCounters * sizeof(int));
4173 }
4174
4175 /*
4176 * Failed to find a way out
4177 */
4178 exec->determinist = 0;
4179 xmlFARegExecRollBack(exec);
4180 if ((exec->inputStack != NULL ) && (exec->status == 0)) {
4181 value = exec->inputStack[exec->index].value;
4182 data = exec->inputStack[exec->index].data;
4183#ifdef DEBUG_PUSH
4184 printf("value loaded: %s\n", value);
4185#endif
4186 }
4187 }
4188 continue;
4189progress:
4190 progress = 1;
4191 continue;
4192 }
4193 if (exec->status == 0) {
4194 return(exec->state->type == XML_REGEXP_FINAL_STATE);
4195 }
4196#ifdef DEBUG_ERR
4197 if (exec->status < 0) {
4198 testerr(exec);
4199 }
4200#endif
4201 return(exec->status);
4202}
4203
4204/**
4205 * xmlRegExecPushString:
4206 * @exec: a regexp execution context or NULL to indicate the end
4207 * @value: a string token input
4208 * @data: data associated to the token to reuse in callbacks
4209 *
4210 * Push one input token in the execution context
4211 *
4212 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
4213 * a negative value in case of error.
4214 */
4215int
4216xmlRegExecPushString(xmlRegExecCtxtPtr exec, const xmlChar *value,
4217 void *data) {
4218 return(xmlRegExecPushStringInternal(exec, value, data, 0));
4219}
4220
4221/**
4222 * xmlRegExecPushString2:
4223 * @exec: a regexp execution context or NULL to indicate the end
4224 * @value: the first string token input
4225 * @value2: the second string token input
4226 * @data: data associated to the token to reuse in callbacks
4227 *
4228 * Push one input token in the execution context
4229 *
4230 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
4231 * a negative value in case of error.
4232 */
4233int
4234xmlRegExecPushString2(xmlRegExecCtxtPtr exec, const xmlChar *value,
4235 const xmlChar *value2, void *data) {
4236 xmlChar buf[150];
4237 int lenn, lenp, ret;
4238 xmlChar *str;
4239
4240 if (exec == NULL)
4241 return(-1);
4242 if (exec->comp == NULL)
4243 return(-1);
4244 if (exec->status != 0)
4245 return(exec->status);
4246
4247 if (value2 == NULL)
4248 return(xmlRegExecPushString(exec, value, data));
4249
4250 lenn = strlen((char *) value2);
4251 lenp = strlen((char *) value);
4252
4253 if (150 < lenn + lenp + 2) {
4254 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
4255 if (str == NULL) {
4256 exec->status = -1;
4257 return(-1);
4258 }
4259 } else {
4260 str = buf;
4261 }
4262 memcpy(&str[0], value, lenp);
4263 str[lenp] = XML_REG_STRING_SEPARATOR;
4264 memcpy(&str[lenp + 1], value2, lenn);
4265 str[lenn + lenp + 1] = 0;
4266
4267 if (exec->comp->compact != NULL)
4268 ret = xmlRegCompactPushString(exec, exec->comp, str, data);
4269 else
4270 ret = xmlRegExecPushStringInternal(exec, str, data, 1);
4271
4272 if (str != buf)
4273 xmlFree(str);
4274 return(ret);
4275}
4276
4277/**
4278 * xmlRegExecGetValues:
4279 * @exec: a regexp execution context
4280 * @err: error extraction or normal one
4281 * @nbval: pointer to the number of accepted values IN/OUT
4282 * @nbneg: return number of negative transitions
4283 * @values: pointer to the array of acceptable values
4284 * @terminal: return value if this was a terminal state
4285 *
4286 * Extract information from the regexp execution, internal routine to
4287 * implement xmlRegExecNextValues() and xmlRegExecErrInfo()
4288 *
4289 * Returns: 0 in case of success or -1 in case of error.
4290 */
4291static int
4292xmlRegExecGetValues(xmlRegExecCtxtPtr exec, int err,
4293 int *nbval, int *nbneg,
4294 xmlChar **values, int *terminal) {
4295 int maxval;
4296 int nb = 0;
4297
4298 if ((exec == NULL) || (nbval == NULL) || (nbneg == NULL) ||
4299 (values == NULL) || (*nbval <= 0))
4300 return(-1);
4301
4302 maxval = *nbval;
4303 *nbval = 0;
4304 *nbneg = 0;
4305 if ((exec->comp != NULL) && (exec->comp->compact != NULL)) {
4306 xmlRegexpPtr comp;
4307 int target, i, state;
4308
4309 comp = exec->comp;
4310
4311 if (err) {
4312 if (exec->errStateNo == -1) return(-1);
4313 state = exec->errStateNo;
4314 } else {
4315 state = exec->index;
4316 }
4317 if (terminal != NULL) {
4318 if (comp->compact[state * (comp->nbstrings + 1)] ==
4319 XML_REGEXP_FINAL_STATE)
4320 *terminal = 1;
4321 else
4322 *terminal = 0;
4323 }
4324 for (i = 0;(i < comp->nbstrings) && (nb < maxval);i++) {
4325 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
4326 if ((target > 0) && (target <= comp->nbstates) &&
4327 (comp->compact[(target - 1) * (comp->nbstrings + 1)] !=
4328 XML_REGEXP_SINK_STATE)) {
4329 values[nb++] = comp->stringMap[i];
4330 (*nbval)++;
4331 }
4332 }
4333 for (i = 0;(i < comp->nbstrings) && (nb < maxval);i++) {
4334 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
4335 if ((target > 0) && (target <= comp->nbstates) &&
4336 (comp->compact[(target - 1) * (comp->nbstrings + 1)] ==
4337 XML_REGEXP_SINK_STATE)) {
4338 values[nb++] = comp->stringMap[i];
4339 (*nbneg)++;
4340 }
4341 }
4342 } else {
4343 int transno;
4344 xmlRegTransPtr trans;
4345 xmlRegAtomPtr atom;
4346 xmlRegStatePtr state;
4347
4348 if (terminal != NULL) {
4349 if (exec->state->type == XML_REGEXP_FINAL_STATE)
4350 *terminal = 1;
4351 else
4352 *terminal = 0;
4353 }
4354
4355 if (err) {
4356 if (exec->errState == NULL) return(-1);
4357 state = exec->errState;
4358 } else {
4359 if (exec->state == NULL) return(-1);
4360 state = exec->state;
4361 }
4362 for (transno = 0;
4363 (transno < state->nbTrans) && (nb < maxval);
4364 transno++) {
4365 trans = &state->trans[transno];
4366 if (trans->to < 0)
4367 continue;
4368 atom = trans->atom;
4369 if ((atom == NULL) || (atom->valuep == NULL))
4370 continue;
4371 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
4372 /* this should not be reached but ... */
4373 TODO;
4374 } else if (trans->count == REGEXP_ALL_COUNTER) {
4375 /* this should not be reached but ... */
4376 TODO;
4377 } else if (trans->counter >= 0) {
4378 xmlRegCounterPtr counter = NULL;
4379 int count;
4380
4381 if (err)
4382 count = exec->errCounts[trans->counter];
4383 else
4384 count = exec->counts[trans->counter];
4385 if (exec->comp != NULL)
4386 counter = &exec->comp->counters[trans->counter];
4387 if ((counter == NULL) || (count < counter->max)) {
4388 if (atom->neg)
4389 values[nb++] = (xmlChar *) atom->valuep2;
4390 else
4391 values[nb++] = (xmlChar *) atom->valuep;
4392 (*nbval)++;
4393 }
4394 } else {
4395 if ((exec->comp != NULL) && (exec->comp->states[trans->to] != NULL) &&
4396 (exec->comp->states[trans->to]->type !=
4397 XML_REGEXP_SINK_STATE)) {
4398 if (atom->neg)
4399 values[nb++] = (xmlChar *) atom->valuep2;
4400 else
4401 values[nb++] = (xmlChar *) atom->valuep;
4402 (*nbval)++;
4403 }
4404 }
4405 }
4406 for (transno = 0;
4407 (transno < state->nbTrans) && (nb < maxval);
4408 transno++) {
4409 trans = &state->trans[transno];
4410 if (trans->to < 0)
4411 continue;
4412 atom = trans->atom;
4413 if ((atom == NULL) || (atom->valuep == NULL))
4414 continue;
4415 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
4416 continue;
4417 } else if (trans->count == REGEXP_ALL_COUNTER) {
4418 continue;
4419 } else if (trans->counter >= 0) {
4420 continue;
4421 } else {
4422 if ((exec->comp->states[trans->to] != NULL) &&
4423 (exec->comp->states[trans->to]->type ==
4424 XML_REGEXP_SINK_STATE)) {
4425 if (atom->neg)
4426 values[nb++] = (xmlChar *) atom->valuep2;
4427 else
4428 values[nb++] = (xmlChar *) atom->valuep;
4429 (*nbneg)++;
4430 }
4431 }
4432 }
4433 }
4434 return(0);
4435}
4436
4437/**
4438 * xmlRegExecNextValues:
4439 * @exec: a regexp execution context
4440 * @nbval: pointer to the number of accepted values IN/OUT
4441 * @nbneg: return number of negative transitions
4442 * @values: pointer to the array of acceptable values
4443 * @terminal: return value if this was a terminal state
4444 *
4445 * Extract information from the regexp execution,
4446 * the parameter @values must point to an array of @nbval string pointers
4447 * on return nbval will contain the number of possible strings in that
4448 * state and the @values array will be updated with them. The string values
4449 * returned will be freed with the @exec context and don't need to be
4450 * deallocated.
4451 *
4452 * Returns: 0 in case of success or -1 in case of error.
4453 */
4454int
4455xmlRegExecNextValues(xmlRegExecCtxtPtr exec, int *nbval, int *nbneg,
4456 xmlChar **values, int *terminal) {
4457 return(xmlRegExecGetValues(exec, 0, nbval, nbneg, values, terminal));
4458}
4459
4460/**
4461 * xmlRegExecErrInfo:
4462 * @exec: a regexp execution context generating an error
4463 * @string: return value for the error string
4464 * @nbval: pointer to the number of accepted values IN/OUT
4465 * @nbneg: return number of negative transitions
4466 * @values: pointer to the array of acceptable values
4467 * @terminal: return value if this was a terminal state
4468 *
4469 * Extract error information from the regexp execution, the parameter
4470 * @string will be updated with the value pushed and not accepted,
4471 * the parameter @values must point to an array of @nbval string pointers
4472 * on return nbval will contain the number of possible strings in that
4473 * state and the @values array will be updated with them. The string values
4474 * returned will be freed with the @exec context and don't need to be
4475 * deallocated.
4476 *
4477 * Returns: 0 in case of success or -1 in case of error.
4478 */
4479int
4480xmlRegExecErrInfo(xmlRegExecCtxtPtr exec, const xmlChar **string,
4481 int *nbval, int *nbneg, xmlChar **values, int *terminal) {
4482 if (exec == NULL)
4483 return(-1);
4484 if (string != NULL) {
4485 if (exec->status != 0)
4486 *string = exec->errString;
4487 else
4488 *string = NULL;
4489 }
4490 return(xmlRegExecGetValues(exec, 1, nbval, nbneg, values, terminal));
4491}
4492
4493#ifdef DEBUG_ERR
4494static void testerr(xmlRegExecCtxtPtr exec) {
4495 const xmlChar *string;
4496 xmlChar *values[5];
4497 int nb = 5;
4498 int nbneg;
4499 int terminal;
4500 xmlRegExecErrInfo(exec, &string, &nb, &nbneg, &values[0], &terminal);
4501}
4502#endif
4503
4504#if 0
4505static int
4506xmlRegExecPushChar(xmlRegExecCtxtPtr exec, int UCS) {
4507 xmlRegTransPtr trans;
4508 xmlRegAtomPtr atom;
4509 int ret;
4510 int codepoint, len;
4511
4512 if (exec == NULL)
4513 return(-1);
4514 if (exec->status != 0)
4515 return(exec->status);
4516
4517 while ((exec->status == 0) &&
4518 ((exec->inputString[exec->index] != 0) ||
4519 (exec->state->type != XML_REGEXP_FINAL_STATE))) {
4520
4521 /*
4522 * End of input on non-terminal state, rollback, however we may
4523 * still have epsilon like transition for counted transitions
4524 * on counters, in that case don't break too early.
4525 */
4526 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL))
4527 goto rollback;
4528
4529 exec->transcount = 0;
4530 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
4531 trans = &exec->state->trans[exec->transno];
4532 if (trans->to < 0)
4533 continue;
4534 atom = trans->atom;
4535 ret = 0;
4536 if (trans->count >= 0) {
4537 int count;
4538 xmlRegCounterPtr counter;
4539
4540 /*
4541 * A counted transition.
4542 */
4543
4544 count = exec->counts[trans->count];
4545 counter = &exec->comp->counters[trans->count];
4546#ifdef DEBUG_REGEXP_EXEC
4547 printf("testing count %d: val %d, min %d, max %d\n",
4548 trans->count, count, counter->min, counter->max);
4549#endif
4550 ret = ((count >= counter->min) && (count <= counter->max));
4551 } else if (atom == NULL) {
4552 fprintf(stderr, "epsilon transition left at runtime\n");
4553 exec->status = -2;
4554 break;
4555 } else if (exec->inputString[exec->index] != 0) {
4556 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
4557 ret = xmlRegCheckCharacter(atom, codepoint);
4558 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
4559 xmlRegStatePtr to = exec->comp->states[trans->to];
4560
4561 /*
4562 * this is a multiple input sequence
4563 */
4564 if (exec->state->nbTrans > exec->transno + 1) {
4565 xmlFARegExecSave(exec);
4566 }
4567 exec->transcount = 1;
4568 do {
4569 /*
4570 * Try to progress as much as possible on the input
4571 */
4572 if (exec->transcount == atom->max) {
4573 break;
4574 }
4575 exec->index += len;
4576 /*
4577 * End of input: stop here
4578 */
4579 if (exec->inputString[exec->index] == 0) {
4580 exec->index -= len;
4581 break;
4582 }
4583 if (exec->transcount >= atom->min) {
4584 int transno = exec->transno;
4585 xmlRegStatePtr state = exec->state;
4586
4587 /*
4588 * The transition is acceptable save it
4589 */
4590 exec->transno = -1; /* trick */
4591 exec->state = to;
4592 xmlFARegExecSave(exec);
4593 exec->transno = transno;
4594 exec->state = state;
4595 }
4596 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
4597 len);
4598 ret = xmlRegCheckCharacter(atom, codepoint);
4599 exec->transcount++;
4600 } while (ret == 1);
4601 if (exec->transcount < atom->min)
4602 ret = 0;
4603
4604 /*
4605 * If the last check failed but one transition was found
4606 * possible, rollback
4607 */
4608 if (ret < 0)
4609 ret = 0;
4610 if (ret == 0) {
4611 goto rollback;
4612 }
4613 }
4614 }
4615 if (ret == 1) {
4616 if (exec->state->nbTrans > exec->transno + 1) {
4617 xmlFARegExecSave(exec);
4618 }
4619 /*
4620 * restart count for expressions like this ((abc){2})*
4621 */
4622 if (trans->count >= 0) {
4623#ifdef DEBUG_REGEXP_EXEC
4624 printf("Reset count %d\n", trans->count);
4625#endif
4626 exec->counts[trans->count] = 0;
4627 }
4628 if (trans->counter >= 0) {
4629#ifdef DEBUG_REGEXP_EXEC
4630 printf("Increasing count %d\n", trans->counter);
4631#endif
4632 exec->counts[trans->counter]++;
4633 }
4634#ifdef DEBUG_REGEXP_EXEC
4635 printf("entering state %d\n", trans->to);
4636#endif
4637 exec->state = exec->comp->states[trans->to];
4638 exec->transno = 0;
4639 if (trans->atom != NULL) {
4640 exec->index += len;
4641 }
4642 goto progress;
4643 } else if (ret < 0) {
4644 exec->status = -4;
4645 break;
4646 }
4647 }
4648 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
4649rollback:
4650 /*
4651 * Failed to find a way out
4652 */
4653 exec->determinist = 0;
4654 xmlFARegExecRollBack(exec);
4655 }
4656progress:
4657 continue;
4658 }
4659}
4660#endif
4661/************************************************************************
4662 * *
4663 * Parser for the Schemas Datatype Regular Expressions *
4664 * http://www.w3.org/TR/2001/REC-xmlschema-2-20010502/#regexs *
4665 * *
4666 ************************************************************************/
4667
4668/**
4669 * xmlFAIsChar:
4670 * @ctxt: a regexp parser context
4671 *
4672 * [10] Char ::= [^.\?*+()|#x5B#x5D]
4673 */
4674static int
4675xmlFAIsChar(xmlRegParserCtxtPtr ctxt) {
4676 int cur;
4677 int len;
4678
4679 cur = CUR_SCHAR(ctxt->cur, len);
4680 if ((cur == '.') || (cur == '\\') || (cur == '?') ||
4681 (cur == '*') || (cur == '+') || (cur == '(') ||
4682 (cur == ')') || (cur == '|') || (cur == 0x5B) ||
4683 (cur == 0x5D) || (cur == 0))
4684 return(-1);
4685 return(cur);
4686}
4687
4688/**
4689 * xmlFAParseCharProp:
4690 * @ctxt: a regexp parser context
4691 *
4692 * [27] charProp ::= IsCategory | IsBlock
4693 * [28] IsCategory ::= Letters | Marks | Numbers | Punctuation |
4694 * Separators | Symbols | Others
4695 * [29] Letters ::= 'L' [ultmo]?
4696 * [30] Marks ::= 'M' [nce]?
4697 * [31] Numbers ::= 'N' [dlo]?
4698 * [32] Punctuation ::= 'P' [cdseifo]?
4699 * [33] Separators ::= 'Z' [slp]?
4700 * [34] Symbols ::= 'S' [mcko]?
4701 * [35] Others ::= 'C' [cfon]?
4702 * [36] IsBlock ::= 'Is' [a-zA-Z0-9#x2D]+
4703 */
4704static void
4705xmlFAParseCharProp(xmlRegParserCtxtPtr ctxt) {
4706 int cur;
4707 xmlRegAtomType type = (xmlRegAtomType) 0;
4708 xmlChar *blockName = NULL;
4709
4710 cur = CUR;
4711 if (cur == 'L') {
4712 NEXT;
4713 cur = CUR;
4714 if (cur == 'u') {
4715 NEXT;
4716 type = XML_REGEXP_LETTER_UPPERCASE;
4717 } else if (cur == 'l') {
4718 NEXT;
4719 type = XML_REGEXP_LETTER_LOWERCASE;
4720 } else if (cur == 't') {
4721 NEXT;
4722 type = XML_REGEXP_LETTER_TITLECASE;
4723 } else if (cur == 'm') {
4724 NEXT;
4725 type = XML_REGEXP_LETTER_MODIFIER;
4726 } else if (cur == 'o') {
4727 NEXT;
4728 type = XML_REGEXP_LETTER_OTHERS;
4729 } else {
4730 type = XML_REGEXP_LETTER;
4731 }
4732 } else if (cur == 'M') {
4733 NEXT;
4734 cur = CUR;
4735 if (cur == 'n') {
4736 NEXT;
4737 /* nonspacing */
4738 type = XML_REGEXP_MARK_NONSPACING;
4739 } else if (cur == 'c') {
4740 NEXT;
4741 /* spacing combining */
4742 type = XML_REGEXP_MARK_SPACECOMBINING;
4743 } else if (cur == 'e') {
4744 NEXT;
4745 /* enclosing */
4746 type = XML_REGEXP_MARK_ENCLOSING;
4747 } else {
4748 /* all marks */
4749 type = XML_REGEXP_MARK;
4750 }
4751 } else if (cur == 'N') {
4752 NEXT;
4753 cur = CUR;
4754 if (cur == 'd') {
4755 NEXT;
4756 /* digital */
4757 type = XML_REGEXP_NUMBER_DECIMAL;
4758 } else if (cur == 'l') {
4759 NEXT;
4760 /* letter */
4761 type = XML_REGEXP_NUMBER_LETTER;
4762 } else if (cur == 'o') {
4763 NEXT;
4764 /* other */
4765 type = XML_REGEXP_NUMBER_OTHERS;
4766 } else {
4767 /* all numbers */
4768 type = XML_REGEXP_NUMBER;
4769 }
4770 } else if (cur == 'P') {
4771 NEXT;
4772 cur = CUR;
4773 if (cur == 'c') {
4774 NEXT;
4775 /* connector */
4776 type = XML_REGEXP_PUNCT_CONNECTOR;
4777 } else if (cur == 'd') {
4778 NEXT;
4779 /* dash */
4780 type = XML_REGEXP_PUNCT_DASH;
4781 } else if (cur == 's') {
4782 NEXT;
4783 /* open */
4784 type = XML_REGEXP_PUNCT_OPEN;
4785 } else if (cur == 'e') {
4786 NEXT;
4787 /* close */
4788 type = XML_REGEXP_PUNCT_CLOSE;
4789 } else if (cur == 'i') {
4790 NEXT;
4791 /* initial quote */
4792 type = XML_REGEXP_PUNCT_INITQUOTE;
4793 } else if (cur == 'f') {
4794 NEXT;
4795 /* final quote */
4796 type = XML_REGEXP_PUNCT_FINQUOTE;
4797 } else if (cur == 'o') {
4798 NEXT;
4799 /* other */
4800 type = XML_REGEXP_PUNCT_OTHERS;
4801 } else {
4802 /* all punctuation */
4803 type = XML_REGEXP_PUNCT;
4804 }
4805 } else if (cur == 'Z') {
4806 NEXT;
4807 cur = CUR;
4808 if (cur == 's') {
4809 NEXT;
4810 /* space */
4811 type = XML_REGEXP_SEPAR_SPACE;
4812 } else if (cur == 'l') {
4813 NEXT;
4814 /* line */
4815 type = XML_REGEXP_SEPAR_LINE;
4816 } else if (cur == 'p') {
4817 NEXT;
4818 /* paragraph */
4819 type = XML_REGEXP_SEPAR_PARA;
4820 } else {
4821 /* all separators */
4822 type = XML_REGEXP_SEPAR;
4823 }
4824 } else if (cur == 'S') {
4825 NEXT;
4826 cur = CUR;
4827 if (cur == 'm') {
4828 NEXT;
4829 type = XML_REGEXP_SYMBOL_MATH;
4830 /* math */
4831 } else if (cur == 'c') {
4832 NEXT;
4833 type = XML_REGEXP_SYMBOL_CURRENCY;
4834 /* currency */
4835 } else if (cur == 'k') {
4836 NEXT;
4837 type = XML_REGEXP_SYMBOL_MODIFIER;
4838 /* modifiers */
4839 } else if (cur == 'o') {
4840 NEXT;
4841 type = XML_REGEXP_SYMBOL_OTHERS;
4842 /* other */
4843 } else {
4844 /* all symbols */
4845 type = XML_REGEXP_SYMBOL;
4846 }
4847 } else if (cur == 'C') {
4848 NEXT;
4849 cur = CUR;
4850 if (cur == 'c') {
4851 NEXT;
4852 /* control */
4853 type = XML_REGEXP_OTHER_CONTROL;
4854 } else if (cur == 'f') {
4855 NEXT;
4856 /* format */
4857 type = XML_REGEXP_OTHER_FORMAT;
4858 } else if (cur == 'o') {
4859 NEXT;
4860 /* private use */
4861 type = XML_REGEXP_OTHER_PRIVATE;
4862 } else if (cur == 'n') {
4863 NEXT;
4864 /* not assigned */
4865 type = XML_REGEXP_OTHER_NA;
4866 } else {
4867 /* all others */
4868 type = XML_REGEXP_OTHER;
4869 }
4870 } else if (cur == 'I') {
4871 const xmlChar *start;
4872 NEXT;
4873 cur = CUR;
4874 if (cur != 's') {
4875 ERROR("IsXXXX expected");
4876 return;
4877 }
4878 NEXT;
4879 start = ctxt->cur;
4880 cur = CUR;
4881 if (((cur >= 'a') && (cur <= 'z')) ||
4882 ((cur >= 'A') && (cur <= 'Z')) ||
4883 ((cur >= '0') && (cur <= '9')) ||
4884 (cur == 0x2D)) {
4885 NEXT;
4886 cur = CUR;
4887 while (((cur >= 'a') && (cur <= 'z')) ||
4888 ((cur >= 'A') && (cur <= 'Z')) ||
4889 ((cur >= '0') && (cur <= '9')) ||
4890 (cur == 0x2D)) {
4891 NEXT;
4892 cur = CUR;
4893 }
4894 }
4895 type = XML_REGEXP_BLOCK_NAME;
4896 blockName = xmlStrndup(start, ctxt->cur - start);
4897 } else {
4898 ERROR("Unknown char property");
4899 return;
4900 }
4901 if (ctxt->atom == NULL) {
4902 ctxt->atom = xmlRegNewAtom(ctxt, type);
4903 if (ctxt->atom != NULL)
4904 ctxt->atom->valuep = blockName;
4905 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4906 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4907 type, 0, 0, blockName);
4908 }
4909}
4910
4911/**
4912 * xmlFAParseCharClassEsc:
4913 * @ctxt: a regexp parser context
4914 *
4915 * [23] charClassEsc ::= ( SingleCharEsc | MultiCharEsc | catEsc | complEsc )
4916 * [24] SingleCharEsc ::= '\' [nrt\|.?*+(){}#x2D#x5B#x5D#x5E]
4917 * [25] catEsc ::= '\p{' charProp '}'
4918 * [26] complEsc ::= '\P{' charProp '}'
4919 * [37] MultiCharEsc ::= '.' | ('\' [sSiIcCdDwW])
4920 */
4921static void
4922xmlFAParseCharClassEsc(xmlRegParserCtxtPtr ctxt) {
4923 int cur;
4924
4925 if (CUR == '.') {
4926 if (ctxt->atom == NULL) {
4927 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_ANYCHAR);
4928 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4929 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4930 XML_REGEXP_ANYCHAR, 0, 0, NULL);
4931 }
4932 NEXT;
4933 return;
4934 }
4935 if (CUR != '\\') {
4936 ERROR("Escaped sequence: expecting \\");
4937 return;
4938 }
4939 NEXT;
4940 cur = CUR;
4941 if (cur == 'p') {
4942 NEXT;
4943 if (CUR != '{') {
4944 ERROR("Expecting '{'");
4945 return;
4946 }
4947 NEXT;
4948 xmlFAParseCharProp(ctxt);
4949 if (CUR != '}') {
4950 ERROR("Expecting '}'");
4951 return;
4952 }
4953 NEXT;
4954 } else if (cur == 'P') {
4955 NEXT;
4956 if (CUR != '{') {
4957 ERROR("Expecting '{'");
4958 return;
4959 }
4960 NEXT;
4961 xmlFAParseCharProp(ctxt);
4962 if (ctxt->atom != NULL)
4963 ctxt->atom->neg = 1;
4964 if (CUR != '}') {
4965 ERROR("Expecting '}'");
4966 return;
4967 }
4968 NEXT;
4969 } else if ((cur == 'n') || (cur == 'r') || (cur == 't') || (cur == '\\') ||
4970 (cur == '|') || (cur == '.') || (cur == '?') || (cur == '*') ||
4971 (cur == '+') || (cur == '(') || (cur == ')') || (cur == '{') ||
4972 (cur == '}') || (cur == 0x2D) || (cur == 0x5B) || (cur == 0x5D) ||
4973 (cur == 0x5E)) {
4974 if (ctxt->atom == NULL) {
4975 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
4976 if (ctxt->atom != NULL) {
4977 switch (cur) {
4978 case 'n':
4979 ctxt->atom->codepoint = '\n';
4980 break;
4981 case 'r':
4982 ctxt->atom->codepoint = '\r';
4983 break;
4984 case 't':
4985 ctxt->atom->codepoint = '\t';
4986 break;
4987 default:
4988 ctxt->atom->codepoint = cur;
4989 }
4990 }
4991 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4992 switch (cur) {
4993 case 'n':
4994 cur = '\n';
4995 break;
4996 case 'r':
4997 cur = '\r';
4998 break;
4999 case 't':
5000 cur = '\t';
5001 break;
5002 }
5003 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5004 XML_REGEXP_CHARVAL, cur, cur, NULL);
5005 }
5006 NEXT;
5007 } else if ((cur == 's') || (cur == 'S') || (cur == 'i') || (cur == 'I') ||
5008 (cur == 'c') || (cur == 'C') || (cur == 'd') || (cur == 'D') ||
5009 (cur == 'w') || (cur == 'W')) {
5010 xmlRegAtomType type = XML_REGEXP_ANYSPACE;
5011
5012 switch (cur) {
5013 case 's':
5014 type = XML_REGEXP_ANYSPACE;
5015 break;
5016 case 'S':
5017 type = XML_REGEXP_NOTSPACE;
5018 break;
5019 case 'i':
5020 type = XML_REGEXP_INITNAME;
5021 break;
5022 case 'I':
5023 type = XML_REGEXP_NOTINITNAME;
5024 break;
5025 case 'c':
5026 type = XML_REGEXP_NAMECHAR;
5027 break;
5028 case 'C':
5029 type = XML_REGEXP_NOTNAMECHAR;
5030 break;
5031 case 'd':
5032 type = XML_REGEXP_DECIMAL;
5033 break;
5034 case 'D':
5035 type = XML_REGEXP_NOTDECIMAL;
5036 break;
5037 case 'w':
5038 type = XML_REGEXP_REALCHAR;
5039 break;
5040 case 'W':
5041 type = XML_REGEXP_NOTREALCHAR;
5042 break;
5043 }
5044 NEXT;
5045 if (ctxt->atom == NULL) {
5046 ctxt->atom = xmlRegNewAtom(ctxt, type);
5047 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
5048 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5049 type, 0, 0, NULL);
5050 }
5051 } else {
5052 ERROR("Wrong escape sequence, misuse of character '\\'");
5053 }
5054}
5055
5056/**
5057 * xmlFAParseCharRange:
5058 * @ctxt: a regexp parser context
5059 *
5060 * [17] charRange ::= seRange | XmlCharRef | XmlCharIncDash
5061 * [18] seRange ::= charOrEsc '-' charOrEsc
5062 * [20] charOrEsc ::= XmlChar | SingleCharEsc
5063 * [21] XmlChar ::= [^\#x2D#x5B#x5D]
5064 * [22] XmlCharIncDash ::= [^\#x5B#x5D]
5065 */
5066static void
5067xmlFAParseCharRange(xmlRegParserCtxtPtr ctxt) {
5068 int cur, len;
5069 int start = -1;
5070 int end = -1;
5071
5072 if (CUR == '\0') {
5073 ERROR("Expecting ']'");
5074 return;
5075 }
5076
5077 cur = CUR;
5078 if (cur == '\\') {
5079 NEXT;
5080 cur = CUR;
5081 switch (cur) {
5082 case 'n': start = 0xA; break;
5083 case 'r': start = 0xD; break;
5084 case 't': start = 0x9; break;
5085 case '\\': case '|': case '.': case '-': case '^': case '?':
5086 case '*': case '+': case '{': case '}': case '(': case ')':
5087 case '[': case ']':
5088 start = cur; break;
5089 default:
5090 ERROR("Invalid escape value");
5091 return;
5092 }
5093 end = start;
5094 len = 1;
5095 } else if ((cur != 0x5B) && (cur != 0x5D)) {
5096 end = start = CUR_SCHAR(ctxt->cur, len);
5097 } else {
5098 ERROR("Expecting a char range");
5099 return;
5100 }
5101 /*
5102 * Since we are "inside" a range, we can assume ctxt->cur is past
5103 * the start of ctxt->string, and PREV should be safe
5104 */
5105 if ((start == '-') && (NXT(1) != ']') && (PREV != '[') && (PREV != '^')) {
5106 NEXTL(len);
5107 return;
5108 }
5109 NEXTL(len);
5110 cur = CUR;
5111 if ((cur != '-') || (NXT(1) == '[') || (NXT(1) == ']')) {
5112 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5113 XML_REGEXP_CHARVAL, start, end, NULL);
5114 return;
5115 }
5116 NEXT;
5117 cur = CUR;
5118 if (cur == '\\') {
5119 NEXT;
5120 cur = CUR;
5121 switch (cur) {
5122 case 'n': end = 0xA; break;
5123 case 'r': end = 0xD; break;
5124 case 't': end = 0x9; break;
5125 case '\\': case '|': case '.': case '-': case '^': case '?':
5126 case '*': case '+': case '{': case '}': case '(': case ')':
5127 case '[': case ']':
5128 end = cur; break;
5129 default:
5130 ERROR("Invalid escape value");
5131 return;
5132 }
5133 len = 1;
5134 } else if ((cur != '\0') && (cur != 0x5B) && (cur != 0x5D)) {
5135 end = CUR_SCHAR(ctxt->cur, len);
5136 } else {
5137 ERROR("Expecting the end of a char range");
5138 return;
5139 }
5140
5141 /* TODO check that the values are acceptable character ranges for XML */
5142 if (end < start) {
5143 ERROR("End of range is before start of range");
5144 } else {
5145 NEXTL(len);
5146 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5147 XML_REGEXP_CHARVAL, start, end, NULL);
5148 }
5149 return;
5150}
5151
5152/**
5153 * xmlFAParsePosCharGroup:
5154 * @ctxt: a regexp parser context
5155 *
5156 * [14] posCharGroup ::= ( charRange | charClassEsc )+
5157 */
5158static void
5159xmlFAParsePosCharGroup(xmlRegParserCtxtPtr ctxt) {
5160 do {
5161 if (CUR == '\\') {
5162 xmlFAParseCharClassEsc(ctxt);
5163 } else {
5164 xmlFAParseCharRange(ctxt);
5165 }
5166 } while ((CUR != ']') && (CUR != '-') &&
5167 (CUR != 0) && (ctxt->error == 0));
5168}
5169
5170/**
5171 * xmlFAParseCharGroup:
5172 * @ctxt: a regexp parser context
5173 *
5174 * [13] charGroup ::= posCharGroup | negCharGroup | charClassSub
5175 * [15] negCharGroup ::= '^' posCharGroup
5176 * [16] charClassSub ::= ( posCharGroup | negCharGroup ) '-' charClassExpr
5177 * [12] charClassExpr ::= '[' charGroup ']'
5178 */
5179static void
5180xmlFAParseCharGroup(xmlRegParserCtxtPtr ctxt) {
5181 int neg = ctxt->neg;
5182
5183 if (CUR == '^') {
5184 NEXT;
5185 ctxt->neg = !ctxt->neg;
5186 xmlFAParsePosCharGroup(ctxt);
5187 ctxt->neg = neg;
5188 }
5189 while ((CUR != ']') && (ctxt->error == 0)) {
5190 if ((CUR == '-') && (NXT(1) == '[')) {
5191 NEXT; /* eat the '-' */
5192 NEXT; /* eat the '[' */
5193 ctxt->neg = 2;
5194 xmlFAParseCharGroup(ctxt);
5195 ctxt->neg = neg;
5196 if (CUR == ']') {
5197 NEXT;
5198 } else {
5199 ERROR("charClassExpr: ']' expected");
5200 }
5201 break;
5202 } else {
5203 xmlFAParsePosCharGroup(ctxt);
5204 }
5205 }
5206}
5207
5208/**
5209 * xmlFAParseCharClass:
5210 * @ctxt: a regexp parser context
5211 *
5212 * [11] charClass ::= charClassEsc | charClassExpr
5213 * [12] charClassExpr ::= '[' charGroup ']'
5214 */
5215static void
5216xmlFAParseCharClass(xmlRegParserCtxtPtr ctxt) {
5217 if (CUR == '[') {
5218 NEXT;
5219 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_RANGES);
5220 if (ctxt->atom == NULL)
5221 return;
5222 xmlFAParseCharGroup(ctxt);
5223 if (CUR == ']') {
5224 NEXT;
5225 } else {
5226 ERROR("xmlFAParseCharClass: ']' expected");
5227 }
5228 } else {
5229 xmlFAParseCharClassEsc(ctxt);
5230 }
5231}
5232
5233/**
5234 * xmlFAParseQuantExact:
5235 * @ctxt: a regexp parser context
5236 *
5237 * [8] QuantExact ::= [0-9]+
5238 *
5239 * Returns 0 if success or -1 in case of error
5240 */
5241static int
5242xmlFAParseQuantExact(xmlRegParserCtxtPtr ctxt) {
5243 int ret = 0;
5244 int ok = 0;
5245 int overflow = 0;
5246
5247 while ((CUR >= '0') && (CUR <= '9')) {
5248 if (ret > INT_MAX / 10) {
5249 overflow = 1;
5250 } else {
5251 int digit = CUR - '0';
5252
5253 ret *= 10;
5254 if (ret > INT_MAX - digit)
5255 overflow = 1;
5256 else
5257 ret += digit;
5258 }
5259 ok = 1;
5260 NEXT;
5261 }
5262 if ((ok != 1) || (overflow == 1)) {
5263 return(-1);
5264 }
5265 return(ret);
5266}
5267
5268/**
5269 * xmlFAParseQuantifier:
5270 * @ctxt: a regexp parser context
5271 *
5272 * [4] quantifier ::= [?*+] | ( '{' quantity '}' )
5273 * [5] quantity ::= quantRange | quantMin | QuantExact
5274 * [6] quantRange ::= QuantExact ',' QuantExact
5275 * [7] quantMin ::= QuantExact ','
5276 * [8] QuantExact ::= [0-9]+
5277 */
5278static int
5279xmlFAParseQuantifier(xmlRegParserCtxtPtr ctxt) {
5280 int cur;
5281
5282 cur = CUR;
5283 if ((cur == '?') || (cur == '*') || (cur == '+')) {
5284 if (ctxt->atom != NULL) {
5285 if (cur == '?')
5286 ctxt->atom->quant = XML_REGEXP_QUANT_OPT;
5287 else if (cur == '*')
5288 ctxt->atom->quant = XML_REGEXP_QUANT_MULT;
5289 else if (cur == '+')
5290 ctxt->atom->quant = XML_REGEXP_QUANT_PLUS;
5291 }
5292 NEXT;
5293 return(1);
5294 }
5295 if (cur == '{') {
5296 int min = 0, max = 0;
5297
5298 NEXT;
5299 cur = xmlFAParseQuantExact(ctxt);
5300 if (cur >= 0)
5301 min = cur;
5302 else {
5303 ERROR("Improper quantifier");
5304 }
5305 if (CUR == ',') {
5306 NEXT;
5307 if (CUR == '}')
5308 max = INT_MAX;
5309 else {
5310 cur = xmlFAParseQuantExact(ctxt);
5311 if (cur >= 0)
5312 max = cur;
5313 else {
5314 ERROR("Improper quantifier");
5315 }
5316 }
5317 }
5318 if (CUR == '}') {
5319 NEXT;
5320 } else {
5321 ERROR("Unterminated quantifier");
5322 }
5323 if (max == 0)
5324 max = min;
5325 if (ctxt->atom != NULL) {
5326 ctxt->atom->quant = XML_REGEXP_QUANT_RANGE;
5327 ctxt->atom->min = min;
5328 ctxt->atom->max = max;
5329 }
5330 return(1);
5331 }
5332 return(0);
5333}
5334
5335/**
5336 * xmlFAParseAtom:
5337 * @ctxt: a regexp parser context
5338 *
5339 * [9] atom ::= Char | charClass | ( '(' regExp ')' )
5340 */
5341static int
5342xmlFAParseAtom(xmlRegParserCtxtPtr ctxt) {
5343 int codepoint, len;
5344
5345 codepoint = xmlFAIsChar(ctxt);
5346 if (codepoint > 0) {
5347 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
5348 if (ctxt->atom == NULL)
5349 return(-1);
5350 codepoint = CUR_SCHAR(ctxt->cur, len);
5351 ctxt->atom->codepoint = codepoint;
5352 NEXTL(len);
5353 return(1);
5354 } else if (CUR == '|') {
5355 return(0);
5356 } else if (CUR == 0) {
5357 return(0);
5358 } else if (CUR == ')') {
5359 return(0);
5360 } else if (CUR == '(') {
5361 xmlRegStatePtr start, oldend, start0;
5362
5363 NEXT;
5364 if (ctxt->depth >= 50) {
5365 ERROR("xmlFAParseAtom: maximum nesting depth exceeded");
5366 return(-1);
5367 }
5368 /*
5369 * this extra Epsilon transition is needed if we count with 0 allowed
5370 * unfortunately this can't be known at that point
5371 */
5372 xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, NULL);
5373 start0 = ctxt->state;
5374 xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, NULL);
5375 start = ctxt->state;
5376 oldend = ctxt->end;
5377 ctxt->end = NULL;
5378 ctxt->atom = NULL;
5379 ctxt->depth++;
5380 xmlFAParseRegExp(ctxt, 0);
5381 ctxt->depth--;
5382 if (CUR == ')') {
5383 NEXT;
5384 } else {
5385 ERROR("xmlFAParseAtom: expecting ')'");
5386 }
5387 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_SUBREG);
5388 if (ctxt->atom == NULL)
5389 return(-1);
5390 ctxt->atom->start = start;
5391 ctxt->atom->start0 = start0;
5392 ctxt->atom->stop = ctxt->state;
5393 ctxt->end = oldend;
5394 return(1);
5395 } else if ((CUR == '[') || (CUR == '\\') || (CUR == '.')) {
5396 xmlFAParseCharClass(ctxt);
5397 return(1);
5398 }
5399 return(0);
5400}
5401
5402/**
5403 * xmlFAParsePiece:
5404 * @ctxt: a regexp parser context
5405 *
5406 * [3] piece ::= atom quantifier?
5407 */
5408static int
5409xmlFAParsePiece(xmlRegParserCtxtPtr ctxt) {
5410 int ret;
5411
5412 ctxt->atom = NULL;
5413 ret = xmlFAParseAtom(ctxt);
5414 if (ret == 0)
5415 return(0);
5416 if (ctxt->atom == NULL) {
5417 ERROR("internal: no atom generated");
5418 }
5419 xmlFAParseQuantifier(ctxt);
5420 return(1);
5421}
5422
5423/**
5424 * xmlFAParseBranch:
5425 * @ctxt: a regexp parser context
5426 * @to: optional target to the end of the branch
5427 *
5428 * @to is used to optimize by removing duplicate path in automata
5429 * in expressions like (a|b)(c|d)
5430 *
5431 * [2] branch ::= piece*
5432 */
5433static int
5434xmlFAParseBranch(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr to) {
5435 xmlRegStatePtr previous;
5436 int ret;
5437
5438 previous = ctxt->state;
5439 ret = xmlFAParsePiece(ctxt);
5440 if (ret == 0) {
5441 /* Empty branch */
5442 xmlFAGenerateEpsilonTransition(ctxt, previous, to);
5443 } else {
5444 if (xmlFAGenerateTransitions(ctxt, previous,
5445 (CUR=='|' || CUR==')' || CUR==0) ? to : NULL, ctxt->atom) < 0)
5446 return(-1);
5447 previous = ctxt->state;
5448 ctxt->atom = NULL;
5449 }
5450 while ((ret != 0) && (ctxt->error == 0)) {
5451 ret = xmlFAParsePiece(ctxt);
5452 if (ret != 0) {
5453 if (xmlFAGenerateTransitions(ctxt, previous,
5454 (CUR=='|' || CUR==')' || CUR==0) ? to : NULL,
5455 ctxt->atom) < 0)
5456 return(-1);
5457 previous = ctxt->state;
5458 ctxt->atom = NULL;
5459 }
5460 }
5461 return(0);
5462}
5463
5464/**
5465 * xmlFAParseRegExp:
5466 * @ctxt: a regexp parser context
5467 * @top: is this the top-level expression ?
5468 *
5469 * [1] regExp ::= branch ( '|' branch )*
5470 */
5471static void
5472xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top) {
5473 xmlRegStatePtr start, end;
5474
5475 /* if not top start should have been generated by an epsilon trans */
5476 start = ctxt->state;
5477 ctxt->end = NULL;
5478 xmlFAParseBranch(ctxt, NULL);
5479 if (top) {
5480#ifdef DEBUG_REGEXP_GRAPH
5481 printf("State %d is final\n", ctxt->state->no);
5482#endif
5483 ctxt->state->type = XML_REGEXP_FINAL_STATE;
5484 }
5485 if (CUR != '|') {
5486 ctxt->end = ctxt->state;
5487 return;
5488 }
5489 end = ctxt->state;
5490 while ((CUR == '|') && (ctxt->error == 0)) {
5491 NEXT;
5492 ctxt->state = start;
5493 ctxt->end = NULL;
5494 xmlFAParseBranch(ctxt, end);
5495 }
5496 if (!top) {
5497 ctxt->state = end;
5498 ctxt->end = end;
5499 }
5500}
5501
5502/************************************************************************
5503 * *
5504 * The basic API *
5505 * *
5506 ************************************************************************/
5507
5508/**
5509 * xmlRegexpPrint:
5510 * @output: the file for the output debug
5511 * @regexp: the compiled regexp
5512 *
5513 * Print the content of the compiled regular expression
5514 */
5515void
5516xmlRegexpPrint(FILE *output, xmlRegexpPtr regexp) {
5517 int i;
5518
5519 if (output == NULL)
5520 return;
5521 fprintf(output, " regexp: ");
5522 if (regexp == NULL) {
5523 fprintf(output, "NULL\n");
5524 return;
5525 }
5526 fprintf(output, "'%s' ", regexp->string);
5527 fprintf(output, "\n");
5528 fprintf(output, "%d atoms:\n", regexp->nbAtoms);
5529 for (i = 0;i < regexp->nbAtoms; i++) {
5530 fprintf(output, " %02d ", i);
5531 xmlRegPrintAtom(output, regexp->atoms[i]);
5532 }
5533 fprintf(output, "%d states:", regexp->nbStates);
5534 fprintf(output, "\n");
5535 for (i = 0;i < regexp->nbStates; i++) {
5536 xmlRegPrintState(output, regexp->states[i]);
5537 }
5538 fprintf(output, "%d counters:\n", regexp->nbCounters);
5539 for (i = 0;i < regexp->nbCounters; i++) {
5540 fprintf(output, " %d: min %d max %d\n", i, regexp->counters[i].min,
5541 regexp->counters[i].max);
5542 }
5543}
5544
5545/**
5546 * xmlRegexpCompile:
5547 * @regexp: a regular expression string
5548 *
5549 * Parses a regular expression conforming to XML Schemas Part 2 Datatype
5550 * Appendix F and builds an automata suitable for testing strings against
5551 * that regular expression
5552 *
5553 * Returns the compiled expression or NULL in case of error
5554 */
5555xmlRegexpPtr
5556xmlRegexpCompile(const xmlChar *regexp) {
5557 xmlRegexpPtr ret;
5558 xmlRegParserCtxtPtr ctxt;
5559
5560 ctxt = xmlRegNewParserCtxt(regexp);
5561 if (ctxt == NULL)
5562 return(NULL);
5563
5564 /* initialize the parser */
5565 ctxt->end = NULL;
5566 ctxt->start = ctxt->state = xmlRegNewState(ctxt);
5567 xmlRegStatePush(ctxt, ctxt->start);
5568
5569 /* parse the expression building an automata */
5570 xmlFAParseRegExp(ctxt, 1);
5571 if (CUR != 0) {
5572 ERROR("xmlFAParseRegExp: extra characters");
5573 }
5574 if (ctxt->error != 0) {
5575 xmlRegFreeParserCtxt(ctxt);
5576 return(NULL);
5577 }
5578 ctxt->end = ctxt->state;
5579 ctxt->start->type = XML_REGEXP_START_STATE;
5580 ctxt->end->type = XML_REGEXP_FINAL_STATE;
5581
5582 /* remove the Epsilon except for counted transitions */
5583 xmlFAEliminateEpsilonTransitions(ctxt);
5584
5585
5586 if (ctxt->error != 0) {
5587 xmlRegFreeParserCtxt(ctxt);
5588 return(NULL);
5589 }
5590 ret = xmlRegEpxFromParse(ctxt);
5591 xmlRegFreeParserCtxt(ctxt);
5592 return(ret);
5593}
5594
5595/**
5596 * xmlRegexpExec:
5597 * @comp: the compiled regular expression
5598 * @content: the value to check against the regular expression
5599 *
5600 * Check if the regular expression generates the value
5601 *
5602 * Returns 1 if it matches, 0 if not and a negative value in case of error
5603 */
5604int
5605xmlRegexpExec(xmlRegexpPtr comp, const xmlChar *content) {
5606 if ((comp == NULL) || (content == NULL))
5607 return(-1);
5608 return(xmlFARegExec(comp, content));
5609}
5610
5611/**
5612 * xmlRegexpIsDeterminist:
5613 * @comp: the compiled regular expression
5614 *
5615 * Check if the regular expression is determinist
5616 *
5617 * Returns 1 if it yes, 0 if not and a negative value in case of error
5618 */
5619int
5620xmlRegexpIsDeterminist(xmlRegexpPtr comp) {
5621 xmlAutomataPtr am;
5622 int ret;
5623
5624 if (comp == NULL)
5625 return(-1);
5626 if (comp->determinist != -1)
5627 return(comp->determinist);
5628
5629 am = xmlNewAutomata();
5630 if (am == NULL)
5631 return(-1);
5632 if (am->states != NULL) {
5633 int i;
5634
5635 for (i = 0;i < am->nbStates;i++)
5636 xmlRegFreeState(am->states[i]);
5637 xmlFree(am->states);
5638 }
5639 am->nbAtoms = comp->nbAtoms;
5640 am->atoms = comp->atoms;
5641 am->nbStates = comp->nbStates;
5642 am->states = comp->states;
5643 am->determinist = -1;
5644 am->flags = comp->flags;
5645 ret = xmlFAComputesDeterminism(am);
5646 am->atoms = NULL;
5647 am->states = NULL;
5648 xmlFreeAutomata(am);
5649 comp->determinist = ret;
5650 return(ret);
5651}
5652
5653/**
5654 * xmlRegFreeRegexp:
5655 * @regexp: the regexp
5656 *
5657 * Free a regexp
5658 */
5659void
5660xmlRegFreeRegexp(xmlRegexpPtr regexp) {
5661 int i;
5662 if (regexp == NULL)
5663 return;
5664
5665 if (regexp->string != NULL)
5666 xmlFree(regexp->string);
5667 if (regexp->states != NULL) {
5668 for (i = 0;i < regexp->nbStates;i++)
5669 xmlRegFreeState(regexp->states[i]);
5670 xmlFree(regexp->states);
5671 }
5672 if (regexp->atoms != NULL) {
5673 for (i = 0;i < regexp->nbAtoms;i++)
5674 xmlRegFreeAtom(regexp->atoms[i]);
5675 xmlFree(regexp->atoms);
5676 }
5677 if (regexp->counters != NULL)
5678 xmlFree(regexp->counters);
5679 if (regexp->compact != NULL)
5680 xmlFree(regexp->compact);
5681 if (regexp->transdata != NULL)
5682 xmlFree(regexp->transdata);
5683 if (regexp->stringMap != NULL) {
5684 for (i = 0; i < regexp->nbstrings;i++)
5685 xmlFree(regexp->stringMap[i]);
5686 xmlFree(regexp->stringMap);
5687 }
5688
5689 xmlFree(regexp);
5690}
5691
5692#ifdef LIBXML_AUTOMATA_ENABLED
5693/************************************************************************
5694 * *
5695 * The Automata interface *
5696 * *
5697 ************************************************************************/
5698
5699/**
5700 * xmlNewAutomata:
5701 *
5702 * Create a new automata
5703 *
5704 * Returns the new object or NULL in case of failure
5705 */
5706xmlAutomataPtr
5707xmlNewAutomata(void) {
5708 xmlAutomataPtr ctxt;
5709
5710 ctxt = xmlRegNewParserCtxt(NULL);
5711 if (ctxt == NULL)
5712 return(NULL);
5713
5714 /* initialize the parser */
5715 ctxt->end = NULL;
5716 ctxt->start = ctxt->state = xmlRegNewState(ctxt);
5717 if (ctxt->start == NULL) {
5718 xmlFreeAutomata(ctxt);
5719 return(NULL);
5720 }
5721 ctxt->start->type = XML_REGEXP_START_STATE;
5722 if (xmlRegStatePush(ctxt, ctxt->start) < 0) {
5723 xmlRegFreeState(ctxt->start);
5724 xmlFreeAutomata(ctxt);
5725 return(NULL);
5726 }
5727 ctxt->flags = 0;
5728
5729 return(ctxt);
5730}
5731
5732/**
5733 * xmlFreeAutomata:
5734 * @am: an automata
5735 *
5736 * Free an automata
5737 */
5738void
5739xmlFreeAutomata(xmlAutomataPtr am) {
5740 if (am == NULL)
5741 return;
5742 xmlRegFreeParserCtxt(am);
5743}
5744
5745/**
5746 * xmlAutomataSetFlags:
5747 * @am: an automata
5748 * @flags: a set of internal flags
5749 *
5750 * Set some flags on the automata
5751 */
5752void
5753xmlAutomataSetFlags(xmlAutomataPtr am, int flags) {
5754 if (am == NULL)
5755 return;
5756 am->flags |= flags;
5757}
5758
5759/**
5760 * xmlAutomataGetInitState:
5761 * @am: an automata
5762 *
5763 * Initial state lookup
5764 *
5765 * Returns the initial state of the automata
5766 */
5767xmlAutomataStatePtr
5768xmlAutomataGetInitState(xmlAutomataPtr am) {
5769 if (am == NULL)
5770 return(NULL);
5771 return(am->start);
5772}
5773
5774/**
5775 * xmlAutomataSetFinalState:
5776 * @am: an automata
5777 * @state: a state in this automata
5778 *
5779 * Makes that state a final state
5780 *
5781 * Returns 0 or -1 in case of error
5782 */
5783int
5784xmlAutomataSetFinalState(xmlAutomataPtr am, xmlAutomataStatePtr state) {
5785 if ((am == NULL) || (state == NULL))
5786 return(-1);
5787 state->type = XML_REGEXP_FINAL_STATE;
5788 return(0);
5789}
5790
5791/**
5792 * xmlAutomataNewTransition:
5793 * @am: an automata
5794 * @from: the starting point of the transition
5795 * @to: the target point of the transition or NULL
5796 * @token: the input string associated to that transition
5797 * @data: data passed to the callback function if the transition is activated
5798 *
5799 * If @to is NULL, this creates first a new target state in the automata
5800 * and then adds a transition from the @from state to the target state
5801 * activated by the value of @token
5802 *
5803 * Returns the target state or NULL in case of error
5804 */
5805xmlAutomataStatePtr
5806xmlAutomataNewTransition(xmlAutomataPtr am, xmlAutomataStatePtr from,
5807 xmlAutomataStatePtr to, const xmlChar *token,
5808 void *data) {
5809 xmlRegAtomPtr atom;
5810
5811 if ((am == NULL) || (from == NULL) || (token == NULL))
5812 return(NULL);
5813 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5814 if (atom == NULL)
5815 return(NULL);
5816 atom->data = data;
5817 atom->valuep = xmlStrdup(token);
5818
5819 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5820 xmlRegFreeAtom(atom);
5821 return(NULL);
5822 }
5823 if (to == NULL)
5824 return(am->state);
5825 return(to);
5826}
5827
5828/**
5829 * xmlAutomataNewTransition2:
5830 * @am: an automata
5831 * @from: the starting point of the transition
5832 * @to: the target point of the transition or NULL
5833 * @token: the first input string associated to that transition
5834 * @token2: the second input string associated to that transition
5835 * @data: data passed to the callback function if the transition is activated
5836 *
5837 * If @to is NULL, this creates first a new target state in the automata
5838 * and then adds a transition from the @from state to the target state
5839 * activated by the value of @token
5840 *
5841 * Returns the target state or NULL in case of error
5842 */
5843xmlAutomataStatePtr
5844xmlAutomataNewTransition2(xmlAutomataPtr am, xmlAutomataStatePtr from,
5845 xmlAutomataStatePtr to, const xmlChar *token,
5846 const xmlChar *token2, void *data) {
5847 xmlRegAtomPtr atom;
5848
5849 if ((am == NULL) || (from == NULL) || (token == NULL))
5850 return(NULL);
5851 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5852 if (atom == NULL)
5853 return(NULL);
5854 atom->data = data;
5855 if ((token2 == NULL) || (*token2 == 0)) {
5856 atom->valuep = xmlStrdup(token);
5857 } else {
5858 int lenn, lenp;
5859 xmlChar *str;
5860
5861 lenn = strlen((char *) token2);
5862 lenp = strlen((char *) token);
5863
5864 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5865 if (str == NULL) {
5866 xmlRegFreeAtom(atom);
5867 return(NULL);
5868 }
5869 memcpy(&str[0], token, lenp);
5870 str[lenp] = '|';
5871 memcpy(&str[lenp + 1], token2, lenn);
5872 str[lenn + lenp + 1] = 0;
5873
5874 atom->valuep = str;
5875 }
5876
5877 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5878 xmlRegFreeAtom(atom);
5879 return(NULL);
5880 }
5881 if (to == NULL)
5882 return(am->state);
5883 return(to);
5884}
5885
5886/**
5887 * xmlAutomataNewNegTrans:
5888 * @am: an automata
5889 * @from: the starting point of the transition
5890 * @to: the target point of the transition or NULL
5891 * @token: the first input string associated to that transition
5892 * @token2: the second input string associated to that transition
5893 * @data: data passed to the callback function if the transition is activated
5894 *
5895 * If @to is NULL, this creates first a new target state in the automata
5896 * and then adds a transition from the @from state to the target state
5897 * activated by any value except (@token,@token2)
5898 * Note that if @token2 is not NULL, then (X, NULL) won't match to follow
5899 # the semantic of XSD ##other
5900 *
5901 * Returns the target state or NULL in case of error
5902 */
5903xmlAutomataStatePtr
5904xmlAutomataNewNegTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
5905 xmlAutomataStatePtr to, const xmlChar *token,
5906 const xmlChar *token2, void *data) {
5907 xmlRegAtomPtr atom;
5908 xmlChar err_msg[200];
5909
5910 if ((am == NULL) || (from == NULL) || (token == NULL))
5911 return(NULL);
5912 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5913 if (atom == NULL)
5914 return(NULL);
5915 atom->data = data;
5916 atom->neg = 1;
5917 if ((token2 == NULL) || (*token2 == 0)) {
5918 atom->valuep = xmlStrdup(token);
5919 } else {
5920 int lenn, lenp;
5921 xmlChar *str;
5922
5923 lenn = strlen((char *) token2);
5924 lenp = strlen((char *) token);
5925
5926 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5927 if (str == NULL) {
5928 xmlRegFreeAtom(atom);
5929 return(NULL);
5930 }
5931 memcpy(&str[0], token, lenp);
5932 str[lenp] = '|';
5933 memcpy(&str[lenp + 1], token2, lenn);
5934 str[lenn + lenp + 1] = 0;
5935
5936 atom->valuep = str;
5937 }
5938 snprintf((char *) err_msg, 199, "not %s", (const char *) atom->valuep);
5939 err_msg[199] = 0;
5940 atom->valuep2 = xmlStrdup(err_msg);
5941
5942 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5943 xmlRegFreeAtom(atom);
5944 return(NULL);
5945 }
5946 am->negs++;
5947 if (to == NULL)
5948 return(am->state);
5949 return(to);
5950}
5951
5952/**
5953 * xmlAutomataNewCountTrans2:
5954 * @am: an automata
5955 * @from: the starting point of the transition
5956 * @to: the target point of the transition or NULL
5957 * @token: the input string associated to that transition
5958 * @token2: the second input string associated to that transition
5959 * @min: the minimum successive occurrences of token
5960 * @max: the maximum successive occurrences of token
5961 * @data: data associated to the transition
5962 *
5963 * If @to is NULL, this creates first a new target state in the automata
5964 * and then adds a transition from the @from state to the target state
5965 * activated by a succession of input of value @token and @token2 and
5966 * whose number is between @min and @max
5967 *
5968 * Returns the target state or NULL in case of error
5969 */
5970xmlAutomataStatePtr
5971xmlAutomataNewCountTrans2(xmlAutomataPtr am, xmlAutomataStatePtr from,
5972 xmlAutomataStatePtr to, const xmlChar *token,
5973 const xmlChar *token2,
5974 int min, int max, void *data) {
5975 xmlRegAtomPtr atom;
5976 int counter;
5977
5978 if ((am == NULL) || (from == NULL) || (token == NULL))
5979 return(NULL);
5980 if (min < 0)
5981 return(NULL);
5982 if ((max < min) || (max < 1))
5983 return(NULL);
5984 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5985 if (atom == NULL)
5986 return(NULL);
5987 if ((token2 == NULL) || (*token2 == 0)) {
5988 atom->valuep = xmlStrdup(token);
5989 } else {
5990 int lenn, lenp;
5991 xmlChar *str;
5992
5993 lenn = strlen((char *) token2);
5994 lenp = strlen((char *) token);
5995
5996 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5997 if (str == NULL) {
5998 xmlRegFreeAtom(atom);
5999 return(NULL);
6000 }
6001 memcpy(&str[0], token, lenp);
6002 str[lenp] = '|';
6003 memcpy(&str[lenp + 1], token2, lenn);
6004 str[lenn + lenp + 1] = 0;
6005
6006 atom->valuep = str;
6007 }
6008 atom->data = data;
6009 if (min == 0)
6010 atom->min = 1;
6011 else
6012 atom->min = min;
6013 atom->max = max;
6014
6015 /*
6016 * associate a counter to the transition.
6017 */
6018 counter = xmlRegGetCounter(am);
6019 am->counters[counter].min = min;
6020 am->counters[counter].max = max;
6021
6022 /* xmlFAGenerateTransitions(am, from, to, atom); */
6023 if (to == NULL) {
6024 to = xmlRegNewState(am);
6025 xmlRegStatePush(am, to);
6026 }
6027 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6028 xmlRegAtomPush(am, atom);
6029 am->state = to;
6030
6031 if (to == NULL)
6032 to = am->state;
6033 if (to == NULL)
6034 return(NULL);
6035 if (min == 0)
6036 xmlFAGenerateEpsilonTransition(am, from, to);
6037 return(to);
6038}
6039
6040/**
6041 * xmlAutomataNewCountTrans:
6042 * @am: an automata
6043 * @from: the starting point of the transition
6044 * @to: the target point of the transition or NULL
6045 * @token: the input string associated to that transition
6046 * @min: the minimum successive occurrences of token
6047 * @max: the maximum successive occurrences of token
6048 * @data: data associated to the transition
6049 *
6050 * If @to is NULL, this creates first a new target state in the automata
6051 * and then adds a transition from the @from state to the target state
6052 * activated by a succession of input of value @token and whose number
6053 * is between @min and @max
6054 *
6055 * Returns the target state or NULL in case of error
6056 */
6057xmlAutomataStatePtr
6058xmlAutomataNewCountTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6059 xmlAutomataStatePtr to, const xmlChar *token,
6060 int min, int max, void *data) {
6061 xmlRegAtomPtr atom;
6062 int counter;
6063
6064 if ((am == NULL) || (from == NULL) || (token == NULL))
6065 return(NULL);
6066 if (min < 0)
6067 return(NULL);
6068 if ((max < min) || (max < 1))
6069 return(NULL);
6070 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6071 if (atom == NULL)
6072 return(NULL);
6073 atom->valuep = xmlStrdup(token);
6074 atom->data = data;
6075 if (min == 0)
6076 atom->min = 1;
6077 else
6078 atom->min = min;
6079 atom->max = max;
6080
6081 /*
6082 * associate a counter to the transition.
6083 */
6084 counter = xmlRegGetCounter(am);
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 = xmlRegNewState(am);
6091 xmlRegStatePush(am, to);
6092 }
6093 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6094 xmlRegAtomPush(am, atom);
6095 am->state = to;
6096
6097 if (to == NULL)
6098 to = am->state;
6099 if (to == NULL)
6100 return(NULL);
6101 if (min == 0)
6102 xmlFAGenerateEpsilonTransition(am, from, to);
6103 return(to);
6104}
6105
6106/**
6107 * xmlAutomataNewOnceTrans2:
6108 * @am: an automata
6109 * @from: the starting point of the transition
6110 * @to: the target point of the transition or NULL
6111 * @token: the input string associated to that transition
6112 * @token2: the second input string associated to that transition
6113 * @min: the minimum successive occurrences of token
6114 * @max: the maximum successive occurrences of token
6115 * @data: data associated to the transition
6116 *
6117 * If @to is NULL, this creates first a new target state in the automata
6118 * and then adds a transition from the @from state to the target state
6119 * activated by a succession of input of value @token and @token2 and whose
6120 * number is between @min and @max, moreover that transition can only be
6121 * crossed once.
6122 *
6123 * Returns the target state or NULL in case of error
6124 */
6125xmlAutomataStatePtr
6126xmlAutomataNewOnceTrans2(xmlAutomataPtr am, xmlAutomataStatePtr from,
6127 xmlAutomataStatePtr to, const xmlChar *token,
6128 const xmlChar *token2,
6129 int min, int max, void *data) {
6130 xmlRegAtomPtr atom;
6131 int counter;
6132
6133 if ((am == NULL) || (from == NULL) || (token == NULL))
6134 return(NULL);
6135 if (min < 1)
6136 return(NULL);
6137 if (max < min)
6138 return(NULL);
6139 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6140 if (atom == NULL)
6141 return(NULL);
6142 if ((token2 == NULL) || (*token2 == 0)) {
6143 atom->valuep = xmlStrdup(token);
6144 } else {
6145 int lenn, lenp;
6146 xmlChar *str;
6147
6148 lenn = strlen((char *) token2);
6149 lenp = strlen((char *) token);
6150
6151 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
6152 if (str == NULL) {
6153 xmlRegFreeAtom(atom);
6154 return(NULL);
6155 }
6156 memcpy(&str[0], token, lenp);
6157 str[lenp] = '|';
6158 memcpy(&str[lenp + 1], token2, lenn);
6159 str[lenn + lenp + 1] = 0;
6160
6161 atom->valuep = str;
6162 }
6163 atom->data = data;
6164 atom->quant = XML_REGEXP_QUANT_ONCEONLY;
6165 atom->min = min;
6166 atom->max = max;
6167 /*
6168 * associate a counter to the transition.
6169 */
6170 counter = xmlRegGetCounter(am);
6171 am->counters[counter].min = 1;
6172 am->counters[counter].max = 1;
6173
6174 /* xmlFAGenerateTransitions(am, from, to, atom); */
6175 if (to == NULL) {
6176 to = xmlRegNewState(am);
6177 xmlRegStatePush(am, to);
6178 }
6179 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6180 xmlRegAtomPush(am, atom);
6181 am->state = to;
6182 return(to);
6183}
6184
6185
6186
6187/**
6188 * xmlAutomataNewOnceTrans:
6189 * @am: an automata
6190 * @from: the starting point of the transition
6191 * @to: the target point of the transition or NULL
6192 * @token: the input string associated to that transition
6193 * @min: the minimum successive occurrences of token
6194 * @max: the maximum successive occurrences of token
6195 * @data: data associated to the transition
6196 *
6197 * If @to is NULL, this creates first a new target state in the automata
6198 * and then adds a transition from the @from state to the target state
6199 * activated by a succession of input of value @token and whose number
6200 * is between @min and @max, moreover that transition can only be crossed
6201 * once.
6202 *
6203 * Returns the target state or NULL in case of error
6204 */
6205xmlAutomataStatePtr
6206xmlAutomataNewOnceTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6207 xmlAutomataStatePtr to, const xmlChar *token,
6208 int min, int max, void *data) {
6209 xmlRegAtomPtr atom;
6210 int counter;
6211
6212 if ((am == NULL) || (from == NULL) || (token == NULL))
6213 return(NULL);
6214 if (min < 1)
6215 return(NULL);
6216 if (max < min)
6217 return(NULL);
6218 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6219 if (atom == NULL)
6220 return(NULL);
6221 atom->valuep = xmlStrdup(token);
6222 atom->data = data;
6223 atom->quant = XML_REGEXP_QUANT_ONCEONLY;
6224 atom->min = min;
6225 atom->max = max;
6226 /*
6227 * associate a counter to the transition.
6228 */
6229 counter = xmlRegGetCounter(am);
6230 am->counters[counter].min = 1;
6231 am->counters[counter].max = 1;
6232
6233 /* xmlFAGenerateTransitions(am, from, to, atom); */
6234 if (to == NULL) {
6235 to = xmlRegNewState(am);
6236 xmlRegStatePush(am, to);
6237 }
6238 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6239 xmlRegAtomPush(am, atom);
6240 am->state = to;
6241 return(to);
6242}
6243
6244/**
6245 * xmlAutomataNewState:
6246 * @am: an automata
6247 *
6248 * Create a new disconnected state in the automata
6249 *
6250 * Returns the new state or NULL in case of error
6251 */
6252xmlAutomataStatePtr
6253xmlAutomataNewState(xmlAutomataPtr am) {
6254 xmlAutomataStatePtr to;
6255
6256 if (am == NULL)
6257 return(NULL);
6258 to = xmlRegNewState(am);
6259 xmlRegStatePush(am, to);
6260 return(to);
6261}
6262
6263/**
6264 * xmlAutomataNewEpsilon:
6265 * @am: an automata
6266 * @from: the starting point of the transition
6267 * @to: the target point of the transition or NULL
6268 *
6269 * If @to is NULL, this creates first a new target state in the automata
6270 * and then adds an epsilon transition from the @from state to the
6271 * target state
6272 *
6273 * Returns the target state or NULL in case of error
6274 */
6275xmlAutomataStatePtr
6276xmlAutomataNewEpsilon(xmlAutomataPtr am, xmlAutomataStatePtr from,
6277 xmlAutomataStatePtr to) {
6278 if ((am == NULL) || (from == NULL))
6279 return(NULL);
6280 xmlFAGenerateEpsilonTransition(am, from, to);
6281 if (to == NULL)
6282 return(am->state);
6283 return(to);
6284}
6285
6286/**
6287 * xmlAutomataNewAllTrans:
6288 * @am: an automata
6289 * @from: the starting point of the transition
6290 * @to: the target point of the transition or NULL
6291 * @lax: allow to transition if not all all transitions have been activated
6292 *
6293 * If @to is NULL, this creates first a new target state in the automata
6294 * and then adds a an ALL transition from the @from state to the
6295 * target state. That transition is an epsilon transition allowed only when
6296 * all transitions from the @from node have been activated.
6297 *
6298 * Returns the target state or NULL in case of error
6299 */
6300xmlAutomataStatePtr
6301xmlAutomataNewAllTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6302 xmlAutomataStatePtr to, int lax) {
6303 if ((am == NULL) || (from == NULL))
6304 return(NULL);
6305 xmlFAGenerateAllTransition(am, from, to, lax);
6306 if (to == NULL)
6307 return(am->state);
6308 return(to);
6309}
6310
6311/**
6312 * xmlAutomataNewCounter:
6313 * @am: an automata
6314 * @min: the minimal value on the counter
6315 * @max: the maximal value on the counter
6316 *
6317 * Create a new counter
6318 *
6319 * Returns the counter number or -1 in case of error
6320 */
6321int
6322xmlAutomataNewCounter(xmlAutomataPtr am, int min, int max) {
6323 int ret;
6324
6325 if (am == NULL)
6326 return(-1);
6327
6328 ret = xmlRegGetCounter(am);
6329 if (ret < 0)
6330 return(-1);
6331 am->counters[ret].min = min;
6332 am->counters[ret].max = max;
6333 return(ret);
6334}
6335
6336/**
6337 * xmlAutomataNewCountedTrans:
6338 * @am: an automata
6339 * @from: the starting point of the transition
6340 * @to: the target point of the transition or NULL
6341 * @counter: the counter associated to that transition
6342 *
6343 * If @to is NULL, this creates first a new target state in the automata
6344 * and then adds an epsilon transition from the @from state to the target state
6345 * which will increment the counter provided
6346 *
6347 * Returns the target state or NULL in case of error
6348 */
6349xmlAutomataStatePtr
6350xmlAutomataNewCountedTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6351 xmlAutomataStatePtr to, int counter) {
6352 if ((am == NULL) || (from == NULL) || (counter < 0))
6353 return(NULL);
6354 xmlFAGenerateCountedEpsilonTransition(am, from, to, counter);
6355 if (to == NULL)
6356 return(am->state);
6357 return(to);
6358}
6359
6360/**
6361 * xmlAutomataNewCounterTrans:
6362 * @am: an automata
6363 * @from: the starting point of the transition
6364 * @to: the target point of the transition or NULL
6365 * @counter: the counter associated to that transition
6366 *
6367 * If @to is NULL, this creates first a new target state in the automata
6368 * and then adds an epsilon transition from the @from state to the target state
6369 * which will be allowed only if the counter is within the right range.
6370 *
6371 * Returns the target state or NULL in case of error
6372 */
6373xmlAutomataStatePtr
6374xmlAutomataNewCounterTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6375 xmlAutomataStatePtr to, int counter) {
6376 if ((am == NULL) || (from == NULL) || (counter < 0))
6377 return(NULL);
6378 xmlFAGenerateCountedTransition(am, from, to, counter);
6379 if (to == NULL)
6380 return(am->state);
6381 return(to);
6382}
6383
6384/**
6385 * xmlAutomataCompile:
6386 * @am: an automata
6387 *
6388 * Compile the automata into a Reg Exp ready for being executed.
6389 * The automata should be free after this point.
6390 *
6391 * Returns the compiled regexp or NULL in case of error
6392 */
6393xmlRegexpPtr
6394xmlAutomataCompile(xmlAutomataPtr am) {
6395 xmlRegexpPtr ret;
6396
6397 if ((am == NULL) || (am->error != 0)) return(NULL);
6398 xmlFAEliminateEpsilonTransitions(am);
6399 /* xmlFAComputesDeterminism(am); */
6400 ret = xmlRegEpxFromParse(am);
6401
6402 return(ret);
6403}
6404
6405/**
6406 * xmlAutomataIsDeterminist:
6407 * @am: an automata
6408 *
6409 * Checks if an automata is determinist.
6410 *
6411 * Returns 1 if true, 0 if not, and -1 in case of error
6412 */
6413int
6414xmlAutomataIsDeterminist(xmlAutomataPtr am) {
6415 int ret;
6416
6417 if (am == NULL)
6418 return(-1);
6419
6420 ret = xmlFAComputesDeterminism(am);
6421 return(ret);
6422}
6423#endif /* LIBXML_AUTOMATA_ENABLED */
6424
6425#ifdef LIBXML_EXPR_ENABLED
6426/************************************************************************
6427 * *
6428 * Formal Expression handling code *
6429 * *
6430 ************************************************************************/
6431/************************************************************************
6432 * *
6433 * Expression handling context *
6434 * *
6435 ************************************************************************/
6436
6437struct _xmlExpCtxt {
6438 xmlDictPtr dict;
6439 xmlExpNodePtr *table;
6440 int size;
6441 int nbElems;
6442 int nb_nodes;
6443 int maxNodes;
6444 const char *expr;
6445 const char *cur;
6446 int nb_cons;
6447 int tabSize;
6448};
6449
6450/**
6451 * xmlExpNewCtxt:
6452 * @maxNodes: the maximum number of nodes
6453 * @dict: optional dictionary to use internally
6454 *
6455 * Creates a new context for manipulating expressions
6456 *
6457 * Returns the context or NULL in case of error
6458 */
6459xmlExpCtxtPtr
6460xmlExpNewCtxt(int maxNodes, xmlDictPtr dict) {
6461 xmlExpCtxtPtr ret;
6462 int size = 256;
6463
6464 if (maxNodes <= 4096)
6465 maxNodes = 4096;
6466
6467 ret = (xmlExpCtxtPtr) xmlMalloc(sizeof(xmlExpCtxt));
6468 if (ret == NULL)
6469 return(NULL);
6470 memset(ret, 0, sizeof(xmlExpCtxt));
6471 ret->size = size;
6472 ret->nbElems = 0;
6473 ret->maxNodes = maxNodes;
6474 ret->table = xmlMalloc(size * sizeof(xmlExpNodePtr));
6475 if (ret->table == NULL) {
6476 xmlFree(ret);
6477 return(NULL);
6478 }
6479 memset(ret->table, 0, size * sizeof(xmlExpNodePtr));
6480 if (dict == NULL) {
6481 ret->dict = xmlDictCreate();
6482 if (ret->dict == NULL) {
6483 xmlFree(ret->table);
6484 xmlFree(ret);
6485 return(NULL);
6486 }
6487 } else {
6488 ret->dict = dict;
6489 xmlDictReference(ret->dict);
6490 }
6491 return(ret);
6492}
6493
6494/**
6495 * xmlExpFreeCtxt:
6496 * @ctxt: an expression context
6497 *
6498 * Free an expression context
6499 */
6500void
6501xmlExpFreeCtxt(xmlExpCtxtPtr ctxt) {
6502 if (ctxt == NULL)
6503 return;
6504 xmlDictFree(ctxt->dict);
6505 if (ctxt->table != NULL)
6506 xmlFree(ctxt->table);
6507 xmlFree(ctxt);
6508}
6509
6510/************************************************************************
6511 * *
6512 * Structure associated to an expression node *
6513 * *
6514 ************************************************************************/
6515#define MAX_NODES 10000
6516
6517/* #define DEBUG_DERIV */
6518
6519/*
6520 * TODO:
6521 * - Wildcards
6522 * - public API for creation
6523 *
6524 * Started
6525 * - regression testing
6526 *
6527 * Done
6528 * - split into module and test tool
6529 * - memleaks
6530 */
6531
6532typedef enum {
6533 XML_EXP_NILABLE = (1 << 0)
6534} xmlExpNodeInfo;
6535
6536#define IS_NILLABLE(node) ((node)->info & XML_EXP_NILABLE)
6537
6538struct _xmlExpNode {
6539 unsigned char type;/* xmlExpNodeType */
6540 unsigned char info;/* OR of xmlExpNodeInfo */
6541 unsigned short key; /* the hash key */
6542 unsigned int ref; /* The number of references */
6543 int c_max; /* the maximum length it can consume */
6544 xmlExpNodePtr exp_left;
6545 xmlExpNodePtr next;/* the next node in the hash table or free list */
6546 union {
6547 struct {
6548 int f_min;
6549 int f_max;
6550 } count;
6551 struct {
6552 xmlExpNodePtr f_right;
6553 } children;
6554 const xmlChar *f_str;
6555 } field;
6556};
6557
6558#define exp_min field.count.f_min
6559#define exp_max field.count.f_max
6560/* #define exp_left field.children.f_left */
6561#define exp_right field.children.f_right
6562#define exp_str field.f_str
6563
6564static xmlExpNodePtr xmlExpNewNode(xmlExpCtxtPtr ctxt, xmlExpNodeType type);
6565static xmlExpNode forbiddenExpNode = {
6566 XML_EXP_FORBID, 0, 0, 0, 0, NULL, NULL, {{ 0, 0}}
6567};
6568xmlExpNodePtr forbiddenExp = &forbiddenExpNode;
6569static xmlExpNode emptyExpNode = {
6570 XML_EXP_EMPTY, 1, 0, 0, 0, NULL, NULL, {{ 0, 0}}
6571};
6572xmlExpNodePtr emptyExp = &emptyExpNode;
6573
6574/************************************************************************
6575 * *
6576 * The custom hash table for unicity and canonicalization *
6577 * of sub-expressions pointers *
6578 * *
6579 ************************************************************************/
6580/*
6581 * xmlExpHashNameComputeKey:
6582 * Calculate the hash key for a token
6583 */
6584static unsigned short
6585xmlExpHashNameComputeKey(const xmlChar *name) {
6586 unsigned short value = 0L;
6587 char ch;
6588
6589 if (name != NULL) {
6590 value += 30 * (*name);
6591 while ((ch = *name++) != 0) {
6592 value = value ^ ((value << 5) + (value >> 3) + (unsigned long)ch);
6593 }
6594 }
6595 return (value);
6596}
6597
6598/*
6599 * xmlExpHashComputeKey:
6600 * Calculate the hash key for a compound expression
6601 */
6602static unsigned short
6603xmlExpHashComputeKey(xmlExpNodeType type, xmlExpNodePtr left,
6604 xmlExpNodePtr right) {
6605 unsigned long value;
6606 unsigned short ret;
6607
6608 switch (type) {
6609 case XML_EXP_SEQ:
6610 value = left->key;
6611 value += right->key;
6612 value *= 3;
6613 ret = (unsigned short) value;
6614 break;
6615 case XML_EXP_OR:
6616 value = left->key;
6617 value += right->key;
6618 value *= 7;
6619 ret = (unsigned short) value;
6620 break;
6621 case XML_EXP_COUNT:
6622 value = left->key;
6623 value += right->key;
6624 ret = (unsigned short) value;
6625 break;
6626 default:
6627 ret = 0;
6628 }
6629 return(ret);
6630}
6631
6632
6633static xmlExpNodePtr
6634xmlExpNewNode(xmlExpCtxtPtr ctxt, xmlExpNodeType type) {
6635 xmlExpNodePtr ret;
6636
6637 if (ctxt->nb_nodes >= MAX_NODES)
6638 return(NULL);
6639 ret = (xmlExpNodePtr) xmlMalloc(sizeof(xmlExpNode));
6640 if (ret == NULL)
6641 return(NULL);
6642 memset(ret, 0, sizeof(xmlExpNode));
6643 ret->type = type;
6644 ret->next = NULL;
6645 ctxt->nb_nodes++;
6646 ctxt->nb_cons++;
6647 return(ret);
6648}
6649
6650/**
6651 * xmlExpHashGetEntry:
6652 * @table: the hash table
6653 *
6654 * Get the unique entry from the hash table. The entry is created if
6655 * needed. @left and @right are consumed, i.e. their ref count will
6656 * be decremented by the operation.
6657 *
6658 * Returns the pointer or NULL in case of error
6659 */
6660static xmlExpNodePtr
6661xmlExpHashGetEntry(xmlExpCtxtPtr ctxt, xmlExpNodeType type,
6662 xmlExpNodePtr left, xmlExpNodePtr right,
6663 const xmlChar *name, int min, int max) {
6664 unsigned short kbase, key;
6665 xmlExpNodePtr entry;
6666 xmlExpNodePtr insert;
6667
6668 if (ctxt == NULL)
6669 return(NULL);
6670
6671 /*
6672 * Check for duplicate and insertion location.
6673 */
6674 if (type == XML_EXP_ATOM) {
6675 kbase = xmlExpHashNameComputeKey(name);
6676 } else if (type == XML_EXP_COUNT) {
6677 /* COUNT reduction rule 1 */
6678 /* a{1} -> a */
6679 if (min == max) {
6680 if (min == 1) {
6681 return(left);
6682 }
6683 if (min == 0) {
6684 xmlExpFree(ctxt, left);
6685 return(emptyExp);
6686 }
6687 }
6688 if (min < 0) {
6689 xmlExpFree(ctxt, left);
6690 return(forbiddenExp);
6691 }
6692 if (max == -1)
6693 kbase = min + 79;
6694 else
6695 kbase = max - min;
6696 kbase += left->key;
6697 } else if (type == XML_EXP_OR) {
6698 /* Forbid reduction rules */
6699 if (left->type == XML_EXP_FORBID) {
6700 xmlExpFree(ctxt, left);
6701 return(right);
6702 }
6703 if (right->type == XML_EXP_FORBID) {
6704 xmlExpFree(ctxt, right);
6705 return(left);
6706 }
6707
6708 /* OR reduction rule 1 */
6709 /* a | a reduced to a */
6710 if (left == right) {
6711 xmlExpFree(ctxt, right);
6712 return(left);
6713 }
6714 /* OR canonicalization rule 1 */
6715 /* linearize (a | b) | c into a | (b | c) */
6716 if ((left->type == XML_EXP_OR) && (right->type != XML_EXP_OR)) {
6717 xmlExpNodePtr tmp = left;
6718 left = right;
6719 right = tmp;
6720 }
6721 /* OR reduction rule 2 */
6722 /* a | (a | b) and b | (a | b) are reduced to a | b */
6723 if (right->type == XML_EXP_OR) {
6724 if ((left == right->exp_left) ||
6725 (left == right->exp_right)) {
6726 xmlExpFree(ctxt, left);
6727 return(right);
6728 }
6729 }
6730 /* OR canonicalization rule 2 */
6731 /* linearize (a | b) | c into a | (b | c) */
6732 if (left->type == XML_EXP_OR) {
6733 xmlExpNodePtr tmp;
6734
6735 /* OR canonicalization rule 2 */
6736 if ((left->exp_right->type != XML_EXP_OR) &&
6737 (left->exp_right->key < left->exp_left->key)) {
6738 tmp = left->exp_right;
6739 left->exp_right = left->exp_left;
6740 left->exp_left = tmp;
6741 }
6742 left->exp_right->ref++;
6743 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left->exp_right, right,
6744 NULL, 0, 0);
6745 left->exp_left->ref++;
6746 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left->exp_left, tmp,
6747 NULL, 0, 0);
6748
6749 xmlExpFree(ctxt, left);
6750 return(tmp);
6751 }
6752 if (right->type == XML_EXP_OR) {
6753 /* Ordering in the tree */
6754 /* C | (A | B) -> A | (B | C) */
6755 if (left->key > right->exp_right->key) {
6756 xmlExpNodePtr tmp;
6757 right->exp_right->ref++;
6758 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_right,
6759 left, NULL, 0, 0);
6760 right->exp_left->ref++;
6761 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_left,
6762 tmp, NULL, 0, 0);
6763 xmlExpFree(ctxt, right);
6764 return(tmp);
6765 }
6766 /* Ordering in the tree */
6767 /* B | (A | C) -> A | (B | C) */
6768 if (left->key > right->exp_left->key) {
6769 xmlExpNodePtr tmp;
6770 right->exp_right->ref++;
6771 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left,
6772 right->exp_right, NULL, 0, 0);
6773 right->exp_left->ref++;
6774 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_left,
6775 tmp, NULL, 0, 0);
6776 xmlExpFree(ctxt, right);
6777 return(tmp);
6778 }
6779 }
6780 /* we know both types are != XML_EXP_OR here */
6781 else if (left->key > right->key) {
6782 xmlExpNodePtr tmp = left;
6783 left = right;
6784 right = tmp;
6785 }
6786 kbase = xmlExpHashComputeKey(type, left, right);
6787 } else if (type == XML_EXP_SEQ) {
6788 /* Forbid reduction rules */
6789 if (left->type == XML_EXP_FORBID) {
6790 xmlExpFree(ctxt, right);
6791 return(left);
6792 }
6793 if (right->type == XML_EXP_FORBID) {
6794 xmlExpFree(ctxt, left);
6795 return(right);
6796 }
6797 /* Empty reduction rules */
6798 if (right->type == XML_EXP_EMPTY) {
6799 return(left);
6800 }
6801 if (left->type == XML_EXP_EMPTY) {
6802 return(right);
6803 }
6804 kbase = xmlExpHashComputeKey(type, left, right);
6805 } else
6806 return(NULL);
6807
6808 key = kbase % ctxt->size;
6809 if (ctxt->table[key] != NULL) {
6810 for (insert = ctxt->table[key]; insert != NULL;
6811 insert = insert->next) {
6812 if ((insert->key == kbase) &&
6813 (insert->type == type)) {
6814 if (type == XML_EXP_ATOM) {
6815 if (name == insert->exp_str) {
6816 insert->ref++;
6817 return(insert);
6818 }
6819 } else if (type == XML_EXP_COUNT) {
6820 if ((insert->exp_min == min) && (insert->exp_max == max) &&
6821 (insert->exp_left == left)) {
6822 insert->ref++;
6823 left->ref--;
6824 return(insert);
6825 }
6826 } else if ((insert->exp_left == left) &&
6827 (insert->exp_right == right)) {
6828 insert->ref++;
6829 left->ref--;
6830 right->ref--;
6831 return(insert);
6832 }
6833 }
6834 }
6835 }
6836
6837 entry = xmlExpNewNode(ctxt, type);
6838 if (entry == NULL)
6839 return(NULL);
6840 entry->key = kbase;
6841 if (type == XML_EXP_ATOM) {
6842 entry->exp_str = name;
6843 entry->c_max = 1;
6844 } else if (type == XML_EXP_COUNT) {
6845 entry->exp_min = min;
6846 entry->exp_max = max;
6847 entry->exp_left = left;
6848 if ((min == 0) || (IS_NILLABLE(left)))
6849 entry->info |= XML_EXP_NILABLE;
6850 if (max < 0)
6851 entry->c_max = -1;
6852 else
6853 entry->c_max = max * entry->exp_left->c_max;
6854 } else {
6855 entry->exp_left = left;
6856 entry->exp_right = right;
6857 if (type == XML_EXP_OR) {
6858 if ((IS_NILLABLE(left)) || (IS_NILLABLE(right)))
6859 entry->info |= XML_EXP_NILABLE;
6860 if ((entry->exp_left->c_max == -1) ||
6861 (entry->exp_right->c_max == -1))
6862 entry->c_max = -1;
6863 else if (entry->exp_left->c_max > entry->exp_right->c_max)
6864 entry->c_max = entry->exp_left->c_max;
6865 else
6866 entry->c_max = entry->exp_right->c_max;
6867 } else {
6868 if ((IS_NILLABLE(left)) && (IS_NILLABLE(right)))
6869 entry->info |= XML_EXP_NILABLE;
6870 if ((entry->exp_left->c_max == -1) ||
6871 (entry->exp_right->c_max == -1))
6872 entry->c_max = -1;
6873 else
6874 entry->c_max = entry->exp_left->c_max + entry->exp_right->c_max;
6875 }
6876 }
6877 entry->ref = 1;
6878 if (ctxt->table[key] != NULL)
6879 entry->next = ctxt->table[key];
6880
6881 ctxt->table[key] = entry;
6882 ctxt->nbElems++;
6883
6884 return(entry);
6885}
6886
6887/**
6888 * xmlExpFree:
6889 * @ctxt: the expression context
6890 * @exp: the expression
6891 *
6892 * Dereference the expression
6893 */
6894void
6895xmlExpFree(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp) {
6896 if ((exp == NULL) || (exp == forbiddenExp) || (exp == emptyExp))
6897 return;
6898 exp->ref--;
6899 if (exp->ref == 0) {
6900 unsigned short key;
6901
6902 /* Unlink it first from the hash table */
6903 key = exp->key % ctxt->size;
6904 if (ctxt->table[key] == exp) {
6905 ctxt->table[key] = exp->next;
6906 } else {
6907 xmlExpNodePtr tmp;
6908
6909 tmp = ctxt->table[key];
6910 while (tmp != NULL) {
6911 if (tmp->next == exp) {
6912 tmp->next = exp->next;
6913 break;
6914 }
6915 tmp = tmp->next;
6916 }
6917 }
6918
6919 if ((exp->type == XML_EXP_SEQ) || (exp->type == XML_EXP_OR)) {
6920 xmlExpFree(ctxt, exp->exp_left);
6921 xmlExpFree(ctxt, exp->exp_right);
6922 } else if (exp->type == XML_EXP_COUNT) {
6923 xmlExpFree(ctxt, exp->exp_left);
6924 }
6925 xmlFree(exp);
6926 ctxt->nb_nodes--;
6927 }
6928}
6929
6930/**
6931 * xmlExpRef:
6932 * @exp: the expression
6933 *
6934 * Increase the reference count of the expression
6935 */
6936void
6937xmlExpRef(xmlExpNodePtr exp) {
6938 if (exp != NULL)
6939 exp->ref++;
6940}
6941
6942/**
6943 * xmlExpNewAtom:
6944 * @ctxt: the expression context
6945 * @name: the atom name
6946 * @len: the atom name length in byte (or -1);
6947 *
6948 * Get the atom associated to this name from that context
6949 *
6950 * Returns the node or NULL in case of error
6951 */
6952xmlExpNodePtr
6953xmlExpNewAtom(xmlExpCtxtPtr ctxt, const xmlChar *name, int len) {
6954 if ((ctxt == NULL) || (name == NULL))
6955 return(NULL);
6956 name = xmlDictLookup(ctxt->dict, name, len);
6957 if (name == NULL)
6958 return(NULL);
6959 return(xmlExpHashGetEntry(ctxt, XML_EXP_ATOM, NULL, NULL, name, 0, 0));
6960}
6961
6962/**
6963 * xmlExpNewOr:
6964 * @ctxt: the expression context
6965 * @left: left expression
6966 * @right: right expression
6967 *
6968 * Get the atom associated to the choice @left | @right
6969 * Note that @left and @right are consumed in the operation, to keep
6970 * an handle on them use xmlExpRef() and use xmlExpFree() to release them,
6971 * this is true even in case of failure (unless ctxt == NULL).
6972 *
6973 * Returns the node or NULL in case of error
6974 */
6975xmlExpNodePtr
6976xmlExpNewOr(xmlExpCtxtPtr ctxt, xmlExpNodePtr left, xmlExpNodePtr right) {
6977 if (ctxt == NULL)
6978 return(NULL);
6979 if ((left == NULL) || (right == NULL)) {
6980 xmlExpFree(ctxt, left);
6981 xmlExpFree(ctxt, right);
6982 return(NULL);
6983 }
6984 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, left, right, NULL, 0, 0));
6985}
6986
6987/**
6988 * xmlExpNewSeq:
6989 * @ctxt: the expression context
6990 * @left: left expression
6991 * @right: right expression
6992 *
6993 * Get the atom associated to the sequence @left , @right
6994 * Note that @left and @right are consumed in the operation, to keep
6995 * an handle on them use xmlExpRef() and use xmlExpFree() to release them,
6996 * this is true even in case of failure (unless ctxt == NULL).
6997 *
6998 * Returns the node or NULL in case of error
6999 */
7000xmlExpNodePtr
7001xmlExpNewSeq(xmlExpCtxtPtr ctxt, xmlExpNodePtr left, xmlExpNodePtr right) {
7002 if (ctxt == NULL)
7003 return(NULL);
7004 if ((left == NULL) || (right == NULL)) {
7005 xmlExpFree(ctxt, left);
7006 xmlExpFree(ctxt, right);
7007 return(NULL);
7008 }
7009 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, left, right, NULL, 0, 0));
7010}
7011
7012/**
7013 * xmlExpNewRange:
7014 * @ctxt: the expression context
7015 * @subset: the expression to be repeated
7016 * @min: the lower bound for the repetition
7017 * @max: the upper bound for the repetition, -1 means infinite
7018 *
7019 * Get the atom associated to the range (@subset){@min, @max}
7020 * Note that @subset is consumed in the operation, to keep
7021 * an handle on it use xmlExpRef() and use xmlExpFree() to release it,
7022 * this is true even in case of failure (unless ctxt == NULL).
7023 *
7024 * Returns the node or NULL in case of error
7025 */
7026xmlExpNodePtr
7027xmlExpNewRange(xmlExpCtxtPtr ctxt, xmlExpNodePtr subset, int min, int max) {
7028 if (ctxt == NULL)
7029 return(NULL);
7030 if ((subset == NULL) || (min < 0) || (max < -1) ||
7031 ((max >= 0) && (min > max))) {
7032 xmlExpFree(ctxt, subset);
7033 return(NULL);
7034 }
7035 return(xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, subset,
7036 NULL, NULL, min, max));
7037}
7038
7039/************************************************************************
7040 * *
7041 * Public API for operations on expressions *
7042 * *
7043 ************************************************************************/
7044
7045static int
7046xmlExpGetLanguageInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7047 const xmlChar**list, int len, int nb) {
7048 int tmp, tmp2;
7049tail:
7050 switch (exp->type) {
7051 case XML_EXP_EMPTY:
7052 return(0);
7053 case XML_EXP_ATOM:
7054 for (tmp = 0;tmp < nb;tmp++)
7055 if (list[tmp] == exp->exp_str)
7056 return(0);
7057 if (nb >= len)
7058 return(-2);
7059 list[nb] = exp->exp_str;
7060 return(1);
7061 case XML_EXP_COUNT:
7062 exp = exp->exp_left;
7063 goto tail;
7064 case XML_EXP_SEQ:
7065 case XML_EXP_OR:
7066 tmp = xmlExpGetLanguageInt(ctxt, exp->exp_left, list, len, nb);
7067 if (tmp < 0)
7068 return(tmp);
7069 tmp2 = xmlExpGetLanguageInt(ctxt, exp->exp_right, list, len,
7070 nb + tmp);
7071 if (tmp2 < 0)
7072 return(tmp2);
7073 return(tmp + tmp2);
7074 }
7075 return(-1);
7076}
7077
7078/**
7079 * xmlExpGetLanguage:
7080 * @ctxt: the expression context
7081 * @exp: the expression
7082 * @langList: where to store the tokens
7083 * @len: the allocated length of @list
7084 *
7085 * Find all the strings used in @exp and store them in @list
7086 *
7087 * Returns the number of unique strings found, -1 in case of errors and
7088 * -2 if there is more than @len strings
7089 */
7090int
7091xmlExpGetLanguage(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7092 const xmlChar**langList, int len) {
7093 if ((ctxt == NULL) || (exp == NULL) || (langList == NULL) || (len <= 0))
7094 return(-1);
7095 return(xmlExpGetLanguageInt(ctxt, exp, langList, len, 0));
7096}
7097
7098static int
7099xmlExpGetStartInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7100 const xmlChar**list, int len, int nb) {
7101 int tmp, tmp2;
7102tail:
7103 switch (exp->type) {
7104 case XML_EXP_FORBID:
7105 return(0);
7106 case XML_EXP_EMPTY:
7107 return(0);
7108 case XML_EXP_ATOM:
7109 for (tmp = 0;tmp < nb;tmp++)
7110 if (list[tmp] == exp->exp_str)
7111 return(0);
7112 if (nb >= len)
7113 return(-2);
7114 list[nb] = exp->exp_str;
7115 return(1);
7116 case XML_EXP_COUNT:
7117 exp = exp->exp_left;
7118 goto tail;
7119 case XML_EXP_SEQ:
7120 tmp = xmlExpGetStartInt(ctxt, exp->exp_left, list, len, nb);
7121 if (tmp < 0)
7122 return(tmp);
7123 if (IS_NILLABLE(exp->exp_left)) {
7124 tmp2 = xmlExpGetStartInt(ctxt, exp->exp_right, list, len,
7125 nb + tmp);
7126 if (tmp2 < 0)
7127 return(tmp2);
7128 tmp += tmp2;
7129 }
7130 return(tmp);
7131 case XML_EXP_OR:
7132 tmp = xmlExpGetStartInt(ctxt, exp->exp_left, list, len, nb);
7133 if (tmp < 0)
7134 return(tmp);
7135 tmp2 = xmlExpGetStartInt(ctxt, exp->exp_right, list, len,
7136 nb + tmp);
7137 if (tmp2 < 0)
7138 return(tmp2);
7139 return(tmp + tmp2);
7140 }
7141 return(-1);
7142}
7143
7144/**
7145 * xmlExpGetStart:
7146 * @ctxt: the expression context
7147 * @exp: the expression
7148 * @tokList: where to store the tokens
7149 * @len: the allocated length of @list
7150 *
7151 * Find all the strings that appears at the start of the languages
7152 * accepted by @exp and store them in @list. E.g. for (a, b) | c
7153 * it will return the list [a, c]
7154 *
7155 * Returns the number of unique strings found, -1 in case of errors and
7156 * -2 if there is more than @len strings
7157 */
7158int
7159xmlExpGetStart(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7160 const xmlChar**tokList, int len) {
7161 if ((ctxt == NULL) || (exp == NULL) || (tokList == NULL) || (len <= 0))
7162 return(-1);
7163 return(xmlExpGetStartInt(ctxt, exp, tokList, len, 0));
7164}
7165
7166/**
7167 * xmlExpIsNillable:
7168 * @exp: the expression
7169 *
7170 * Finds if the expression is nillable, i.e. if it accepts the empty sequence
7171 *
7172 * Returns 1 if nillable, 0 if not and -1 in case of error
7173 */
7174int
7175xmlExpIsNillable(xmlExpNodePtr exp) {
7176 if (exp == NULL)
7177 return(-1);
7178 return(IS_NILLABLE(exp) != 0);
7179}
7180
7181static xmlExpNodePtr
7182xmlExpStringDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, const xmlChar *str)
7183{
7184 xmlExpNodePtr ret;
7185
7186 switch (exp->type) {
7187 case XML_EXP_EMPTY:
7188 return(forbiddenExp);
7189 case XML_EXP_FORBID:
7190 return(forbiddenExp);
7191 case XML_EXP_ATOM:
7192 if (exp->exp_str == str) {
7193#ifdef DEBUG_DERIV
7194 printf("deriv atom: equal => Empty\n");
7195#endif
7196 ret = emptyExp;
7197 } else {
7198#ifdef DEBUG_DERIV
7199 printf("deriv atom: mismatch => forbid\n");
7200#endif
7201 /* TODO wildcards here */
7202 ret = forbiddenExp;
7203 }
7204 return(ret);
7205 case XML_EXP_OR: {
7206 xmlExpNodePtr tmp;
7207
7208#ifdef DEBUG_DERIV
7209 printf("deriv or: => or(derivs)\n");
7210#endif
7211 tmp = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7212 if (tmp == NULL) {
7213 return(NULL);
7214 }
7215 ret = xmlExpStringDeriveInt(ctxt, exp->exp_right, str);
7216 if (ret == NULL) {
7217 xmlExpFree(ctxt, tmp);
7218 return(NULL);
7219 }
7220 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, tmp, ret,
7221 NULL, 0, 0);
7222 return(ret);
7223 }
7224 case XML_EXP_SEQ:
7225#ifdef DEBUG_DERIV
7226 printf("deriv seq: starting with left\n");
7227#endif
7228 ret = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7229 if (ret == NULL) {
7230 return(NULL);
7231 } else if (ret == forbiddenExp) {
7232 if (IS_NILLABLE(exp->exp_left)) {
7233#ifdef DEBUG_DERIV
7234 printf("deriv seq: left failed but nillable\n");
7235#endif
7236 ret = xmlExpStringDeriveInt(ctxt, exp->exp_right, str);
7237 }
7238 } else {
7239#ifdef DEBUG_DERIV
7240 printf("deriv seq: left match => sequence\n");
7241#endif
7242 exp->exp_right->ref++;
7243 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, exp->exp_right,
7244 NULL, 0, 0);
7245 }
7246 return(ret);
7247 case XML_EXP_COUNT: {
7248 int min, max;
7249 xmlExpNodePtr tmp;
7250
7251 if (exp->exp_max == 0)
7252 return(forbiddenExp);
7253 ret = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7254 if (ret == NULL)
7255 return(NULL);
7256 if (ret == forbiddenExp) {
7257#ifdef DEBUG_DERIV
7258 printf("deriv count: pattern mismatch => forbid\n");
7259#endif
7260 return(ret);
7261 }
7262 if (exp->exp_max == 1)
7263 return(ret);
7264 if (exp->exp_max < 0) /* unbounded */
7265 max = -1;
7266 else
7267 max = exp->exp_max - 1;
7268 if (exp->exp_min > 0)
7269 min = exp->exp_min - 1;
7270 else
7271 min = 0;
7272 exp->exp_left->ref++;
7273 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left, NULL,
7274 NULL, min, max);
7275 if (ret == emptyExp) {
7276#ifdef DEBUG_DERIV
7277 printf("deriv count: match to empty => new count\n");
7278#endif
7279 return(tmp);
7280 }
7281#ifdef DEBUG_DERIV
7282 printf("deriv count: match => sequence with new count\n");
7283#endif
7284 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, tmp,
7285 NULL, 0, 0));
7286 }
7287 }
7288 return(NULL);
7289}
7290
7291/**
7292 * xmlExpStringDerive:
7293 * @ctxt: the expression context
7294 * @exp: the expression
7295 * @str: the string
7296 * @len: the string len in bytes if available
7297 *
7298 * Do one step of Brzozowski derivation of the expression @exp with
7299 * respect to the input string
7300 *
7301 * Returns the resulting expression or NULL in case of internal error
7302 */
7303xmlExpNodePtr
7304xmlExpStringDerive(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7305 const xmlChar *str, int len) {
7306 const xmlChar *input;
7307
7308 if ((exp == NULL) || (ctxt == NULL) || (str == NULL)) {
7309 return(NULL);
7310 }
7311 /*
7312 * check the string is in the dictionary, if yes use an interned
7313 * copy, otherwise we know it's not an acceptable input
7314 */
7315 input = xmlDictExists(ctxt->dict, str, len);
7316 if (input == NULL) {
7317 return(forbiddenExp);
7318 }
7319 return(xmlExpStringDeriveInt(ctxt, exp, input));
7320}
7321
7322static int
7323xmlExpCheckCard(xmlExpNodePtr exp, xmlExpNodePtr sub) {
7324 int ret = 1;
7325
7326 if (sub->c_max == -1) {
7327 if (exp->c_max != -1)
7328 ret = 0;
7329 } else if ((exp->c_max >= 0) && (exp->c_max < sub->c_max)) {
7330 ret = 0;
7331 }
7332#if 0
7333 if ((IS_NILLABLE(sub)) && (!IS_NILLABLE(exp)))
7334 ret = 0;
7335#endif
7336 return(ret);
7337}
7338
7339static xmlExpNodePtr xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7340 xmlExpNodePtr sub);
7341/**
7342 * xmlExpDivide:
7343 * @ctxt: the expressions context
7344 * @exp: the englobing expression
7345 * @sub: the subexpression
7346 * @mult: the multiple expression
7347 * @remain: the remain from the derivation of the multiple
7348 *
7349 * Check if exp is a multiple of sub, i.e. if there is a finite number n
7350 * so that sub{n} subsume exp
7351 *
7352 * Returns the multiple value if successful, 0 if it is not a multiple
7353 * and -1 in case of internal error.
7354 */
7355
7356static int
7357xmlExpDivide(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub,
7358 xmlExpNodePtr *mult, xmlExpNodePtr *remain) {
7359 int i;
7360 xmlExpNodePtr tmp, tmp2;
7361
7362 if (mult != NULL) *mult = NULL;
7363 if (remain != NULL) *remain = NULL;
7364 if (exp->c_max == -1) return(0);
7365 if (IS_NILLABLE(exp) && (!IS_NILLABLE(sub))) return(0);
7366
7367 for (i = 1;i <= exp->c_max;i++) {
7368 sub->ref++;
7369 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT,
7370 sub, NULL, NULL, i, i);
7371 if (tmp == NULL) {
7372 return(-1);
7373 }
7374 if (!xmlExpCheckCard(tmp, exp)) {
7375 xmlExpFree(ctxt, tmp);
7376 continue;
7377 }
7378 tmp2 = xmlExpExpDeriveInt(ctxt, tmp, exp);
7379 if (tmp2 == NULL) {
7380 xmlExpFree(ctxt, tmp);
7381 return(-1);
7382 }
7383 if ((tmp2 != forbiddenExp) && (IS_NILLABLE(tmp2))) {
7384 if (remain != NULL)
7385 *remain = tmp2;
7386 else
7387 xmlExpFree(ctxt, tmp2);
7388 if (mult != NULL)
7389 *mult = tmp;
7390 else
7391 xmlExpFree(ctxt, tmp);
7392#ifdef DEBUG_DERIV
7393 printf("Divide succeeded %d\n", i);
7394#endif
7395 return(i);
7396 }
7397 xmlExpFree(ctxt, tmp);
7398 xmlExpFree(ctxt, tmp2);
7399 }
7400#ifdef DEBUG_DERIV
7401 printf("Divide failed\n");
7402#endif
7403 return(0);
7404}
7405
7406/**
7407 * xmlExpExpDeriveInt:
7408 * @ctxt: the expressions context
7409 * @exp: the englobing expression
7410 * @sub: the subexpression
7411 *
7412 * Try to do a step of Brzozowski derivation but at a higher level
7413 * the input being a subexpression.
7414 *
7415 * Returns the resulting expression or NULL in case of internal error
7416 */
7417static xmlExpNodePtr
7418xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7419 xmlExpNodePtr ret, tmp, tmp2, tmp3;
7420 const xmlChar **tab;
7421 int len, i;
7422
7423 /*
7424 * In case of equality and if the expression can only consume a finite
7425 * amount, then the derivation is empty
7426 */
7427 if ((exp == sub) && (exp->c_max >= 0)) {
7428#ifdef DEBUG_DERIV
7429 printf("Equal(exp, sub) and finite -> Empty\n");
7430#endif
7431 return(emptyExp);
7432 }
7433 /*
7434 * decompose sub sequence first
7435 */
7436 if (sub->type == XML_EXP_EMPTY) {
7437#ifdef DEBUG_DERIV
7438 printf("Empty(sub) -> Empty\n");
7439#endif
7440 exp->ref++;
7441 return(exp);
7442 }
7443 if (sub->type == XML_EXP_SEQ) {
7444#ifdef DEBUG_DERIV
7445 printf("Seq(sub) -> decompose\n");
7446#endif
7447 tmp = xmlExpExpDeriveInt(ctxt, exp, sub->exp_left);
7448 if (tmp == NULL)
7449 return(NULL);
7450 if (tmp == forbiddenExp)
7451 return(tmp);
7452 ret = xmlExpExpDeriveInt(ctxt, tmp, sub->exp_right);
7453 xmlExpFree(ctxt, tmp);
7454 return(ret);
7455 }
7456 if (sub->type == XML_EXP_OR) {
7457#ifdef DEBUG_DERIV
7458 printf("Or(sub) -> decompose\n");
7459#endif
7460 tmp = xmlExpExpDeriveInt(ctxt, exp, sub->exp_left);
7461 if (tmp == forbiddenExp)
7462 return(tmp);
7463 if (tmp == NULL)
7464 return(NULL);
7465 ret = xmlExpExpDeriveInt(ctxt, exp, sub->exp_right);
7466 if ((ret == NULL) || (ret == forbiddenExp)) {
7467 xmlExpFree(ctxt, tmp);
7468 return(ret);
7469 }
7470 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, tmp, ret, NULL, 0, 0));
7471 }
7472 if (!xmlExpCheckCard(exp, sub)) {
7473#ifdef DEBUG_DERIV
7474 printf("CheckCard(exp, sub) failed -> Forbid\n");
7475#endif
7476 return(forbiddenExp);
7477 }
7478 switch (exp->type) {
7479 case XML_EXP_EMPTY:
7480 if (sub == emptyExp)
7481 return(emptyExp);
7482#ifdef DEBUG_DERIV
7483 printf("Empty(exp) -> Forbid\n");
7484#endif
7485 return(forbiddenExp);
7486 case XML_EXP_FORBID:
7487#ifdef DEBUG_DERIV
7488 printf("Forbid(exp) -> Forbid\n");
7489#endif
7490 return(forbiddenExp);
7491 case XML_EXP_ATOM:
7492 if (sub->type == XML_EXP_ATOM) {
7493 /* TODO: handle wildcards */
7494 if (exp->exp_str == sub->exp_str) {
7495#ifdef DEBUG_DERIV
7496 printf("Atom match -> Empty\n");
7497#endif
7498 return(emptyExp);
7499 }
7500#ifdef DEBUG_DERIV
7501 printf("Atom mismatch -> Forbid\n");
7502#endif
7503 return(forbiddenExp);
7504 }
7505 if ((sub->type == XML_EXP_COUNT) &&
7506 (sub->exp_max == 1) &&
7507 (sub->exp_left->type == XML_EXP_ATOM)) {
7508 /* TODO: handle wildcards */
7509 if (exp->exp_str == sub->exp_left->exp_str) {
7510#ifdef DEBUG_DERIV
7511 printf("Atom match -> Empty\n");
7512#endif
7513 return(emptyExp);
7514 }
7515#ifdef DEBUG_DERIV
7516 printf("Atom mismatch -> Forbid\n");
7517#endif
7518 return(forbiddenExp);
7519 }
7520#ifdef DEBUG_DERIV
7521 printf("Complex exp vs Atom -> Forbid\n");
7522#endif
7523 return(forbiddenExp);
7524 case XML_EXP_SEQ:
7525 /* try to get the sequence consumed only if possible */
7526 if (xmlExpCheckCard(exp->exp_left, sub)) {
7527 /* See if the sequence can be consumed directly */
7528#ifdef DEBUG_DERIV
7529 printf("Seq trying left only\n");
7530#endif
7531 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7532 if ((ret != forbiddenExp) && (ret != NULL)) {
7533#ifdef DEBUG_DERIV
7534 printf("Seq trying left only worked\n");
7535#endif
7536 /*
7537 * TODO: assumption here that we are determinist
7538 * i.e. we won't get to a nillable exp left
7539 * subset which could be matched by the right
7540 * part too.
7541 * e.g.: (a | b)+,(a | c) and 'a+,a'
7542 */
7543 exp->exp_right->ref++;
7544 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret,
7545 exp->exp_right, NULL, 0, 0));
7546 }
7547#ifdef DEBUG_DERIV
7548 } else {
7549 printf("Seq: left too short\n");
7550#endif
7551 }
7552 /* Try instead to decompose */
7553 if (sub->type == XML_EXP_COUNT) {
7554 int min, max;
7555
7556#ifdef DEBUG_DERIV
7557 printf("Seq: sub is a count\n");
7558#endif
7559 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub->exp_left);
7560 if (ret == NULL)
7561 return(NULL);
7562 if (ret != forbiddenExp) {
7563#ifdef DEBUG_DERIV
7564 printf("Seq , Count match on left\n");
7565#endif
7566 if (sub->exp_max < 0)
7567 max = -1;
7568 else
7569 max = sub->exp_max -1;
7570 if (sub->exp_min > 0)
7571 min = sub->exp_min -1;
7572 else
7573 min = 0;
7574 exp->exp_right->ref++;
7575 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret,
7576 exp->exp_right, NULL, 0, 0);
7577 if (tmp == NULL)
7578 return(NULL);
7579
7580 sub->exp_left->ref++;
7581 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT,
7582 sub->exp_left, NULL, NULL, min, max);
7583 if (tmp2 == NULL) {
7584 xmlExpFree(ctxt, tmp);
7585 return(NULL);
7586 }
7587 ret = xmlExpExpDeriveInt(ctxt, tmp, tmp2);
7588 xmlExpFree(ctxt, tmp);
7589 xmlExpFree(ctxt, tmp2);
7590 return(ret);
7591 }
7592 }
7593 /* we made no progress on structured operations */
7594 break;
7595 case XML_EXP_OR:
7596#ifdef DEBUG_DERIV
7597 printf("Or , trying both side\n");
7598#endif
7599 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7600 if (ret == NULL)
7601 return(NULL);
7602 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_right, sub);
7603 if (tmp == NULL) {
7604 xmlExpFree(ctxt, ret);
7605 return(NULL);
7606 }
7607 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, tmp, NULL, 0, 0));
7608 case XML_EXP_COUNT: {
7609 int min, max;
7610
7611 if (sub->type == XML_EXP_COUNT) {
7612 /*
7613 * Try to see if the loop is completely subsumed
7614 */
7615 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub->exp_left);
7616 if (tmp == NULL)
7617 return(NULL);
7618 if (tmp == forbiddenExp) {
7619 int mult;
7620
7621#ifdef DEBUG_DERIV
7622 printf("Count, Count inner don't subsume\n");
7623#endif
7624 mult = xmlExpDivide(ctxt, sub->exp_left, exp->exp_left,
7625 NULL, &tmp);
7626 if (mult <= 0) {
7627#ifdef DEBUG_DERIV
7628 printf("Count, Count not multiple => forbidden\n");
7629#endif
7630 return(forbiddenExp);
7631 }
7632 if (sub->exp_max == -1) {
7633 max = -1;
7634 if (exp->exp_max == -1) {
7635 if (exp->exp_min <= sub->exp_min * mult)
7636 min = 0;
7637 else
7638 min = exp->exp_min - sub->exp_min * mult;
7639 } else {
7640#ifdef DEBUG_DERIV
7641 printf("Count, Count finite can't subsume infinite\n");
7642#endif
7643 xmlExpFree(ctxt, tmp);
7644 return(forbiddenExp);
7645 }
7646 } else {
7647 if (exp->exp_max == -1) {
7648#ifdef DEBUG_DERIV
7649 printf("Infinite loop consume mult finite loop\n");
7650#endif
7651 if (exp->exp_min > sub->exp_min * mult) {
7652 max = -1;
7653 min = exp->exp_min - sub->exp_min * mult;
7654 } else {
7655 max = -1;
7656 min = 0;
7657 }
7658 } else {
7659 if (exp->exp_max < sub->exp_max * mult) {
7660#ifdef DEBUG_DERIV
7661 printf("loops max mult mismatch => forbidden\n");
7662#endif
7663 xmlExpFree(ctxt, tmp);
7664 return(forbiddenExp);
7665 }
7666 if (sub->exp_max * mult > exp->exp_min)
7667 min = 0;
7668 else
7669 min = exp->exp_min - sub->exp_max * mult;
7670 max = exp->exp_max - sub->exp_max * mult;
7671 }
7672 }
7673 } else if (!IS_NILLABLE(tmp)) {
7674 /*
7675 * TODO: loop here to try to grow if working on finite
7676 * blocks.
7677 */
7678#ifdef DEBUG_DERIV
7679 printf("Count, Count remain not nillable => forbidden\n");
7680#endif
7681 xmlExpFree(ctxt, tmp);
7682 return(forbiddenExp);
7683 } else if (sub->exp_max == -1) {
7684 if (exp->exp_max == -1) {
7685 if (exp->exp_min <= sub->exp_min) {
7686#ifdef DEBUG_DERIV
7687 printf("Infinite loops Okay => COUNT(0,Inf)\n");
7688#endif
7689 max = -1;
7690 min = 0;
7691 } else {
7692#ifdef DEBUG_DERIV
7693 printf("Infinite loops min => Count(X,Inf)\n");
7694#endif
7695 max = -1;
7696 min = exp->exp_min - sub->exp_min;
7697 }
7698 } else if (exp->exp_min > sub->exp_min) {
7699#ifdef DEBUG_DERIV
7700 printf("loops min mismatch 1 => forbidden ???\n");
7701#endif
7702 xmlExpFree(ctxt, tmp);
7703 return(forbiddenExp);
7704 } else {
7705 max = -1;
7706 min = 0;
7707 }
7708 } else {
7709 if (exp->exp_max == -1) {
7710#ifdef DEBUG_DERIV
7711 printf("Infinite loop consume finite loop\n");
7712#endif
7713 if (exp->exp_min > sub->exp_min) {
7714 max = -1;
7715 min = exp->exp_min - sub->exp_min;
7716 } else {
7717 max = -1;
7718 min = 0;
7719 }
7720 } else {
7721 if (exp->exp_max < sub->exp_max) {
7722#ifdef DEBUG_DERIV
7723 printf("loops max mismatch => forbidden\n");
7724#endif
7725 xmlExpFree(ctxt, tmp);
7726 return(forbiddenExp);
7727 }
7728 if (sub->exp_max > exp->exp_min)
7729 min = 0;
7730 else
7731 min = exp->exp_min - sub->exp_max;
7732 max = exp->exp_max - sub->exp_max;
7733 }
7734 }
7735#ifdef DEBUG_DERIV
7736 printf("loops match => SEQ(COUNT())\n");
7737#endif
7738 exp->exp_left->ref++;
7739 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left,
7740 NULL, NULL, min, max);
7741 if (tmp2 == NULL) {
7742 return(NULL);
7743 }
7744 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, tmp, tmp2,
7745 NULL, 0, 0);
7746 return(ret);
7747 }
7748 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7749 if (tmp == NULL)
7750 return(NULL);
7751 if (tmp == forbiddenExp) {
7752#ifdef DEBUG_DERIV
7753 printf("loop mismatch => forbidden\n");
7754#endif
7755 return(forbiddenExp);
7756 }
7757 if (exp->exp_min > 0)
7758 min = exp->exp_min - 1;
7759 else
7760 min = 0;
7761 if (exp->exp_max < 0)
7762 max = -1;
7763 else
7764 max = exp->exp_max - 1;
7765
7766#ifdef DEBUG_DERIV
7767 printf("loop match => SEQ(COUNT())\n");
7768#endif
7769 exp->exp_left->ref++;
7770 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left,
7771 NULL, NULL, min, max);
7772 if (tmp2 == NULL)
7773 return(NULL);
7774 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, tmp, tmp2,
7775 NULL, 0, 0);
7776 return(ret);
7777 }
7778 }
7779
7780#ifdef DEBUG_DERIV
7781 printf("Fallback to derivative\n");
7782#endif
7783 if (IS_NILLABLE(sub)) {
7784 if (!(IS_NILLABLE(exp)))
7785 return(forbiddenExp);
7786 else
7787 ret = emptyExp;
7788 } else
7789 ret = NULL;
7790 /*
7791 * here the structured derivation made no progress so
7792 * we use the default token based derivation to force one more step
7793 */
7794 if (ctxt->tabSize == 0)
7795 ctxt->tabSize = 40;
7796
7797 tab = (const xmlChar **) xmlMalloc(ctxt->tabSize *
7798 sizeof(const xmlChar *));
7799 if (tab == NULL) {
7800 return(NULL);
7801 }
7802
7803 /*
7804 * collect all the strings accepted by the subexpression on input
7805 */
7806 len = xmlExpGetStartInt(ctxt, sub, tab, ctxt->tabSize, 0);
7807 while (len < 0) {
7808 const xmlChar **temp;
7809 temp = (const xmlChar **) xmlRealloc((xmlChar **) tab, ctxt->tabSize * 2 *
7810 sizeof(const xmlChar *));
7811 if (temp == NULL) {
7812 xmlFree((xmlChar **) tab);
7813 return(NULL);
7814 }
7815 tab = temp;
7816 ctxt->tabSize *= 2;
7817 len = xmlExpGetStartInt(ctxt, sub, tab, ctxt->tabSize, 0);
7818 }
7819 for (i = 0;i < len;i++) {
7820 tmp = xmlExpStringDeriveInt(ctxt, exp, tab[i]);
7821 if ((tmp == NULL) || (tmp == forbiddenExp)) {
7822 xmlExpFree(ctxt, ret);
7823 xmlFree((xmlChar **) tab);
7824 return(tmp);
7825 }
7826 tmp2 = xmlExpStringDeriveInt(ctxt, sub, tab[i]);
7827 if ((tmp2 == NULL) || (tmp2 == forbiddenExp)) {
7828 xmlExpFree(ctxt, tmp);
7829 xmlExpFree(ctxt, ret);
7830 xmlFree((xmlChar **) tab);
7831 return(tmp);
7832 }
7833 tmp3 = xmlExpExpDeriveInt(ctxt, tmp, tmp2);
7834 xmlExpFree(ctxt, tmp);
7835 xmlExpFree(ctxt, tmp2);
7836
7837 if ((tmp3 == NULL) || (tmp3 == forbiddenExp)) {
7838 xmlExpFree(ctxt, ret);
7839 xmlFree((xmlChar **) tab);
7840 return(tmp3);
7841 }
7842
7843 if (ret == NULL)
7844 ret = tmp3;
7845 else {
7846 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, tmp3, NULL, 0, 0);
7847 if (ret == NULL) {
7848 xmlFree((xmlChar **) tab);
7849 return(NULL);
7850 }
7851 }
7852 }
7853 xmlFree((xmlChar **) tab);
7854 return(ret);
7855}
7856
7857/**
7858 * xmlExpExpDerive:
7859 * @ctxt: the expressions context
7860 * @exp: the englobing expression
7861 * @sub: the subexpression
7862 *
7863 * Evaluates the expression resulting from @exp consuming a sub expression @sub
7864 * Based on algebraic derivation and sometimes direct Brzozowski derivation
7865 * it usually takes less than linear time and can handle expressions generating
7866 * infinite languages.
7867 *
7868 * Returns the resulting expression or NULL in case of internal error, the
7869 * result must be freed
7870 */
7871xmlExpNodePtr
7872xmlExpExpDerive(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7873 if ((exp == NULL) || (ctxt == NULL) || (sub == NULL))
7874 return(NULL);
7875
7876 /*
7877 * O(1) speedups
7878 */
7879 if (IS_NILLABLE(sub) && (!IS_NILLABLE(exp))) {
7880#ifdef DEBUG_DERIV
7881 printf("Sub nillable and not exp : can't subsume\n");
7882#endif
7883 return(forbiddenExp);
7884 }
7885 if (xmlExpCheckCard(exp, sub) == 0) {
7886#ifdef DEBUG_DERIV
7887 printf("sub generate longer sequences than exp : can't subsume\n");
7888#endif
7889 return(forbiddenExp);
7890 }
7891 return(xmlExpExpDeriveInt(ctxt, exp, sub));
7892}
7893
7894/**
7895 * xmlExpSubsume:
7896 * @ctxt: the expressions context
7897 * @exp: the englobing expression
7898 * @sub: the subexpression
7899 *
7900 * Check whether @exp accepts all the languages accepted by @sub
7901 * the input being a subexpression.
7902 *
7903 * Returns 1 if true 0 if false and -1 in case of failure.
7904 */
7905int
7906xmlExpSubsume(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7907 xmlExpNodePtr tmp;
7908
7909 if ((exp == NULL) || (ctxt == NULL) || (sub == NULL))
7910 return(-1);
7911
7912 /*
7913 * TODO: speedup by checking the language of sub is a subset of the
7914 * language of exp
7915 */
7916 /*
7917 * O(1) speedups
7918 */
7919 if (IS_NILLABLE(sub) && (!IS_NILLABLE(exp))) {
7920#ifdef DEBUG_DERIV
7921 printf("Sub nillable and not exp : can't subsume\n");
7922#endif
7923 return(0);
7924 }
7925 if (xmlExpCheckCard(exp, sub) == 0) {
7926#ifdef DEBUG_DERIV
7927 printf("sub generate longer sequences than exp : can't subsume\n");
7928#endif
7929 return(0);
7930 }
7931 tmp = xmlExpExpDeriveInt(ctxt, exp, sub);
7932#ifdef DEBUG_DERIV
7933 printf("Result derivation :\n");
7934 PRINT_EXP(tmp);
7935#endif
7936 if (tmp == NULL)
7937 return(-1);
7938 if (tmp == forbiddenExp)
7939 return(0);
7940 if (tmp == emptyExp)
7941 return(1);
7942 if ((tmp != NULL) && (IS_NILLABLE(tmp))) {
7943 xmlExpFree(ctxt, tmp);
7944 return(1);
7945 }
7946 xmlExpFree(ctxt, tmp);
7947 return(0);
7948}
7949
7950/************************************************************************
7951 * *
7952 * Parsing expression *
7953 * *
7954 ************************************************************************/
7955
7956static xmlExpNodePtr xmlExpParseExpr(xmlExpCtxtPtr ctxt);
7957
7958#undef CUR
7959#define CUR (*ctxt->cur)
7960#undef NEXT
7961#define NEXT ctxt->cur++;
7962#undef IS_BLANK
7963#define IS_BLANK(c) ((c == ' ') || (c == '\n') || (c == '\r') || (c == '\t'))
7964#define SKIP_BLANKS while (IS_BLANK(*ctxt->cur)) ctxt->cur++;
7965
7966static int
7967xmlExpParseNumber(xmlExpCtxtPtr ctxt) {
7968 int ret = 0;
7969
7970 SKIP_BLANKS
7971 if (CUR == '*') {
7972 NEXT
7973 return(-1);
7974 }
7975 if ((CUR < '0') || (CUR > '9'))
7976 return(-1);
7977 while ((CUR >= '0') && (CUR <= '9')) {
7978 ret = ret * 10 + (CUR - '0');
7979 NEXT
7980 }
7981 return(ret);
7982}
7983
7984static xmlExpNodePtr
7985xmlExpParseOr(xmlExpCtxtPtr ctxt) {
7986 const char *base;
7987 xmlExpNodePtr ret;
7988 const xmlChar *val;
7989
7990 SKIP_BLANKS
7991 base = ctxt->cur;
7992 if (*ctxt->cur == '(') {
7993 NEXT
7994 ret = xmlExpParseExpr(ctxt);
7995 SKIP_BLANKS
7996 if (*ctxt->cur != ')') {
7997 fprintf(stderr, "unbalanced '(' : %s\n", base);
7998 xmlExpFree(ctxt, ret);
7999 return(NULL);
8000 }
8001 NEXT;
8002 SKIP_BLANKS
8003 goto parse_quantifier;
8004 }
8005 while ((CUR != 0) && (!(IS_BLANK(CUR))) && (CUR != '(') &&
8006 (CUR != ')') && (CUR != '|') && (CUR != ',') && (CUR != '{') &&
8007 (CUR != '*') && (CUR != '+') && (CUR != '?') && (CUR != '}'))
8008 NEXT;
8009 val = xmlDictLookup(ctxt->dict, BAD_CAST base, ctxt->cur - base);
8010 if (val == NULL)
8011 return(NULL);
8012 ret = xmlExpHashGetEntry(ctxt, XML_EXP_ATOM, NULL, NULL, val, 0, 0);
8013 if (ret == NULL)
8014 return(NULL);
8015 SKIP_BLANKS
8016parse_quantifier:
8017 if (CUR == '{') {
8018 int min, max;
8019
8020 NEXT
8021 min = xmlExpParseNumber(ctxt);
8022 if (min < 0) {
8023 xmlExpFree(ctxt, ret);
8024 return(NULL);
8025 }
8026 SKIP_BLANKS
8027 if (CUR == ',') {
8028 NEXT
8029 max = xmlExpParseNumber(ctxt);
8030 SKIP_BLANKS
8031 } else
8032 max = min;
8033 if (CUR != '}') {
8034 xmlExpFree(ctxt, ret);
8035 return(NULL);
8036 }
8037 NEXT
8038 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
8039 min, max);
8040 SKIP_BLANKS
8041 } else if (CUR == '?') {
8042 NEXT
8043 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
8044 0, 1);
8045 SKIP_BLANKS
8046 } else if (CUR == '+') {
8047 NEXT
8048 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
8049 1, -1);
8050 SKIP_BLANKS
8051 } else if (CUR == '*') {
8052 NEXT
8053 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
8054 0, -1);
8055 SKIP_BLANKS
8056 }
8057 return(ret);
8058}
8059
8060
8061static xmlExpNodePtr
8062xmlExpParseSeq(xmlExpCtxtPtr ctxt) {
8063 xmlExpNodePtr ret, right;
8064
8065 ret = xmlExpParseOr(ctxt);
8066 SKIP_BLANKS
8067 while (CUR == '|') {
8068 NEXT
8069 right = xmlExpParseOr(ctxt);
8070 if (right == NULL) {
8071 xmlExpFree(ctxt, ret);
8072 return(NULL);
8073 }
8074 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, right, NULL, 0, 0);
8075 if (ret == NULL)
8076 return(NULL);
8077 }
8078 return(ret);
8079}
8080
8081static xmlExpNodePtr
8082xmlExpParseExpr(xmlExpCtxtPtr ctxt) {
8083 xmlExpNodePtr ret, right;
8084
8085 ret = xmlExpParseSeq(ctxt);
8086 SKIP_BLANKS
8087 while (CUR == ',') {
8088 NEXT
8089 right = xmlExpParseSeq(ctxt);
8090 if (right == NULL) {
8091 xmlExpFree(ctxt, ret);
8092 return(NULL);
8093 }
8094 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, right, NULL, 0, 0);
8095 if (ret == NULL)
8096 return(NULL);
8097 }
8098 return(ret);
8099}
8100
8101/**
8102 * xmlExpParse:
8103 * @ctxt: the expressions context
8104 * @expr: the 0 terminated string
8105 *
8106 * Minimal parser for regexps, it understand the following constructs
8107 * - string terminals
8108 * - choice operator |
8109 * - sequence operator ,
8110 * - subexpressions (...)
8111 * - usual cardinality operators + * and ?
8112 * - finite sequences { min, max }
8113 * - infinite sequences { min, * }
8114 * There is minimal checkings made especially no checking on strings values
8115 *
8116 * Returns a new expression or NULL in case of failure
8117 */
8118xmlExpNodePtr
8119xmlExpParse(xmlExpCtxtPtr ctxt, const char *expr) {
8120 xmlExpNodePtr ret;
8121
8122 ctxt->expr = expr;
8123 ctxt->cur = expr;
8124
8125 ret = xmlExpParseExpr(ctxt);
8126 SKIP_BLANKS
8127 if (*ctxt->cur != 0) {
8128 xmlExpFree(ctxt, ret);
8129 return(NULL);
8130 }
8131 return(ret);
8132}
8133
8134static void
8135xmlExpDumpInt(xmlBufferPtr buf, xmlExpNodePtr expr, int glob) {
8136 xmlExpNodePtr c;
8137
8138 if (expr == NULL) return;
8139 if (glob) xmlBufferWriteChar(buf, "(");
8140 switch (expr->type) {
8141 case XML_EXP_EMPTY:
8142 xmlBufferWriteChar(buf, "empty");
8143 break;
8144 case XML_EXP_FORBID:
8145 xmlBufferWriteChar(buf, "forbidden");
8146 break;
8147 case XML_EXP_ATOM:
8148 xmlBufferWriteCHAR(buf, expr->exp_str);
8149 break;
8150 case XML_EXP_SEQ:
8151 c = expr->exp_left;
8152 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8153 xmlExpDumpInt(buf, c, 1);
8154 else
8155 xmlExpDumpInt(buf, c, 0);
8156 xmlBufferWriteChar(buf, " , ");
8157 c = expr->exp_right;
8158 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8159 xmlExpDumpInt(buf, c, 1);
8160 else
8161 xmlExpDumpInt(buf, c, 0);
8162 break;
8163 case XML_EXP_OR:
8164 c = expr->exp_left;
8165 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8166 xmlExpDumpInt(buf, c, 1);
8167 else
8168 xmlExpDumpInt(buf, c, 0);
8169 xmlBufferWriteChar(buf, " | ");
8170 c = expr->exp_right;
8171 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8172 xmlExpDumpInt(buf, c, 1);
8173 else
8174 xmlExpDumpInt(buf, c, 0);
8175 break;
8176 case XML_EXP_COUNT: {
8177 char rep[40];
8178
8179 c = expr->exp_left;
8180 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8181 xmlExpDumpInt(buf, c, 1);
8182 else
8183 xmlExpDumpInt(buf, c, 0);
8184 if ((expr->exp_min == 0) && (expr->exp_max == 1)) {
8185 rep[0] = '?';
8186 rep[1] = 0;
8187 } else if ((expr->exp_min == 0) && (expr->exp_max == -1)) {
8188 rep[0] = '*';
8189 rep[1] = 0;
8190 } else if ((expr->exp_min == 1) && (expr->exp_max == -1)) {
8191 rep[0] = '+';
8192 rep[1] = 0;
8193 } else if (expr->exp_max == expr->exp_min) {
8194 snprintf(rep, 39, "{%d}", expr->exp_min);
8195 } else if (expr->exp_max < 0) {
8196 snprintf(rep, 39, "{%d,inf}", expr->exp_min);
8197 } else {
8198 snprintf(rep, 39, "{%d,%d}", expr->exp_min, expr->exp_max);
8199 }
8200 rep[39] = 0;
8201 xmlBufferWriteChar(buf, rep);
8202 break;
8203 }
8204 default:
8205 fprintf(stderr, "Error in tree\n");
8206 }
8207 if (glob)
8208 xmlBufferWriteChar(buf, ")");
8209}
8210/**
8211 * xmlExpDump:
8212 * @buf: a buffer to receive the output
8213 * @expr: the compiled expression
8214 *
8215 * Serialize the expression as compiled to the buffer
8216 */
8217void
8218xmlExpDump(xmlBufferPtr buf, xmlExpNodePtr expr) {
8219 if ((buf == NULL) || (expr == NULL))
8220 return;
8221 xmlExpDumpInt(buf, expr, 0);
8222}
8223
8224/**
8225 * xmlExpMaxToken:
8226 * @expr: a compiled expression
8227 *
8228 * Indicate the maximum number of input a expression can accept
8229 *
8230 * Returns the maximum length or -1 in case of error
8231 */
8232int
8233xmlExpMaxToken(xmlExpNodePtr expr) {
8234 if (expr == NULL)
8235 return(-1);
8236 return(expr->c_max);
8237}
8238
8239/**
8240 * xmlExpCtxtNbNodes:
8241 * @ctxt: an expression context
8242 *
8243 * Debugging facility provides the number of allocated nodes at a that point
8244 *
8245 * Returns the number of nodes in use or -1 in case of error
8246 */
8247int
8248xmlExpCtxtNbNodes(xmlExpCtxtPtr ctxt) {
8249 if (ctxt == NULL)
8250 return(-1);
8251 return(ctxt->nb_nodes);
8252}
8253
8254/**
8255 * xmlExpCtxtNbCons:
8256 * @ctxt: an expression context
8257 *
8258 * Debugging facility provides the number of allocated nodes over lifetime
8259 *
8260 * Returns the number of nodes ever allocated or -1 in case of error
8261 */
8262int
8263xmlExpCtxtNbCons(xmlExpCtxtPtr ctxt) {
8264 if (ctxt == NULL)
8265 return(-1);
8266 return(ctxt->nb_cons);
8267}
8268
8269#endif /* LIBXML_EXPR_ENABLED */
8270#define bottom_xmlregexp
8271#include "elfgcchack.h"
8272#endif /* LIBXML_REGEXP_ENABLED */
Note: See TracBrowser for help on using the repository browser.

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