1 /* libxml2 - Library for parsing XML documents
2 * Copyright (C) 2006-2019 Free Software Foundation, Inc.
3 *
4 * This file is not part of the GNU gettext program, but is used with
5 * GNU gettext.
6 *
7 * The original copyright notice is as follows:
8 */
9
10 /*
11 * Copyright (C) 1998-2012 Daniel Veillard. All Rights Reserved.
12 *
13 * Permission is hereby granted, free of charge, to any person obtaining a copy
14 * of this software and associated documentation files (the "Software"), to deal
15 * in the Software without restriction, including without limitation the rights
16 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
17 * copies of the Software, and to permit persons to whom the Software is fur-
18 * nished to do so, subject to the following conditions:
19 *
20 * The above copyright notice and this permission notice shall be included in
21 * all copies or substantial portions of the Software.
22 *
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
24 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FIT-
25 * NESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
26 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
27 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
28 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
29 * THE SOFTWARE.
30 *
31 * Daniel Veillard <veillard@redhat.com>
32 */
33
34 /*
35 * regexp.c: generic and extensible Regular Expression engine
36 *
37 * Basically designed with the purpose of compiling regexps for
38 * the variety of validation/shemas mechanisms now available in
39 * XML related specifications these include:
40 * - XML-1.0 DTD validation
41 * - XML Schemas structure part 1
42 * - XML Schemas Datatypes part 2 especially Appendix F
43 * - RELAX-NG/TREX i.e. the counter proposal
44 */
45
46 #define IN_LIBXML
47 #include "libxml.h"
48
49 #ifdef LIBXML_REGEXP_ENABLED
50
51 /* #define DEBUG_ERR */
52
53 #include <stdio.h>
54 #include <string.h>
55 #ifdef HAVE_LIMITS_H
56 #include <limits.h>
57 #endif
58
59 #include <libxml/tree.h>
60 #include <libxml/parserInternals.h>
61 #include <libxml/xmlregexp.h>
62 #include <libxml/xmlautomata.h>
63 #include <libxml/xmlunicode.h>
64
65 #ifndef INT_MAX
66 #define INT_MAX 123456789 /* easy to flag and big enough for our needs */
67 #endif
68
69 /* #define DEBUG_REGEXP_GRAPH */
70 /* #define DEBUG_REGEXP_EXEC */
71 /* #define DEBUG_PUSH */
72 /* #define DEBUG_COMPACTION */
73
74 #define MAX_PUSH 10000000
75
76 #ifdef ERROR
77 #undef ERROR
78 #endif
79 #define ERROR(str) \
80 ctxt->error = XML_REGEXP_COMPILE_ERROR; \
81 xmlRegexpErrCompile(ctxt, str);
82 #define NEXT ctxt->cur++
83 #define CUR (*(ctxt->cur))
84 #define NXT(index) (ctxt->cur[index])
85
86 #define CUR_SCHAR(s, l) xmlStringCurrentChar(NULL, s, &l)
87 #define NEXTL(l) ctxt->cur += l;
88 #define XML_REG_STRING_SEPARATOR '|'
89 /*
90 * Need PREV to check on a '-' within a Character Group. May only be used
91 * when it's guaranteed that cur is not at the beginning of ctxt->string!
92 */
93 #define PREV (ctxt->cur[-1])
94
95 /**
96 * TODO:
97 *
98 * macro to flag unimplemented blocks
99 */
100 #define TODO \
101 xmlGenericError(xmlGenericErrorContext, \
102 "Unimplemented block at %s:%d\n", \
103 __FILE__, __LINE__);
104
105 /************************************************************************
106 * *
107 * Datatypes and structures *
108 * *
109 ************************************************************************/
110
111 /*
112 * Note: the order of the enums below is significant, do not shuffle
113 */
114 typedef enum {
115 XML_REGEXP_EPSILON = 1,
116 XML_REGEXP_CHARVAL,
117 XML_REGEXP_RANGES,
118 XML_REGEXP_SUBREG, /* used for () sub regexps */
119 XML_REGEXP_STRING,
120 XML_REGEXP_ANYCHAR, /* . */
121 XML_REGEXP_ANYSPACE, /* \s */
122 XML_REGEXP_NOTSPACE, /* \S */
123 XML_REGEXP_INITNAME, /* \l */
124 XML_REGEXP_NOTINITNAME, /* \L */
125 XML_REGEXP_NAMECHAR, /* \c */
126 XML_REGEXP_NOTNAMECHAR, /* \C */
127 XML_REGEXP_DECIMAL, /* \d */
128 XML_REGEXP_NOTDECIMAL, /* \D */
129 XML_REGEXP_REALCHAR, /* \w */
130 XML_REGEXP_NOTREALCHAR, /* \W */
131 XML_REGEXP_LETTER = 100,
132 XML_REGEXP_LETTER_UPPERCASE,
133 XML_REGEXP_LETTER_LOWERCASE,
134 XML_REGEXP_LETTER_TITLECASE,
135 XML_REGEXP_LETTER_MODIFIER,
136 XML_REGEXP_LETTER_OTHERS,
137 XML_REGEXP_MARK,
138 XML_REGEXP_MARK_NONSPACING,
139 XML_REGEXP_MARK_SPACECOMBINING,
140 XML_REGEXP_MARK_ENCLOSING,
141 XML_REGEXP_NUMBER,
142 XML_REGEXP_NUMBER_DECIMAL,
143 XML_REGEXP_NUMBER_LETTER,
144 XML_REGEXP_NUMBER_OTHERS,
145 XML_REGEXP_PUNCT,
146 XML_REGEXP_PUNCT_CONNECTOR,
147 XML_REGEXP_PUNCT_DASH,
148 XML_REGEXP_PUNCT_OPEN,
149 XML_REGEXP_PUNCT_CLOSE,
150 XML_REGEXP_PUNCT_INITQUOTE,
151 XML_REGEXP_PUNCT_FINQUOTE,
152 XML_REGEXP_PUNCT_OTHERS,
153 XML_REGEXP_SEPAR,
154 XML_REGEXP_SEPAR_SPACE,
155 XML_REGEXP_SEPAR_LINE,
156 XML_REGEXP_SEPAR_PARA,
157 XML_REGEXP_SYMBOL,
158 XML_REGEXP_SYMBOL_MATH,
159 XML_REGEXP_SYMBOL_CURRENCY,
160 XML_REGEXP_SYMBOL_MODIFIER,
161 XML_REGEXP_SYMBOL_OTHERS,
162 XML_REGEXP_OTHER,
163 XML_REGEXP_OTHER_CONTROL,
164 XML_REGEXP_OTHER_FORMAT,
165 XML_REGEXP_OTHER_PRIVATE,
166 XML_REGEXP_OTHER_NA,
167 XML_REGEXP_BLOCK_NAME
168 } xmlRegAtomType;
169
170 typedef enum {
171 XML_REGEXP_QUANT_EPSILON = 1,
172 XML_REGEXP_QUANT_ONCE,
173 XML_REGEXP_QUANT_OPT,
174 XML_REGEXP_QUANT_MULT,
175 XML_REGEXP_QUANT_PLUS,
176 XML_REGEXP_QUANT_ONCEONLY,
177 XML_REGEXP_QUANT_ALL,
178 XML_REGEXP_QUANT_RANGE
179 } xmlRegQuantType;
180
181 typedef enum {
182 XML_REGEXP_START_STATE = 1,
183 XML_REGEXP_FINAL_STATE,
184 XML_REGEXP_TRANS_STATE,
185 XML_REGEXP_SINK_STATE,
186 XML_REGEXP_UNREACH_STATE
187 } xmlRegStateType;
188
189 typedef enum {
190 XML_REGEXP_MARK_NORMAL = 0,
191 XML_REGEXP_MARK_START,
192 XML_REGEXP_MARK_VISITED
193 } xmlRegMarkedType;
194
195 typedef struct _xmlRegRange xmlRegRange;
196 typedef xmlRegRange *xmlRegRangePtr;
197
198 struct _xmlRegRange {
199 int neg; /* 0 normal, 1 not, 2 exclude */
200 xmlRegAtomType type;
201 int start;
202 int end;
203 xmlChar *blockName;
204 };
205
206 typedef struct _xmlRegAtom xmlRegAtom;
207 typedef xmlRegAtom *xmlRegAtomPtr;
208
209 typedef struct _xmlAutomataState xmlRegState;
210 typedef xmlRegState *xmlRegStatePtr;
211
212 struct _xmlRegAtom {
213 int no;
214 xmlRegAtomType type;
215 xmlRegQuantType quant;
216 int min;
217 int max;
218
219 void *valuep;
220 void *valuep2;
221 int neg;
222 int codepoint;
223 xmlRegStatePtr start;
224 xmlRegStatePtr start0;
225 xmlRegStatePtr stop;
226 int maxRanges;
227 int nbRanges;
228 xmlRegRangePtr *ranges;
229 void *data;
230 };
231
232 typedef struct _xmlRegCounter xmlRegCounter;
233 typedef xmlRegCounter *xmlRegCounterPtr;
234
235 struct _xmlRegCounter {
236 int min;
237 int max;
238 };
239
240 typedef struct _xmlRegTrans xmlRegTrans;
241 typedef xmlRegTrans *xmlRegTransPtr;
242
243 struct _xmlRegTrans {
244 xmlRegAtomPtr atom;
245 int to;
246 int counter;
247 int count;
248 int nd;
249 };
250
251 struct _xmlAutomataState {
252 xmlRegStateType type;
253 xmlRegMarkedType mark;
254 xmlRegMarkedType markd;
255 xmlRegMarkedType reached;
256 int no;
257 int maxTrans;
258 int nbTrans;
259 xmlRegTrans *trans;
260 /* knowing states ponting to us can speed things up */
261 int maxTransTo;
262 int nbTransTo;
263 int *transTo;
264 };
265
266 typedef struct _xmlAutomata xmlRegParserCtxt;
267 typedef xmlRegParserCtxt *xmlRegParserCtxtPtr;
268
269 #define AM_AUTOMATA_RNG 1
270
271 struct _xmlAutomata {
272 xmlChar *string;
273 xmlChar *cur;
274
275 int error;
276 int neg;
277
278 xmlRegStatePtr start;
279 xmlRegStatePtr end;
280 xmlRegStatePtr state;
281
282 xmlRegAtomPtr atom;
283
284 int maxAtoms;
285 int nbAtoms;
286 xmlRegAtomPtr *atoms;
287
288 int maxStates;
289 int nbStates;
290 xmlRegStatePtr *states;
291
292 int maxCounters;
293 int nbCounters;
294 xmlRegCounter *counters;
295
296 int determinist;
297 int negs;
298 int flags;
299 };
300
301 struct _xmlRegexp {
302 xmlChar *string;
303 int nbStates;
304 xmlRegStatePtr *states;
305 int nbAtoms;
306 xmlRegAtomPtr *atoms;
307 int nbCounters;
308 xmlRegCounter *counters;
309 int determinist;
310 int flags;
311 /*
312 * That's the compact form for determinists automatas
313 */
314 int nbstates;
315 int *compact;
316 void **transdata;
317 int nbstrings;
318 xmlChar **stringMap;
319 };
320
321 typedef struct _xmlRegExecRollback xmlRegExecRollback;
322 typedef xmlRegExecRollback *xmlRegExecRollbackPtr;
323
324 struct _xmlRegExecRollback {
325 xmlRegStatePtr state;/* the current state */
326 int index; /* the index in the input stack */
327 int nextbranch; /* the next transition to explore in that state */
328 int *counts; /* save the automata state if it has some */
329 };
330
331 typedef struct _xmlRegInputToken xmlRegInputToken;
332 typedef xmlRegInputToken *xmlRegInputTokenPtr;
333
334 struct _xmlRegInputToken {
335 xmlChar *value;
336 void *data;
337 };
338
339 struct _xmlRegExecCtxt {
340 int status; /* execution status != 0 indicate an error */
341 int determinist; /* did we find an indeterministic behaviour */
342 xmlRegexpPtr comp; /* the compiled regexp */
343 xmlRegExecCallbacks callback;
344 void *data;
345
346 xmlRegStatePtr state;/* the current state */
347 int transno; /* the current transition on that state */
348 int transcount; /* the number of chars in char counted transitions */
349
350 /*
351 * A stack of rollback states
352 */
353 int maxRollbacks;
354 int nbRollbacks;
355 xmlRegExecRollback *rollbacks;
356
357 /*
358 * The state of the automata if any
359 */
360 int *counts;
361
362 /*
363 * The input stack
364 */
365 int inputStackMax;
366 int inputStackNr;
367 int index;
368 int *charStack;
369 const xmlChar *inputString; /* when operating on characters */
370 xmlRegInputTokenPtr inputStack;/* when operating on strings */
371
372 /*
373 * error handling
374 */
375 int errStateNo; /* the error state number */
376 xmlRegStatePtr errState; /* the error state */
377 xmlChar *errString; /* the string raising the error */
378 int *errCounts; /* counters at the error state */
379 int nbPush;
380 };
381
382 #define REGEXP_ALL_COUNTER 0x123456
383 #define REGEXP_ALL_LAX_COUNTER 0x123457
384
385 static void xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top);
386 static void xmlRegFreeState(xmlRegStatePtr state);
387 static void xmlRegFreeAtom(xmlRegAtomPtr atom);
388 static int xmlRegStrEqualWildcard(const xmlChar *expStr, const xmlChar *valStr);
389 static int xmlRegCheckCharacter(xmlRegAtomPtr atom, int codepoint);
390 static int xmlRegCheckCharacterRange(xmlRegAtomType type, int codepoint,
391 int neg, int start, int end, const xmlChar *blockName);
392
393 void xmlAutomataSetFlags(xmlAutomataPtr am, int flags);
394
395 /************************************************************************
396 * *
397 * Regexp memory error handler *
398 * *
399 ************************************************************************/
400 /**
401 * xmlRegexpErrMemory:
402 * @extra: extra information
403 *
404 * Handle an out of memory condition
405 */
406 static void
xmlRegexpErrMemory(xmlRegParserCtxtPtr ctxt,const char * extra)407 xmlRegexpErrMemory(xmlRegParserCtxtPtr ctxt, const char *extra)
408 {
409 const char *regexp = NULL;
410 if (ctxt != NULL) {
411 regexp = (const char *) ctxt->string;
412 ctxt->error = XML_ERR_NO_MEMORY;
413 }
414 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_REGEXP,
415 XML_ERR_NO_MEMORY, XML_ERR_FATAL, NULL, 0, extra,
416 regexp, NULL, 0, 0,
417 "Memory allocation failed : %s\n", extra);
418 }
419
420 /**
421 * xmlRegexpErrCompile:
422 * @extra: extra information
423 *
424 * Handle a compilation failure
425 */
426 static void
xmlRegexpErrCompile(xmlRegParserCtxtPtr ctxt,const char * extra)427 xmlRegexpErrCompile(xmlRegParserCtxtPtr ctxt, const char *extra)
428 {
429 const char *regexp = NULL;
430 int idx = 0;
431
432 if (ctxt != NULL) {
433 regexp = (const char *) ctxt->string;
434 idx = ctxt->cur - ctxt->string;
435 ctxt->error = XML_REGEXP_COMPILE_ERROR;
436 }
437 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_REGEXP,
438 XML_REGEXP_COMPILE_ERROR, XML_ERR_FATAL, NULL, 0, extra,
439 regexp, NULL, idx, 0,
440 "failed to compile: %s\n", extra);
441 }
442
443 /************************************************************************
444 * *
445 * Allocation/Deallocation *
446 * *
447 ************************************************************************/
448
449 static int xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt);
450 /**
451 * xmlRegEpxFromParse:
452 * @ctxt: the parser context used to build it
453 *
454 * Allocate a new regexp and fill it with the result from the parser
455 *
456 * Returns the new regexp or NULL in case of error
457 */
458 static xmlRegexpPtr
xmlRegEpxFromParse(xmlRegParserCtxtPtr ctxt)459 xmlRegEpxFromParse(xmlRegParserCtxtPtr ctxt) {
460 xmlRegexpPtr ret;
461
462 ret = (xmlRegexpPtr) xmlMalloc(sizeof(xmlRegexp));
463 if (ret == NULL) {
464 xmlRegexpErrMemory(ctxt, "compiling regexp");
465 return(NULL);
466 }
467 memset(ret, 0, sizeof(xmlRegexp));
468 ret->string = ctxt->string;
469 ret->nbStates = ctxt->nbStates;
470 ret->states = ctxt->states;
471 ret->nbAtoms = ctxt->nbAtoms;
472 ret->atoms = ctxt->atoms;
473 ret->nbCounters = ctxt->nbCounters;
474 ret->counters = ctxt->counters;
475 ret->determinist = ctxt->determinist;
476 ret->flags = ctxt->flags;
477 if (ret->determinist == -1) {
478 xmlRegexpIsDeterminist(ret);
479 }
480
481 if ((ret->determinist != 0) &&
482 (ret->nbCounters == 0) &&
483 (ctxt->negs == 0) &&
484 (ret->atoms != NULL) &&
485 (ret->atoms[0] != NULL) &&
486 (ret->atoms[0]->type == XML_REGEXP_STRING)) {
487 int i, j, nbstates = 0, nbatoms = 0;
488 int *stateRemap;
489 int *stringRemap;
490 int *transitions;
491 void **transdata;
492 xmlChar **stringMap;
493 xmlChar *value;
494
495 /*
496 * Switch to a compact representation
497 * 1/ counting the effective number of states left
498 * 2/ counting the unique number of atoms, and check that
499 * they are all of the string type
500 * 3/ build a table state x atom for the transitions
501 */
502
503 stateRemap = xmlMalloc(ret->nbStates * sizeof(int));
504 if (stateRemap == NULL) {
505 xmlRegexpErrMemory(ctxt, "compiling regexp");
506 xmlFree(ret);
507 return(NULL);
508 }
509 for (i = 0;i < ret->nbStates;i++) {
510 if (ret->states[i] != NULL) {
511 stateRemap[i] = nbstates;
512 nbstates++;
513 } else {
514 stateRemap[i] = -1;
515 }
516 }
517 #ifdef DEBUG_COMPACTION
518 printf("Final: %d states\n", nbstates);
519 #endif
520 stringMap = xmlMalloc(ret->nbAtoms * sizeof(char *));
521 if (stringMap == NULL) {
522 xmlRegexpErrMemory(ctxt, "compiling regexp");
523 xmlFree(stateRemap);
524 xmlFree(ret);
525 return(NULL);
526 }
527 stringRemap = xmlMalloc(ret->nbAtoms * sizeof(int));
528 if (stringRemap == NULL) {
529 xmlRegexpErrMemory(ctxt, "compiling regexp");
530 xmlFree(stringMap);
531 xmlFree(stateRemap);
532 xmlFree(ret);
533 return(NULL);
534 }
535 for (i = 0;i < ret->nbAtoms;i++) {
536 if ((ret->atoms[i]->type == XML_REGEXP_STRING) &&
537 (ret->atoms[i]->quant == XML_REGEXP_QUANT_ONCE)) {
538 value = ret->atoms[i]->valuep;
539 for (j = 0;j < nbatoms;j++) {
540 if (xmlStrEqual(stringMap[j], value)) {
541 stringRemap[i] = j;
542 break;
543 }
544 }
545 if (j >= nbatoms) {
546 stringRemap[i] = nbatoms;
547 stringMap[nbatoms] = xmlStrdup(value);
548 if (stringMap[nbatoms] == NULL) {
549 for (i = 0;i < nbatoms;i++)
550 xmlFree(stringMap[i]);
551 xmlFree(stringRemap);
552 xmlFree(stringMap);
553 xmlFree(stateRemap);
554 xmlFree(ret);
555 return(NULL);
556 }
557 nbatoms++;
558 }
559 } else {
560 xmlFree(stateRemap);
561 xmlFree(stringRemap);
562 for (i = 0;i < nbatoms;i++)
563 xmlFree(stringMap[i]);
564 xmlFree(stringMap);
565 xmlFree(ret);
566 return(NULL);
567 }
568 }
569 #ifdef DEBUG_COMPACTION
570 printf("Final: %d atoms\n", nbatoms);
571 #endif
572 transitions = (int *) xmlMalloc((nbstates + 1) *
573 (nbatoms + 1) * sizeof(int));
574 if (transitions == NULL) {
575 xmlFree(stateRemap);
576 xmlFree(stringRemap);
577 xmlFree(stringMap);
578 xmlFree(ret);
579 return(NULL);
580 }
581 memset(transitions, 0, (nbstates + 1) * (nbatoms + 1) * sizeof(int));
582
583 /*
584 * Allocate the transition table. The first entry for each
585 * state corresponds to the state type.
586 */
587 transdata = NULL;
588
589 for (i = 0;i < ret->nbStates;i++) {
590 int stateno, atomno, targetno, prev;
591 xmlRegStatePtr state;
592 xmlRegTransPtr trans;
593
594 stateno = stateRemap[i];
595 if (stateno == -1)
596 continue;
597 state = ret->states[i];
598
599 transitions[stateno * (nbatoms + 1)] = state->type;
600
601 for (j = 0;j < state->nbTrans;j++) {
602 trans = &(state->trans[j]);
603 if ((trans->to == -1) || (trans->atom == NULL))
604 continue;
605 atomno = stringRemap[trans->atom->no];
606 if ((trans->atom->data != NULL) && (transdata == NULL)) {
607 transdata = (void **) xmlMalloc(nbstates * nbatoms *
608 sizeof(void *));
609 if (transdata != NULL)
610 memset(transdata, 0,
611 nbstates * nbatoms * sizeof(void *));
612 else {
613 xmlRegexpErrMemory(ctxt, "compiling regexp");
614 break;
615 }
616 }
617 targetno = stateRemap[trans->to];
618 /*
619 * if the same atom can generate transitions to 2 different
620 * states then it means the automata is not determinist and
621 * the compact form can't be used !
622 */
623 prev = transitions[stateno * (nbatoms + 1) + atomno + 1];
624 if (prev != 0) {
625 if (prev != targetno + 1) {
626 ret->determinist = 0;
627 #ifdef DEBUG_COMPACTION
628 printf("Indet: state %d trans %d, atom %d to %d : %d to %d\n",
629 i, j, trans->atom->no, trans->to, atomno, targetno);
630 printf(" previous to is %d\n", prev);
631 #endif
632 if (transdata != NULL)
633 xmlFree(transdata);
634 xmlFree(transitions);
635 xmlFree(stateRemap);
636 xmlFree(stringRemap);
637 for (i = 0;i < nbatoms;i++)
638 xmlFree(stringMap[i]);
639 xmlFree(stringMap);
640 goto not_determ;
641 }
642 } else {
643 #if 0
644 printf("State %d trans %d: atom %d to %d : %d to %d\n",
645 i, j, trans->atom->no, trans->to, atomno, targetno);
646 #endif
647 transitions[stateno * (nbatoms + 1) + atomno + 1] =
648 targetno + 1; /* to avoid 0 */
649 if (transdata != NULL)
650 transdata[stateno * nbatoms + atomno] =
651 trans->atom->data;
652 }
653 }
654 }
655 ret->determinist = 1;
656 #ifdef DEBUG_COMPACTION
657 /*
658 * Debug
659 */
660 for (i = 0;i < nbstates;i++) {
661 for (j = 0;j < nbatoms + 1;j++) {
662 printf("%02d ", transitions[i * (nbatoms + 1) + j]);
663 }
664 printf("\n");
665 }
666 printf("\n");
667 #endif
668 /*
669 * Cleanup of the old data
670 */
671 if (ret->states != NULL) {
672 for (i = 0;i < ret->nbStates;i++)
673 xmlRegFreeState(ret->states[i]);
674 xmlFree(ret->states);
675 }
676 ret->states = NULL;
677 ret->nbStates = 0;
678 if (ret->atoms != NULL) {
679 for (i = 0;i < ret->nbAtoms;i++)
680 xmlRegFreeAtom(ret->atoms[i]);
681 xmlFree(ret->atoms);
682 }
683 ret->atoms = NULL;
684 ret->nbAtoms = 0;
685
686 ret->compact = transitions;
687 ret->transdata = transdata;
688 ret->stringMap = stringMap;
689 ret->nbstrings = nbatoms;
690 ret->nbstates = nbstates;
691 xmlFree(stateRemap);
692 xmlFree(stringRemap);
693 }
694 not_determ:
695 ctxt->string = NULL;
696 ctxt->nbStates = 0;
697 ctxt->states = NULL;
698 ctxt->nbAtoms = 0;
699 ctxt->atoms = NULL;
700 ctxt->nbCounters = 0;
701 ctxt->counters = NULL;
702 return(ret);
703 }
704
705 /**
706 * xmlRegNewParserCtxt:
707 * @string: the string to parse
708 *
709 * Allocate a new regexp parser context
710 *
711 * Returns the new context or NULL in case of error
712 */
713 static xmlRegParserCtxtPtr
xmlRegNewParserCtxt(const xmlChar * string)714 xmlRegNewParserCtxt(const xmlChar *string) {
715 xmlRegParserCtxtPtr ret;
716
717 ret = (xmlRegParserCtxtPtr) xmlMalloc(sizeof(xmlRegParserCtxt));
718 if (ret == NULL)
719 return(NULL);
720 memset(ret, 0, sizeof(xmlRegParserCtxt));
721 if (string != NULL)
722 ret->string = xmlStrdup(string);
723 ret->cur = ret->string;
724 ret->neg = 0;
725 ret->negs = 0;
726 ret->error = 0;
727 ret->determinist = -1;
728 return(ret);
729 }
730
731 /**
732 * xmlRegNewRange:
733 * @ctxt: the regexp parser context
734 * @neg: is that negative
735 * @type: the type of range
736 * @start: the start codepoint
737 * @end: the end codepoint
738 *
739 * Allocate a new regexp range
740 *
741 * Returns the new range or NULL in case of error
742 */
743 static xmlRegRangePtr
xmlRegNewRange(xmlRegParserCtxtPtr ctxt,int neg,xmlRegAtomType type,int start,int end)744 xmlRegNewRange(xmlRegParserCtxtPtr ctxt,
745 int neg, xmlRegAtomType type, int start, int end) {
746 xmlRegRangePtr ret;
747
748 ret = (xmlRegRangePtr) xmlMalloc(sizeof(xmlRegRange));
749 if (ret == NULL) {
750 xmlRegexpErrMemory(ctxt, "allocating range");
751 return(NULL);
752 }
753 ret->neg = neg;
754 ret->type = type;
755 ret->start = start;
756 ret->end = end;
757 return(ret);
758 }
759
760 /**
761 * xmlRegFreeRange:
762 * @range: the regexp range
763 *
764 * Free a regexp range
765 */
766 static void
xmlRegFreeRange(xmlRegRangePtr range)767 xmlRegFreeRange(xmlRegRangePtr range) {
768 if (range == NULL)
769 return;
770
771 if (range->blockName != NULL)
772 xmlFree(range->blockName);
773 xmlFree(range);
774 }
775
776 /**
777 * xmlRegCopyRange:
778 * @range: the regexp range
779 *
780 * Copy a regexp range
781 *
782 * Returns the new copy or NULL in case of error.
783 */
784 static xmlRegRangePtr
xmlRegCopyRange(xmlRegParserCtxtPtr ctxt,xmlRegRangePtr range)785 xmlRegCopyRange(xmlRegParserCtxtPtr ctxt, xmlRegRangePtr range) {
786 xmlRegRangePtr ret;
787
788 if (range == NULL)
789 return(NULL);
790
791 ret = xmlRegNewRange(ctxt, range->neg, range->type, range->start,
792 range->end);
793 if (ret == NULL)
794 return(NULL);
795 if (range->blockName != NULL) {
796 ret->blockName = xmlStrdup(range->blockName);
797 if (ret->blockName == NULL) {
798 xmlRegexpErrMemory(ctxt, "allocating range");
799 xmlRegFreeRange(ret);
800 return(NULL);
801 }
802 }
803 return(ret);
804 }
805
806 /**
807 * xmlRegNewAtom:
808 * @ctxt: the regexp parser context
809 * @type: the type of atom
810 *
811 * Allocate a new atom
812 *
813 * Returns the new atom or NULL in case of error
814 */
815 static xmlRegAtomPtr
xmlRegNewAtom(xmlRegParserCtxtPtr ctxt,xmlRegAtomType type)816 xmlRegNewAtom(xmlRegParserCtxtPtr ctxt, xmlRegAtomType type) {
817 xmlRegAtomPtr ret;
818
819 ret = (xmlRegAtomPtr) xmlMalloc(sizeof(xmlRegAtom));
820 if (ret == NULL) {
821 xmlRegexpErrMemory(ctxt, "allocating atom");
822 return(NULL);
823 }
824 memset(ret, 0, sizeof(xmlRegAtom));
825 ret->type = type;
826 ret->quant = XML_REGEXP_QUANT_ONCE;
827 ret->min = 0;
828 ret->max = 0;
829 return(ret);
830 }
831
832 /**
833 * xmlRegFreeAtom:
834 * @atom: the regexp atom
835 *
836 * Free a regexp atom
837 */
838 static void
xmlRegFreeAtom(xmlRegAtomPtr atom)839 xmlRegFreeAtom(xmlRegAtomPtr atom) {
840 int i;
841
842 if (atom == NULL)
843 return;
844
845 for (i = 0;i < atom->nbRanges;i++)
846 xmlRegFreeRange(atom->ranges[i]);
847 if (atom->ranges != NULL)
848 xmlFree(atom->ranges);
849 if ((atom->type == XML_REGEXP_STRING) && (atom->valuep != NULL))
850 xmlFree(atom->valuep);
851 if ((atom->type == XML_REGEXP_STRING) && (atom->valuep2 != NULL))
852 xmlFree(atom->valuep2);
853 if ((atom->type == XML_REGEXP_BLOCK_NAME) && (atom->valuep != NULL))
854 xmlFree(atom->valuep);
855 xmlFree(atom);
856 }
857
858 /**
859 * xmlRegCopyAtom:
860 * @ctxt: the regexp parser context
861 * @atom: the oiginal atom
862 *
863 * Allocate a new regexp range
864 *
865 * Returns the new atom or NULL in case of error
866 */
867 static xmlRegAtomPtr
xmlRegCopyAtom(xmlRegParserCtxtPtr ctxt,xmlRegAtomPtr atom)868 xmlRegCopyAtom(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom) {
869 xmlRegAtomPtr ret;
870
871 ret = (xmlRegAtomPtr) xmlMalloc(sizeof(xmlRegAtom));
872 if (ret == NULL) {
873 xmlRegexpErrMemory(ctxt, "copying atom");
874 return(NULL);
875 }
876 memset(ret, 0, sizeof(xmlRegAtom));
877 ret->type = atom->type;
878 ret->quant = atom->quant;
879 ret->min = atom->min;
880 ret->max = atom->max;
881 if (atom->nbRanges > 0) {
882 int i;
883
884 ret->ranges = (xmlRegRangePtr *) xmlMalloc(sizeof(xmlRegRangePtr) *
885 atom->nbRanges);
886 if (ret->ranges == NULL) {
887 xmlRegexpErrMemory(ctxt, "copying atom");
888 goto error;
889 }
890 for (i = 0;i < atom->nbRanges;i++) {
891 ret->ranges[i] = xmlRegCopyRange(ctxt, atom->ranges[i]);
892 if (ret->ranges[i] == NULL)
893 goto error;
894 ret->nbRanges = i + 1;
895 }
896 }
897 return(ret);
898
899 error:
900 xmlRegFreeAtom(ret);
901 return(NULL);
902 }
903
904 static xmlRegStatePtr
xmlRegNewState(xmlRegParserCtxtPtr ctxt)905 xmlRegNewState(xmlRegParserCtxtPtr ctxt) {
906 xmlRegStatePtr ret;
907
908 ret = (xmlRegStatePtr) xmlMalloc(sizeof(xmlRegState));
909 if (ret == NULL) {
910 xmlRegexpErrMemory(ctxt, "allocating state");
911 return(NULL);
912 }
913 memset(ret, 0, sizeof(xmlRegState));
914 ret->type = XML_REGEXP_TRANS_STATE;
915 ret->mark = XML_REGEXP_MARK_NORMAL;
916 return(ret);
917 }
918
919 /**
920 * xmlRegFreeState:
921 * @state: the regexp state
922 *
923 * Free a regexp state
924 */
925 static void
xmlRegFreeState(xmlRegStatePtr state)926 xmlRegFreeState(xmlRegStatePtr state) {
927 if (state == NULL)
928 return;
929
930 if (state->trans != NULL)
931 xmlFree(state->trans);
932 if (state->transTo != NULL)
933 xmlFree(state->transTo);
934 xmlFree(state);
935 }
936
937 /**
938 * xmlRegFreeParserCtxt:
939 * @ctxt: the regexp parser context
940 *
941 * Free a regexp parser context
942 */
943 static void
xmlRegFreeParserCtxt(xmlRegParserCtxtPtr ctxt)944 xmlRegFreeParserCtxt(xmlRegParserCtxtPtr ctxt) {
945 int i;
946 if (ctxt == NULL)
947 return;
948
949 if (ctxt->string != NULL)
950 xmlFree(ctxt->string);
951 if (ctxt->states != NULL) {
952 for (i = 0;i < ctxt->nbStates;i++)
953 xmlRegFreeState(ctxt->states[i]);
954 xmlFree(ctxt->states);
955 }
956 if (ctxt->atoms != NULL) {
957 for (i = 0;i < ctxt->nbAtoms;i++)
958 xmlRegFreeAtom(ctxt->atoms[i]);
959 xmlFree(ctxt->atoms);
960 }
961 if (ctxt->counters != NULL)
962 xmlFree(ctxt->counters);
963 xmlFree(ctxt);
964 }
965
966 /************************************************************************
967 * *
968 * Display of Data structures *
969 * *
970 ************************************************************************/
971
972 static void
xmlRegPrintAtomType(FILE * output,xmlRegAtomType type)973 xmlRegPrintAtomType(FILE *output, xmlRegAtomType type) {
974 switch (type) {
975 case XML_REGEXP_EPSILON:
976 fprintf(output, "epsilon "); break;
977 case XML_REGEXP_CHARVAL:
978 fprintf(output, "charval "); break;
979 case XML_REGEXP_RANGES:
980 fprintf(output, "ranges "); break;
981 case XML_REGEXP_SUBREG:
982 fprintf(output, "subexpr "); break;
983 case XML_REGEXP_STRING:
984 fprintf(output, "string "); break;
985 case XML_REGEXP_ANYCHAR:
986 fprintf(output, "anychar "); break;
987 case XML_REGEXP_ANYSPACE:
988 fprintf(output, "anyspace "); break;
989 case XML_REGEXP_NOTSPACE:
990 fprintf(output, "notspace "); break;
991 case XML_REGEXP_INITNAME:
992 fprintf(output, "initname "); break;
993 case XML_REGEXP_NOTINITNAME:
994 fprintf(output, "notinitname "); break;
995 case XML_REGEXP_NAMECHAR:
996 fprintf(output, "namechar "); break;
997 case XML_REGEXP_NOTNAMECHAR:
998 fprintf(output, "notnamechar "); break;
999 case XML_REGEXP_DECIMAL:
1000 fprintf(output, "decimal "); break;
1001 case XML_REGEXP_NOTDECIMAL:
1002 fprintf(output, "notdecimal "); break;
1003 case XML_REGEXP_REALCHAR:
1004 fprintf(output, "realchar "); break;
1005 case XML_REGEXP_NOTREALCHAR:
1006 fprintf(output, "notrealchar "); break;
1007 case XML_REGEXP_LETTER:
1008 fprintf(output, "LETTER "); break;
1009 case XML_REGEXP_LETTER_UPPERCASE:
1010 fprintf(output, "LETTER_UPPERCASE "); break;
1011 case XML_REGEXP_LETTER_LOWERCASE:
1012 fprintf(output, "LETTER_LOWERCASE "); break;
1013 case XML_REGEXP_LETTER_TITLECASE:
1014 fprintf(output, "LETTER_TITLECASE "); break;
1015 case XML_REGEXP_LETTER_MODIFIER:
1016 fprintf(output, "LETTER_MODIFIER "); break;
1017 case XML_REGEXP_LETTER_OTHERS:
1018 fprintf(output, "LETTER_OTHERS "); break;
1019 case XML_REGEXP_MARK:
1020 fprintf(output, "MARK "); break;
1021 case XML_REGEXP_MARK_NONSPACING:
1022 fprintf(output, "MARK_NONSPACING "); break;
1023 case XML_REGEXP_MARK_SPACECOMBINING:
1024 fprintf(output, "MARK_SPACECOMBINING "); break;
1025 case XML_REGEXP_MARK_ENCLOSING:
1026 fprintf(output, "MARK_ENCLOSING "); break;
1027 case XML_REGEXP_NUMBER:
1028 fprintf(output, "NUMBER "); break;
1029 case XML_REGEXP_NUMBER_DECIMAL:
1030 fprintf(output, "NUMBER_DECIMAL "); break;
1031 case XML_REGEXP_NUMBER_LETTER:
1032 fprintf(output, "NUMBER_LETTER "); break;
1033 case XML_REGEXP_NUMBER_OTHERS:
1034 fprintf(output, "NUMBER_OTHERS "); break;
1035 case XML_REGEXP_PUNCT:
1036 fprintf(output, "PUNCT "); break;
1037 case XML_REGEXP_PUNCT_CONNECTOR:
1038 fprintf(output, "PUNCT_CONNECTOR "); break;
1039 case XML_REGEXP_PUNCT_DASH:
1040 fprintf(output, "PUNCT_DASH "); break;
1041 case XML_REGEXP_PUNCT_OPEN:
1042 fprintf(output, "PUNCT_OPEN "); break;
1043 case XML_REGEXP_PUNCT_CLOSE:
1044 fprintf(output, "PUNCT_CLOSE "); break;
1045 case XML_REGEXP_PUNCT_INITQUOTE:
1046 fprintf(output, "PUNCT_INITQUOTE "); break;
1047 case XML_REGEXP_PUNCT_FINQUOTE:
1048 fprintf(output, "PUNCT_FINQUOTE "); break;
1049 case XML_REGEXP_PUNCT_OTHERS:
1050 fprintf(output, "PUNCT_OTHERS "); break;
1051 case XML_REGEXP_SEPAR:
1052 fprintf(output, "SEPAR "); break;
1053 case XML_REGEXP_SEPAR_SPACE:
1054 fprintf(output, "SEPAR_SPACE "); break;
1055 case XML_REGEXP_SEPAR_LINE:
1056 fprintf(output, "SEPAR_LINE "); break;
1057 case XML_REGEXP_SEPAR_PARA:
1058 fprintf(output, "SEPAR_PARA "); break;
1059 case XML_REGEXP_SYMBOL:
1060 fprintf(output, "SYMBOL "); break;
1061 case XML_REGEXP_SYMBOL_MATH:
1062 fprintf(output, "SYMBOL_MATH "); break;
1063 case XML_REGEXP_SYMBOL_CURRENCY:
1064 fprintf(output, "SYMBOL_CURRENCY "); break;
1065 case XML_REGEXP_SYMBOL_MODIFIER:
1066 fprintf(output, "SYMBOL_MODIFIER "); break;
1067 case XML_REGEXP_SYMBOL_OTHERS:
1068 fprintf(output, "SYMBOL_OTHERS "); break;
1069 case XML_REGEXP_OTHER:
1070 fprintf(output, "OTHER "); break;
1071 case XML_REGEXP_OTHER_CONTROL:
1072 fprintf(output, "OTHER_CONTROL "); break;
1073 case XML_REGEXP_OTHER_FORMAT:
1074 fprintf(output, "OTHER_FORMAT "); break;
1075 case XML_REGEXP_OTHER_PRIVATE:
1076 fprintf(output, "OTHER_PRIVATE "); break;
1077 case XML_REGEXP_OTHER_NA:
1078 fprintf(output, "OTHER_NA "); break;
1079 case XML_REGEXP_BLOCK_NAME:
1080 fprintf(output, "BLOCK "); break;
1081 }
1082 }
1083
1084 static void
xmlRegPrintQuantType(FILE * output,xmlRegQuantType type)1085 xmlRegPrintQuantType(FILE *output, xmlRegQuantType type) {
1086 switch (type) {
1087 case XML_REGEXP_QUANT_EPSILON:
1088 fprintf(output, "epsilon "); break;
1089 case XML_REGEXP_QUANT_ONCE:
1090 fprintf(output, "once "); break;
1091 case XML_REGEXP_QUANT_OPT:
1092 fprintf(output, "? "); break;
1093 case XML_REGEXP_QUANT_MULT:
1094 fprintf(output, "* "); break;
1095 case XML_REGEXP_QUANT_PLUS:
1096 fprintf(output, "+ "); break;
1097 case XML_REGEXP_QUANT_RANGE:
1098 fprintf(output, "range "); break;
1099 case XML_REGEXP_QUANT_ONCEONLY:
1100 fprintf(output, "onceonly "); break;
1101 case XML_REGEXP_QUANT_ALL:
1102 fprintf(output, "all "); break;
1103 }
1104 }
1105 static void
xmlRegPrintRange(FILE * output,xmlRegRangePtr range)1106 xmlRegPrintRange(FILE *output, xmlRegRangePtr range) {
1107 fprintf(output, " range: ");
1108 if (range->neg)
1109 fprintf(output, "negative ");
1110 xmlRegPrintAtomType(output, range->type);
1111 fprintf(output, "%c - %c\n", range->start, range->end);
1112 }
1113
1114 static void
xmlRegPrintAtom(FILE * output,xmlRegAtomPtr atom)1115 xmlRegPrintAtom(FILE *output, xmlRegAtomPtr atom) {
1116 fprintf(output, " atom: ");
1117 if (atom == NULL) {
1118 fprintf(output, "NULL\n");
1119 return;
1120 }
1121 if (atom->neg)
1122 fprintf(output, "not ");
1123 xmlRegPrintAtomType(output, atom->type);
1124 xmlRegPrintQuantType(output, atom->quant);
1125 if (atom->quant == XML_REGEXP_QUANT_RANGE)
1126 fprintf(output, "%d-%d ", atom->min, atom->max);
1127 if (atom->type == XML_REGEXP_STRING)
1128 fprintf(output, "'%s' ", (char *) atom->valuep);
1129 if (atom->type == XML_REGEXP_CHARVAL)
1130 fprintf(output, "char %c\n", atom->codepoint);
1131 else if (atom->type == XML_REGEXP_RANGES) {
1132 int i;
1133 fprintf(output, "%d entries\n", atom->nbRanges);
1134 for (i = 0; i < atom->nbRanges;i++)
1135 xmlRegPrintRange(output, atom->ranges[i]);
1136 } else if (atom->type == XML_REGEXP_SUBREG) {
1137 fprintf(output, "start %d end %d\n", atom->start->no, atom->stop->no);
1138 } else {
1139 fprintf(output, "\n");
1140 }
1141 }
1142
1143 static void
xmlRegPrintTrans(FILE * output,xmlRegTransPtr trans)1144 xmlRegPrintTrans(FILE *output, xmlRegTransPtr trans) {
1145 fprintf(output, " trans: ");
1146 if (trans == NULL) {
1147 fprintf(output, "NULL\n");
1148 return;
1149 }
1150 if (trans->to < 0) {
1151 fprintf(output, "removed\n");
1152 return;
1153 }
1154 if (trans->nd != 0) {
1155 if (trans->nd == 2)
1156 fprintf(output, "last not determinist, ");
1157 else
1158 fprintf(output, "not determinist, ");
1159 }
1160 if (trans->counter >= 0) {
1161 fprintf(output, "counted %d, ", trans->counter);
1162 }
1163 if (trans->count == REGEXP_ALL_COUNTER) {
1164 fprintf(output, "all transition, ");
1165 } else if (trans->count >= 0) {
1166 fprintf(output, "count based %d, ", trans->count);
1167 }
1168 if (trans->atom == NULL) {
1169 fprintf(output, "epsilon to %d\n", trans->to);
1170 return;
1171 }
1172 if (trans->atom->type == XML_REGEXP_CHARVAL)
1173 fprintf(output, "char %c ", trans->atom->codepoint);
1174 fprintf(output, "atom %d, to %d\n", trans->atom->no, trans->to);
1175 }
1176
1177 static void
xmlRegPrintState(FILE * output,xmlRegStatePtr state)1178 xmlRegPrintState(FILE *output, xmlRegStatePtr state) {
1179 int i;
1180
1181 fprintf(output, " state: ");
1182 if (state == NULL) {
1183 fprintf(output, "NULL\n");
1184 return;
1185 }
1186 if (state->type == XML_REGEXP_START_STATE)
1187 fprintf(output, "START ");
1188 if (state->type == XML_REGEXP_FINAL_STATE)
1189 fprintf(output, "FINAL ");
1190
1191 fprintf(output, "%d, %d transitions:\n", state->no, state->nbTrans);
1192 for (i = 0;i < state->nbTrans; i++) {
1193 xmlRegPrintTrans(output, &(state->trans[i]));
1194 }
1195 }
1196
1197 #ifdef DEBUG_REGEXP_GRAPH
1198 static void
xmlRegPrintCtxt(FILE * output,xmlRegParserCtxtPtr ctxt)1199 xmlRegPrintCtxt(FILE *output, xmlRegParserCtxtPtr ctxt) {
1200 int i;
1201
1202 fprintf(output, " ctxt: ");
1203 if (ctxt == NULL) {
1204 fprintf(output, "NULL\n");
1205 return;
1206 }
1207 fprintf(output, "'%s' ", ctxt->string);
1208 if (ctxt->error)
1209 fprintf(output, "error ");
1210 if (ctxt->neg)
1211 fprintf(output, "neg ");
1212 fprintf(output, "\n");
1213 fprintf(output, "%d atoms:\n", ctxt->nbAtoms);
1214 for (i = 0;i < ctxt->nbAtoms; i++) {
1215 fprintf(output, " %02d ", i);
1216 xmlRegPrintAtom(output, ctxt->atoms[i]);
1217 }
1218 if (ctxt->atom != NULL) {
1219 fprintf(output, "current atom:\n");
1220 xmlRegPrintAtom(output, ctxt->atom);
1221 }
1222 fprintf(output, "%d states:", ctxt->nbStates);
1223 if (ctxt->start != NULL)
1224 fprintf(output, " start: %d", ctxt->start->no);
1225 if (ctxt->end != NULL)
1226 fprintf(output, " end: %d", ctxt->end->no);
1227 fprintf(output, "\n");
1228 for (i = 0;i < ctxt->nbStates; i++) {
1229 xmlRegPrintState(output, ctxt->states[i]);
1230 }
1231 fprintf(output, "%d counters:\n", ctxt->nbCounters);
1232 for (i = 0;i < ctxt->nbCounters; i++) {
1233 fprintf(output, " %d: min %d max %d\n", i, ctxt->counters[i].min,
1234 ctxt->counters[i].max);
1235 }
1236 }
1237 #endif
1238
1239 /************************************************************************
1240 * *
1241 * Finite Automata structures manipulations *
1242 * *
1243 ************************************************************************/
1244
1245 static void
xmlRegAtomAddRange(xmlRegParserCtxtPtr ctxt,xmlRegAtomPtr atom,int neg,xmlRegAtomType type,int start,int end,xmlChar * blockName)1246 xmlRegAtomAddRange(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom,
1247 int neg, xmlRegAtomType type, int start, int end,
1248 xmlChar *blockName) {
1249 xmlRegRangePtr range;
1250
1251 if (atom == NULL) {
1252 ERROR("add range: atom is NULL");
1253 return;
1254 }
1255 if (atom->type != XML_REGEXP_RANGES) {
1256 ERROR("add range: atom is not ranges");
1257 return;
1258 }
1259 if (atom->maxRanges == 0) {
1260 atom->maxRanges = 4;
1261 atom->ranges = (xmlRegRangePtr *) xmlMalloc(atom->maxRanges *
1262 sizeof(xmlRegRangePtr));
1263 if (atom->ranges == NULL) {
1264 xmlRegexpErrMemory(ctxt, "adding ranges");
1265 atom->maxRanges = 0;
1266 return;
1267 }
1268 } else if (atom->nbRanges >= atom->maxRanges) {
1269 xmlRegRangePtr *tmp;
1270 atom->maxRanges *= 2;
1271 tmp = (xmlRegRangePtr *) xmlRealloc(atom->ranges, atom->maxRanges *
1272 sizeof(xmlRegRangePtr));
1273 if (tmp == NULL) {
1274 xmlRegexpErrMemory(ctxt, "adding ranges");
1275 atom->maxRanges /= 2;
1276 return;
1277 }
1278 atom->ranges = tmp;
1279 }
1280 range = xmlRegNewRange(ctxt, neg, type, start, end);
1281 if (range == NULL)
1282 return;
1283 range->blockName = blockName;
1284 atom->ranges[atom->nbRanges++] = range;
1285
1286 }
1287
1288 static int
xmlRegGetCounter(xmlRegParserCtxtPtr ctxt)1289 xmlRegGetCounter(xmlRegParserCtxtPtr ctxt) {
1290 if (ctxt->maxCounters == 0) {
1291 ctxt->maxCounters = 4;
1292 ctxt->counters = (xmlRegCounter *) xmlMalloc(ctxt->maxCounters *
1293 sizeof(xmlRegCounter));
1294 if (ctxt->counters == NULL) {
1295 xmlRegexpErrMemory(ctxt, "allocating counter");
1296 ctxt->maxCounters = 0;
1297 return(-1);
1298 }
1299 } else if (ctxt->nbCounters >= ctxt->maxCounters) {
1300 xmlRegCounter *tmp;
1301 ctxt->maxCounters *= 2;
1302 tmp = (xmlRegCounter *) xmlRealloc(ctxt->counters, ctxt->maxCounters *
1303 sizeof(xmlRegCounter));
1304 if (tmp == NULL) {
1305 xmlRegexpErrMemory(ctxt, "allocating counter");
1306 ctxt->maxCounters /= 2;
1307 return(-1);
1308 }
1309 ctxt->counters = tmp;
1310 }
1311 ctxt->counters[ctxt->nbCounters].min = -1;
1312 ctxt->counters[ctxt->nbCounters].max = -1;
1313 return(ctxt->nbCounters++);
1314 }
1315
1316 static int
xmlRegAtomPush(xmlRegParserCtxtPtr ctxt,xmlRegAtomPtr atom)1317 xmlRegAtomPush(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom) {
1318 if (atom == NULL) {
1319 ERROR("atom push: atom is NULL");
1320 return(-1);
1321 }
1322 if (ctxt->maxAtoms == 0) {
1323 ctxt->maxAtoms = 4;
1324 ctxt->atoms = (xmlRegAtomPtr *) xmlMalloc(ctxt->maxAtoms *
1325 sizeof(xmlRegAtomPtr));
1326 if (ctxt->atoms == NULL) {
1327 xmlRegexpErrMemory(ctxt, "pushing atom");
1328 ctxt->maxAtoms = 0;
1329 return(-1);
1330 }
1331 } else if (ctxt->nbAtoms >= ctxt->maxAtoms) {
1332 xmlRegAtomPtr *tmp;
1333 ctxt->maxAtoms *= 2;
1334 tmp = (xmlRegAtomPtr *) xmlRealloc(ctxt->atoms, ctxt->maxAtoms *
1335 sizeof(xmlRegAtomPtr));
1336 if (tmp == NULL) {
1337 xmlRegexpErrMemory(ctxt, "allocating counter");
1338 ctxt->maxAtoms /= 2;
1339 return(-1);
1340 }
1341 ctxt->atoms = tmp;
1342 }
1343 atom->no = ctxt->nbAtoms;
1344 ctxt->atoms[ctxt->nbAtoms++] = atom;
1345 return(0);
1346 }
1347
1348 static void
xmlRegStateAddTransTo(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr target,int from)1349 xmlRegStateAddTransTo(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr target,
1350 int from) {
1351 if (target->maxTransTo == 0) {
1352 target->maxTransTo = 8;
1353 target->transTo = (int *) xmlMalloc(target->maxTransTo *
1354 sizeof(int));
1355 if (target->transTo == NULL) {
1356 xmlRegexpErrMemory(ctxt, "adding transition");
1357 target->maxTransTo = 0;
1358 return;
1359 }
1360 } else if (target->nbTransTo >= target->maxTransTo) {
1361 int *tmp;
1362 target->maxTransTo *= 2;
1363 tmp = (int *) xmlRealloc(target->transTo, target->maxTransTo *
1364 sizeof(int));
1365 if (tmp == NULL) {
1366 xmlRegexpErrMemory(ctxt, "adding transition");
1367 target->maxTransTo /= 2;
1368 return;
1369 }
1370 target->transTo = tmp;
1371 }
1372 target->transTo[target->nbTransTo] = from;
1373 target->nbTransTo++;
1374 }
1375
1376 static void
xmlRegStateAddTrans(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr state,xmlRegAtomPtr atom,xmlRegStatePtr target,int counter,int count)1377 xmlRegStateAddTrans(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
1378 xmlRegAtomPtr atom, xmlRegStatePtr target,
1379 int counter, int count) {
1380
1381 int nrtrans;
1382
1383 if (state == NULL) {
1384 ERROR("add state: state is NULL");
1385 return;
1386 }
1387 if (target == NULL) {
1388 ERROR("add state: target is NULL");
1389 return;
1390 }
1391 /*
1392 * Other routines follow the philosophy 'When in doubt, add a transition'
1393 * so we check here whether such a transition is already present and, if
1394 * so, silently ignore this request.
1395 */
1396
1397 for (nrtrans = state->nbTrans - 1; nrtrans >= 0; nrtrans--) {
1398 xmlRegTransPtr trans = &(state->trans[nrtrans]);
1399 if ((trans->atom == atom) &&
1400 (trans->to == target->no) &&
1401 (trans->counter == counter) &&
1402 (trans->count == count)) {
1403 #ifdef DEBUG_REGEXP_GRAPH
1404 printf("Ignoring duplicate transition from %d to %d\n",
1405 state->no, target->no);
1406 #endif
1407 return;
1408 }
1409 }
1410
1411 if (state->maxTrans == 0) {
1412 state->maxTrans = 8;
1413 state->trans = (xmlRegTrans *) xmlMalloc(state->maxTrans *
1414 sizeof(xmlRegTrans));
1415 if (state->trans == NULL) {
1416 xmlRegexpErrMemory(ctxt, "adding transition");
1417 state->maxTrans = 0;
1418 return;
1419 }
1420 } else if (state->nbTrans >= state->maxTrans) {
1421 xmlRegTrans *tmp;
1422 state->maxTrans *= 2;
1423 tmp = (xmlRegTrans *) xmlRealloc(state->trans, state->maxTrans *
1424 sizeof(xmlRegTrans));
1425 if (tmp == NULL) {
1426 xmlRegexpErrMemory(ctxt, "adding transition");
1427 state->maxTrans /= 2;
1428 return;
1429 }
1430 state->trans = tmp;
1431 }
1432 #ifdef DEBUG_REGEXP_GRAPH
1433 printf("Add trans from %d to %d ", state->no, target->no);
1434 if (count == REGEXP_ALL_COUNTER)
1435 printf("all transition\n");
1436 else if (count >= 0)
1437 printf("count based %d\n", count);
1438 else if (counter >= 0)
1439 printf("counted %d\n", counter);
1440 else if (atom == NULL)
1441 printf("epsilon transition\n");
1442 else if (atom != NULL)
1443 xmlRegPrintAtom(stdout, atom);
1444 #endif
1445
1446 state->trans[state->nbTrans].atom = atom;
1447 state->trans[state->nbTrans].to = target->no;
1448 state->trans[state->nbTrans].counter = counter;
1449 state->trans[state->nbTrans].count = count;
1450 state->trans[state->nbTrans].nd = 0;
1451 state->nbTrans++;
1452 xmlRegStateAddTransTo(ctxt, target, state->no);
1453 }
1454
1455 static int
xmlRegStatePush(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr state)1456 xmlRegStatePush(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state) {
1457 if (state == NULL) return(-1);
1458 if (ctxt->maxStates == 0) {
1459 ctxt->maxStates = 4;
1460 ctxt->states = (xmlRegStatePtr *) xmlMalloc(ctxt->maxStates *
1461 sizeof(xmlRegStatePtr));
1462 if (ctxt->states == NULL) {
1463 xmlRegexpErrMemory(ctxt, "adding state");
1464 ctxt->maxStates = 0;
1465 return(-1);
1466 }
1467 } else if (ctxt->nbStates >= ctxt->maxStates) {
1468 xmlRegStatePtr *tmp;
1469 ctxt->maxStates *= 2;
1470 tmp = (xmlRegStatePtr *) xmlRealloc(ctxt->states, ctxt->maxStates *
1471 sizeof(xmlRegStatePtr));
1472 if (tmp == NULL) {
1473 xmlRegexpErrMemory(ctxt, "adding state");
1474 ctxt->maxStates /= 2;
1475 return(-1);
1476 }
1477 ctxt->states = tmp;
1478 }
1479 state->no = ctxt->nbStates;
1480 ctxt->states[ctxt->nbStates++] = state;
1481 return(0);
1482 }
1483
1484 /**
1485 * xmlFAGenerateAllTransition:
1486 * @ctxt: a regexp parser context
1487 * @from: the from state
1488 * @to: the target state or NULL for building a new one
1489 * @lax:
1490 *
1491 */
1492 static void
xmlFAGenerateAllTransition(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr from,xmlRegStatePtr to,int lax)1493 xmlFAGenerateAllTransition(xmlRegParserCtxtPtr ctxt,
1494 xmlRegStatePtr from, xmlRegStatePtr to,
1495 int lax) {
1496 if (to == NULL) {
1497 to = xmlRegNewState(ctxt);
1498 xmlRegStatePush(ctxt, to);
1499 ctxt->state = to;
1500 }
1501 if (lax)
1502 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_LAX_COUNTER);
1503 else
1504 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_COUNTER);
1505 }
1506
1507 /**
1508 * xmlFAGenerateEpsilonTransition:
1509 * @ctxt: a regexp parser context
1510 * @from: the from state
1511 * @to: the target state or NULL for building a new one
1512 *
1513 */
1514 static void
xmlFAGenerateEpsilonTransition(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr from,xmlRegStatePtr to)1515 xmlFAGenerateEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1516 xmlRegStatePtr from, xmlRegStatePtr to) {
1517 if (to == NULL) {
1518 to = xmlRegNewState(ctxt);
1519 xmlRegStatePush(ctxt, to);
1520 ctxt->state = to;
1521 }
1522 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, -1);
1523 }
1524
1525 /**
1526 * xmlFAGenerateCountedEpsilonTransition:
1527 * @ctxt: a regexp parser context
1528 * @from: the from state
1529 * @to: the target state or NULL for building a new one
1530 * counter: the counter for that transition
1531 *
1532 */
1533 static void
xmlFAGenerateCountedEpsilonTransition(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr from,xmlRegStatePtr to,int counter)1534 xmlFAGenerateCountedEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1535 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1536 if (to == NULL) {
1537 to = xmlRegNewState(ctxt);
1538 xmlRegStatePush(ctxt, to);
1539 ctxt->state = to;
1540 }
1541 xmlRegStateAddTrans(ctxt, from, NULL, to, counter, -1);
1542 }
1543
1544 /**
1545 * xmlFAGenerateCountedTransition:
1546 * @ctxt: a regexp parser context
1547 * @from: the from state
1548 * @to: the target state or NULL for building a new one
1549 * counter: the counter for that transition
1550 *
1551 */
1552 static void
xmlFAGenerateCountedTransition(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr from,xmlRegStatePtr to,int counter)1553 xmlFAGenerateCountedTransition(xmlRegParserCtxtPtr ctxt,
1554 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1555 if (to == NULL) {
1556 to = xmlRegNewState(ctxt);
1557 xmlRegStatePush(ctxt, to);
1558 ctxt->state = to;
1559 }
1560 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, counter);
1561 }
1562
1563 /**
1564 * xmlFAGenerateTransitions:
1565 * @ctxt: a regexp parser context
1566 * @from: the from state
1567 * @to: the target state or NULL for building a new one
1568 * @atom: the atom generating the transition
1569 *
1570 * Returns 0 if success and -1 in case of error.
1571 */
1572 static int
xmlFAGenerateTransitions(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr from,xmlRegStatePtr to,xmlRegAtomPtr atom)1573 xmlFAGenerateTransitions(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr from,
1574 xmlRegStatePtr to, xmlRegAtomPtr atom) {
1575 xmlRegStatePtr end;
1576 int nullable = 0;
1577
1578 if (atom == NULL) {
1579 ERROR("genrate transition: atom == NULL");
1580 return(-1);
1581 }
1582 if (atom->type == XML_REGEXP_SUBREG) {
1583 /*
1584 * this is a subexpression handling one should not need to
1585 * create a new node except for XML_REGEXP_QUANT_RANGE.
1586 */
1587 if (xmlRegAtomPush(ctxt, atom) < 0) {
1588 return(-1);
1589 }
1590 if ((to != NULL) && (atom->stop != to) &&
1591 (atom->quant != XML_REGEXP_QUANT_RANGE)) {
1592 /*
1593 * Generate an epsilon transition to link to the target
1594 */
1595 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, to);
1596 #ifdef DV
1597 } else if ((to == NULL) && (atom->quant != XML_REGEXP_QUANT_RANGE) &&
1598 (atom->quant != XML_REGEXP_QUANT_ONCE)) {
1599 to = xmlRegNewState(ctxt);
1600 xmlRegStatePush(ctxt, to);
1601 ctxt->state = to;
1602 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, to);
1603 #endif
1604 }
1605 switch (atom->quant) {
1606 case XML_REGEXP_QUANT_OPT:
1607 atom->quant = XML_REGEXP_QUANT_ONCE;
1608 /*
1609 * transition done to the state after end of atom.
1610 * 1. set transition from atom start to new state
1611 * 2. set transition from atom end to this state.
1612 */
1613 if (to == NULL) {
1614 xmlFAGenerateEpsilonTransition(ctxt, atom->start, 0);
1615 xmlFAGenerateEpsilonTransition(ctxt, atom->stop,
1616 ctxt->state);
1617 } else {
1618 xmlFAGenerateEpsilonTransition(ctxt, atom->start, to);
1619 }
1620 break;
1621 case XML_REGEXP_QUANT_MULT:
1622 atom->quant = XML_REGEXP_QUANT_ONCE;
1623 xmlFAGenerateEpsilonTransition(ctxt, atom->start, atom->stop);
1624 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1625 break;
1626 case XML_REGEXP_QUANT_PLUS:
1627 atom->quant = XML_REGEXP_QUANT_ONCE;
1628 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1629 break;
1630 case XML_REGEXP_QUANT_RANGE: {
1631 int counter;
1632 xmlRegStatePtr inter, newstate;
1633
1634 /*
1635 * create the final state now if needed
1636 */
1637 if (to != NULL) {
1638 newstate = to;
1639 } else {
1640 newstate = xmlRegNewState(ctxt);
1641 xmlRegStatePush(ctxt, newstate);
1642 }
1643
1644 /*
1645 * The principle here is to use counted transition
1646 * to avoid explosion in the number of states in the
1647 * graph. This is clearly more complex but should not
1648 * be exploitable at runtime.
1649 */
1650 if ((atom->min == 0) && (atom->start0 == NULL)) {
1651 xmlRegAtomPtr copy;
1652 /*
1653 * duplicate a transition based on atom to count next
1654 * occurences after 1. We cannot loop to atom->start
1655 * directly because we need an epsilon transition to
1656 * newstate.
1657 */
1658 /* ???? For some reason it seems we never reach that
1659 case, I suppose this got optimized out before when
1660 building the automata */
1661 copy = xmlRegCopyAtom(ctxt, atom);
1662 if (copy == NULL)
1663 return(-1);
1664 copy->quant = XML_REGEXP_QUANT_ONCE;
1665 copy->min = 0;
1666 copy->max = 0;
1667
1668 if (xmlFAGenerateTransitions(ctxt, atom->start, NULL, copy)
1669 < 0)
1670 return(-1);
1671 inter = ctxt->state;
1672 counter = xmlRegGetCounter(ctxt);
1673 ctxt->counters[counter].min = atom->min - 1;
1674 ctxt->counters[counter].max = atom->max - 1;
1675 /* count the number of times we see it again */
1676 xmlFAGenerateCountedEpsilonTransition(ctxt, inter,
1677 atom->stop, counter);
1678 /* allow a way out based on the count */
1679 xmlFAGenerateCountedTransition(ctxt, inter,
1680 newstate, counter);
1681 /* and also allow a direct exit for 0 */
1682 xmlFAGenerateEpsilonTransition(ctxt, atom->start,
1683 newstate);
1684 } else {
1685 /*
1686 * either we need the atom at least once or there
1687 * is an atom->start0 allowing to easilly plug the
1688 * epsilon transition.
1689 */
1690 counter = xmlRegGetCounter(ctxt);
1691 ctxt->counters[counter].min = atom->min - 1;
1692 ctxt->counters[counter].max = atom->max - 1;
1693 /* count the number of times we see it again */
1694 xmlFAGenerateCountedEpsilonTransition(ctxt, atom->stop,
1695 atom->start, counter);
1696 /* allow a way out based on the count */
1697 xmlFAGenerateCountedTransition(ctxt, atom->stop,
1698 newstate, counter);
1699 /* and if needed allow a direct exit for 0 */
1700 if (atom->min == 0)
1701 xmlFAGenerateEpsilonTransition(ctxt, atom->start0,
1702 newstate);
1703
1704 }
1705 atom->min = 0;
1706 atom->max = 0;
1707 atom->quant = XML_REGEXP_QUANT_ONCE;
1708 ctxt->state = newstate;
1709 }
1710 default:
1711 break;
1712 }
1713 return(0);
1714 }
1715 if ((atom->min == 0) && (atom->max == 0) &&
1716 (atom->quant == XML_REGEXP_QUANT_RANGE)) {
1717 /*
1718 * we can discard the atom and generate an epsilon transition instead
1719 */
1720 if (to == NULL) {
1721 to = xmlRegNewState(ctxt);
1722 if (to != NULL)
1723 xmlRegStatePush(ctxt, to);
1724 else {
1725 return(-1);
1726 }
1727 }
1728 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1729 ctxt->state = to;
1730 xmlRegFreeAtom(atom);
1731 return(0);
1732 }
1733 if (to == NULL) {
1734 to = xmlRegNewState(ctxt);
1735 if (to != NULL)
1736 xmlRegStatePush(ctxt, to);
1737 else {
1738 return(-1);
1739 }
1740 }
1741 end = to;
1742 if ((atom->quant == XML_REGEXP_QUANT_MULT) ||
1743 (atom->quant == XML_REGEXP_QUANT_PLUS)) {
1744 /*
1745 * Do not pollute the target state by adding transitions from
1746 * it as it is likely to be the shared target of multiple branches.
1747 * So isolate with an epsilon transition.
1748 */
1749 xmlRegStatePtr tmp;
1750
1751 tmp = xmlRegNewState(ctxt);
1752 if (tmp != NULL)
1753 xmlRegStatePush(ctxt, tmp);
1754 else {
1755 return(-1);
1756 }
1757 xmlFAGenerateEpsilonTransition(ctxt, tmp, to);
1758 to = tmp;
1759 }
1760 if (xmlRegAtomPush(ctxt, atom) < 0) {
1761 return(-1);
1762 }
1763 if ((atom->quant == XML_REGEXP_QUANT_RANGE) &&
1764 (atom->min == 0) && (atom->max > 0)) {
1765 nullable = 1;
1766 atom->min = 1;
1767 if (atom->max == 1)
1768 atom->quant = XML_REGEXP_QUANT_OPT;
1769 }
1770 xmlRegStateAddTrans(ctxt, from, atom, to, -1, -1);
1771 ctxt->state = end;
1772 switch (atom->quant) {
1773 case XML_REGEXP_QUANT_OPT:
1774 atom->quant = XML_REGEXP_QUANT_ONCE;
1775 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1776 break;
1777 case XML_REGEXP_QUANT_MULT:
1778 atom->quant = XML_REGEXP_QUANT_ONCE;
1779 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1780 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1781 break;
1782 case XML_REGEXP_QUANT_PLUS:
1783 atom->quant = XML_REGEXP_QUANT_ONCE;
1784 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1785 break;
1786 case XML_REGEXP_QUANT_RANGE:
1787 if (nullable)
1788 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1789 break;
1790 default:
1791 break;
1792 }
1793 return(0);
1794 }
1795
1796 /**
1797 * xmlFAReduceEpsilonTransitions:
1798 * @ctxt: a regexp parser context
1799 * @fromnr: the from state
1800 * @tonr: the to state
1801 * @counter: should that transition be associated to a counted
1802 *
1803 */
1804 static void
xmlFAReduceEpsilonTransitions(xmlRegParserCtxtPtr ctxt,int fromnr,int tonr,int counter)1805 xmlFAReduceEpsilonTransitions(xmlRegParserCtxtPtr ctxt, int fromnr,
1806 int tonr, int counter) {
1807 int transnr;
1808 xmlRegStatePtr from;
1809 xmlRegStatePtr to;
1810
1811 #ifdef DEBUG_REGEXP_GRAPH
1812 printf("xmlFAReduceEpsilonTransitions(%d, %d)\n", fromnr, tonr);
1813 #endif
1814 from = ctxt->states[fromnr];
1815 if (from == NULL)
1816 return;
1817 to = ctxt->states[tonr];
1818 if (to == NULL)
1819 return;
1820 if ((to->mark == XML_REGEXP_MARK_START) ||
1821 (to->mark == XML_REGEXP_MARK_VISITED))
1822 return;
1823
1824 to->mark = XML_REGEXP_MARK_VISITED;
1825 if (to->type == XML_REGEXP_FINAL_STATE) {
1826 #ifdef DEBUG_REGEXP_GRAPH
1827 printf("State %d is final, so %d becomes final\n", tonr, fromnr);
1828 #endif
1829 from->type = XML_REGEXP_FINAL_STATE;
1830 }
1831 for (transnr = 0;transnr < to->nbTrans;transnr++) {
1832 if (to->trans[transnr].to < 0)
1833 continue;
1834 if (to->trans[transnr].atom == NULL) {
1835 /*
1836 * Don't remove counted transitions
1837 * Don't loop either
1838 */
1839 if (to->trans[transnr].to != fromnr) {
1840 if (to->trans[transnr].count >= 0) {
1841 int newto = to->trans[transnr].to;
1842
1843 xmlRegStateAddTrans(ctxt, from, NULL,
1844 ctxt->states[newto],
1845 -1, to->trans[transnr].count);
1846 } else {
1847 #ifdef DEBUG_REGEXP_GRAPH
1848 printf("Found epsilon trans %d from %d to %d\n",
1849 transnr, tonr, to->trans[transnr].to);
1850 #endif
1851 if (to->trans[transnr].counter >= 0) {
1852 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1853 to->trans[transnr].to,
1854 to->trans[transnr].counter);
1855 } else {
1856 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1857 to->trans[transnr].to,
1858 counter);
1859 }
1860 }
1861 }
1862 } else {
1863 int newto = to->trans[transnr].to;
1864
1865 if (to->trans[transnr].counter >= 0) {
1866 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1867 ctxt->states[newto],
1868 to->trans[transnr].counter, -1);
1869 } else {
1870 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1871 ctxt->states[newto], counter, -1);
1872 }
1873 }
1874 }
1875 to->mark = XML_REGEXP_MARK_NORMAL;
1876 }
1877
1878 /**
1879 * xmlFAEliminateSimpleEpsilonTransitions:
1880 * @ctxt: a regexp parser context
1881 *
1882 * Eliminating general epsilon transitions can get costly in the general
1883 * algorithm due to the large amount of generated new transitions and
1884 * associated comparisons. However for simple epsilon transition used just
1885 * to separate building blocks when generating the automata this can be
1886 * reduced to state elimination:
1887 * - if there exists an epsilon from X to Y
1888 * - if there is no other transition from X
1889 * then X and Y are semantically equivalent and X can be eliminated
1890 * If X is the start state then make Y the start state, else replace the
1891 * target of all transitions to X by transitions to Y.
1892 */
1893 static void
xmlFAEliminateSimpleEpsilonTransitions(xmlRegParserCtxtPtr ctxt)1894 xmlFAEliminateSimpleEpsilonTransitions(xmlRegParserCtxtPtr ctxt) {
1895 int statenr, i, j, newto;
1896 xmlRegStatePtr state, tmp;
1897
1898 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1899 state = ctxt->states[statenr];
1900 if (state == NULL)
1901 continue;
1902 if (state->nbTrans != 1)
1903 continue;
1904 if (state->type == XML_REGEXP_UNREACH_STATE)
1905 continue;
1906 /* is the only transition out a basic transition */
1907 if ((state->trans[0].atom == NULL) &&
1908 (state->trans[0].to >= 0) &&
1909 (state->trans[0].to != statenr) &&
1910 (state->trans[0].counter < 0) &&
1911 (state->trans[0].count < 0)) {
1912 newto = state->trans[0].to;
1913
1914 if (state->type == XML_REGEXP_START_STATE) {
1915 #ifdef DEBUG_REGEXP_GRAPH
1916 printf("Found simple epsilon trans from start %d to %d\n",
1917 statenr, newto);
1918 #endif
1919 } else {
1920 #ifdef DEBUG_REGEXP_GRAPH
1921 printf("Found simple epsilon trans from %d to %d\n",
1922 statenr, newto);
1923 #endif
1924 for (i = 0;i < state->nbTransTo;i++) {
1925 tmp = ctxt->states[state->transTo[i]];
1926 for (j = 0;j < tmp->nbTrans;j++) {
1927 if (tmp->trans[j].to == statenr) {
1928 #ifdef DEBUG_REGEXP_GRAPH
1929 printf("Changed transition %d on %d to go to %d\n",
1930 j, tmp->no, newto);
1931 #endif
1932 tmp->trans[j].to = -1;
1933 xmlRegStateAddTrans(ctxt, tmp, tmp->trans[j].atom,
1934 ctxt->states[newto],
1935 tmp->trans[j].counter,
1936 tmp->trans[j].count);
1937 }
1938 }
1939 }
1940 if (state->type == XML_REGEXP_FINAL_STATE)
1941 ctxt->states[newto]->type = XML_REGEXP_FINAL_STATE;
1942 /* eliminate the transition completely */
1943 state->nbTrans = 0;
1944
1945 state->type = XML_REGEXP_UNREACH_STATE;
1946
1947 }
1948
1949 }
1950 }
1951 }
1952 /**
1953 * xmlFAEliminateEpsilonTransitions:
1954 * @ctxt: a regexp parser context
1955 *
1956 */
1957 static void
xmlFAEliminateEpsilonTransitions(xmlRegParserCtxtPtr ctxt)1958 xmlFAEliminateEpsilonTransitions(xmlRegParserCtxtPtr ctxt) {
1959 int statenr, transnr;
1960 xmlRegStatePtr state;
1961 int has_epsilon;
1962
1963 if (ctxt->states == NULL) return;
1964
1965 /*
1966 * Eliminate simple epsilon transition and the associated unreachable
1967 * states.
1968 */
1969 xmlFAEliminateSimpleEpsilonTransitions(ctxt);
1970 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1971 state = ctxt->states[statenr];
1972 if ((state != NULL) && (state->type == XML_REGEXP_UNREACH_STATE)) {
1973 #ifdef DEBUG_REGEXP_GRAPH
1974 printf("Removed unreachable state %d\n", statenr);
1975 #endif
1976 xmlRegFreeState(state);
1977 ctxt->states[statenr] = NULL;
1978 }
1979 }
1980
1981 has_epsilon = 0;
1982
1983 /*
1984 * Build the completed transitions bypassing the epsilons
1985 * Use a marking algorithm to avoid loops
1986 * Mark sink states too.
1987 * Process from the latests states backward to the start when
1988 * there is long cascading epsilon chains this minimize the
1989 * recursions and transition compares when adding the new ones
1990 */
1991 for (statenr = ctxt->nbStates - 1;statenr >= 0;statenr--) {
1992 state = ctxt->states[statenr];
1993 if (state == NULL)
1994 continue;
1995 if ((state->nbTrans == 0) &&
1996 (state->type != XML_REGEXP_FINAL_STATE)) {
1997 state->type = XML_REGEXP_SINK_STATE;
1998 }
1999 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2000 if ((state->trans[transnr].atom == NULL) &&
2001 (state->trans[transnr].to >= 0)) {
2002 if (state->trans[transnr].to == statenr) {
2003 state->trans[transnr].to = -1;
2004 #ifdef DEBUG_REGEXP_GRAPH
2005 printf("Removed loopback epsilon trans %d on %d\n",
2006 transnr, statenr);
2007 #endif
2008 } else if (state->trans[transnr].count < 0) {
2009 int newto = state->trans[transnr].to;
2010
2011 #ifdef DEBUG_REGEXP_GRAPH
2012 printf("Found epsilon trans %d from %d to %d\n",
2013 transnr, statenr, newto);
2014 #endif
2015 has_epsilon = 1;
2016 state->trans[transnr].to = -2;
2017 state->mark = XML_REGEXP_MARK_START;
2018 xmlFAReduceEpsilonTransitions(ctxt, statenr,
2019 newto, state->trans[transnr].counter);
2020 state->mark = XML_REGEXP_MARK_NORMAL;
2021 #ifdef DEBUG_REGEXP_GRAPH
2022 } else {
2023 printf("Found counted transition %d on %d\n",
2024 transnr, statenr);
2025 #endif
2026 }
2027 }
2028 }
2029 }
2030 /*
2031 * Eliminate the epsilon transitions
2032 */
2033 if (has_epsilon) {
2034 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2035 state = ctxt->states[statenr];
2036 if (state == NULL)
2037 continue;
2038 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2039 xmlRegTransPtr trans = &(state->trans[transnr]);
2040 if ((trans->atom == NULL) &&
2041 (trans->count < 0) &&
2042 (trans->to >= 0)) {
2043 trans->to = -1;
2044 }
2045 }
2046 }
2047 }
2048
2049 /*
2050 * Use this pass to detect unreachable states too
2051 */
2052 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2053 state = ctxt->states[statenr];
2054 if (state != NULL)
2055 state->reached = XML_REGEXP_MARK_NORMAL;
2056 }
2057 state = ctxt->states[0];
2058 if (state != NULL)
2059 state->reached = XML_REGEXP_MARK_START;
2060 while (state != NULL) {
2061 xmlRegStatePtr target = NULL;
2062 state->reached = XML_REGEXP_MARK_VISITED;
2063 /*
2064 * Mark all states reachable from the current reachable state
2065 */
2066 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2067 if ((state->trans[transnr].to >= 0) &&
2068 ((state->trans[transnr].atom != NULL) ||
2069 (state->trans[transnr].count >= 0))) {
2070 int newto = state->trans[transnr].to;
2071
2072 if (ctxt->states[newto] == NULL)
2073 continue;
2074 if (ctxt->states[newto]->reached == XML_REGEXP_MARK_NORMAL) {
2075 ctxt->states[newto]->reached = XML_REGEXP_MARK_START;
2076 target = ctxt->states[newto];
2077 }
2078 }
2079 }
2080
2081 /*
2082 * find the next accessible state not explored
2083 */
2084 if (target == NULL) {
2085 for (statenr = 1;statenr < ctxt->nbStates;statenr++) {
2086 state = ctxt->states[statenr];
2087 if ((state != NULL) && (state->reached ==
2088 XML_REGEXP_MARK_START)) {
2089 target = state;
2090 break;
2091 }
2092 }
2093 }
2094 state = target;
2095 }
2096 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2097 state = ctxt->states[statenr];
2098 if ((state != NULL) && (state->reached == XML_REGEXP_MARK_NORMAL)) {
2099 #ifdef DEBUG_REGEXP_GRAPH
2100 printf("Removed unreachable state %d\n", statenr);
2101 #endif
2102 xmlRegFreeState(state);
2103 ctxt->states[statenr] = NULL;
2104 }
2105 }
2106
2107 }
2108
2109 static int
xmlFACompareRanges(xmlRegRangePtr range1,xmlRegRangePtr range2)2110 xmlFACompareRanges(xmlRegRangePtr range1, xmlRegRangePtr range2) {
2111 int ret = 0;
2112
2113 if ((range1->type == XML_REGEXP_RANGES) ||
2114 (range2->type == XML_REGEXP_RANGES) ||
2115 (range2->type == XML_REGEXP_SUBREG) ||
2116 (range1->type == XML_REGEXP_SUBREG) ||
2117 (range1->type == XML_REGEXP_STRING) ||
2118 (range2->type == XML_REGEXP_STRING))
2119 return(-1);
2120
2121 /* put them in order */
2122 if (range1->type > range2->type) {
2123 xmlRegRangePtr tmp;
2124
2125 tmp = range1;
2126 range1 = range2;
2127 range2 = tmp;
2128 }
2129 if ((range1->type == XML_REGEXP_ANYCHAR) ||
2130 (range2->type == XML_REGEXP_ANYCHAR)) {
2131 ret = 1;
2132 } else if ((range1->type == XML_REGEXP_EPSILON) ||
2133 (range2->type == XML_REGEXP_EPSILON)) {
2134 return(0);
2135 } else if (range1->type == range2->type) {
2136 if (range1->type != XML_REGEXP_CHARVAL)
2137 ret = 1;
2138 else if ((range1->end < range2->start) ||
2139 (range2->end < range1->start))
2140 ret = 0;
2141 else
2142 ret = 1;
2143 } else if (range1->type == XML_REGEXP_CHARVAL) {
2144 int codepoint;
2145 int neg = 0;
2146
2147 /*
2148 * just check all codepoints in the range for acceptance,
2149 * this is usually way cheaper since done only once at
2150 * compilation than testing over and over at runtime or
2151 * pushing too many states when evaluating.
2152 */
2153 if (((range1->neg == 0) && (range2->neg != 0)) ||
2154 ((range1->neg != 0) && (range2->neg == 0)))
2155 neg = 1;
2156
2157 for (codepoint = range1->start;codepoint <= range1->end ;codepoint++) {
2158 ret = xmlRegCheckCharacterRange(range2->type, codepoint,
2159 0, range2->start, range2->end,
2160 range2->blockName);
2161 if (ret < 0)
2162 return(-1);
2163 if (((neg == 1) && (ret == 0)) ||
2164 ((neg == 0) && (ret == 1)))
2165 return(1);
2166 }
2167 return(0);
2168 } else if ((range1->type == XML_REGEXP_BLOCK_NAME) ||
2169 (range2->type == XML_REGEXP_BLOCK_NAME)) {
2170 if (range1->type == range2->type) {
2171 ret = xmlStrEqual(range1->blockName, range2->blockName);
2172 } else {
2173 /*
2174 * comparing a block range with anything else is way
2175 * too costly, and maintining the table is like too much
2176 * memory too, so let's force the automata to save state
2177 * here.
2178 */
2179 return(1);
2180 }
2181 } else if ((range1->type < XML_REGEXP_LETTER) ||
2182 (range2->type < XML_REGEXP_LETTER)) {
2183 if ((range1->type == XML_REGEXP_ANYSPACE) &&
2184 (range2->type == XML_REGEXP_NOTSPACE))
2185 ret = 0;
2186 else if ((range1->type == XML_REGEXP_INITNAME) &&
2187 (range2->type == XML_REGEXP_NOTINITNAME))
2188 ret = 0;
2189 else if ((range1->type == XML_REGEXP_NAMECHAR) &&
2190 (range2->type == XML_REGEXP_NOTNAMECHAR))
2191 ret = 0;
2192 else if ((range1->type == XML_REGEXP_DECIMAL) &&
2193 (range2->type == XML_REGEXP_NOTDECIMAL))
2194 ret = 0;
2195 else if ((range1->type == XML_REGEXP_REALCHAR) &&
2196 (range2->type == XML_REGEXP_NOTREALCHAR))
2197 ret = 0;
2198 else {
2199 /* same thing to limit complexity */
2200 return(1);
2201 }
2202 } else {
2203 ret = 0;
2204 /* range1->type < range2->type here */
2205 switch (range1->type) {
2206 case XML_REGEXP_LETTER:
2207 /* all disjoint except in the subgroups */
2208 if ((range2->type == XML_REGEXP_LETTER_UPPERCASE) ||
2209 (range2->type == XML_REGEXP_LETTER_LOWERCASE) ||
2210 (range2->type == XML_REGEXP_LETTER_TITLECASE) ||
2211 (range2->type == XML_REGEXP_LETTER_MODIFIER) ||
2212 (range2->type == XML_REGEXP_LETTER_OTHERS))
2213 ret = 1;
2214 break;
2215 case XML_REGEXP_MARK:
2216 if ((range2->type == XML_REGEXP_MARK_NONSPACING) ||
2217 (range2->type == XML_REGEXP_MARK_SPACECOMBINING) ||
2218 (range2->type == XML_REGEXP_MARK_ENCLOSING))
2219 ret = 1;
2220 break;
2221 case XML_REGEXP_NUMBER:
2222 if ((range2->type == XML_REGEXP_NUMBER_DECIMAL) ||
2223 (range2->type == XML_REGEXP_NUMBER_LETTER) ||
2224 (range2->type == XML_REGEXP_NUMBER_OTHERS))
2225 ret = 1;
2226 break;
2227 case XML_REGEXP_PUNCT:
2228 if ((range2->type == XML_REGEXP_PUNCT_CONNECTOR) ||
2229 (range2->type == XML_REGEXP_PUNCT_DASH) ||
2230 (range2->type == XML_REGEXP_PUNCT_OPEN) ||
2231 (range2->type == XML_REGEXP_PUNCT_CLOSE) ||
2232 (range2->type == XML_REGEXP_PUNCT_INITQUOTE) ||
2233 (range2->type == XML_REGEXP_PUNCT_FINQUOTE) ||
2234 (range2->type == XML_REGEXP_PUNCT_OTHERS))
2235 ret = 1;
2236 break;
2237 case XML_REGEXP_SEPAR:
2238 if ((range2->type == XML_REGEXP_SEPAR_SPACE) ||
2239 (range2->type == XML_REGEXP_SEPAR_LINE) ||
2240 (range2->type == XML_REGEXP_SEPAR_PARA))
2241 ret = 1;
2242 break;
2243 case XML_REGEXP_SYMBOL:
2244 if ((range2->type == XML_REGEXP_SYMBOL_MATH) ||
2245 (range2->type == XML_REGEXP_SYMBOL_CURRENCY) ||
2246 (range2->type == XML_REGEXP_SYMBOL_MODIFIER) ||
2247 (range2->type == XML_REGEXP_SYMBOL_OTHERS))
2248 ret = 1;
2249 break;
2250 case XML_REGEXP_OTHER:
2251 if ((range2->type == XML_REGEXP_OTHER_CONTROL) ||
2252 (range2->type == XML_REGEXP_OTHER_FORMAT) ||
2253 (range2->type == XML_REGEXP_OTHER_PRIVATE))
2254 ret = 1;
2255 break;
2256 default:
2257 if ((range2->type >= XML_REGEXP_LETTER) &&
2258 (range2->type < XML_REGEXP_BLOCK_NAME))
2259 ret = 0;
2260 else {
2261 /* safety net ! */
2262 return(1);
2263 }
2264 }
2265 }
2266 if (((range1->neg == 0) && (range2->neg != 0)) ||
2267 ((range1->neg != 0) && (range2->neg == 0)))
2268 ret = !ret;
2269 return(ret);
2270 }
2271
2272 /**
2273 * xmlFACompareAtomTypes:
2274 * @type1: an atom type
2275 * @type2: an atom type
2276 *
2277 * Compares two atoms type to check whether they intersect in some ways,
2278 * this is used by xmlFACompareAtoms only
2279 *
2280 * Returns 1 if they may intersect and 0 otherwise
2281 */
2282 static int
xmlFACompareAtomTypes(xmlRegAtomType type1,xmlRegAtomType type2)2283 xmlFACompareAtomTypes(xmlRegAtomType type1, xmlRegAtomType type2) {
2284 if ((type1 == XML_REGEXP_EPSILON) ||
2285 (type1 == XML_REGEXP_CHARVAL) ||
2286 (type1 == XML_REGEXP_RANGES) ||
2287 (type1 == XML_REGEXP_SUBREG) ||
2288 (type1 == XML_REGEXP_STRING) ||
2289 (type1 == XML_REGEXP_ANYCHAR))
2290 return(1);
2291 if ((type2 == XML_REGEXP_EPSILON) ||
2292 (type2 == XML_REGEXP_CHARVAL) ||
2293 (type2 == XML_REGEXP_RANGES) ||
2294 (type2 == XML_REGEXP_SUBREG) ||
2295 (type2 == XML_REGEXP_STRING) ||
2296 (type2 == XML_REGEXP_ANYCHAR))
2297 return(1);
2298
2299 if (type1 == type2) return(1);
2300
2301 /* simplify subsequent compares by making sure type1 < type2 */
2302 if (type1 > type2) {
2303 xmlRegAtomType tmp = type1;
2304 type1 = type2;
2305 type2 = tmp;
2306 }
2307 switch (type1) {
2308 case XML_REGEXP_ANYSPACE: /* \s */
2309 /* can't be a letter, number, mark, pontuation, symbol */
2310 if ((type2 == XML_REGEXP_NOTSPACE) ||
2311 ((type2 >= XML_REGEXP_LETTER) &&
2312 (type2 <= XML_REGEXP_LETTER_OTHERS)) ||
2313 ((type2 >= XML_REGEXP_NUMBER) &&
2314 (type2 <= XML_REGEXP_NUMBER_OTHERS)) ||
2315 ((type2 >= XML_REGEXP_MARK) &&
2316 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2317 ((type2 >= XML_REGEXP_PUNCT) &&
2318 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2319 ((type2 >= XML_REGEXP_SYMBOL) &&
2320 (type2 <= XML_REGEXP_SYMBOL_OTHERS))
2321 ) return(0);
2322 break;
2323 case XML_REGEXP_NOTSPACE: /* \S */
2324 break;
2325 case XML_REGEXP_INITNAME: /* \l */
2326 /* can't be a number, mark, separator, pontuation, symbol or other */
2327 if ((type2 == XML_REGEXP_NOTINITNAME) ||
2328 ((type2 >= XML_REGEXP_NUMBER) &&
2329 (type2 <= XML_REGEXP_NUMBER_OTHERS)) ||
2330 ((type2 >= XML_REGEXP_MARK) &&
2331 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2332 ((type2 >= XML_REGEXP_SEPAR) &&
2333 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2334 ((type2 >= XML_REGEXP_PUNCT) &&
2335 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2336 ((type2 >= XML_REGEXP_SYMBOL) &&
2337 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2338 ((type2 >= XML_REGEXP_OTHER) &&
2339 (type2 <= XML_REGEXP_OTHER_NA))
2340 ) return(0);
2341 break;
2342 case XML_REGEXP_NOTINITNAME: /* \L */
2343 break;
2344 case XML_REGEXP_NAMECHAR: /* \c */
2345 /* can't be a mark, separator, pontuation, symbol or other */
2346 if ((type2 == XML_REGEXP_NOTNAMECHAR) ||
2347 ((type2 >= XML_REGEXP_MARK) &&
2348 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2349 ((type2 >= XML_REGEXP_PUNCT) &&
2350 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2351 ((type2 >= XML_REGEXP_SEPAR) &&
2352 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2353 ((type2 >= XML_REGEXP_SYMBOL) &&
2354 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2355 ((type2 >= XML_REGEXP_OTHER) &&
2356 (type2 <= XML_REGEXP_OTHER_NA))
2357 ) return(0);
2358 break;
2359 case XML_REGEXP_NOTNAMECHAR: /* \C */
2360 break;
2361 case XML_REGEXP_DECIMAL: /* \d */
2362 /* can't be a letter, mark, separator, pontuation, symbol or other */
2363 if ((type2 == XML_REGEXP_NOTDECIMAL) ||
2364 (type2 == XML_REGEXP_REALCHAR) ||
2365 ((type2 >= XML_REGEXP_LETTER) &&
2366 (type2 <= XML_REGEXP_LETTER_OTHERS)) ||
2367 ((type2 >= XML_REGEXP_MARK) &&
2368 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2369 ((type2 >= XML_REGEXP_PUNCT) &&
2370 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2371 ((type2 >= XML_REGEXP_SEPAR) &&
2372 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2373 ((type2 >= XML_REGEXP_SYMBOL) &&
2374 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2375 ((type2 >= XML_REGEXP_OTHER) &&
2376 (type2 <= XML_REGEXP_OTHER_NA))
2377 )return(0);
2378 break;
2379 case XML_REGEXP_NOTDECIMAL: /* \D */
2380 break;
2381 case XML_REGEXP_REALCHAR: /* \w */
2382 /* can't be a mark, separator, pontuation, symbol or other */
2383 if ((type2 == XML_REGEXP_NOTDECIMAL) ||
2384 ((type2 >= XML_REGEXP_MARK) &&
2385 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2386 ((type2 >= XML_REGEXP_PUNCT) &&
2387 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2388 ((type2 >= XML_REGEXP_SEPAR) &&
2389 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2390 ((type2 >= XML_REGEXP_SYMBOL) &&
2391 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2392 ((type2 >= XML_REGEXP_OTHER) &&
2393 (type2 <= XML_REGEXP_OTHER_NA))
2394 )return(0);
2395 break;
2396 case XML_REGEXP_NOTREALCHAR: /* \W */
2397 break;
2398 /*
2399 * at that point we know both type 1 and type2 are from
2400 * character categories are ordered and are different,
2401 * it becomes simple because this is a partition
2402 */
2403 case XML_REGEXP_LETTER:
2404 if (type2 <= XML_REGEXP_LETTER_OTHERS)
2405 return(1);
2406 return(0);
2407 case XML_REGEXP_LETTER_UPPERCASE:
2408 case XML_REGEXP_LETTER_LOWERCASE:
2409 case XML_REGEXP_LETTER_TITLECASE:
2410 case XML_REGEXP_LETTER_MODIFIER:
2411 case XML_REGEXP_LETTER_OTHERS:
2412 return(0);
2413 case XML_REGEXP_MARK:
2414 if (type2 <= XML_REGEXP_MARK_ENCLOSING)
2415 return(1);
2416 return(0);
2417 case XML_REGEXP_MARK_NONSPACING:
2418 case XML_REGEXP_MARK_SPACECOMBINING:
2419 case XML_REGEXP_MARK_ENCLOSING:
2420 return(0);
2421 case XML_REGEXP_NUMBER:
2422 if (type2 <= XML_REGEXP_NUMBER_OTHERS)
2423 return(1);
2424 return(0);
2425 case XML_REGEXP_NUMBER_DECIMAL:
2426 case XML_REGEXP_NUMBER_LETTER:
2427 case XML_REGEXP_NUMBER_OTHERS:
2428 return(0);
2429 case XML_REGEXP_PUNCT:
2430 if (type2 <= XML_REGEXP_PUNCT_OTHERS)
2431 return(1);
2432 return(0);
2433 case XML_REGEXP_PUNCT_CONNECTOR:
2434 case XML_REGEXP_PUNCT_DASH:
2435 case XML_REGEXP_PUNCT_OPEN:
2436 case XML_REGEXP_PUNCT_CLOSE:
2437 case XML_REGEXP_PUNCT_INITQUOTE:
2438 case XML_REGEXP_PUNCT_FINQUOTE:
2439 case XML_REGEXP_PUNCT_OTHERS:
2440 return(0);
2441 case XML_REGEXP_SEPAR:
2442 if (type2 <= XML_REGEXP_SEPAR_PARA)
2443 return(1);
2444 return(0);
2445 case XML_REGEXP_SEPAR_SPACE:
2446 case XML_REGEXP_SEPAR_LINE:
2447 case XML_REGEXP_SEPAR_PARA:
2448 return(0);
2449 case XML_REGEXP_SYMBOL:
2450 if (type2 <= XML_REGEXP_SYMBOL_OTHERS)
2451 return(1);
2452 return(0);
2453 case XML_REGEXP_SYMBOL_MATH:
2454 case XML_REGEXP_SYMBOL_CURRENCY:
2455 case XML_REGEXP_SYMBOL_MODIFIER:
2456 case XML_REGEXP_SYMBOL_OTHERS:
2457 return(0);
2458 case XML_REGEXP_OTHER:
2459 if (type2 <= XML_REGEXP_OTHER_NA)
2460 return(1);
2461 return(0);
2462 case XML_REGEXP_OTHER_CONTROL:
2463 case XML_REGEXP_OTHER_FORMAT:
2464 case XML_REGEXP_OTHER_PRIVATE:
2465 case XML_REGEXP_OTHER_NA:
2466 return(0);
2467 default:
2468 break;
2469 }
2470 return(1);
2471 }
2472
2473 /**
2474 * xmlFAEqualAtoms:
2475 * @atom1: an atom
2476 * @atom2: an atom
2477 * @deep: if not set only compare string pointers
2478 *
2479 * Compares two atoms to check whether they are the same exactly
2480 * this is used to remove equivalent transitions
2481 *
2482 * Returns 1 if same and 0 otherwise
2483 */
2484 static int
xmlFAEqualAtoms(xmlRegAtomPtr atom1,xmlRegAtomPtr atom2,int deep)2485 xmlFAEqualAtoms(xmlRegAtomPtr atom1, xmlRegAtomPtr atom2, int deep) {
2486 int ret = 0;
2487
2488 if (atom1 == atom2)
2489 return(1);
2490 if ((atom1 == NULL) || (atom2 == NULL))
2491 return(0);
2492
2493 if (atom1->type != atom2->type)
2494 return(0);
2495 switch (atom1->type) {
2496 case XML_REGEXP_EPSILON:
2497 ret = 0;
2498 break;
2499 case XML_REGEXP_STRING:
2500 if (!deep)
2501 ret = (atom1->valuep == atom2->valuep);
2502 else
2503 ret = xmlStrEqual((xmlChar *)atom1->valuep,
2504 (xmlChar *)atom2->valuep);
2505 break;
2506 case XML_REGEXP_CHARVAL:
2507 ret = (atom1->codepoint == atom2->codepoint);
2508 break;
2509 case XML_REGEXP_RANGES:
2510 /* too hard to do in the general case */
2511 ret = 0;
2512 default:
2513 break;
2514 }
2515 return(ret);
2516 }
2517
2518 /**
2519 * xmlFACompareAtoms:
2520 * @atom1: an atom
2521 * @atom2: an atom
2522 * @deep: if not set only compare string pointers
2523 *
2524 * Compares two atoms to check whether they intersect in some ways,
2525 * this is used by xmlFAComputesDeterminism and xmlFARecurseDeterminism only
2526 *
2527 * Returns 1 if yes and 0 otherwise
2528 */
2529 static int
xmlFACompareAtoms(xmlRegAtomPtr atom1,xmlRegAtomPtr atom2,int deep)2530 xmlFACompareAtoms(xmlRegAtomPtr atom1, xmlRegAtomPtr atom2, int deep) {
2531 int ret = 1;
2532
2533 if (atom1 == atom2)
2534 return(1);
2535 if ((atom1 == NULL) || (atom2 == NULL))
2536 return(0);
2537
2538 if ((atom1->type == XML_REGEXP_ANYCHAR) ||
2539 (atom2->type == XML_REGEXP_ANYCHAR))
2540 return(1);
2541
2542 if (atom1->type > atom2->type) {
2543 xmlRegAtomPtr tmp;
2544 tmp = atom1;
2545 atom1 = atom2;
2546 atom2 = tmp;
2547 }
2548 if (atom1->type != atom2->type) {
2549 ret = xmlFACompareAtomTypes(atom1->type, atom2->type);
2550 /* if they can't intersect at the type level break now */
2551 if (ret == 0)
2552 return(0);
2553 }
2554 switch (atom1->type) {
2555 case XML_REGEXP_STRING:
2556 if (!deep)
2557 ret = (atom1->valuep != atom2->valuep);
2558 else
2559 ret = xmlRegStrEqualWildcard((xmlChar *)atom1->valuep,
2560 (xmlChar *)atom2->valuep);
2561 break;
2562 case XML_REGEXP_EPSILON:
2563 goto not_determinist;
2564 case XML_REGEXP_CHARVAL:
2565 if (atom2->type == XML_REGEXP_CHARVAL) {
2566 ret = (atom1->codepoint == atom2->codepoint);
2567 } else {
2568 ret = xmlRegCheckCharacter(atom2, atom1->codepoint);
2569 if (ret < 0)
2570 ret = 1;
2571 }
2572 break;
2573 case XML_REGEXP_RANGES:
2574 if (atom2->type == XML_REGEXP_RANGES) {
2575 int i, j, res;
2576 xmlRegRangePtr r1, r2;
2577
2578 /*
2579 * need to check that none of the ranges eventually matches
2580 */
2581 for (i = 0;i < atom1->nbRanges;i++) {
2582 for (j = 0;j < atom2->nbRanges;j++) {
2583 r1 = atom1->ranges[i];
2584 r2 = atom2->ranges[j];
2585 res = xmlFACompareRanges(r1, r2);
2586 if (res == 1) {
2587 ret = 1;
2588 goto done;
2589 }
2590 }
2591 }
2592 ret = 0;
2593 }
2594 break;
2595 default:
2596 goto not_determinist;
2597 }
2598 done:
2599 if (atom1->neg != atom2->neg) {
2600 ret = !ret;
2601 }
2602 if (ret == 0)
2603 return(0);
2604 not_determinist:
2605 return(1);
2606 }
2607
2608 /**
2609 * xmlFARecurseDeterminism:
2610 * @ctxt: a regexp parser context
2611 *
2612 * Check whether the associated regexp is determinist,
2613 * should be called after xmlFAEliminateEpsilonTransitions()
2614 *
2615 */
2616 static int
xmlFARecurseDeterminism(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr state,int to,xmlRegAtomPtr atom)2617 xmlFARecurseDeterminism(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
2618 int to, xmlRegAtomPtr atom) {
2619 int ret = 1;
2620 int res;
2621 int transnr, nbTrans;
2622 xmlRegTransPtr t1;
2623 int deep = 1;
2624
2625 if (state == NULL)
2626 return(ret);
2627 if (state->markd == XML_REGEXP_MARK_VISITED)
2628 return(ret);
2629
2630 if (ctxt->flags & AM_AUTOMATA_RNG)
2631 deep = 0;
2632
2633 /*
2634 * don't recurse on transitions potentially added in the course of
2635 * the elimination.
2636 */
2637 nbTrans = state->nbTrans;
2638 for (transnr = 0;transnr < nbTrans;transnr++) {
2639 t1 = &(state->trans[transnr]);
2640 /*
2641 * check transitions conflicting with the one looked at
2642 */
2643 if (t1->atom == NULL) {
2644 if (t1->to < 0)
2645 continue;
2646 state->markd = XML_REGEXP_MARK_VISITED;
2647 res = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
2648 to, atom);
2649 state->markd = 0;
2650 if (res == 0) {
2651 ret = 0;
2652 /* t1->nd = 1; */
2653 }
2654 continue;
2655 }
2656 if (t1->to != to)
2657 continue;
2658 if (xmlFACompareAtoms(t1->atom, atom, deep)) {
2659 ret = 0;
2660 /* mark the transition as non-deterministic */
2661 t1->nd = 1;
2662 }
2663 }
2664 return(ret);
2665 }
2666
2667 /**
2668 * xmlFAComputesDeterminism:
2669 * @ctxt: a regexp parser context
2670 *
2671 * Check whether the associated regexp is determinist,
2672 * should be called after xmlFAEliminateEpsilonTransitions()
2673 *
2674 */
2675 static int
xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt)2676 xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt) {
2677 int statenr, transnr;
2678 xmlRegStatePtr state;
2679 xmlRegTransPtr t1, t2, last;
2680 int i;
2681 int ret = 1;
2682 int deep = 1;
2683
2684 #ifdef DEBUG_REGEXP_GRAPH
2685 printf("xmlFAComputesDeterminism\n");
2686 xmlRegPrintCtxt(stdout, ctxt);
2687 #endif
2688 if (ctxt->determinist != -1)
2689 return(ctxt->determinist);
2690
2691 if (ctxt->flags & AM_AUTOMATA_RNG)
2692 deep = 0;
2693
2694 /*
2695 * First cleanup the automata removing cancelled transitions
2696 */
2697 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2698 state = ctxt->states[statenr];
2699 if (state == NULL)
2700 continue;
2701 if (state->nbTrans < 2)
2702 continue;
2703 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2704 t1 = &(state->trans[transnr]);
2705 /*
2706 * Determinism checks in case of counted or all transitions
2707 * will have to be handled separately
2708 */
2709 if (t1->atom == NULL) {
2710 /* t1->nd = 1; */
2711 continue;
2712 }
2713 if (t1->to == -1) /* eliminated */
2714 continue;
2715 for (i = 0;i < transnr;i++) {
2716 t2 = &(state->trans[i]);
2717 if (t2->to == -1) /* eliminated */
2718 continue;
2719 if (t2->atom != NULL) {
2720 if (t1->to == t2->to) {
2721 /*
2722 * Here we use deep because we want to keep the
2723 * transitions which indicate a conflict
2724 */
2725 if (xmlFAEqualAtoms(t1->atom, t2->atom, deep) &&
2726 (t1->counter == t2->counter) &&
2727 (t1->count == t2->count))
2728 t2->to = -1; /* eliminated */
2729 }
2730 }
2731 }
2732 }
2733 }
2734
2735 /*
2736 * Check for all states that there aren't 2 transitions
2737 * with the same atom and a different target.
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 last = NULL;
2746 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2747 t1 = &(state->trans[transnr]);
2748 /*
2749 * Determinism checks in case of counted or all transitions
2750 * will have to be handled separately
2751 */
2752 if (t1->atom == NULL) {
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 /*
2763 * But here we don't use deep because we want to
2764 * find transitions which indicate a conflict
2765 */
2766 if (xmlFACompareAtoms(t1->atom, t2->atom, 1)) {
2767 ret = 0;
2768 /* mark the transitions as non-deterministic ones */
2769 t1->nd = 1;
2770 t2->nd = 1;
2771 last = t1;
2772 }
2773 } else if (t1->to != -1) {
2774 /*
2775 * do the closure in case of remaining specific
2776 * epsilon transitions like choices or all
2777 */
2778 ret = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
2779 t2->to, t2->atom);
2780 /* don't shortcut the computation so all non deterministic
2781 transition get marked down
2782 if (ret == 0)
2783 return(0);
2784 */
2785 if (ret == 0) {
2786 t1->nd = 1;
2787 /* t2->nd = 1; */
2788 last = t1;
2789 }
2790 }
2791 }
2792 /* don't shortcut the computation so all non deterministic
2793 transition get marked down
2794 if (ret == 0)
2795 break; */
2796 }
2797
2798 /*
2799 * mark specifically the last non-deterministic transition
2800 * from a state since there is no need to set-up rollback
2801 * from it
2802 */
2803 if (last != NULL) {
2804 last->nd = 2;
2805 }
2806
2807 /* don't shortcut the computation so all non deterministic
2808 transition get marked down
2809 if (ret == 0)
2810 break; */
2811 }
2812
2813 ctxt->determinist = ret;
2814 return(ret);
2815 }
2816
2817 /************************************************************************
2818 * *
2819 * Routines to check input against transition atoms *
2820 * *
2821 ************************************************************************/
2822
2823 static int
xmlRegCheckCharacterRange(xmlRegAtomType type,int codepoint,int neg,int start,int end,const xmlChar * blockName)2824 xmlRegCheckCharacterRange(xmlRegAtomType type, int codepoint, int neg,
2825 int start, int end, const xmlChar *blockName) {
2826 int ret = 0;
2827
2828 switch (type) {
2829 case XML_REGEXP_STRING:
2830 case XML_REGEXP_SUBREG:
2831 case XML_REGEXP_RANGES:
2832 case XML_REGEXP_EPSILON:
2833 return(-1);
2834 case XML_REGEXP_ANYCHAR:
2835 ret = ((codepoint != '\n') && (codepoint != '\r'));
2836 break;
2837 case XML_REGEXP_CHARVAL:
2838 ret = ((codepoint >= start) && (codepoint <= end));
2839 break;
2840 case XML_REGEXP_NOTSPACE:
2841 neg = !neg;
2842 /* Falls through. */
2843 case XML_REGEXP_ANYSPACE:
2844 ret = ((codepoint == '\n') || (codepoint == '\r') ||
2845 (codepoint == '\t') || (codepoint == ' '));
2846 break;
2847 case XML_REGEXP_NOTINITNAME:
2848 neg = !neg;
2849 /* Falls through. */
2850 case XML_REGEXP_INITNAME:
2851 ret = (IS_LETTER(codepoint) ||
2852 (codepoint == '_') || (codepoint == ':'));
2853 break;
2854 case XML_REGEXP_NOTNAMECHAR:
2855 neg = !neg;
2856 /* Falls through. */
2857 case XML_REGEXP_NAMECHAR:
2858 ret = (IS_LETTER(codepoint) || IS_DIGIT(codepoint) ||
2859 (codepoint == '.') || (codepoint == '-') ||
2860 (codepoint == '_') || (codepoint == ':') ||
2861 IS_COMBINING(codepoint) || IS_EXTENDER(codepoint));
2862 break;
2863 case XML_REGEXP_NOTDECIMAL:
2864 neg = !neg;
2865 /* Falls through. */
2866 case XML_REGEXP_DECIMAL:
2867 ret = xmlUCSIsCatNd(codepoint);
2868 break;
2869 case XML_REGEXP_REALCHAR:
2870 neg = !neg;
2871 /* Falls through. */
2872 case XML_REGEXP_NOTREALCHAR:
2873 ret = xmlUCSIsCatP(codepoint);
2874 if (ret == 0)
2875 ret = xmlUCSIsCatZ(codepoint);
2876 if (ret == 0)
2877 ret = xmlUCSIsCatC(codepoint);
2878 break;
2879 case XML_REGEXP_LETTER:
2880 ret = xmlUCSIsCatL(codepoint);
2881 break;
2882 case XML_REGEXP_LETTER_UPPERCASE:
2883 ret = xmlUCSIsCatLu(codepoint);
2884 break;
2885 case XML_REGEXP_LETTER_LOWERCASE:
2886 ret = xmlUCSIsCatLl(codepoint);
2887 break;
2888 case XML_REGEXP_LETTER_TITLECASE:
2889 ret = xmlUCSIsCatLt(codepoint);
2890 break;
2891 case XML_REGEXP_LETTER_MODIFIER:
2892 ret = xmlUCSIsCatLm(codepoint);
2893 break;
2894 case XML_REGEXP_LETTER_OTHERS:
2895 ret = xmlUCSIsCatLo(codepoint);
2896 break;
2897 case XML_REGEXP_MARK:
2898 ret = xmlUCSIsCatM(codepoint);
2899 break;
2900 case XML_REGEXP_MARK_NONSPACING:
2901 ret = xmlUCSIsCatMn(codepoint);
2902 break;
2903 case XML_REGEXP_MARK_SPACECOMBINING:
2904 ret = xmlUCSIsCatMc(codepoint);
2905 break;
2906 case XML_REGEXP_MARK_ENCLOSING:
2907 ret = xmlUCSIsCatMe(codepoint);
2908 break;
2909 case XML_REGEXP_NUMBER:
2910 ret = xmlUCSIsCatN(codepoint);
2911 break;
2912 case XML_REGEXP_NUMBER_DECIMAL:
2913 ret = xmlUCSIsCatNd(codepoint);
2914 break;
2915 case XML_REGEXP_NUMBER_LETTER:
2916 ret = xmlUCSIsCatNl(codepoint);
2917 break;
2918 case XML_REGEXP_NUMBER_OTHERS:
2919 ret = xmlUCSIsCatNo(codepoint);
2920 break;
2921 case XML_REGEXP_PUNCT:
2922 ret = xmlUCSIsCatP(codepoint);
2923 break;
2924 case XML_REGEXP_PUNCT_CONNECTOR:
2925 ret = xmlUCSIsCatPc(codepoint);
2926 break;
2927 case XML_REGEXP_PUNCT_DASH:
2928 ret = xmlUCSIsCatPd(codepoint);
2929 break;
2930 case XML_REGEXP_PUNCT_OPEN:
2931 ret = xmlUCSIsCatPs(codepoint);
2932 break;
2933 case XML_REGEXP_PUNCT_CLOSE:
2934 ret = xmlUCSIsCatPe(codepoint);
2935 break;
2936 case XML_REGEXP_PUNCT_INITQUOTE:
2937 ret = xmlUCSIsCatPi(codepoint);
2938 break;
2939 case XML_REGEXP_PUNCT_FINQUOTE:
2940 ret = xmlUCSIsCatPf(codepoint);
2941 break;
2942 case XML_REGEXP_PUNCT_OTHERS:
2943 ret = xmlUCSIsCatPo(codepoint);
2944 break;
2945 case XML_REGEXP_SEPAR:
2946 ret = xmlUCSIsCatZ(codepoint);
2947 break;
2948 case XML_REGEXP_SEPAR_SPACE:
2949 ret = xmlUCSIsCatZs(codepoint);
2950 break;
2951 case XML_REGEXP_SEPAR_LINE:
2952 ret = xmlUCSIsCatZl(codepoint);
2953 break;
2954 case XML_REGEXP_SEPAR_PARA:
2955 ret = xmlUCSIsCatZp(codepoint);
2956 break;
2957 case XML_REGEXP_SYMBOL:
2958 ret = xmlUCSIsCatS(codepoint);
2959 break;
2960 case XML_REGEXP_SYMBOL_MATH:
2961 ret = xmlUCSIsCatSm(codepoint);
2962 break;
2963 case XML_REGEXP_SYMBOL_CURRENCY:
2964 ret = xmlUCSIsCatSc(codepoint);
2965 break;
2966 case XML_REGEXP_SYMBOL_MODIFIER:
2967 ret = xmlUCSIsCatSk(codepoint);
2968 break;
2969 case XML_REGEXP_SYMBOL_OTHERS:
2970 ret = xmlUCSIsCatSo(codepoint);
2971 break;
2972 case XML_REGEXP_OTHER:
2973 ret = xmlUCSIsCatC(codepoint);
2974 break;
2975 case XML_REGEXP_OTHER_CONTROL:
2976 ret = xmlUCSIsCatCc(codepoint);
2977 break;
2978 case XML_REGEXP_OTHER_FORMAT:
2979 ret = xmlUCSIsCatCf(codepoint);
2980 break;
2981 case XML_REGEXP_OTHER_PRIVATE:
2982 ret = xmlUCSIsCatCo(codepoint);
2983 break;
2984 case XML_REGEXP_OTHER_NA:
2985 /* ret = xmlUCSIsCatCn(codepoint); */
2986 /* Seems it doesn't exist anymore in recent Unicode releases */
2987 ret = 0;
2988 break;
2989 case XML_REGEXP_BLOCK_NAME:
2990 ret = xmlUCSIsBlock(codepoint, (const char *) blockName);
2991 break;
2992 }
2993 if (neg)
2994 return(!ret);
2995 return(ret);
2996 }
2997
2998 static int
xmlRegCheckCharacter(xmlRegAtomPtr atom,int codepoint)2999 xmlRegCheckCharacter(xmlRegAtomPtr atom, int codepoint) {
3000 int i, ret = 0;
3001 xmlRegRangePtr range;
3002
3003 if ((atom == NULL) || (!IS_CHAR(codepoint)))
3004 return(-1);
3005
3006 switch (atom->type) {
3007 case XML_REGEXP_SUBREG:
3008 case XML_REGEXP_EPSILON:
3009 return(-1);
3010 case XML_REGEXP_CHARVAL:
3011 return(codepoint == atom->codepoint);
3012 case XML_REGEXP_RANGES: {
3013 int accept = 0;
3014
3015 for (i = 0;i < atom->nbRanges;i++) {
3016 range = atom->ranges[i];
3017 if (range->neg == 2) {
3018 ret = xmlRegCheckCharacterRange(range->type, codepoint,
3019 0, range->start, range->end,
3020 range->blockName);
3021 if (ret != 0)
3022 return(0); /* excluded char */
3023 } else if (range->neg) {
3024 ret = xmlRegCheckCharacterRange(range->type, codepoint,
3025 0, range->start, range->end,
3026 range->blockName);
3027 if (ret == 0)
3028 accept = 1;
3029 else
3030 return(0);
3031 } else {
3032 ret = xmlRegCheckCharacterRange(range->type, codepoint,
3033 0, range->start, range->end,
3034 range->blockName);
3035 if (ret != 0)
3036 accept = 1; /* might still be excluded */
3037 }
3038 }
3039 return(accept);
3040 }
3041 case XML_REGEXP_STRING:
3042 printf("TODO: XML_REGEXP_STRING\n");
3043 return(-1);
3044 case XML_REGEXP_ANYCHAR:
3045 case XML_REGEXP_ANYSPACE:
3046 case XML_REGEXP_NOTSPACE:
3047 case XML_REGEXP_INITNAME:
3048 case XML_REGEXP_NOTINITNAME:
3049 case XML_REGEXP_NAMECHAR:
3050 case XML_REGEXP_NOTNAMECHAR:
3051 case XML_REGEXP_DECIMAL:
3052 case XML_REGEXP_NOTDECIMAL:
3053 case XML_REGEXP_REALCHAR:
3054 case XML_REGEXP_NOTREALCHAR:
3055 case XML_REGEXP_LETTER:
3056 case XML_REGEXP_LETTER_UPPERCASE:
3057 case XML_REGEXP_LETTER_LOWERCASE:
3058 case XML_REGEXP_LETTER_TITLECASE:
3059 case XML_REGEXP_LETTER_MODIFIER:
3060 case XML_REGEXP_LETTER_OTHERS:
3061 case XML_REGEXP_MARK:
3062 case XML_REGEXP_MARK_NONSPACING:
3063 case XML_REGEXP_MARK_SPACECOMBINING:
3064 case XML_REGEXP_MARK_ENCLOSING:
3065 case XML_REGEXP_NUMBER:
3066 case XML_REGEXP_NUMBER_DECIMAL:
3067 case XML_REGEXP_NUMBER_LETTER:
3068 case XML_REGEXP_NUMBER_OTHERS:
3069 case XML_REGEXP_PUNCT:
3070 case XML_REGEXP_PUNCT_CONNECTOR:
3071 case XML_REGEXP_PUNCT_DASH:
3072 case XML_REGEXP_PUNCT_OPEN:
3073 case XML_REGEXP_PUNCT_CLOSE:
3074 case XML_REGEXP_PUNCT_INITQUOTE:
3075 case XML_REGEXP_PUNCT_FINQUOTE:
3076 case XML_REGEXP_PUNCT_OTHERS:
3077 case XML_REGEXP_SEPAR:
3078 case XML_REGEXP_SEPAR_SPACE:
3079 case XML_REGEXP_SEPAR_LINE:
3080 case XML_REGEXP_SEPAR_PARA:
3081 case XML_REGEXP_SYMBOL:
3082 case XML_REGEXP_SYMBOL_MATH:
3083 case XML_REGEXP_SYMBOL_CURRENCY:
3084 case XML_REGEXP_SYMBOL_MODIFIER:
3085 case XML_REGEXP_SYMBOL_OTHERS:
3086 case XML_REGEXP_OTHER:
3087 case XML_REGEXP_OTHER_CONTROL:
3088 case XML_REGEXP_OTHER_FORMAT:
3089 case XML_REGEXP_OTHER_PRIVATE:
3090 case XML_REGEXP_OTHER_NA:
3091 case XML_REGEXP_BLOCK_NAME:
3092 ret = xmlRegCheckCharacterRange(atom->type, codepoint, 0, 0, 0,
3093 (const xmlChar *)atom->valuep);
3094 if (atom->neg)
3095 ret = !ret;
3096 break;
3097 }
3098 return(ret);
3099 }
3100
3101 /************************************************************************
3102 * *
3103 * Saving and restoring state of an execution context *
3104 * *
3105 ************************************************************************/
3106
3107 #ifdef DEBUG_REGEXP_EXEC
3108 static void
xmlFARegDebugExec(xmlRegExecCtxtPtr exec)3109 xmlFARegDebugExec(xmlRegExecCtxtPtr exec) {
3110 printf("state: %d:%d:idx %d", exec->state->no, exec->transno, exec->index);
3111 if (exec->inputStack != NULL) {
3112 int i;
3113 printf(": ");
3114 for (i = 0;(i < 3) && (i < exec->inputStackNr);i++)
3115 printf("%s ", (const char *)
3116 exec->inputStack[exec->inputStackNr - (i + 1)].value);
3117 } else {
3118 printf(": %s", &(exec->inputString[exec->index]));
3119 }
3120 printf("\n");
3121 }
3122 #endif
3123
3124 static void
xmlFARegExecSave(xmlRegExecCtxtPtr exec)3125 xmlFARegExecSave(xmlRegExecCtxtPtr exec) {
3126 #ifdef DEBUG_REGEXP_EXEC
3127 printf("saving ");
3128 exec->transno++;
3129 xmlFARegDebugExec(exec);
3130 exec->transno--;
3131 #endif
3132 #ifdef MAX_PUSH
3133 if (exec->nbPush > MAX_PUSH) {
3134 return;
3135 }
3136 exec->nbPush++;
3137 #endif
3138
3139 if (exec->maxRollbacks == 0) {
3140 exec->maxRollbacks = 4;
3141 exec->rollbacks = (xmlRegExecRollback *) xmlMalloc(exec->maxRollbacks *
3142 sizeof(xmlRegExecRollback));
3143 if (exec->rollbacks == NULL) {
3144 xmlRegexpErrMemory(NULL, "saving regexp");
3145 exec->maxRollbacks = 0;
3146 return;
3147 }
3148 memset(exec->rollbacks, 0,
3149 exec->maxRollbacks * sizeof(xmlRegExecRollback));
3150 } else if (exec->nbRollbacks >= exec->maxRollbacks) {
3151 xmlRegExecRollback *tmp;
3152 int len = exec->maxRollbacks;
3153
3154 exec->maxRollbacks *= 2;
3155 tmp = (xmlRegExecRollback *) xmlRealloc(exec->rollbacks,
3156 exec->maxRollbacks * sizeof(xmlRegExecRollback));
3157 if (tmp == NULL) {
3158 xmlRegexpErrMemory(NULL, "saving regexp");
3159 exec->maxRollbacks /= 2;
3160 return;
3161 }
3162 exec->rollbacks = tmp;
3163 tmp = &exec->rollbacks[len];
3164 memset(tmp, 0, (exec->maxRollbacks - len) * sizeof(xmlRegExecRollback));
3165 }
3166 exec->rollbacks[exec->nbRollbacks].state = exec->state;
3167 exec->rollbacks[exec->nbRollbacks].index = exec->index;
3168 exec->rollbacks[exec->nbRollbacks].nextbranch = exec->transno + 1;
3169 if (exec->comp->nbCounters > 0) {
3170 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3171 exec->rollbacks[exec->nbRollbacks].counts = (int *)
3172 xmlMalloc(exec->comp->nbCounters * sizeof(int));
3173 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3174 xmlRegexpErrMemory(NULL, "saving regexp");
3175 exec->status = -5;
3176 return;
3177 }
3178 }
3179 memcpy(exec->rollbacks[exec->nbRollbacks].counts, exec->counts,
3180 exec->comp->nbCounters * sizeof(int));
3181 }
3182 exec->nbRollbacks++;
3183 }
3184
3185 static void
xmlFARegExecRollBack(xmlRegExecCtxtPtr exec)3186 xmlFARegExecRollBack(xmlRegExecCtxtPtr exec) {
3187 if (exec->nbRollbacks <= 0) {
3188 exec->status = -1;
3189 #ifdef DEBUG_REGEXP_EXEC
3190 printf("rollback failed on empty stack\n");
3191 #endif
3192 return;
3193 }
3194 exec->nbRollbacks--;
3195 exec->state = exec->rollbacks[exec->nbRollbacks].state;
3196 exec->index = exec->rollbacks[exec->nbRollbacks].index;
3197 exec->transno = exec->rollbacks[exec->nbRollbacks].nextbranch;
3198 if (exec->comp->nbCounters > 0) {
3199 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3200 fprintf(stderr, "exec save: allocation failed");
3201 exec->status = -6;
3202 return;
3203 }
3204 if (exec->counts) {
3205 memcpy(exec->counts, exec->rollbacks[exec->nbRollbacks].counts,
3206 exec->comp->nbCounters * sizeof(int));
3207 }
3208 }
3209
3210 #ifdef DEBUG_REGEXP_EXEC
3211 printf("restored ");
3212 xmlFARegDebugExec(exec);
3213 #endif
3214 }
3215
3216 /************************************************************************
3217 * *
3218 * Verifier, running an input against a compiled regexp *
3219 * *
3220 ************************************************************************/
3221
3222 static int
xmlFARegExec(xmlRegexpPtr comp,const xmlChar * content)3223 xmlFARegExec(xmlRegexpPtr comp, const xmlChar *content) {
3224 xmlRegExecCtxt execval;
3225 xmlRegExecCtxtPtr exec = &execval;
3226 int ret, codepoint = 0, len, deter;
3227
3228 exec->inputString = content;
3229 exec->index = 0;
3230 exec->nbPush = 0;
3231 exec->determinist = 1;
3232 exec->maxRollbacks = 0;
3233 exec->nbRollbacks = 0;
3234 exec->rollbacks = NULL;
3235 exec->status = 0;
3236 exec->comp = comp;
3237 exec->state = comp->states[0];
3238 exec->transno = 0;
3239 exec->transcount = 0;
3240 exec->inputStack = NULL;
3241 exec->inputStackMax = 0;
3242 if (comp->nbCounters > 0) {
3243 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int));
3244 if (exec->counts == NULL) {
3245 xmlRegexpErrMemory(NULL, "running regexp");
3246 return(-1);
3247 }
3248 memset(exec->counts, 0, comp->nbCounters * sizeof(int));
3249 } else
3250 exec->counts = NULL;
3251 while ((exec->status == 0) && (exec->state != NULL) &&
3252 ((exec->inputString[exec->index] != 0) ||
3253 ((exec->state != NULL) &&
3254 (exec->state->type != XML_REGEXP_FINAL_STATE)))) {
3255 xmlRegTransPtr trans;
3256 xmlRegAtomPtr atom;
3257
3258 /*
3259 * If end of input on non-terminal state, rollback, however we may
3260 * still have epsilon like transition for counted transitions
3261 * on counters, in that case don't break too early. Additionally,
3262 * if we are working on a range like "AB{0,2}", where B is not present,
3263 * we don't want to break.
3264 */
3265 len = 1;
3266 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL)) {
3267 /*
3268 * if there is a transition, we must check if
3269 * atom allows minOccurs of 0
3270 */
3271 if (exec->transno < exec->state->nbTrans) {
3272 trans = &exec->state->trans[exec->transno];
3273 if (trans->to >=0) {
3274 atom = trans->atom;
3275 if (!((atom->min == 0) && (atom->max > 0)))
3276 goto rollback;
3277 }
3278 } else
3279 goto rollback;
3280 }
3281
3282 exec->transcount = 0;
3283 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
3284 trans = &exec->state->trans[exec->transno];
3285 if (trans->to < 0)
3286 continue;
3287 atom = trans->atom;
3288 ret = 0;
3289 deter = 1;
3290 if (trans->count >= 0) {
3291 int count;
3292 xmlRegCounterPtr counter;
3293
3294 if (exec->counts == NULL) {
3295 exec->status = -1;
3296 goto error;
3297 }
3298 /*
3299 * A counted transition.
3300 */
3301
3302 count = exec->counts[trans->count];
3303 counter = &exec->comp->counters[trans->count];
3304 #ifdef DEBUG_REGEXP_EXEC
3305 printf("testing count %d: val %d, min %d, max %d\n",
3306 trans->count, count, counter->min, counter->max);
3307 #endif
3308 ret = ((count >= counter->min) && (count <= counter->max));
3309 if ((ret) && (counter->min != counter->max))
3310 deter = 0;
3311 } else if (atom == NULL) {
3312 fprintf(stderr, "epsilon transition left at runtime\n");
3313 exec->status = -2;
3314 break;
3315 } else if (exec->inputString[exec->index] != 0) {
3316 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
3317 ret = xmlRegCheckCharacter(atom, codepoint);
3318 if ((ret == 1) && (atom->min >= 0) && (atom->max > 0)) {
3319 xmlRegStatePtr to = comp->states[trans->to];
3320
3321 /*
3322 * this is a multiple input sequence
3323 * If there is a counter associated increment it now.
3324 * before potentially saving and rollback
3325 * do not increment if the counter is already over the
3326 * maximum limit in which case get to next transition
3327 */
3328 if (trans->counter >= 0) {
3329 xmlRegCounterPtr counter;
3330
3331 if ((exec->counts == NULL) ||
3332 (exec->comp == NULL) ||
3333 (exec->comp->counters == NULL)) {
3334 exec->status = -1;
3335 goto error;
3336 }
3337 counter = &exec->comp->counters[trans->counter];
3338 if (exec->counts[trans->counter] >= counter->max)
3339 continue; /* for loop on transitions */
3340
3341 #ifdef DEBUG_REGEXP_EXEC
3342 printf("Increasing count %d\n", trans->counter);
3343 #endif
3344 exec->counts[trans->counter]++;
3345 }
3346 if (exec->state->nbTrans > exec->transno + 1) {
3347 xmlFARegExecSave(exec);
3348 }
3349 exec->transcount = 1;
3350 do {
3351 /*
3352 * Try to progress as much as possible on the input
3353 */
3354 if (exec->transcount == atom->max) {
3355 break;
3356 }
3357 exec->index += len;
3358 /*
3359 * End of input: stop here
3360 */
3361 if (exec->inputString[exec->index] == 0) {
3362 exec->index -= len;
3363 break;
3364 }
3365 if (exec->transcount >= atom->min) {
3366 int transno = exec->transno;
3367 xmlRegStatePtr state = exec->state;
3368
3369 /*
3370 * The transition is acceptable save it
3371 */
3372 exec->transno = -1; /* trick */
3373 exec->state = to;
3374 xmlFARegExecSave(exec);
3375 exec->transno = transno;
3376 exec->state = state;
3377 }
3378 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
3379 len);
3380 ret = xmlRegCheckCharacter(atom, codepoint);
3381 exec->transcount++;
3382 } while (ret == 1);
3383 if (exec->transcount < atom->min)
3384 ret = 0;
3385
3386 /*
3387 * If the last check failed but one transition was found
3388 * possible, rollback
3389 */
3390 if (ret < 0)
3391 ret = 0;
3392 if (ret == 0) {
3393 goto rollback;
3394 }
3395 if (trans->counter >= 0) {
3396 if (exec->counts == NULL) {
3397 exec->status = -1;
3398 goto error;
3399 }
3400 #ifdef DEBUG_REGEXP_EXEC
3401 printf("Decreasing count %d\n", trans->counter);
3402 #endif
3403 exec->counts[trans->counter]--;
3404 }
3405 } else if ((ret == 0) && (atom->min == 0) && (atom->max > 0)) {
3406 /*
3407 * we don't match on the codepoint, but minOccurs of 0
3408 * says that's ok. Setting len to 0 inhibits stepping
3409 * over the codepoint.
3410 */
3411 exec->transcount = 1;
3412 len = 0;
3413 ret = 1;
3414 }
3415 } else if ((atom->min == 0) && (atom->max > 0)) {
3416 /* another spot to match when minOccurs is 0 */
3417 exec->transcount = 1;
3418 len = 0;
3419 ret = 1;
3420 }
3421 if (ret == 1) {
3422 if ((trans->nd == 1) ||
3423 ((trans->count >= 0) && (deter == 0) &&
3424 (exec->state->nbTrans > exec->transno + 1))) {
3425 #ifdef DEBUG_REGEXP_EXEC
3426 if (trans->nd == 1)
3427 printf("Saving on nd transition atom %d for %c at %d\n",
3428 trans->atom->no, codepoint, exec->index);
3429 else
3430 printf("Saving on counted transition count %d for %c at %d\n",
3431 trans->count, codepoint, exec->index);
3432 #endif
3433 xmlFARegExecSave(exec);
3434 }
3435 if (trans->counter >= 0) {
3436 xmlRegCounterPtr counter;
3437
3438 /* make sure we don't go over the counter maximum value */
3439 if ((exec->counts == NULL) ||
3440 (exec->comp == NULL) ||
3441 (exec->comp->counters == NULL)) {
3442 exec->status = -1;
3443 goto error;
3444 }
3445 counter = &exec->comp->counters[trans->counter];
3446 if (exec->counts[trans->counter] >= counter->max)
3447 continue; /* for loop on transitions */
3448 #ifdef DEBUG_REGEXP_EXEC
3449 printf("Increasing count %d\n", trans->counter);
3450 #endif
3451 exec->counts[trans->counter]++;
3452 }
3453 if ((trans->count >= 0) &&
3454 (trans->count < REGEXP_ALL_COUNTER)) {
3455 if (exec->counts == NULL) {
3456 exec->status = -1;
3457 goto error;
3458 }
3459 #ifdef DEBUG_REGEXP_EXEC
3460 printf("resetting count %d on transition\n",
3461 trans->count);
3462 #endif
3463 exec->counts[trans->count] = 0;
3464 }
3465 #ifdef DEBUG_REGEXP_EXEC
3466 printf("entering state %d\n", trans->to);
3467 #endif
3468 exec->state = comp->states[trans->to];
3469 exec->transno = 0;
3470 if (trans->atom != NULL) {
3471 exec->index += len;
3472 }
3473 goto progress;
3474 } else if (ret < 0) {
3475 exec->status = -4;
3476 break;
3477 }
3478 }
3479 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
3480 rollback:
3481 /*
3482 * Failed to find a way out
3483 */
3484 exec->determinist = 0;
3485 #ifdef DEBUG_REGEXP_EXEC
3486 printf("rollback from state %d on %d:%c\n", exec->state->no,
3487 codepoint,codepoint);
3488 #endif
3489 xmlFARegExecRollBack(exec);
3490 }
3491 progress:
3492 continue;
3493 }
3494 error:
3495 if (exec->rollbacks != NULL) {
3496 if (exec->counts != NULL) {
3497 int i;
3498
3499 for (i = 0;i < exec->maxRollbacks;i++)
3500 if (exec->rollbacks[i].counts != NULL)
3501 xmlFree(exec->rollbacks[i].counts);
3502 }
3503 xmlFree(exec->rollbacks);
3504 }
3505 if (exec->state == NULL)
3506 return(-1);
3507 if (exec->counts != NULL)
3508 xmlFree(exec->counts);
3509 if (exec->status == 0)
3510 return(1);
3511 if (exec->status == -1) {
3512 if (exec->nbPush > MAX_PUSH)
3513 return(-1);
3514 return(0);
3515 }
3516 return(exec->status);
3517 }
3518
3519 /************************************************************************
3520 * *
3521 * Progressive interface to the verifier one atom at a time *
3522 * *
3523 ************************************************************************/
3524 #ifdef DEBUG_ERR
3525 static void testerr(xmlRegExecCtxtPtr exec);
3526 #endif
3527
3528 /**
3529 * xmlRegNewExecCtxt:
3530 * @comp: a precompiled regular expression
3531 * @callback: a callback function used for handling progresses in the
3532 * automata matching phase
3533 * @data: the context data associated to the callback in this context
3534 *
3535 * Build a context used for progressive evaluation of a regexp.
3536 *
3537 * Returns the new context
3538 */
3539 xmlRegExecCtxtPtr
xmlRegNewExecCtxt(xmlRegexpPtr comp,xmlRegExecCallbacks callback,void * data)3540 xmlRegNewExecCtxt(xmlRegexpPtr comp, xmlRegExecCallbacks callback, void *data) {
3541 xmlRegExecCtxtPtr exec;
3542
3543 if (comp == NULL)
3544 return(NULL);
3545 if ((comp->compact == NULL) && (comp->states == NULL))
3546 return(NULL);
3547 exec = (xmlRegExecCtxtPtr) xmlMalloc(sizeof(xmlRegExecCtxt));
3548 if (exec == NULL) {
3549 xmlRegexpErrMemory(NULL, "creating execution context");
3550 return(NULL);
3551 }
3552 memset(exec, 0, sizeof(xmlRegExecCtxt));
3553 exec->inputString = NULL;
3554 exec->index = 0;
3555 exec->determinist = 1;
3556 exec->maxRollbacks = 0;
3557 exec->nbRollbacks = 0;
3558 exec->rollbacks = NULL;
3559 exec->status = 0;
3560 exec->comp = comp;
3561 if (comp->compact == NULL)
3562 exec->state = comp->states[0];
3563 exec->transno = 0;
3564 exec->transcount = 0;
3565 exec->callback = callback;
3566 exec->data = data;
3567 if (comp->nbCounters > 0) {
3568 /*
3569 * For error handling, exec->counts is allocated twice the size
3570 * the second half is used to store the data in case of rollback
3571 */
3572 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int)
3573 * 2);
3574 if (exec->counts == NULL) {
3575 xmlRegexpErrMemory(NULL, "creating execution context");
3576 xmlFree(exec);
3577 return(NULL);
3578 }
3579 memset(exec->counts, 0, comp->nbCounters * sizeof(int) * 2);
3580 exec->errCounts = &exec->counts[comp->nbCounters];
3581 } else {
3582 exec->counts = NULL;
3583 exec->errCounts = NULL;
3584 }
3585 exec->inputStackMax = 0;
3586 exec->inputStackNr = 0;
3587 exec->inputStack = NULL;
3588 exec->errStateNo = -1;
3589 exec->errString = NULL;
3590 exec->nbPush = 0;
3591 return(exec);
3592 }
3593
3594 /**
3595 * xmlRegFreeExecCtxt:
3596 * @exec: a regular expression evaulation context
3597 *
3598 * Free the structures associated to a regular expression evaulation context.
3599 */
3600 void
xmlRegFreeExecCtxt(xmlRegExecCtxtPtr exec)3601 xmlRegFreeExecCtxt(xmlRegExecCtxtPtr exec) {
3602 if (exec == NULL)
3603 return;
3604
3605 if (exec->rollbacks != NULL) {
3606 if (exec->counts != NULL) {
3607 int i;
3608
3609 for (i = 0;i < exec->maxRollbacks;i++)
3610 if (exec->rollbacks[i].counts != NULL)
3611 xmlFree(exec->rollbacks[i].counts);
3612 }
3613 xmlFree(exec->rollbacks);
3614 }
3615 if (exec->counts != NULL)
3616 xmlFree(exec->counts);
3617 if (exec->inputStack != NULL) {
3618 int i;
3619
3620 for (i = 0;i < exec->inputStackNr;i++) {
3621 if (exec->inputStack[i].value != NULL)
3622 xmlFree(exec->inputStack[i].value);
3623 }
3624 xmlFree(exec->inputStack);
3625 }
3626 if (exec->errString != NULL)
3627 xmlFree(exec->errString);
3628 xmlFree(exec);
3629 }
3630
3631 static void
xmlFARegExecSaveInputString(xmlRegExecCtxtPtr exec,const xmlChar * value,void * data)3632 xmlFARegExecSaveInputString(xmlRegExecCtxtPtr exec, const xmlChar *value,
3633 void *data) {
3634 #ifdef DEBUG_PUSH
3635 printf("saving value: %d:%s\n", exec->inputStackNr, value);
3636 #endif
3637 if (exec->inputStackMax == 0) {
3638 exec->inputStackMax = 4;
3639 exec->inputStack = (xmlRegInputTokenPtr)
3640 xmlMalloc(exec->inputStackMax * sizeof(xmlRegInputToken));
3641 if (exec->inputStack == NULL) {
3642 xmlRegexpErrMemory(NULL, "pushing input string");
3643 exec->inputStackMax = 0;
3644 return;
3645 }
3646 } else if (exec->inputStackNr + 1 >= exec->inputStackMax) {
3647 xmlRegInputTokenPtr tmp;
3648
3649 exec->inputStackMax *= 2;
3650 tmp = (xmlRegInputTokenPtr) xmlRealloc(exec->inputStack,
3651 exec->inputStackMax * sizeof(xmlRegInputToken));
3652 if (tmp == NULL) {
3653 xmlRegexpErrMemory(NULL, "pushing input string");
3654 exec->inputStackMax /= 2;
3655 return;
3656 }
3657 exec->inputStack = tmp;
3658 }
3659 exec->inputStack[exec->inputStackNr].value = xmlStrdup(value);
3660 exec->inputStack[exec->inputStackNr].data = data;
3661 exec->inputStackNr++;
3662 exec->inputStack[exec->inputStackNr].value = NULL;
3663 exec->inputStack[exec->inputStackNr].data = NULL;
3664 }
3665
3666 /**
3667 * xmlRegStrEqualWildcard:
3668 * @expStr: the string to be evaluated
3669 * @valStr: the validation string
3670 *
3671 * Checks if both strings are equal or have the same content. "*"
3672 * can be used as a wildcard in @valStr; "|" is used as a seperator of
3673 * substrings in both @expStr and @valStr.
3674 *
3675 * Returns 1 if the comparison is satisfied and the number of substrings
3676 * is equal, 0 otherwise.
3677 */
3678
3679 static int
xmlRegStrEqualWildcard(const xmlChar * expStr,const xmlChar * valStr)3680 xmlRegStrEqualWildcard(const xmlChar *expStr, const xmlChar *valStr) {
3681 if (expStr == valStr) return(1);
3682 if (expStr == NULL) return(0);
3683 if (valStr == NULL) return(0);
3684 do {
3685 /*
3686 * Eval if we have a wildcard for the current item.
3687 */
3688 if (*expStr != *valStr) {
3689 /* if one of them starts with a wildcard make valStr be it */
3690 if (*valStr == '*') {
3691 const xmlChar *tmp;
3692
3693 tmp = valStr;
3694 valStr = expStr;
3695 expStr = tmp;
3696 }
3697 if ((*valStr != 0) && (*expStr != 0) && (*expStr++ == '*')) {
3698 do {
3699 if (*valStr == XML_REG_STRING_SEPARATOR)
3700 break;
3701 valStr++;
3702 } while (*valStr != 0);
3703 continue;
3704 } else
3705 return(0);
3706 }
3707 expStr++;
3708 valStr++;
3709 } while (*valStr != 0);
3710 if (*expStr != 0)
3711 return (0);
3712 else
3713 return (1);
3714 }
3715
3716 /**
3717 * xmlRegCompactPushString:
3718 * @exec: a regexp execution context
3719 * @comp: the precompiled exec with a compact table
3720 * @value: a string token input
3721 * @data: data associated to the token to reuse in callbacks
3722 *
3723 * Push one input token in the execution context
3724 *
3725 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
3726 * a negative value in case of error.
3727 */
3728 static int
xmlRegCompactPushString(xmlRegExecCtxtPtr exec,xmlRegexpPtr comp,const xmlChar * value,void * data)3729 xmlRegCompactPushString(xmlRegExecCtxtPtr exec,
3730 xmlRegexpPtr comp,
3731 const xmlChar *value,
3732 void *data) {
3733 int state = exec->index;
3734 int i, target;
3735
3736 if ((comp == NULL) || (comp->compact == NULL) || (comp->stringMap == NULL))
3737 return(-1);
3738
3739 if (value == NULL) {
3740 /*
3741 * are we at a final state ?
3742 */
3743 if (comp->compact[state * (comp->nbstrings + 1)] ==
3744 XML_REGEXP_FINAL_STATE)
3745 return(1);
3746 return(0);
3747 }
3748
3749 #ifdef DEBUG_PUSH
3750 printf("value pushed: %s\n", value);
3751 #endif
3752
3753 /*
3754 * Examine all outside transitions from current state
3755 */
3756 for (i = 0;i < comp->nbstrings;i++) {
3757 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
3758 if ((target > 0) && (target <= comp->nbstates)) {
3759 target--; /* to avoid 0 */
3760 if (xmlRegStrEqualWildcard(comp->stringMap[i], value)) {
3761 exec->index = target;
3762 if ((exec->callback != NULL) && (comp->transdata != NULL)) {
3763 exec->callback(exec->data, value,
3764 comp->transdata[state * comp->nbstrings + i], data);
3765 }
3766 #ifdef DEBUG_PUSH
3767 printf("entering state %d\n", target);
3768 #endif
3769 if (comp->compact[target * (comp->nbstrings + 1)] ==
3770 XML_REGEXP_SINK_STATE)
3771 goto error;
3772
3773 if (comp->compact[target * (comp->nbstrings + 1)] ==
3774 XML_REGEXP_FINAL_STATE)
3775 return(1);
3776 return(0);
3777 }
3778 }
3779 }
3780 /*
3781 * Failed to find an exit transition out from current state for the
3782 * current token
3783 */
3784 #ifdef DEBUG_PUSH
3785 printf("failed to find a transition for %s on state %d\n", value, state);
3786 #endif
3787 error:
3788 if (exec->errString != NULL)
3789 xmlFree(exec->errString);
3790 exec->errString = xmlStrdup(value);
3791 exec->errStateNo = state;
3792 exec->status = -1;
3793 #ifdef DEBUG_ERR
3794 testerr(exec);
3795 #endif
3796 return(-1);
3797 }
3798
3799 /**
3800 * xmlRegExecPushStringInternal:
3801 * @exec: a regexp execution context or NULL to indicate the end
3802 * @value: a string token input
3803 * @data: data associated to the token to reuse in callbacks
3804 * @compound: value was assembled from 2 strings
3805 *
3806 * Push one input token in the execution context
3807 *
3808 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
3809 * a negative value in case of error.
3810 */
3811 static int
xmlRegExecPushStringInternal(xmlRegExecCtxtPtr exec,const xmlChar * value,void * data,int compound)3812 xmlRegExecPushStringInternal(xmlRegExecCtxtPtr exec, const xmlChar *value,
3813 void *data, int compound) {
3814 xmlRegTransPtr trans;
3815 xmlRegAtomPtr atom;
3816 int ret;
3817 int final = 0;
3818 int progress = 1;
3819
3820 if (exec == NULL)
3821 return(-1);
3822 if (exec->comp == NULL)
3823 return(-1);
3824 if (exec->status != 0)
3825 return(exec->status);
3826
3827 if (exec->comp->compact != NULL)
3828 return(xmlRegCompactPushString(exec, exec->comp, value, data));
3829
3830 if (value == NULL) {
3831 if (exec->state->type == XML_REGEXP_FINAL_STATE)
3832 return(1);
3833 final = 1;
3834 }
3835
3836 #ifdef DEBUG_PUSH
3837 printf("value pushed: %s\n", value);
3838 #endif
3839 /*
3840 * If we have an active rollback stack push the new value there
3841 * and get back to where we were left
3842 */
3843 if ((value != NULL) && (exec->inputStackNr > 0)) {
3844 xmlFARegExecSaveInputString(exec, value, data);
3845 value = exec->inputStack[exec->index].value;
3846 data = exec->inputStack[exec->index].data;
3847 #ifdef DEBUG_PUSH
3848 printf("value loaded: %s\n", value);
3849 #endif
3850 }
3851
3852 while ((exec->status == 0) &&
3853 ((value != NULL) ||
3854 ((final == 1) &&
3855 (exec->state->type != XML_REGEXP_FINAL_STATE)))) {
3856
3857 /*
3858 * End of input on non-terminal state, rollback, however we may
3859 * still have epsilon like transition for counted transitions
3860 * on counters, in that case don't break too early.
3861 */
3862 if ((value == NULL) && (exec->counts == NULL))
3863 goto rollback;
3864
3865 exec->transcount = 0;
3866 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
3867 trans = &exec->state->trans[exec->transno];
3868 if (trans->to < 0)
3869 continue;
3870 atom = trans->atom;
3871 ret = 0;
3872 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
3873 int i;
3874 int count;
3875 xmlRegTransPtr t;
3876 xmlRegCounterPtr counter;
3877
3878 ret = 0;
3879
3880 #ifdef DEBUG_PUSH
3881 printf("testing all lax %d\n", trans->count);
3882 #endif
3883 /*
3884 * Check all counted transitions from the current state
3885 */
3886 if ((value == NULL) && (final)) {
3887 ret = 1;
3888 } else if (value != NULL) {
3889 for (i = 0;i < exec->state->nbTrans;i++) {
3890 t = &exec->state->trans[i];
3891 if ((t->counter < 0) || (t == trans))
3892 continue;
3893 counter = &exec->comp->counters[t->counter];
3894 count = exec->counts[t->counter];
3895 if ((count < counter->max) &&
3896 (t->atom != NULL) &&
3897 (xmlStrEqual(value, t->atom->valuep))) {
3898 ret = 0;
3899 break;
3900 }
3901 if ((count >= counter->min) &&
3902 (count < counter->max) &&
3903 (t->atom != NULL) &&
3904 (xmlStrEqual(value, t->atom->valuep))) {
3905 ret = 1;
3906 break;
3907 }
3908 }
3909 }
3910 } else if (trans->count == REGEXP_ALL_COUNTER) {
3911 int i;
3912 int count;
3913 xmlRegTransPtr t;
3914 xmlRegCounterPtr counter;
3915
3916 ret = 1;
3917
3918 #ifdef DEBUG_PUSH
3919 printf("testing all %d\n", trans->count);
3920 #endif
3921 /*
3922 * Check all counted transitions from the current state
3923 */
3924 for (i = 0;i < exec->state->nbTrans;i++) {
3925 t = &exec->state->trans[i];
3926 if ((t->counter < 0) || (t == trans))
3927 continue;
3928 counter = &exec->comp->counters[t->counter];
3929 count = exec->counts[t->counter];
3930 if ((count < counter->min) || (count > counter->max)) {
3931 ret = 0;
3932 break;
3933 }
3934 }
3935 } else if (trans->count >= 0) {
3936 int count;
3937 xmlRegCounterPtr counter;
3938
3939 /*
3940 * A counted transition.
3941 */
3942
3943 count = exec->counts[trans->count];
3944 counter = &exec->comp->counters[trans->count];
3945 #ifdef DEBUG_PUSH
3946 printf("testing count %d: val %d, min %d, max %d\n",
3947 trans->count, count, counter->min, counter->max);
3948 #endif
3949 ret = ((count >= counter->min) && (count <= counter->max));
3950 } else if (atom == NULL) {
3951 fprintf(stderr, "epsilon transition left at runtime\n");
3952 exec->status = -2;
3953 break;
3954 } else if (value != NULL) {
3955 ret = xmlRegStrEqualWildcard(atom->valuep, value);
3956 if (atom->neg) {
3957 ret = !ret;
3958 if (!compound)
3959 ret = 0;
3960 }
3961 if ((ret == 1) && (trans->counter >= 0)) {
3962 xmlRegCounterPtr counter;
3963 int count;
3964
3965 count = exec->counts[trans->counter];
3966 counter = &exec->comp->counters[trans->counter];
3967 if (count >= counter->max)
3968 ret = 0;
3969 }
3970
3971 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
3972 xmlRegStatePtr to = exec->comp->states[trans->to];
3973
3974 /*
3975 * this is a multiple input sequence
3976 */
3977 if (exec->state->nbTrans > exec->transno + 1) {
3978 if (exec->inputStackNr <= 0) {
3979 xmlFARegExecSaveInputString(exec, value, data);
3980 }
3981 xmlFARegExecSave(exec);
3982 }
3983 exec->transcount = 1;
3984 do {
3985 /*
3986 * Try to progress as much as possible on the input
3987 */
3988 if (exec->transcount == atom->max) {
3989 break;
3990 }
3991 exec->index++;
3992 value = exec->inputStack[exec->index].value;
3993 data = exec->inputStack[exec->index].data;
3994 #ifdef DEBUG_PUSH
3995 printf("value loaded: %s\n", value);
3996 #endif
3997
3998 /*
3999 * End of input: stop here
4000 */
4001 if (value == NULL) {
4002 exec->index --;
4003 break;
4004 }
4005 if (exec->transcount >= atom->min) {
4006 int transno = exec->transno;
4007 xmlRegStatePtr state = exec->state;
4008
4009 /*
4010 * The transition is acceptable save it
4011 */
4012 exec->transno = -1; /* trick */
4013 exec->state = to;
4014 if (exec->inputStackNr <= 0) {
4015 xmlFARegExecSaveInputString(exec, value, data);
4016 }
4017 xmlFARegExecSave(exec);
4018 exec->transno = transno;
4019 exec->state = state;
4020 }
4021 ret = xmlStrEqual(value, atom->valuep);
4022 exec->transcount++;
4023 } while (ret == 1);
4024 if (exec->transcount < atom->min)
4025 ret = 0;
4026
4027 /*
4028 * If the last check failed but one transition was found
4029 * possible, rollback
4030 */
4031 if (ret < 0)
4032 ret = 0;
4033 if (ret == 0) {
4034 goto rollback;
4035 }
4036 }
4037 }
4038 if (ret == 1) {
4039 if ((exec->callback != NULL) && (atom != NULL) &&
4040 (data != NULL)) {
4041 exec->callback(exec->data, atom->valuep,
4042 atom->data, data);
4043 }
4044 if (exec->state->nbTrans > exec->transno + 1) {
4045 if (exec->inputStackNr <= 0) {
4046 xmlFARegExecSaveInputString(exec, value, data);
4047 }
4048 xmlFARegExecSave(exec);
4049 }
4050 if (trans->counter >= 0) {
4051 #ifdef DEBUG_PUSH
4052 printf("Increasing count %d\n", trans->counter);
4053 #endif
4054 exec->counts[trans->counter]++;
4055 }
4056 if ((trans->count >= 0) &&
4057 (trans->count < REGEXP_ALL_COUNTER)) {
4058 #ifdef DEBUG_REGEXP_EXEC
4059 printf("resetting count %d on transition\n",
4060 trans->count);
4061 #endif
4062 exec->counts[trans->count] = 0;
4063 }
4064 #ifdef DEBUG_PUSH
4065 printf("entering state %d\n", trans->to);
4066 #endif
4067 if ((exec->comp->states[trans->to] != NULL) &&
4068 (exec->comp->states[trans->to]->type ==
4069 XML_REGEXP_SINK_STATE)) {
4070 /*
4071 * entering a sink state, save the current state as error
4072 * state.
4073 */
4074 if (exec->errString != NULL)
4075 xmlFree(exec->errString);
4076 exec->errString = xmlStrdup(value);
4077 exec->errState = exec->state;
4078 memcpy(exec->errCounts, exec->counts,
4079 exec->comp->nbCounters * sizeof(int));
4080 }
4081 exec->state = exec->comp->states[trans->to];
4082 exec->transno = 0;
4083 if (trans->atom != NULL) {
4084 if (exec->inputStack != NULL) {
4085 exec->index++;
4086 if (exec->index < exec->inputStackNr) {
4087 value = exec->inputStack[exec->index].value;
4088 data = exec->inputStack[exec->index].data;
4089 #ifdef DEBUG_PUSH
4090 printf("value loaded: %s\n", value);
4091 #endif
4092 } else {
4093 value = NULL;
4094 data = NULL;
4095 #ifdef DEBUG_PUSH
4096 printf("end of input\n");
4097 #endif
4098 }
4099 } else {
4100 value = NULL;
4101 data = NULL;
4102 #ifdef DEBUG_PUSH
4103 printf("end of input\n");
4104 #endif
4105 }
4106 }
4107 goto progress;
4108 } else if (ret < 0) {
4109 exec->status = -4;
4110 break;
4111 }
4112 }
4113 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
4114 rollback:
4115 /*
4116 * if we didn't yet rollback on the current input
4117 * store the current state as the error state.
4118 */
4119 if ((progress) && (exec->state != NULL) &&
4120 (exec->state->type != XML_REGEXP_SINK_STATE)) {
4121 progress = 0;
4122 if (exec->errString != NULL)
4123 xmlFree(exec->errString);
4124 exec->errString = xmlStrdup(value);
4125 exec->errState = exec->state;
4126 if (exec->comp->nbCounters)
4127 memcpy(exec->errCounts, exec->counts,
4128 exec->comp->nbCounters * sizeof(int));
4129 }
4130
4131 /*
4132 * Failed to find a way out
4133 */
4134 exec->determinist = 0;
4135 xmlFARegExecRollBack(exec);
4136 if ((exec->inputStack != NULL ) && (exec->status == 0)) {
4137 value = exec->inputStack[exec->index].value;
4138 data = exec->inputStack[exec->index].data;
4139 #ifdef DEBUG_PUSH
4140 printf("value loaded: %s\n", value);
4141 #endif
4142 }
4143 }
4144 continue;
4145 progress:
4146 progress = 1;
4147 continue;
4148 }
4149 if (exec->status == 0) {
4150 return(exec->state->type == XML_REGEXP_FINAL_STATE);
4151 }
4152 #ifdef DEBUG_ERR
4153 if (exec->status < 0) {
4154 testerr(exec);
4155 }
4156 #endif
4157 return(exec->status);
4158 }
4159
4160 /**
4161 * xmlRegExecPushString:
4162 * @exec: a regexp execution context or NULL to indicate the end
4163 * @value: a string token input
4164 * @data: data associated to the token to reuse in callbacks
4165 *
4166 * Push one input token in the execution context
4167 *
4168 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
4169 * a negative value in case of error.
4170 */
4171 int
xmlRegExecPushString(xmlRegExecCtxtPtr exec,const xmlChar * value,void * data)4172 xmlRegExecPushString(xmlRegExecCtxtPtr exec, const xmlChar *value,
4173 void *data) {
4174 return(xmlRegExecPushStringInternal(exec, value, data, 0));
4175 }
4176
4177 /**
4178 * xmlRegExecPushString2:
4179 * @exec: a regexp execution context or NULL to indicate the end
4180 * @value: the first string token input
4181 * @value2: the second string token input
4182 * @data: data associated to the token to reuse in callbacks
4183 *
4184 * Push one input token in the execution context
4185 *
4186 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
4187 * a negative value in case of error.
4188 */
4189 int
xmlRegExecPushString2(xmlRegExecCtxtPtr exec,const xmlChar * value,const xmlChar * value2,void * data)4190 xmlRegExecPushString2(xmlRegExecCtxtPtr exec, const xmlChar *value,
4191 const xmlChar *value2, void *data) {
4192 xmlChar buf[150];
4193 int lenn, lenp, ret;
4194 xmlChar *str;
4195
4196 if (exec == NULL)
4197 return(-1);
4198 if (exec->comp == NULL)
4199 return(-1);
4200 if (exec->status != 0)
4201 return(exec->status);
4202
4203 if (value2 == NULL)
4204 return(xmlRegExecPushString(exec, value, data));
4205
4206 lenn = strlen((char *) value2);
4207 lenp = strlen((char *) value);
4208
4209 if (150 < lenn + lenp + 2) {
4210 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
4211 if (str == NULL) {
4212 exec->status = -1;
4213 return(-1);
4214 }
4215 } else {
4216 str = buf;
4217 }
4218 memcpy(&str[0], value, lenp);
4219 str[lenp] = XML_REG_STRING_SEPARATOR;
4220 memcpy(&str[lenp + 1], value2, lenn);
4221 str[lenn + lenp + 1] = 0;
4222
4223 if (exec->comp->compact != NULL)
4224 ret = xmlRegCompactPushString(exec, exec->comp, str, data);
4225 else
4226 ret = xmlRegExecPushStringInternal(exec, str, data, 1);
4227
4228 if (str != buf)
4229 xmlFree(str);
4230 return(ret);
4231 }
4232
4233 /**
4234 * xmlRegExecGetValues:
4235 * @exec: a regexp execution context
4236 * @err: error extraction or normal one
4237 * @nbval: pointer to the number of accepted values IN/OUT
4238 * @nbneg: return number of negative transitions
4239 * @values: pointer to the array of acceptable values
4240 * @terminal: return value if this was a terminal state
4241 *
4242 * Extract informations from the regexp execution, internal routine to
4243 * implement xmlRegExecNextValues() and xmlRegExecErrInfo()
4244 *
4245 * Returns: 0 in case of success or -1 in case of error.
4246 */
4247 static int
xmlRegExecGetValues(xmlRegExecCtxtPtr exec,int err,int * nbval,int * nbneg,xmlChar ** values,int * terminal)4248 xmlRegExecGetValues(xmlRegExecCtxtPtr exec, int err,
4249 int *nbval, int *nbneg,
4250 xmlChar **values, int *terminal) {
4251 int maxval;
4252 int nb = 0;
4253
4254 if ((exec == NULL) || (nbval == NULL) || (nbneg == NULL) ||
4255 (values == NULL) || (*nbval <= 0))
4256 return(-1);
4257
4258 maxval = *nbval;
4259 *nbval = 0;
4260 *nbneg = 0;
4261 if ((exec->comp != NULL) && (exec->comp->compact != NULL)) {
4262 xmlRegexpPtr comp;
4263 int target, i, state;
4264
4265 comp = exec->comp;
4266
4267 if (err) {
4268 if (exec->errStateNo == -1) return(-1);
4269 state = exec->errStateNo;
4270 } else {
4271 state = exec->index;
4272 }
4273 if (terminal != NULL) {
4274 if (comp->compact[state * (comp->nbstrings + 1)] ==
4275 XML_REGEXP_FINAL_STATE)
4276 *terminal = 1;
4277 else
4278 *terminal = 0;
4279 }
4280 for (i = 0;(i < comp->nbstrings) && (nb < maxval);i++) {
4281 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
4282 if ((target > 0) && (target <= comp->nbstates) &&
4283 (comp->compact[(target - 1) * (comp->nbstrings + 1)] !=
4284 XML_REGEXP_SINK_STATE)) {
4285 values[nb++] = comp->stringMap[i];
4286 (*nbval)++;
4287 }
4288 }
4289 for (i = 0;(i < comp->nbstrings) && (nb < maxval);i++) {
4290 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
4291 if ((target > 0) && (target <= comp->nbstates) &&
4292 (comp->compact[(target - 1) * (comp->nbstrings + 1)] ==
4293 XML_REGEXP_SINK_STATE)) {
4294 values[nb++] = comp->stringMap[i];
4295 (*nbneg)++;
4296 }
4297 }
4298 } else {
4299 int transno;
4300 xmlRegTransPtr trans;
4301 xmlRegAtomPtr atom;
4302 xmlRegStatePtr state;
4303
4304 if (terminal != NULL) {
4305 if (exec->state->type == XML_REGEXP_FINAL_STATE)
4306 *terminal = 1;
4307 else
4308 *terminal = 0;
4309 }
4310
4311 if (err) {
4312 if (exec->errState == NULL) return(-1);
4313 state = exec->errState;
4314 } else {
4315 if (exec->state == NULL) return(-1);
4316 state = exec->state;
4317 }
4318 for (transno = 0;
4319 (transno < state->nbTrans) && (nb < maxval);
4320 transno++) {
4321 trans = &state->trans[transno];
4322 if (trans->to < 0)
4323 continue;
4324 atom = trans->atom;
4325 if ((atom == NULL) || (atom->valuep == NULL))
4326 continue;
4327 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
4328 /* this should not be reached but ... */
4329 TODO;
4330 } else if (trans->count == REGEXP_ALL_COUNTER) {
4331 /* this should not be reached but ... */
4332 TODO;
4333 } else if (trans->counter >= 0) {
4334 xmlRegCounterPtr counter = NULL;
4335 int count;
4336
4337 if (err)
4338 count = exec->errCounts[trans->counter];
4339 else
4340 count = exec->counts[trans->counter];
4341 if (exec->comp != NULL)
4342 counter = &exec->comp->counters[trans->counter];
4343 if ((counter == NULL) || (count < counter->max)) {
4344 if (atom->neg)
4345 values[nb++] = (xmlChar *) atom->valuep2;
4346 else
4347 values[nb++] = (xmlChar *) atom->valuep;
4348 (*nbval)++;
4349 }
4350 } else {
4351 if ((exec->comp != NULL) && (exec->comp->states[trans->to] != NULL) &&
4352 (exec->comp->states[trans->to]->type !=
4353 XML_REGEXP_SINK_STATE)) {
4354 if (atom->neg)
4355 values[nb++] = (xmlChar *) atom->valuep2;
4356 else
4357 values[nb++] = (xmlChar *) atom->valuep;
4358 (*nbval)++;
4359 }
4360 }
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 continue;
4373 } else if (trans->count == REGEXP_ALL_COUNTER) {
4374 continue;
4375 } else if (trans->counter >= 0) {
4376 continue;
4377 } else {
4378 if ((exec->comp->states[trans->to] != NULL) &&
4379 (exec->comp->states[trans->to]->type ==
4380 XML_REGEXP_SINK_STATE)) {
4381 if (atom->neg)
4382 values[nb++] = (xmlChar *) atom->valuep2;
4383 else
4384 values[nb++] = (xmlChar *) atom->valuep;
4385 (*nbneg)++;
4386 }
4387 }
4388 }
4389 }
4390 return(0);
4391 }
4392
4393 /**
4394 * xmlRegExecNextValues:
4395 * @exec: a regexp execution context
4396 * @nbval: pointer to the number of accepted values IN/OUT
4397 * @nbneg: return number of negative transitions
4398 * @values: pointer to the array of acceptable values
4399 * @terminal: return value if this was a terminal state
4400 *
4401 * Extract informations from the regexp execution,
4402 * the parameter @values must point to an array of @nbval string pointers
4403 * on return nbval will contain the number of possible strings in that
4404 * state and the @values array will be updated with them. The string values
4405 * returned will be freed with the @exec context and don't need to be
4406 * deallocated.
4407 *
4408 * Returns: 0 in case of success or -1 in case of error.
4409 */
4410 int
xmlRegExecNextValues(xmlRegExecCtxtPtr exec,int * nbval,int * nbneg,xmlChar ** values,int * terminal)4411 xmlRegExecNextValues(xmlRegExecCtxtPtr exec, int *nbval, int *nbneg,
4412 xmlChar **values, int *terminal) {
4413 return(xmlRegExecGetValues(exec, 0, nbval, nbneg, values, terminal));
4414 }
4415
4416 /**
4417 * xmlRegExecErrInfo:
4418 * @exec: a regexp execution context generating an error
4419 * @string: return value for the error string
4420 * @nbval: pointer to the number of accepted values IN/OUT
4421 * @nbneg: return number of negative transitions
4422 * @values: pointer to the array of acceptable values
4423 * @terminal: return value if this was a terminal state
4424 *
4425 * Extract error informations from the regexp execution, the parameter
4426 * @string will be updated with the value pushed and not accepted,
4427 * the parameter @values must point to an array of @nbval string pointers
4428 * on return nbval will contain the number of possible strings in that
4429 * state and the @values array will be updated with them. The string values
4430 * returned will be freed with the @exec context and don't need to be
4431 * deallocated.
4432 *
4433 * Returns: 0 in case of success or -1 in case of error.
4434 */
4435 int
xmlRegExecErrInfo(xmlRegExecCtxtPtr exec,const xmlChar ** string,int * nbval,int * nbneg,xmlChar ** values,int * terminal)4436 xmlRegExecErrInfo(xmlRegExecCtxtPtr exec, const xmlChar **string,
4437 int *nbval, int *nbneg, xmlChar **values, int *terminal) {
4438 if (exec == NULL)
4439 return(-1);
4440 if (string != NULL) {
4441 if (exec->status != 0)
4442 *string = exec->errString;
4443 else
4444 *string = NULL;
4445 }
4446 return(xmlRegExecGetValues(exec, 1, nbval, nbneg, values, terminal));
4447 }
4448
4449 #ifdef DEBUG_ERR
testerr(xmlRegExecCtxtPtr exec)4450 static void testerr(xmlRegExecCtxtPtr exec) {
4451 const xmlChar *string;
4452 xmlChar *values[5];
4453 int nb = 5;
4454 int nbneg;
4455 int terminal;
4456 xmlRegExecErrInfo(exec, &string, &nb, &nbneg, &values[0], &terminal);
4457 }
4458 #endif
4459
4460 #if 0
4461 static int
4462 xmlRegExecPushChar(xmlRegExecCtxtPtr exec, int UCS) {
4463 xmlRegTransPtr trans;
4464 xmlRegAtomPtr atom;
4465 int ret;
4466 int codepoint, len;
4467
4468 if (exec == NULL)
4469 return(-1);
4470 if (exec->status != 0)
4471 return(exec->status);
4472
4473 while ((exec->status == 0) &&
4474 ((exec->inputString[exec->index] != 0) ||
4475 (exec->state->type != XML_REGEXP_FINAL_STATE))) {
4476
4477 /*
4478 * End of input on non-terminal state, rollback, however we may
4479 * still have epsilon like transition for counted transitions
4480 * on counters, in that case don't break too early.
4481 */
4482 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL))
4483 goto rollback;
4484
4485 exec->transcount = 0;
4486 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
4487 trans = &exec->state->trans[exec->transno];
4488 if (trans->to < 0)
4489 continue;
4490 atom = trans->atom;
4491 ret = 0;
4492 if (trans->count >= 0) {
4493 int count;
4494 xmlRegCounterPtr counter;
4495
4496 /*
4497 * A counted transition.
4498 */
4499
4500 count = exec->counts[trans->count];
4501 counter = &exec->comp->counters[trans->count];
4502 #ifdef DEBUG_REGEXP_EXEC
4503 printf("testing count %d: val %d, min %d, max %d\n",
4504 trans->count, count, counter->min, counter->max);
4505 #endif
4506 ret = ((count >= counter->min) && (count <= counter->max));
4507 } else if (atom == NULL) {
4508 fprintf(stderr, "epsilon transition left at runtime\n");
4509 exec->status = -2;
4510 break;
4511 } else if (exec->inputString[exec->index] != 0) {
4512 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
4513 ret = xmlRegCheckCharacter(atom, codepoint);
4514 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
4515 xmlRegStatePtr to = exec->comp->states[trans->to];
4516
4517 /*
4518 * this is a multiple input sequence
4519 */
4520 if (exec->state->nbTrans > exec->transno + 1) {
4521 xmlFARegExecSave(exec);
4522 }
4523 exec->transcount = 1;
4524 do {
4525 /*
4526 * Try to progress as much as possible on the input
4527 */
4528 if (exec->transcount == atom->max) {
4529 break;
4530 }
4531 exec->index += len;
4532 /*
4533 * End of input: stop here
4534 */
4535 if (exec->inputString[exec->index] == 0) {
4536 exec->index -= len;
4537 break;
4538 }
4539 if (exec->transcount >= atom->min) {
4540 int transno = exec->transno;
4541 xmlRegStatePtr state = exec->state;
4542
4543 /*
4544 * The transition is acceptable save it
4545 */
4546 exec->transno = -1; /* trick */
4547 exec->state = to;
4548 xmlFARegExecSave(exec);
4549 exec->transno = transno;
4550 exec->state = state;
4551 }
4552 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
4553 len);
4554 ret = xmlRegCheckCharacter(atom, codepoint);
4555 exec->transcount++;
4556 } while (ret == 1);
4557 if (exec->transcount < atom->min)
4558 ret = 0;
4559
4560 /*
4561 * If the last check failed but one transition was found
4562 * possible, rollback
4563 */
4564 if (ret < 0)
4565 ret = 0;
4566 if (ret == 0) {
4567 goto rollback;
4568 }
4569 }
4570 }
4571 if (ret == 1) {
4572 if (exec->state->nbTrans > exec->transno + 1) {
4573 xmlFARegExecSave(exec);
4574 }
4575 /*
4576 * restart count for expressions like this ((abc){2})*
4577 */
4578 if (trans->count >= 0) {
4579 #ifdef DEBUG_REGEXP_EXEC
4580 printf("Reset count %d\n", trans->count);
4581 #endif
4582 exec->counts[trans->count] = 0;
4583 }
4584 if (trans->counter >= 0) {
4585 #ifdef DEBUG_REGEXP_EXEC
4586 printf("Increasing count %d\n", trans->counter);
4587 #endif
4588 exec->counts[trans->counter]++;
4589 }
4590 #ifdef DEBUG_REGEXP_EXEC
4591 printf("entering state %d\n", trans->to);
4592 #endif
4593 exec->state = exec->comp->states[trans->to];
4594 exec->transno = 0;
4595 if (trans->atom != NULL) {
4596 exec->index += len;
4597 }
4598 goto progress;
4599 } else if (ret < 0) {
4600 exec->status = -4;
4601 break;
4602 }
4603 }
4604 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
4605 rollback:
4606 /*
4607 * Failed to find a way out
4608 */
4609 exec->determinist = 0;
4610 xmlFARegExecRollBack(exec);
4611 }
4612 progress:
4613 continue;
4614 }
4615 }
4616 #endif
4617 /************************************************************************
4618 * *
4619 * Parser for the Schemas Datatype Regular Expressions *
4620 * http://www.w3.org/TR/2001/REC-xmlschema-2-20010502/#regexs *
4621 * *
4622 ************************************************************************/
4623
4624 /**
4625 * xmlFAIsChar:
4626 * @ctxt: a regexp parser context
4627 *
4628 * [10] Char ::= [^.\?*+()|#x5B#x5D]
4629 */
4630 static int
xmlFAIsChar(xmlRegParserCtxtPtr ctxt)4631 xmlFAIsChar(xmlRegParserCtxtPtr ctxt) {
4632 int cur;
4633 int len;
4634
4635 cur = CUR_SCHAR(ctxt->cur, len);
4636 if ((cur == '.') || (cur == '\\') || (cur == '?') ||
4637 (cur == '*') || (cur == '+') || (cur == '(') ||
4638 (cur == ')') || (cur == '|') || (cur == 0x5B) ||
4639 (cur == 0x5D) || (cur == 0))
4640 return(-1);
4641 return(cur);
4642 }
4643
4644 /**
4645 * xmlFAParseCharProp:
4646 * @ctxt: a regexp parser context
4647 *
4648 * [27] charProp ::= IsCategory | IsBlock
4649 * [28] IsCategory ::= Letters | Marks | Numbers | Punctuation |
4650 * Separators | Symbols | Others
4651 * [29] Letters ::= 'L' [ultmo]?
4652 * [30] Marks ::= 'M' [nce]?
4653 * [31] Numbers ::= 'N' [dlo]?
4654 * [32] Punctuation ::= 'P' [cdseifo]?
4655 * [33] Separators ::= 'Z' [slp]?
4656 * [34] Symbols ::= 'S' [mcko]?
4657 * [35] Others ::= 'C' [cfon]?
4658 * [36] IsBlock ::= 'Is' [a-zA-Z0-9#x2D]+
4659 */
4660 static void
xmlFAParseCharProp(xmlRegParserCtxtPtr ctxt)4661 xmlFAParseCharProp(xmlRegParserCtxtPtr ctxt) {
4662 int cur;
4663 xmlRegAtomType type = (xmlRegAtomType) 0;
4664 xmlChar *blockName = NULL;
4665
4666 cur = CUR;
4667 if (cur == 'L') {
4668 NEXT;
4669 cur = CUR;
4670 if (cur == 'u') {
4671 NEXT;
4672 type = XML_REGEXP_LETTER_UPPERCASE;
4673 } else if (cur == 'l') {
4674 NEXT;
4675 type = XML_REGEXP_LETTER_LOWERCASE;
4676 } else if (cur == 't') {
4677 NEXT;
4678 type = XML_REGEXP_LETTER_TITLECASE;
4679 } else if (cur == 'm') {
4680 NEXT;
4681 type = XML_REGEXP_LETTER_MODIFIER;
4682 } else if (cur == 'o') {
4683 NEXT;
4684 type = XML_REGEXP_LETTER_OTHERS;
4685 } else {
4686 type = XML_REGEXP_LETTER;
4687 }
4688 } else if (cur == 'M') {
4689 NEXT;
4690 cur = CUR;
4691 if (cur == 'n') {
4692 NEXT;
4693 /* nonspacing */
4694 type = XML_REGEXP_MARK_NONSPACING;
4695 } else if (cur == 'c') {
4696 NEXT;
4697 /* spacing combining */
4698 type = XML_REGEXP_MARK_SPACECOMBINING;
4699 } else if (cur == 'e') {
4700 NEXT;
4701 /* enclosing */
4702 type = XML_REGEXP_MARK_ENCLOSING;
4703 } else {
4704 /* all marks */
4705 type = XML_REGEXP_MARK;
4706 }
4707 } else if (cur == 'N') {
4708 NEXT;
4709 cur = CUR;
4710 if (cur == 'd') {
4711 NEXT;
4712 /* digital */
4713 type = XML_REGEXP_NUMBER_DECIMAL;
4714 } else if (cur == 'l') {
4715 NEXT;
4716 /* letter */
4717 type = XML_REGEXP_NUMBER_LETTER;
4718 } else if (cur == 'o') {
4719 NEXT;
4720 /* other */
4721 type = XML_REGEXP_NUMBER_OTHERS;
4722 } else {
4723 /* all numbers */
4724 type = XML_REGEXP_NUMBER;
4725 }
4726 } else if (cur == 'P') {
4727 NEXT;
4728 cur = CUR;
4729 if (cur == 'c') {
4730 NEXT;
4731 /* connector */
4732 type = XML_REGEXP_PUNCT_CONNECTOR;
4733 } else if (cur == 'd') {
4734 NEXT;
4735 /* dash */
4736 type = XML_REGEXP_PUNCT_DASH;
4737 } else if (cur == 's') {
4738 NEXT;
4739 /* open */
4740 type = XML_REGEXP_PUNCT_OPEN;
4741 } else if (cur == 'e') {
4742 NEXT;
4743 /* close */
4744 type = XML_REGEXP_PUNCT_CLOSE;
4745 } else if (cur == 'i') {
4746 NEXT;
4747 /* initial quote */
4748 type = XML_REGEXP_PUNCT_INITQUOTE;
4749 } else if (cur == 'f') {
4750 NEXT;
4751 /* final quote */
4752 type = XML_REGEXP_PUNCT_FINQUOTE;
4753 } else if (cur == 'o') {
4754 NEXT;
4755 /* other */
4756 type = XML_REGEXP_PUNCT_OTHERS;
4757 } else {
4758 /* all punctuation */
4759 type = XML_REGEXP_PUNCT;
4760 }
4761 } else if (cur == 'Z') {
4762 NEXT;
4763 cur = CUR;
4764 if (cur == 's') {
4765 NEXT;
4766 /* space */
4767 type = XML_REGEXP_SEPAR_SPACE;
4768 } else if (cur == 'l') {
4769 NEXT;
4770 /* line */
4771 type = XML_REGEXP_SEPAR_LINE;
4772 } else if (cur == 'p') {
4773 NEXT;
4774 /* paragraph */
4775 type = XML_REGEXP_SEPAR_PARA;
4776 } else {
4777 /* all separators */
4778 type = XML_REGEXP_SEPAR;
4779 }
4780 } else if (cur == 'S') {
4781 NEXT;
4782 cur = CUR;
4783 if (cur == 'm') {
4784 NEXT;
4785 type = XML_REGEXP_SYMBOL_MATH;
4786 /* math */
4787 } else if (cur == 'c') {
4788 NEXT;
4789 type = XML_REGEXP_SYMBOL_CURRENCY;
4790 /* currency */
4791 } else if (cur == 'k') {
4792 NEXT;
4793 type = XML_REGEXP_SYMBOL_MODIFIER;
4794 /* modifiers */
4795 } else if (cur == 'o') {
4796 NEXT;
4797 type = XML_REGEXP_SYMBOL_OTHERS;
4798 /* other */
4799 } else {
4800 /* all symbols */
4801 type = XML_REGEXP_SYMBOL;
4802 }
4803 } else if (cur == 'C') {
4804 NEXT;
4805 cur = CUR;
4806 if (cur == 'c') {
4807 NEXT;
4808 /* control */
4809 type = XML_REGEXP_OTHER_CONTROL;
4810 } else if (cur == 'f') {
4811 NEXT;
4812 /* format */
4813 type = XML_REGEXP_OTHER_FORMAT;
4814 } else if (cur == 'o') {
4815 NEXT;
4816 /* private use */
4817 type = XML_REGEXP_OTHER_PRIVATE;
4818 } else if (cur == 'n') {
4819 NEXT;
4820 /* not assigned */
4821 type = XML_REGEXP_OTHER_NA;
4822 } else {
4823 /* all others */
4824 type = XML_REGEXP_OTHER;
4825 }
4826 } else if (cur == 'I') {
4827 const xmlChar *start;
4828 NEXT;
4829 cur = CUR;
4830 if (cur != 's') {
4831 ERROR("IsXXXX expected");
4832 return;
4833 }
4834 NEXT;
4835 start = ctxt->cur;
4836 cur = CUR;
4837 if (((cur >= 'a') && (cur <= 'z')) ||
4838 ((cur >= 'A') && (cur <= 'Z')) ||
4839 ((cur >= '0') && (cur <= '9')) ||
4840 (cur == 0x2D)) {
4841 NEXT;
4842 cur = CUR;
4843 while (((cur >= 'a') && (cur <= 'z')) ||
4844 ((cur >= 'A') && (cur <= 'Z')) ||
4845 ((cur >= '0') && (cur <= '9')) ||
4846 (cur == 0x2D)) {
4847 NEXT;
4848 cur = CUR;
4849 }
4850 }
4851 type = XML_REGEXP_BLOCK_NAME;
4852 blockName = xmlStrndup(start, ctxt->cur - start);
4853 } else {
4854 ERROR("Unknown char property");
4855 return;
4856 }
4857 if (ctxt->atom == NULL) {
4858 ctxt->atom = xmlRegNewAtom(ctxt, type);
4859 if (ctxt->atom != NULL)
4860 ctxt->atom->valuep = blockName;
4861 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4862 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4863 type, 0, 0, blockName);
4864 }
4865 }
4866
4867 /**
4868 * xmlFAParseCharClassEsc:
4869 * @ctxt: a regexp parser context
4870 *
4871 * [23] charClassEsc ::= ( SingleCharEsc | MultiCharEsc | catEsc | complEsc )
4872 * [24] SingleCharEsc ::= '\' [nrt\|.?*+(){}#x2D#x5B#x5D#x5E]
4873 * [25] catEsc ::= '\p{' charProp '}'
4874 * [26] complEsc ::= '\P{' charProp '}'
4875 * [37] MultiCharEsc ::= '.' | ('\' [sSiIcCdDwW])
4876 */
4877 static void
xmlFAParseCharClassEsc(xmlRegParserCtxtPtr ctxt)4878 xmlFAParseCharClassEsc(xmlRegParserCtxtPtr ctxt) {
4879 int cur;
4880
4881 if (CUR == '.') {
4882 if (ctxt->atom == NULL) {
4883 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_ANYCHAR);
4884 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4885 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4886 XML_REGEXP_ANYCHAR, 0, 0, NULL);
4887 }
4888 NEXT;
4889 return;
4890 }
4891 if (CUR != '\\') {
4892 ERROR("Escaped sequence: expecting \\");
4893 return;
4894 }
4895 NEXT;
4896 cur = CUR;
4897 if (cur == 'p') {
4898 NEXT;
4899 if (CUR != '{') {
4900 ERROR("Expecting '{'");
4901 return;
4902 }
4903 NEXT;
4904 xmlFAParseCharProp(ctxt);
4905 if (CUR != '}') {
4906 ERROR("Expecting '}'");
4907 return;
4908 }
4909 NEXT;
4910 } else if (cur == 'P') {
4911 NEXT;
4912 if (CUR != '{') {
4913 ERROR("Expecting '{'");
4914 return;
4915 }
4916 NEXT;
4917 xmlFAParseCharProp(ctxt);
4918 if (ctxt->atom != NULL)
4919 ctxt->atom->neg = 1;
4920 if (CUR != '}') {
4921 ERROR("Expecting '}'");
4922 return;
4923 }
4924 NEXT;
4925 } else if ((cur == 'n') || (cur == 'r') || (cur == 't') || (cur == '\\') ||
4926 (cur == '|') || (cur == '.') || (cur == '?') || (cur == '*') ||
4927 (cur == '+') || (cur == '(') || (cur == ')') || (cur == '{') ||
4928 (cur == '}') || (cur == 0x2D) || (cur == 0x5B) || (cur == 0x5D) ||
4929 (cur == 0x5E)) {
4930 if (ctxt->atom == NULL) {
4931 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
4932 if (ctxt->atom != NULL) {
4933 switch (cur) {
4934 case 'n':
4935 ctxt->atom->codepoint = '\n';
4936 break;
4937 case 'r':
4938 ctxt->atom->codepoint = '\r';
4939 break;
4940 case 't':
4941 ctxt->atom->codepoint = '\t';
4942 break;
4943 default:
4944 ctxt->atom->codepoint = cur;
4945 }
4946 }
4947 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4948 switch (cur) {
4949 case 'n':
4950 cur = '\n';
4951 break;
4952 case 'r':
4953 cur = '\r';
4954 break;
4955 case 't':
4956 cur = '\t';
4957 break;
4958 }
4959 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4960 XML_REGEXP_CHARVAL, cur, cur, NULL);
4961 }
4962 NEXT;
4963 } else if ((cur == 's') || (cur == 'S') || (cur == 'i') || (cur == 'I') ||
4964 (cur == 'c') || (cur == 'C') || (cur == 'd') || (cur == 'D') ||
4965 (cur == 'w') || (cur == 'W')) {
4966 xmlRegAtomType type = XML_REGEXP_ANYSPACE;
4967
4968 switch (cur) {
4969 case 's':
4970 type = XML_REGEXP_ANYSPACE;
4971 break;
4972 case 'S':
4973 type = XML_REGEXP_NOTSPACE;
4974 break;
4975 case 'i':
4976 type = XML_REGEXP_INITNAME;
4977 break;
4978 case 'I':
4979 type = XML_REGEXP_NOTINITNAME;
4980 break;
4981 case 'c':
4982 type = XML_REGEXP_NAMECHAR;
4983 break;
4984 case 'C':
4985 type = XML_REGEXP_NOTNAMECHAR;
4986 break;
4987 case 'd':
4988 type = XML_REGEXP_DECIMAL;
4989 break;
4990 case 'D':
4991 type = XML_REGEXP_NOTDECIMAL;
4992 break;
4993 case 'w':
4994 type = XML_REGEXP_REALCHAR;
4995 break;
4996 case 'W':
4997 type = XML_REGEXP_NOTREALCHAR;
4998 break;
4999 }
5000 NEXT;
5001 if (ctxt->atom == NULL) {
5002 ctxt->atom = xmlRegNewAtom(ctxt, type);
5003 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
5004 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5005 type, 0, 0, NULL);
5006 }
5007 } else {
5008 ERROR("Wrong escape sequence, misuse of character '\\'");
5009 }
5010 }
5011
5012 /**
5013 * xmlFAParseCharRange:
5014 * @ctxt: a regexp parser context
5015 *
5016 * [17] charRange ::= seRange | XmlCharRef | XmlCharIncDash
5017 * [18] seRange ::= charOrEsc '-' charOrEsc
5018 * [20] charOrEsc ::= XmlChar | SingleCharEsc
5019 * [21] XmlChar ::= [^\#x2D#x5B#x5D]
5020 * [22] XmlCharIncDash ::= [^\#x5B#x5D]
5021 */
5022 static void
xmlFAParseCharRange(xmlRegParserCtxtPtr ctxt)5023 xmlFAParseCharRange(xmlRegParserCtxtPtr ctxt) {
5024 int cur, len;
5025 int start = -1;
5026 int end = -1;
5027
5028 if (CUR == '\0') {
5029 ERROR("Expecting ']'");
5030 return;
5031 }
5032
5033 cur = CUR;
5034 if (cur == '\\') {
5035 NEXT;
5036 cur = CUR;
5037 switch (cur) {
5038 case 'n': start = 0xA; break;
5039 case 'r': start = 0xD; break;
5040 case 't': start = 0x9; break;
5041 case '\\': case '|': case '.': case '-': case '^': case '?':
5042 case '*': case '+': case '{': case '}': case '(': case ')':
5043 case '[': case ']':
5044 start = cur; break;
5045 default:
5046 ERROR("Invalid escape value");
5047 return;
5048 }
5049 end = start;
5050 len = 1;
5051 } else if ((cur != 0x5B) && (cur != 0x5D)) {
5052 end = start = CUR_SCHAR(ctxt->cur, len);
5053 } else {
5054 ERROR("Expecting a char range");
5055 return;
5056 }
5057 /*
5058 * Since we are "inside" a range, we can assume ctxt->cur is past
5059 * the start of ctxt->string, and PREV should be safe
5060 */
5061 if ((start == '-') && (NXT(1) != ']') && (PREV != '[') && (PREV != '^')) {
5062 NEXTL(len);
5063 return;
5064 }
5065 NEXTL(len);
5066 cur = CUR;
5067 if ((cur != '-') || (NXT(1) == ']')) {
5068 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5069 XML_REGEXP_CHARVAL, start, end, NULL);
5070 return;
5071 }
5072 NEXT;
5073 cur = CUR;
5074 if (cur == '\\') {
5075 NEXT;
5076 cur = CUR;
5077 switch (cur) {
5078 case 'n': end = 0xA; break;
5079 case 'r': end = 0xD; break;
5080 case 't': end = 0x9; break;
5081 case '\\': case '|': case '.': case '-': case '^': case '?':
5082 case '*': case '+': case '{': case '}': case '(': case ')':
5083 case '[': case ']':
5084 end = cur; break;
5085 default:
5086 ERROR("Invalid escape value");
5087 return;
5088 }
5089 len = 1;
5090 } else if ((cur != '\0') && (cur != 0x5B) && (cur != 0x5D)) {
5091 end = CUR_SCHAR(ctxt->cur, len);
5092 } else {
5093 ERROR("Expecting the end of a char range");
5094 return;
5095 }
5096
5097 /* TODO check that the values are acceptable character ranges for XML */
5098 if (end < start) {
5099 ERROR("End of range is before start of range");
5100 } else {
5101 NEXTL(len);
5102 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5103 XML_REGEXP_CHARVAL, start, end, NULL);
5104 }
5105 return;
5106 }
5107
5108 /**
5109 * xmlFAParsePosCharGroup:
5110 * @ctxt: a regexp parser context
5111 *
5112 * [14] posCharGroup ::= ( charRange | charClassEsc )+
5113 */
5114 static void
xmlFAParsePosCharGroup(xmlRegParserCtxtPtr ctxt)5115 xmlFAParsePosCharGroup(xmlRegParserCtxtPtr ctxt) {
5116 do {
5117 if (CUR == '\\') {
5118 xmlFAParseCharClassEsc(ctxt);
5119 } else {
5120 xmlFAParseCharRange(ctxt);
5121 }
5122 } while ((CUR != ']') && (CUR != '^') && (CUR != '-') &&
5123 (CUR != 0) && (ctxt->error == 0));
5124 }
5125
5126 /**
5127 * xmlFAParseCharGroup:
5128 * @ctxt: a regexp parser context
5129 *
5130 * [13] charGroup ::= posCharGroup | negCharGroup | charClassSub
5131 * [15] negCharGroup ::= '^' posCharGroup
5132 * [16] charClassSub ::= ( posCharGroup | negCharGroup ) '-' charClassExpr
5133 * [12] charClassExpr ::= '[' charGroup ']'
5134 */
5135 static void
xmlFAParseCharGroup(xmlRegParserCtxtPtr ctxt)5136 xmlFAParseCharGroup(xmlRegParserCtxtPtr ctxt) {
5137 int n = ctxt->neg;
5138 while ((CUR != ']') && (ctxt->error == 0)) {
5139 if (CUR == '^') {
5140 int neg = ctxt->neg;
5141
5142 NEXT;
5143 ctxt->neg = !ctxt->neg;
5144 xmlFAParsePosCharGroup(ctxt);
5145 ctxt->neg = neg;
5146 } else if ((CUR == '-') && (NXT(1) == '[')) {
5147 int neg = ctxt->neg;
5148 ctxt->neg = 2;
5149 NEXT; /* eat the '-' */
5150 NEXT; /* eat the '[' */
5151 xmlFAParseCharGroup(ctxt);
5152 if (CUR == ']') {
5153 NEXT;
5154 } else {
5155 ERROR("charClassExpr: ']' expected");
5156 break;
5157 }
5158 ctxt->neg = neg;
5159 break;
5160 } else if (CUR != ']') {
5161 xmlFAParsePosCharGroup(ctxt);
5162 }
5163 }
5164 ctxt->neg = n;
5165 }
5166
5167 /**
5168 * xmlFAParseCharClass:
5169 * @ctxt: a regexp parser context
5170 *
5171 * [11] charClass ::= charClassEsc | charClassExpr
5172 * [12] charClassExpr ::= '[' charGroup ']'
5173 */
5174 static void
xmlFAParseCharClass(xmlRegParserCtxtPtr ctxt)5175 xmlFAParseCharClass(xmlRegParserCtxtPtr ctxt) {
5176 if (CUR == '[') {
5177 NEXT;
5178 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_RANGES);
5179 if (ctxt->atom == NULL)
5180 return;
5181 xmlFAParseCharGroup(ctxt);
5182 if (CUR == ']') {
5183 NEXT;
5184 } else {
5185 ERROR("xmlFAParseCharClass: ']' expected");
5186 }
5187 } else {
5188 xmlFAParseCharClassEsc(ctxt);
5189 }
5190 }
5191
5192 /**
5193 * xmlFAParseQuantExact:
5194 * @ctxt: a regexp parser context
5195 *
5196 * [8] QuantExact ::= [0-9]+
5197 *
5198 * Returns 0 if success or -1 in case of error
5199 */
5200 static int
xmlFAParseQuantExact(xmlRegParserCtxtPtr ctxt)5201 xmlFAParseQuantExact(xmlRegParserCtxtPtr ctxt) {
5202 int ret = 0;
5203 int ok = 0;
5204
5205 while ((CUR >= '0') && (CUR <= '9')) {
5206 ret = ret * 10 + (CUR - '0');
5207 ok = 1;
5208 NEXT;
5209 }
5210 if (ok != 1) {
5211 return(-1);
5212 }
5213 return(ret);
5214 }
5215
5216 /**
5217 * xmlFAParseQuantifier:
5218 * @ctxt: a regexp parser context
5219 *
5220 * [4] quantifier ::= [?*+] | ( '{' quantity '}' )
5221 * [5] quantity ::= quantRange | quantMin | QuantExact
5222 * [6] quantRange ::= QuantExact ',' QuantExact
5223 * [7] quantMin ::= QuantExact ','
5224 * [8] QuantExact ::= [0-9]+
5225 */
5226 static int
xmlFAParseQuantifier(xmlRegParserCtxtPtr ctxt)5227 xmlFAParseQuantifier(xmlRegParserCtxtPtr ctxt) {
5228 int cur;
5229
5230 cur = CUR;
5231 if ((cur == '?') || (cur == '*') || (cur == '+')) {
5232 if (ctxt->atom != NULL) {
5233 if (cur == '?')
5234 ctxt->atom->quant = XML_REGEXP_QUANT_OPT;
5235 else if (cur == '*')
5236 ctxt->atom->quant = XML_REGEXP_QUANT_MULT;
5237 else if (cur == '+')
5238 ctxt->atom->quant = XML_REGEXP_QUANT_PLUS;
5239 }
5240 NEXT;
5241 return(1);
5242 }
5243 if (cur == '{') {
5244 int min = 0, max = 0;
5245
5246 NEXT;
5247 cur = xmlFAParseQuantExact(ctxt);
5248 if (cur >= 0)
5249 min = cur;
5250 if (CUR == ',') {
5251 NEXT;
5252 if (CUR == '}')
5253 max = INT_MAX;
5254 else {
5255 cur = xmlFAParseQuantExact(ctxt);
5256 if (cur >= 0)
5257 max = cur;
5258 else {
5259 ERROR("Improper quantifier");
5260 }
5261 }
5262 }
5263 if (CUR == '}') {
5264 NEXT;
5265 } else {
5266 ERROR("Unterminated quantifier");
5267 }
5268 if (max == 0)
5269 max = min;
5270 if (ctxt->atom != NULL) {
5271 ctxt->atom->quant = XML_REGEXP_QUANT_RANGE;
5272 ctxt->atom->min = min;
5273 ctxt->atom->max = max;
5274 }
5275 return(1);
5276 }
5277 return(0);
5278 }
5279
5280 /**
5281 * xmlFAParseAtom:
5282 * @ctxt: a regexp parser context
5283 *
5284 * [9] atom ::= Char | charClass | ( '(' regExp ')' )
5285 */
5286 static int
xmlFAParseAtom(xmlRegParserCtxtPtr ctxt)5287 xmlFAParseAtom(xmlRegParserCtxtPtr ctxt) {
5288 int codepoint, len;
5289
5290 codepoint = xmlFAIsChar(ctxt);
5291 if (codepoint > 0) {
5292 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
5293 if (ctxt->atom == NULL)
5294 return(-1);
5295 codepoint = CUR_SCHAR(ctxt->cur, len);
5296 ctxt->atom->codepoint = codepoint;
5297 NEXTL(len);
5298 return(1);
5299 } else if (CUR == '|') {
5300 return(0);
5301 } else if (CUR == 0) {
5302 return(0);
5303 } else if (CUR == ')') {
5304 return(0);
5305 } else if (CUR == '(') {
5306 xmlRegStatePtr start, oldend, start0;
5307
5308 NEXT;
5309 /*
5310 * this extra Epsilon transition is needed if we count with 0 allowed
5311 * unfortunately this can't be known at that point
5312 */
5313 xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, NULL);
5314 start0 = ctxt->state;
5315 xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, NULL);
5316 start = ctxt->state;
5317 oldend = ctxt->end;
5318 ctxt->end = NULL;
5319 ctxt->atom = NULL;
5320 xmlFAParseRegExp(ctxt, 0);
5321 if (CUR == ')') {
5322 NEXT;
5323 } else {
5324 ERROR("xmlFAParseAtom: expecting ')'");
5325 }
5326 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_SUBREG);
5327 if (ctxt->atom == NULL)
5328 return(-1);
5329 ctxt->atom->start = start;
5330 ctxt->atom->start0 = start0;
5331 ctxt->atom->stop = ctxt->state;
5332 ctxt->end = oldend;
5333 return(1);
5334 } else if ((CUR == '[') || (CUR == '\\') || (CUR == '.')) {
5335 xmlFAParseCharClass(ctxt);
5336 return(1);
5337 }
5338 return(0);
5339 }
5340
5341 /**
5342 * xmlFAParsePiece:
5343 * @ctxt: a regexp parser context
5344 *
5345 * [3] piece ::= atom quantifier?
5346 */
5347 static int
xmlFAParsePiece(xmlRegParserCtxtPtr ctxt)5348 xmlFAParsePiece(xmlRegParserCtxtPtr ctxt) {
5349 int ret;
5350
5351 ctxt->atom = NULL;
5352 ret = xmlFAParseAtom(ctxt);
5353 if (ret == 0)
5354 return(0);
5355 if (ctxt->atom == NULL) {
5356 ERROR("internal: no atom generated");
5357 }
5358 xmlFAParseQuantifier(ctxt);
5359 return(1);
5360 }
5361
5362 /**
5363 * xmlFAParseBranch:
5364 * @ctxt: a regexp parser context
5365 * @to: optional target to the end of the branch
5366 *
5367 * @to is used to optimize by removing duplicate path in automata
5368 * in expressions like (a|b)(c|d)
5369 *
5370 * [2] branch ::= piece*
5371 */
5372 static int
xmlFAParseBranch(xmlRegParserCtxtPtr ctxt,xmlRegStatePtr to)5373 xmlFAParseBranch(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr to) {
5374 xmlRegStatePtr previous;
5375 int ret;
5376
5377 previous = ctxt->state;
5378 ret = xmlFAParsePiece(ctxt);
5379 if (ret != 0) {
5380 if (xmlFAGenerateTransitions(ctxt, previous,
5381 (CUR=='|' || CUR==')') ? to : NULL, ctxt->atom) < 0)
5382 return(-1);
5383 previous = ctxt->state;
5384 ctxt->atom = NULL;
5385 }
5386 while ((ret != 0) && (ctxt->error == 0)) {
5387 ret = xmlFAParsePiece(ctxt);
5388 if (ret != 0) {
5389 if (xmlFAGenerateTransitions(ctxt, previous,
5390 (CUR=='|' || CUR==')') ? to : NULL, ctxt->atom) < 0)
5391 return(-1);
5392 previous = ctxt->state;
5393 ctxt->atom = NULL;
5394 }
5395 }
5396 return(0);
5397 }
5398
5399 /**
5400 * xmlFAParseRegExp:
5401 * @ctxt: a regexp parser context
5402 * @top: is this the top-level expression ?
5403 *
5404 * [1] regExp ::= branch ( '|' branch )*
5405 */
5406 static void
xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt,int top)5407 xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top) {
5408 xmlRegStatePtr start, end;
5409
5410 /* if not top start should have been generated by an epsilon trans */
5411 start = ctxt->state;
5412 ctxt->end = NULL;
5413 xmlFAParseBranch(ctxt, NULL);
5414 if (top) {
5415 #ifdef DEBUG_REGEXP_GRAPH
5416 printf("State %d is final\n", ctxt->state->no);
5417 #endif
5418 ctxt->state->type = XML_REGEXP_FINAL_STATE;
5419 }
5420 if (CUR != '|') {
5421 ctxt->end = ctxt->state;
5422 return;
5423 }
5424 end = ctxt->state;
5425 while ((CUR == '|') && (ctxt->error == 0)) {
5426 NEXT;
5427 if (CUR == 0) {
5428 ERROR("expecting a branch after |")
5429 return;
5430 }
5431 ctxt->state = start;
5432 ctxt->end = NULL;
5433 xmlFAParseBranch(ctxt, end);
5434 }
5435 if (!top) {
5436 ctxt->state = end;
5437 ctxt->end = end;
5438 }
5439 }
5440
5441 /************************************************************************
5442 * *
5443 * The basic API *
5444 * *
5445 ************************************************************************/
5446
5447 /**
5448 * xmlRegexpPrint:
5449 * @output: the file for the output debug
5450 * @regexp: the compiled regexp
5451 *
5452 * Print the content of the compiled regular expression
5453 */
5454 void
xmlRegexpPrint(FILE * output,xmlRegexpPtr regexp)5455 xmlRegexpPrint(FILE *output, xmlRegexpPtr regexp) {
5456 int i;
5457
5458 if (output == NULL)
5459 return;
5460 fprintf(output, " regexp: ");
5461 if (regexp == NULL) {
5462 fprintf(output, "NULL\n");
5463 return;
5464 }
5465 fprintf(output, "'%s' ", regexp->string);
5466 fprintf(output, "\n");
5467 fprintf(output, "%d atoms:\n", regexp->nbAtoms);
5468 for (i = 0;i < regexp->nbAtoms; i++) {
5469 fprintf(output, " %02d ", i);
5470 xmlRegPrintAtom(output, regexp->atoms[i]);
5471 }
5472 fprintf(output, "%d states:", regexp->nbStates);
5473 fprintf(output, "\n");
5474 for (i = 0;i < regexp->nbStates; i++) {
5475 xmlRegPrintState(output, regexp->states[i]);
5476 }
5477 fprintf(output, "%d counters:\n", regexp->nbCounters);
5478 for (i = 0;i < regexp->nbCounters; i++) {
5479 fprintf(output, " %d: min %d max %d\n", i, regexp->counters[i].min,
5480 regexp->counters[i].max);
5481 }
5482 }
5483
5484 /**
5485 * xmlRegexpCompile:
5486 * @regexp: a regular expression string
5487 *
5488 * Parses a regular expression conforming to XML Schemas Part 2 Datatype
5489 * Appendix F and builds an automata suitable for testing strings against
5490 * that regular expression
5491 *
5492 * Returns the compiled expression or NULL in case of error
5493 */
5494 xmlRegexpPtr
xmlRegexpCompile(const xmlChar * regexp)5495 xmlRegexpCompile(const xmlChar *regexp) {
5496 xmlRegexpPtr ret;
5497 xmlRegParserCtxtPtr ctxt;
5498
5499 ctxt = xmlRegNewParserCtxt(regexp);
5500 if (ctxt == NULL)
5501 return(NULL);
5502
5503 /* initialize the parser */
5504 ctxt->end = NULL;
5505 ctxt->start = ctxt->state = xmlRegNewState(ctxt);
5506 xmlRegStatePush(ctxt, ctxt->start);
5507
5508 /* parse the expression building an automata */
5509 xmlFAParseRegExp(ctxt, 1);
5510 if (CUR != 0) {
5511 ERROR("xmlFAParseRegExp: extra characters");
5512 }
5513 if (ctxt->error != 0) {
5514 xmlRegFreeParserCtxt(ctxt);
5515 return(NULL);
5516 }
5517 ctxt->end = ctxt->state;
5518 ctxt->start->type = XML_REGEXP_START_STATE;
5519 ctxt->end->type = XML_REGEXP_FINAL_STATE;
5520
5521 /* remove the Epsilon except for counted transitions */
5522 xmlFAEliminateEpsilonTransitions(ctxt);
5523
5524
5525 if (ctxt->error != 0) {
5526 xmlRegFreeParserCtxt(ctxt);
5527 return(NULL);
5528 }
5529 ret = xmlRegEpxFromParse(ctxt);
5530 xmlRegFreeParserCtxt(ctxt);
5531 return(ret);
5532 }
5533
5534 /**
5535 * xmlRegexpExec:
5536 * @comp: the compiled regular expression
5537 * @content: the value to check against the regular expression
5538 *
5539 * Check if the regular expression generates the value
5540 *
5541 * Returns 1 if it matches, 0 if not and a negative value in case of error
5542 */
5543 int
xmlRegexpExec(xmlRegexpPtr comp,const xmlChar * content)5544 xmlRegexpExec(xmlRegexpPtr comp, const xmlChar *content) {
5545 if ((comp == NULL) || (content == NULL))
5546 return(-1);
5547 return(xmlFARegExec(comp, content));
5548 }
5549
5550 /**
5551 * xmlRegexpIsDeterminist:
5552 * @comp: the compiled regular expression
5553 *
5554 * Check if the regular expression is determinist
5555 *
5556 * Returns 1 if it yes, 0 if not and a negative value in case of error
5557 */
5558 int
xmlRegexpIsDeterminist(xmlRegexpPtr comp)5559 xmlRegexpIsDeterminist(xmlRegexpPtr comp) {
5560 xmlAutomataPtr am;
5561 int ret;
5562
5563 if (comp == NULL)
5564 return(-1);
5565 if (comp->determinist != -1)
5566 return(comp->determinist);
5567
5568 am = xmlNewAutomata();
5569 if (am->states != NULL) {
5570 int i;
5571
5572 for (i = 0;i < am->nbStates;i++)
5573 xmlRegFreeState(am->states[i]);
5574 xmlFree(am->states);
5575 }
5576 am->nbAtoms = comp->nbAtoms;
5577 am->atoms = comp->atoms;
5578 am->nbStates = comp->nbStates;
5579 am->states = comp->states;
5580 am->determinist = -1;
5581 am->flags = comp->flags;
5582 ret = xmlFAComputesDeterminism(am);
5583 am->atoms = NULL;
5584 am->states = NULL;
5585 xmlFreeAutomata(am);
5586 comp->determinist = ret;
5587 return(ret);
5588 }
5589
5590 /**
5591 * xmlRegFreeRegexp:
5592 * @regexp: the regexp
5593 *
5594 * Free a regexp
5595 */
5596 void
xmlRegFreeRegexp(xmlRegexpPtr regexp)5597 xmlRegFreeRegexp(xmlRegexpPtr regexp) {
5598 int i;
5599 if (regexp == NULL)
5600 return;
5601
5602 if (regexp->string != NULL)
5603 xmlFree(regexp->string);
5604 if (regexp->states != NULL) {
5605 for (i = 0;i < regexp->nbStates;i++)
5606 xmlRegFreeState(regexp->states[i]);
5607 xmlFree(regexp->states);
5608 }
5609 if (regexp->atoms != NULL) {
5610 for (i = 0;i < regexp->nbAtoms;i++)
5611 xmlRegFreeAtom(regexp->atoms[i]);
5612 xmlFree(regexp->atoms);
5613 }
5614 if (regexp->counters != NULL)
5615 xmlFree(regexp->counters);
5616 if (regexp->compact != NULL)
5617 xmlFree(regexp->compact);
5618 if (regexp->transdata != NULL)
5619 xmlFree(regexp->transdata);
5620 if (regexp->stringMap != NULL) {
5621 for (i = 0; i < regexp->nbstrings;i++)
5622 xmlFree(regexp->stringMap[i]);
5623 xmlFree(regexp->stringMap);
5624 }
5625
5626 xmlFree(regexp);
5627 }
5628
5629 #ifdef LIBXML_AUTOMATA_ENABLED
5630 /************************************************************************
5631 * *
5632 * The Automata interface *
5633 * *
5634 ************************************************************************/
5635
5636 /**
5637 * xmlNewAutomata:
5638 *
5639 * Create a new automata
5640 *
5641 * Returns the new object or NULL in case of failure
5642 */
5643 xmlAutomataPtr
xmlNewAutomata(void)5644 xmlNewAutomata(void) {
5645 xmlAutomataPtr ctxt;
5646
5647 ctxt = xmlRegNewParserCtxt(NULL);
5648 if (ctxt == NULL)
5649 return(NULL);
5650
5651 /* initialize the parser */
5652 ctxt->end = NULL;
5653 ctxt->start = ctxt->state = xmlRegNewState(ctxt);
5654 if (ctxt->start == NULL) {
5655 xmlFreeAutomata(ctxt);
5656 return(NULL);
5657 }
5658 ctxt->start->type = XML_REGEXP_START_STATE;
5659 if (xmlRegStatePush(ctxt, ctxt->start) < 0) {
5660 xmlRegFreeState(ctxt->start);
5661 xmlFreeAutomata(ctxt);
5662 return(NULL);
5663 }
5664 ctxt->flags = 0;
5665
5666 return(ctxt);
5667 }
5668
5669 /**
5670 * xmlFreeAutomata:
5671 * @am: an automata
5672 *
5673 * Free an automata
5674 */
5675 void
xmlFreeAutomata(xmlAutomataPtr am)5676 xmlFreeAutomata(xmlAutomataPtr am) {
5677 if (am == NULL)
5678 return;
5679 xmlRegFreeParserCtxt(am);
5680 }
5681
5682 /**
5683 * xmlAutomataSetFlags:
5684 * @am: an automata
5685 * @flags: a set of internal flags
5686 *
5687 * Set some flags on the automata
5688 */
5689 void
xmlAutomataSetFlags(xmlAutomataPtr am,int flags)5690 xmlAutomataSetFlags(xmlAutomataPtr am, int flags) {
5691 if (am == NULL)
5692 return;
5693 am->flags |= flags;
5694 }
5695
5696 /**
5697 * xmlAutomataGetInitState:
5698 * @am: an automata
5699 *
5700 * Initial state lookup
5701 *
5702 * Returns the initial state of the automata
5703 */
5704 xmlAutomataStatePtr
xmlAutomataGetInitState(xmlAutomataPtr am)5705 xmlAutomataGetInitState(xmlAutomataPtr am) {
5706 if (am == NULL)
5707 return(NULL);
5708 return(am->start);
5709 }
5710
5711 /**
5712 * xmlAutomataSetFinalState:
5713 * @am: an automata
5714 * @state: a state in this automata
5715 *
5716 * Makes that state a final state
5717 *
5718 * Returns 0 or -1 in case of error
5719 */
5720 int
xmlAutomataSetFinalState(xmlAutomataPtr am,xmlAutomataStatePtr state)5721 xmlAutomataSetFinalState(xmlAutomataPtr am, xmlAutomataStatePtr state) {
5722 if ((am == NULL) || (state == NULL))
5723 return(-1);
5724 state->type = XML_REGEXP_FINAL_STATE;
5725 return(0);
5726 }
5727
5728 /**
5729 * xmlAutomataNewTransition:
5730 * @am: an automata
5731 * @from: the starting point of the transition
5732 * @to: the target point of the transition or NULL
5733 * @token: the input string associated to that transition
5734 * @data: data passed to the callback function if the transition is activated
5735 *
5736 * If @to is NULL, this creates first a new target state in the automata
5737 * and then adds a transition from the @from state to the target state
5738 * activated by the value of @token
5739 *
5740 * Returns the target state or NULL in case of error
5741 */
5742 xmlAutomataStatePtr
xmlAutomataNewTransition(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,void * data)5743 xmlAutomataNewTransition(xmlAutomataPtr am, xmlAutomataStatePtr from,
5744 xmlAutomataStatePtr to, const xmlChar *token,
5745 void *data) {
5746 xmlRegAtomPtr atom;
5747
5748 if ((am == NULL) || (from == NULL) || (token == NULL))
5749 return(NULL);
5750 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5751 if (atom == NULL)
5752 return(NULL);
5753 atom->data = data;
5754 atom->valuep = xmlStrdup(token);
5755
5756 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5757 xmlRegFreeAtom(atom);
5758 return(NULL);
5759 }
5760 if (to == NULL)
5761 return(am->state);
5762 return(to);
5763 }
5764
5765 /**
5766 * xmlAutomataNewTransition2:
5767 * @am: an automata
5768 * @from: the starting point of the transition
5769 * @to: the target point of the transition or NULL
5770 * @token: the first input string associated to that transition
5771 * @token2: the second input string associated to that transition
5772 * @data: data passed to the callback function if the transition is activated
5773 *
5774 * If @to is NULL, this creates first a new target state in the automata
5775 * and then adds a transition from the @from state to the target state
5776 * activated by the value of @token
5777 *
5778 * Returns the target state or NULL in case of error
5779 */
5780 xmlAutomataStatePtr
xmlAutomataNewTransition2(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,const xmlChar * token2,void * data)5781 xmlAutomataNewTransition2(xmlAutomataPtr am, xmlAutomataStatePtr from,
5782 xmlAutomataStatePtr to, const xmlChar *token,
5783 const xmlChar *token2, void *data) {
5784 xmlRegAtomPtr atom;
5785
5786 if ((am == NULL) || (from == NULL) || (token == NULL))
5787 return(NULL);
5788 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5789 if (atom == NULL)
5790 return(NULL);
5791 atom->data = data;
5792 if ((token2 == NULL) || (*token2 == 0)) {
5793 atom->valuep = xmlStrdup(token);
5794 } else {
5795 int lenn, lenp;
5796 xmlChar *str;
5797
5798 lenn = strlen((char *) token2);
5799 lenp = strlen((char *) token);
5800
5801 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5802 if (str == NULL) {
5803 xmlRegFreeAtom(atom);
5804 return(NULL);
5805 }
5806 memcpy(&str[0], token, lenp);
5807 str[lenp] = '|';
5808 memcpy(&str[lenp + 1], token2, lenn);
5809 str[lenn + lenp + 1] = 0;
5810
5811 atom->valuep = str;
5812 }
5813
5814 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5815 xmlRegFreeAtom(atom);
5816 return(NULL);
5817 }
5818 if (to == NULL)
5819 return(am->state);
5820 return(to);
5821 }
5822
5823 /**
5824 * xmlAutomataNewNegTrans:
5825 * @am: an automata
5826 * @from: the starting point of the transition
5827 * @to: the target point of the transition or NULL
5828 * @token: the first input string associated to that transition
5829 * @token2: the second input string associated to that transition
5830 * @data: data passed to the callback function if the transition is activated
5831 *
5832 * If @to is NULL, this creates first a new target state in the automata
5833 * and then adds a transition from the @from state to the target state
5834 * activated by any value except (@token,@token2)
5835 * Note that if @token2 is not NULL, then (X, NULL) won't match to follow
5836 # the semantic of XSD ##other
5837 *
5838 * Returns the target state or NULL in case of error
5839 */
5840 xmlAutomataStatePtr
xmlAutomataNewNegTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,const xmlChar * token2,void * data)5841 xmlAutomataNewNegTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
5842 xmlAutomataStatePtr to, const xmlChar *token,
5843 const xmlChar *token2, void *data) {
5844 xmlRegAtomPtr atom;
5845 xmlChar err_msg[200];
5846
5847 if ((am == NULL) || (from == NULL) || (token == NULL))
5848 return(NULL);
5849 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5850 if (atom == NULL)
5851 return(NULL);
5852 atom->data = data;
5853 atom->neg = 1;
5854 if ((token2 == NULL) || (*token2 == 0)) {
5855 atom->valuep = xmlStrdup(token);
5856 } else {
5857 int lenn, lenp;
5858 xmlChar *str;
5859
5860 lenn = strlen((char *) token2);
5861 lenp = strlen((char *) token);
5862
5863 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5864 if (str == NULL) {
5865 xmlRegFreeAtom(atom);
5866 return(NULL);
5867 }
5868 memcpy(&str[0], token, lenp);
5869 str[lenp] = '|';
5870 memcpy(&str[lenp + 1], token2, lenn);
5871 str[lenn + lenp + 1] = 0;
5872
5873 atom->valuep = str;
5874 }
5875 snprintf((char *) err_msg, 199, "not %s", (const char *) atom->valuep);
5876 err_msg[199] = 0;
5877 atom->valuep2 = xmlStrdup(err_msg);
5878
5879 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5880 xmlRegFreeAtom(atom);
5881 return(NULL);
5882 }
5883 am->negs++;
5884 if (to == NULL)
5885 return(am->state);
5886 return(to);
5887 }
5888
5889 /**
5890 * xmlAutomataNewCountTrans2:
5891 * @am: an automata
5892 * @from: the starting point of the transition
5893 * @to: the target point of the transition or NULL
5894 * @token: the input string associated to that transition
5895 * @token2: the second input string associated to that transition
5896 * @min: the minimum successive occurences of token
5897 * @max: the maximum successive occurences of token
5898 * @data: data associated to the transition
5899 *
5900 * If @to is NULL, this creates first a new target state in the automata
5901 * and then adds a transition from the @from state to the target state
5902 * activated by a succession of input of value @token and @token2 and
5903 * whose number is between @min and @max
5904 *
5905 * Returns the target state or NULL in case of error
5906 */
5907 xmlAutomataStatePtr
xmlAutomataNewCountTrans2(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,const xmlChar * token2,int min,int max,void * data)5908 xmlAutomataNewCountTrans2(xmlAutomataPtr am, xmlAutomataStatePtr from,
5909 xmlAutomataStatePtr to, const xmlChar *token,
5910 const xmlChar *token2,
5911 int min, int max, void *data) {
5912 xmlRegAtomPtr atom;
5913 int counter;
5914
5915 if ((am == NULL) || (from == NULL) || (token == NULL))
5916 return(NULL);
5917 if (min < 0)
5918 return(NULL);
5919 if ((max < min) || (max < 1))
5920 return(NULL);
5921 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5922 if (atom == NULL)
5923 return(NULL);
5924 if ((token2 == NULL) || (*token2 == 0)) {
5925 atom->valuep = xmlStrdup(token);
5926 } else {
5927 int lenn, lenp;
5928 xmlChar *str;
5929
5930 lenn = strlen((char *) token2);
5931 lenp = strlen((char *) token);
5932
5933 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5934 if (str == NULL) {
5935 xmlRegFreeAtom(atom);
5936 return(NULL);
5937 }
5938 memcpy(&str[0], token, lenp);
5939 str[lenp] = '|';
5940 memcpy(&str[lenp + 1], token2, lenn);
5941 str[lenn + lenp + 1] = 0;
5942
5943 atom->valuep = str;
5944 }
5945 atom->data = data;
5946 if (min == 0)
5947 atom->min = 1;
5948 else
5949 atom->min = min;
5950 atom->max = max;
5951
5952 /*
5953 * associate a counter to the transition.
5954 */
5955 counter = xmlRegGetCounter(am);
5956 am->counters[counter].min = min;
5957 am->counters[counter].max = max;
5958
5959 /* xmlFAGenerateTransitions(am, from, to, atom); */
5960 if (to == NULL) {
5961 to = xmlRegNewState(am);
5962 xmlRegStatePush(am, to);
5963 }
5964 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
5965 xmlRegAtomPush(am, atom);
5966 am->state = to;
5967
5968 if (to == NULL)
5969 to = am->state;
5970 if (to == NULL)
5971 return(NULL);
5972 if (min == 0)
5973 xmlFAGenerateEpsilonTransition(am, from, to);
5974 return(to);
5975 }
5976
5977 /**
5978 * xmlAutomataNewCountTrans:
5979 * @am: an automata
5980 * @from: the starting point of the transition
5981 * @to: the target point of the transition or NULL
5982 * @token: the input string associated to that transition
5983 * @min: the minimum successive occurences of token
5984 * @max: the maximum successive occurences of token
5985 * @data: data associated to the transition
5986 *
5987 * If @to is NULL, this creates first a new target state in the automata
5988 * and then adds a transition from the @from state to the target state
5989 * activated by a succession of input of value @token and whose number
5990 * is between @min and @max
5991 *
5992 * Returns the target state or NULL in case of error
5993 */
5994 xmlAutomataStatePtr
xmlAutomataNewCountTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,int min,int max,void * data)5995 xmlAutomataNewCountTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
5996 xmlAutomataStatePtr to, const xmlChar *token,
5997 int min, int max, void *data) {
5998 xmlRegAtomPtr atom;
5999 int counter;
6000
6001 if ((am == NULL) || (from == NULL) || (token == NULL))
6002 return(NULL);
6003 if (min < 0)
6004 return(NULL);
6005 if ((max < min) || (max < 1))
6006 return(NULL);
6007 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6008 if (atom == NULL)
6009 return(NULL);
6010 atom->valuep = xmlStrdup(token);
6011 atom->data = data;
6012 if (min == 0)
6013 atom->min = 1;
6014 else
6015 atom->min = min;
6016 atom->max = max;
6017
6018 /*
6019 * associate a counter to the transition.
6020 */
6021 counter = xmlRegGetCounter(am);
6022 am->counters[counter].min = min;
6023 am->counters[counter].max = max;
6024
6025 /* xmlFAGenerateTransitions(am, from, to, atom); */
6026 if (to == NULL) {
6027 to = xmlRegNewState(am);
6028 xmlRegStatePush(am, to);
6029 }
6030 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6031 xmlRegAtomPush(am, atom);
6032 am->state = to;
6033
6034 if (to == NULL)
6035 to = am->state;
6036 if (to == NULL)
6037 return(NULL);
6038 if (min == 0)
6039 xmlFAGenerateEpsilonTransition(am, from, to);
6040 return(to);
6041 }
6042
6043 /**
6044 * xmlAutomataNewOnceTrans2:
6045 * @am: an automata
6046 * @from: the starting point of the transition
6047 * @to: the target point of the transition or NULL
6048 * @token: the input string associated to that transition
6049 * @token2: the second input string associated to that transition
6050 * @min: the minimum successive occurences of token
6051 * @max: the maximum successive occurences of token
6052 * @data: data associated to the transition
6053 *
6054 * If @to is NULL, this creates first a new target state in the automata
6055 * and then adds a transition from the @from state to the target state
6056 * activated by a succession of input of value @token and @token2 and whose
6057 * number is between @min and @max, moreover that transition can only be
6058 * crossed once.
6059 *
6060 * Returns the target state or NULL in case of error
6061 */
6062 xmlAutomataStatePtr
xmlAutomataNewOnceTrans2(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,const xmlChar * token2,int min,int max,void * data)6063 xmlAutomataNewOnceTrans2(xmlAutomataPtr am, xmlAutomataStatePtr from,
6064 xmlAutomataStatePtr to, const xmlChar *token,
6065 const xmlChar *token2,
6066 int min, int max, void *data) {
6067 xmlRegAtomPtr atom;
6068 int counter;
6069
6070 if ((am == NULL) || (from == NULL) || (token == NULL))
6071 return(NULL);
6072 if (min < 1)
6073 return(NULL);
6074 if ((max < min) || (max < 1))
6075 return(NULL);
6076 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6077 if (atom == NULL)
6078 return(NULL);
6079 if ((token2 == NULL) || (*token2 == 0)) {
6080 atom->valuep = xmlStrdup(token);
6081 } else {
6082 int lenn, lenp;
6083 xmlChar *str;
6084
6085 lenn = strlen((char *) token2);
6086 lenp = strlen((char *) token);
6087
6088 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
6089 if (str == NULL) {
6090 xmlRegFreeAtom(atom);
6091 return(NULL);
6092 }
6093 memcpy(&str[0], token, lenp);
6094 str[lenp] = '|';
6095 memcpy(&str[lenp + 1], token2, lenn);
6096 str[lenn + lenp + 1] = 0;
6097
6098 atom->valuep = str;
6099 }
6100 atom->data = data;
6101 atom->quant = XML_REGEXP_QUANT_ONCEONLY;
6102 atom->min = min;
6103 atom->max = max;
6104 /*
6105 * associate a counter to the transition.
6106 */
6107 counter = xmlRegGetCounter(am);
6108 am->counters[counter].min = 1;
6109 am->counters[counter].max = 1;
6110
6111 /* xmlFAGenerateTransitions(am, from, to, atom); */
6112 if (to == NULL) {
6113 to = xmlRegNewState(am);
6114 xmlRegStatePush(am, to);
6115 }
6116 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6117 xmlRegAtomPush(am, atom);
6118 am->state = to;
6119 return(to);
6120 }
6121
6122
6123
6124 /**
6125 * xmlAutomataNewOnceTrans:
6126 * @am: an automata
6127 * @from: the starting point of the transition
6128 * @to: the target point of the transition or NULL
6129 * @token: the input string associated to that transition
6130 * @min: the minimum successive occurences of token
6131 * @max: the maximum successive occurences of token
6132 * @data: data associated to the transition
6133 *
6134 * If @to is NULL, this creates first a new target state in the automata
6135 * and then adds a transition from the @from state to the target state
6136 * activated by a succession of input of value @token and whose number
6137 * is between @min and @max, moreover that transition can only be crossed
6138 * once.
6139 *
6140 * Returns the target state or NULL in case of error
6141 */
6142 xmlAutomataStatePtr
xmlAutomataNewOnceTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,const xmlChar * token,int min,int max,void * data)6143 xmlAutomataNewOnceTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6144 xmlAutomataStatePtr to, const xmlChar *token,
6145 int min, int max, void *data) {
6146 xmlRegAtomPtr atom;
6147 int counter;
6148
6149 if ((am == NULL) || (from == NULL) || (token == NULL))
6150 return(NULL);
6151 if (min < 1)
6152 return(NULL);
6153 if ((max < min) || (max < 1))
6154 return(NULL);
6155 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6156 if (atom == NULL)
6157 return(NULL);
6158 atom->valuep = xmlStrdup(token);
6159 atom->data = data;
6160 atom->quant = XML_REGEXP_QUANT_ONCEONLY;
6161 atom->min = min;
6162 atom->max = max;
6163 /*
6164 * associate a counter to the transition.
6165 */
6166 counter = xmlRegGetCounter(am);
6167 am->counters[counter].min = 1;
6168 am->counters[counter].max = 1;
6169
6170 /* xmlFAGenerateTransitions(am, from, to, atom); */
6171 if (to == NULL) {
6172 to = xmlRegNewState(am);
6173 xmlRegStatePush(am, to);
6174 }
6175 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6176 xmlRegAtomPush(am, atom);
6177 am->state = to;
6178 return(to);
6179 }
6180
6181 /**
6182 * xmlAutomataNewState:
6183 * @am: an automata
6184 *
6185 * Create a new disconnected state in the automata
6186 *
6187 * Returns the new state or NULL in case of error
6188 */
6189 xmlAutomataStatePtr
xmlAutomataNewState(xmlAutomataPtr am)6190 xmlAutomataNewState(xmlAutomataPtr am) {
6191 xmlAutomataStatePtr to;
6192
6193 if (am == NULL)
6194 return(NULL);
6195 to = xmlRegNewState(am);
6196 xmlRegStatePush(am, to);
6197 return(to);
6198 }
6199
6200 /**
6201 * xmlAutomataNewEpsilon:
6202 * @am: an automata
6203 * @from: the starting point of the transition
6204 * @to: the target point of the transition or NULL
6205 *
6206 * If @to is NULL, this creates first a new target state in the automata
6207 * and then adds an epsilon transition from the @from state to the
6208 * target state
6209 *
6210 * Returns the target state or NULL in case of error
6211 */
6212 xmlAutomataStatePtr
xmlAutomataNewEpsilon(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to)6213 xmlAutomataNewEpsilon(xmlAutomataPtr am, xmlAutomataStatePtr from,
6214 xmlAutomataStatePtr to) {
6215 if ((am == NULL) || (from == NULL))
6216 return(NULL);
6217 xmlFAGenerateEpsilonTransition(am, from, to);
6218 if (to == NULL)
6219 return(am->state);
6220 return(to);
6221 }
6222
6223 /**
6224 * xmlAutomataNewAllTrans:
6225 * @am: an automata
6226 * @from: the starting point of the transition
6227 * @to: the target point of the transition or NULL
6228 * @lax: allow to transition if not all all transitions have been activated
6229 *
6230 * If @to is NULL, this creates first a new target state in the automata
6231 * and then adds a an ALL transition from the @from state to the
6232 * target state. That transition is an epsilon transition allowed only when
6233 * all transitions from the @from node have been activated.
6234 *
6235 * Returns the target state or NULL in case of error
6236 */
6237 xmlAutomataStatePtr
xmlAutomataNewAllTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,int lax)6238 xmlAutomataNewAllTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6239 xmlAutomataStatePtr to, int lax) {
6240 if ((am == NULL) || (from == NULL))
6241 return(NULL);
6242 xmlFAGenerateAllTransition(am, from, to, lax);
6243 if (to == NULL)
6244 return(am->state);
6245 return(to);
6246 }
6247
6248 /**
6249 * xmlAutomataNewCounter:
6250 * @am: an automata
6251 * @min: the minimal value on the counter
6252 * @max: the maximal value on the counter
6253 *
6254 * Create a new counter
6255 *
6256 * Returns the counter number or -1 in case of error
6257 */
6258 int
xmlAutomataNewCounter(xmlAutomataPtr am,int min,int max)6259 xmlAutomataNewCounter(xmlAutomataPtr am, int min, int max) {
6260 int ret;
6261
6262 if (am == NULL)
6263 return(-1);
6264
6265 ret = xmlRegGetCounter(am);
6266 if (ret < 0)
6267 return(-1);
6268 am->counters[ret].min = min;
6269 am->counters[ret].max = max;
6270 return(ret);
6271 }
6272
6273 /**
6274 * xmlAutomataNewCountedTrans:
6275 * @am: an automata
6276 * @from: the starting point of the transition
6277 * @to: the target point of the transition or NULL
6278 * @counter: the counter associated to that transition
6279 *
6280 * If @to is NULL, this creates first a new target state in the automata
6281 * and then adds an epsilon transition from the @from state to the target state
6282 * which will increment the counter provided
6283 *
6284 * Returns the target state or NULL in case of error
6285 */
6286 xmlAutomataStatePtr
xmlAutomataNewCountedTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,int counter)6287 xmlAutomataNewCountedTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6288 xmlAutomataStatePtr to, int counter) {
6289 if ((am == NULL) || (from == NULL) || (counter < 0))
6290 return(NULL);
6291 xmlFAGenerateCountedEpsilonTransition(am, from, to, counter);
6292 if (to == NULL)
6293 return(am->state);
6294 return(to);
6295 }
6296
6297 /**
6298 * xmlAutomataNewCounterTrans:
6299 * @am: an automata
6300 * @from: the starting point of the transition
6301 * @to: the target point of the transition or NULL
6302 * @counter: the counter associated to that transition
6303 *
6304 * If @to is NULL, this creates first a new target state in the automata
6305 * and then adds an epsilon transition from the @from state to the target state
6306 * which will be allowed only if the counter is within the right range.
6307 *
6308 * Returns the target state or NULL in case of error
6309 */
6310 xmlAutomataStatePtr
xmlAutomataNewCounterTrans(xmlAutomataPtr am,xmlAutomataStatePtr from,xmlAutomataStatePtr to,int counter)6311 xmlAutomataNewCounterTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6312 xmlAutomataStatePtr to, int counter) {
6313 if ((am == NULL) || (from == NULL) || (counter < 0))
6314 return(NULL);
6315 xmlFAGenerateCountedTransition(am, from, to, counter);
6316 if (to == NULL)
6317 return(am->state);
6318 return(to);
6319 }
6320
6321 /**
6322 * xmlAutomataCompile:
6323 * @am: an automata
6324 *
6325 * Compile the automata into a Reg Exp ready for being executed.
6326 * The automata should be free after this point.
6327 *
6328 * Returns the compiled regexp or NULL in case of error
6329 */
6330 xmlRegexpPtr
xmlAutomataCompile(xmlAutomataPtr am)6331 xmlAutomataCompile(xmlAutomataPtr am) {
6332 xmlRegexpPtr ret;
6333
6334 if ((am == NULL) || (am->error != 0)) return(NULL);
6335 xmlFAEliminateEpsilonTransitions(am);
6336 /* xmlFAComputesDeterminism(am); */
6337 ret = xmlRegEpxFromParse(am);
6338
6339 return(ret);
6340 }
6341
6342 /**
6343 * xmlAutomataIsDeterminist:
6344 * @am: an automata
6345 *
6346 * Checks if an automata is determinist.
6347 *
6348 * Returns 1 if true, 0 if not, and -1 in case of error
6349 */
6350 int
xmlAutomataIsDeterminist(xmlAutomataPtr am)6351 xmlAutomataIsDeterminist(xmlAutomataPtr am) {
6352 int ret;
6353
6354 if (am == NULL)
6355 return(-1);
6356
6357 ret = xmlFAComputesDeterminism(am);
6358 return(ret);
6359 }
6360 #endif /* LIBXML_AUTOMATA_ENABLED */
6361
6362 #ifdef LIBXML_EXPR_ENABLED
6363 /************************************************************************
6364 * *
6365 * Formal Expression handling code *
6366 * *
6367 ************************************************************************/
6368 /************************************************************************
6369 * *
6370 * Expression handling context *
6371 * *
6372 ************************************************************************/
6373
6374 struct _xmlExpCtxt {
6375 xmlDictPtr dict;
6376 xmlExpNodePtr *table;
6377 int size;
6378 int nbElems;
6379 int nb_nodes;
6380 int maxNodes;
6381 const char *expr;
6382 const char *cur;
6383 int nb_cons;
6384 int tabSize;
6385 };
6386
6387 /**
6388 * xmlExpNewCtxt:
6389 * @maxNodes: the maximum number of nodes
6390 * @dict: optional dictionary to use internally
6391 *
6392 * Creates a new context for manipulating expressions
6393 *
6394 * Returns the context or NULL in case of error
6395 */
6396 xmlExpCtxtPtr
xmlExpNewCtxt(int maxNodes,xmlDictPtr dict)6397 xmlExpNewCtxt(int maxNodes, xmlDictPtr dict) {
6398 xmlExpCtxtPtr ret;
6399 int size = 256;
6400
6401 if (maxNodes <= 4096)
6402 maxNodes = 4096;
6403
6404 ret = (xmlExpCtxtPtr) xmlMalloc(sizeof(xmlExpCtxt));
6405 if (ret == NULL)
6406 return(NULL);
6407 memset(ret, 0, sizeof(xmlExpCtxt));
6408 ret->size = size;
6409 ret->nbElems = 0;
6410 ret->maxNodes = maxNodes;
6411 ret->table = xmlMalloc(size * sizeof(xmlExpNodePtr));
6412 if (ret->table == NULL) {
6413 xmlFree(ret);
6414 return(NULL);
6415 }
6416 memset(ret->table, 0, size * sizeof(xmlExpNodePtr));
6417 if (dict == NULL) {
6418 ret->dict = xmlDictCreate();
6419 if (ret->dict == NULL) {
6420 xmlFree(ret->table);
6421 xmlFree(ret);
6422 return(NULL);
6423 }
6424 } else {
6425 ret->dict = dict;
6426 xmlDictReference(ret->dict);
6427 }
6428 return(ret);
6429 }
6430
6431 /**
6432 * xmlExpFreeCtxt:
6433 * @ctxt: an expression context
6434 *
6435 * Free an expression context
6436 */
6437 void
xmlExpFreeCtxt(xmlExpCtxtPtr ctxt)6438 xmlExpFreeCtxt(xmlExpCtxtPtr ctxt) {
6439 if (ctxt == NULL)
6440 return;
6441 xmlDictFree(ctxt->dict);
6442 if (ctxt->table != NULL)
6443 xmlFree(ctxt->table);
6444 xmlFree(ctxt);
6445 }
6446
6447 /************************************************************************
6448 * *
6449 * Structure associated to an expression node *
6450 * *
6451 ************************************************************************/
6452 #define MAX_NODES 10000
6453
6454 /* #define DEBUG_DERIV */
6455
6456 /*
6457 * TODO:
6458 * - Wildcards
6459 * - public API for creation
6460 *
6461 * Started
6462 * - regression testing
6463 *
6464 * Done
6465 * - split into module and test tool
6466 * - memleaks
6467 */
6468
6469 typedef enum {
6470 XML_EXP_NILABLE = (1 << 0)
6471 } xmlExpNodeInfo;
6472
6473 #define IS_NILLABLE(node) ((node)->info & XML_EXP_NILABLE)
6474
6475 struct _xmlExpNode {
6476 unsigned char type;/* xmlExpNodeType */
6477 unsigned char info;/* OR of xmlExpNodeInfo */
6478 unsigned short key; /* the hash key */
6479 unsigned int ref; /* The number of references */
6480 int c_max; /* the maximum length it can consume */
6481 xmlExpNodePtr exp_left;
6482 xmlExpNodePtr next;/* the next node in the hash table or free list */
6483 union {
6484 struct {
6485 int f_min;
6486 int f_max;
6487 } count;
6488 struct {
6489 xmlExpNodePtr f_right;
6490 } children;
6491 const xmlChar *f_str;
6492 } field;
6493 };
6494
6495 #define exp_min field.count.f_min
6496 #define exp_max field.count.f_max
6497 /* #define exp_left field.children.f_left */
6498 #define exp_right field.children.f_right
6499 #define exp_str field.f_str
6500
6501 static xmlExpNodePtr xmlExpNewNode(xmlExpCtxtPtr ctxt, xmlExpNodeType type);
6502 static xmlExpNode forbiddenExpNode = {
6503 XML_EXP_FORBID, 0, 0, 0, 0, NULL, NULL, {{ 0, 0}}
6504 };
6505 xmlExpNodePtr forbiddenExp = &forbiddenExpNode;
6506 static xmlExpNode emptyExpNode = {
6507 XML_EXP_EMPTY, 1, 0, 0, 0, NULL, NULL, {{ 0, 0}}
6508 };
6509 xmlExpNodePtr emptyExp = &emptyExpNode;
6510
6511 /************************************************************************
6512 * *
6513 * The custom hash table for unicity and canonicalization *
6514 * of sub-expressions pointers *
6515 * *
6516 ************************************************************************/
6517 /*
6518 * xmlExpHashNameComputeKey:
6519 * Calculate the hash key for a token
6520 */
6521 static unsigned short
xmlExpHashNameComputeKey(const xmlChar * name)6522 xmlExpHashNameComputeKey(const xmlChar *name) {
6523 unsigned short value = 0L;
6524 char ch;
6525
6526 if (name != NULL) {
6527 value += 30 * (*name);
6528 while ((ch = *name++) != 0) {
6529 value = value ^ ((value << 5) + (value >> 3) + (unsigned long)ch);
6530 }
6531 }
6532 return (value);
6533 }
6534
6535 /*
6536 * xmlExpHashComputeKey:
6537 * Calculate the hash key for a compound expression
6538 */
6539 static unsigned short
xmlExpHashComputeKey(xmlExpNodeType type,xmlExpNodePtr left,xmlExpNodePtr right)6540 xmlExpHashComputeKey(xmlExpNodeType type, xmlExpNodePtr left,
6541 xmlExpNodePtr right) {
6542 unsigned long value;
6543 unsigned short ret;
6544
6545 switch (type) {
6546 case XML_EXP_SEQ:
6547 value = left->key;
6548 value += right->key;
6549 value *= 3;
6550 ret = (unsigned short) value;
6551 break;
6552 case XML_EXP_OR:
6553 value = left->key;
6554 value += right->key;
6555 value *= 7;
6556 ret = (unsigned short) value;
6557 break;
6558 case XML_EXP_COUNT:
6559 value = left->key;
6560 value += right->key;
6561 ret = (unsigned short) value;
6562 break;
6563 default:
6564 ret = 0;
6565 }
6566 return(ret);
6567 }
6568
6569
6570 static xmlExpNodePtr
xmlExpNewNode(xmlExpCtxtPtr ctxt,xmlExpNodeType type)6571 xmlExpNewNode(xmlExpCtxtPtr ctxt, xmlExpNodeType type) {
6572 xmlExpNodePtr ret;
6573
6574 if (ctxt->nb_nodes >= MAX_NODES)
6575 return(NULL);
6576 ret = (xmlExpNodePtr) xmlMalloc(sizeof(xmlExpNode));
6577 if (ret == NULL)
6578 return(NULL);
6579 memset(ret, 0, sizeof(xmlExpNode));
6580 ret->type = type;
6581 ret->next = NULL;
6582 ctxt->nb_nodes++;
6583 ctxt->nb_cons++;
6584 return(ret);
6585 }
6586
6587 /**
6588 * xmlExpHashGetEntry:
6589 * @table: the hash table
6590 *
6591 * Get the unique entry from the hash table. The entry is created if
6592 * needed. @left and @right are consumed, i.e. their ref count will
6593 * be decremented by the operation.
6594 *
6595 * Returns the pointer or NULL in case of error
6596 */
6597 static xmlExpNodePtr
xmlExpHashGetEntry(xmlExpCtxtPtr ctxt,xmlExpNodeType type,xmlExpNodePtr left,xmlExpNodePtr right,const xmlChar * name,int min,int max)6598 xmlExpHashGetEntry(xmlExpCtxtPtr ctxt, xmlExpNodeType type,
6599 xmlExpNodePtr left, xmlExpNodePtr right,
6600 const xmlChar *name, int min, int max) {
6601 unsigned short kbase, key;
6602 xmlExpNodePtr entry;
6603 xmlExpNodePtr insert;
6604
6605 if (ctxt == NULL)
6606 return(NULL);
6607
6608 /*
6609 * Check for duplicate and insertion location.
6610 */
6611 if (type == XML_EXP_ATOM) {
6612 kbase = xmlExpHashNameComputeKey(name);
6613 } else if (type == XML_EXP_COUNT) {
6614 /* COUNT reduction rule 1 */
6615 /* a{1} -> a */
6616 if (min == max) {
6617 if (min == 1) {
6618 return(left);
6619 }
6620 if (min == 0) {
6621 xmlExpFree(ctxt, left);
6622 return(emptyExp);
6623 }
6624 }
6625 if (min < 0) {
6626 xmlExpFree(ctxt, left);
6627 return(forbiddenExp);
6628 }
6629 if (max == -1)
6630 kbase = min + 79;
6631 else
6632 kbase = max - min;
6633 kbase += left->key;
6634 } else if (type == XML_EXP_OR) {
6635 /* Forbid reduction rules */
6636 if (left->type == XML_EXP_FORBID) {
6637 xmlExpFree(ctxt, left);
6638 return(right);
6639 }
6640 if (right->type == XML_EXP_FORBID) {
6641 xmlExpFree(ctxt, right);
6642 return(left);
6643 }
6644
6645 /* OR reduction rule 1 */
6646 /* a | a reduced to a */
6647 if (left == right) {
6648 left->ref--;
6649 return(left);
6650 }
6651 /* OR canonicalization rule 1 */
6652 /* linearize (a | b) | c into a | (b | c) */
6653 if ((left->type == XML_EXP_OR) && (right->type != XML_EXP_OR)) {
6654 xmlExpNodePtr tmp = left;
6655 left = right;
6656 right = tmp;
6657 }
6658 /* OR reduction rule 2 */
6659 /* a | (a | b) and b | (a | b) are reduced to a | b */
6660 if (right->type == XML_EXP_OR) {
6661 if ((left == right->exp_left) ||
6662 (left == right->exp_right)) {
6663 xmlExpFree(ctxt, left);
6664 return(right);
6665 }
6666 }
6667 /* OR canonicalization rule 2 */
6668 /* linearize (a | b) | c into a | (b | c) */
6669 if (left->type == XML_EXP_OR) {
6670 xmlExpNodePtr tmp;
6671
6672 /* OR canonicalization rule 2 */
6673 if ((left->exp_right->type != XML_EXP_OR) &&
6674 (left->exp_right->key < left->exp_left->key)) {
6675 tmp = left->exp_right;
6676 left->exp_right = left->exp_left;
6677 left->exp_left = tmp;
6678 }
6679 left->exp_right->ref++;
6680 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left->exp_right, right,
6681 NULL, 0, 0);
6682 left->exp_left->ref++;
6683 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left->exp_left, tmp,
6684 NULL, 0, 0);
6685
6686 xmlExpFree(ctxt, left);
6687 return(tmp);
6688 }
6689 if (right->type == XML_EXP_OR) {
6690 /* Ordering in the tree */
6691 /* C | (A | B) -> A | (B | C) */
6692 if (left->key > right->exp_right->key) {
6693 xmlExpNodePtr tmp;
6694 right->exp_right->ref++;
6695 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_right,
6696 left, NULL, 0, 0);
6697 right->exp_left->ref++;
6698 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_left,
6699 tmp, NULL, 0, 0);
6700 xmlExpFree(ctxt, right);
6701 return(tmp);
6702 }
6703 /* Ordering in the tree */
6704 /* B | (A | C) -> A | (B | C) */
6705 if (left->key > right->exp_left->key) {
6706 xmlExpNodePtr tmp;
6707 right->exp_right->ref++;
6708 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left,
6709 right->exp_right, NULL, 0, 0);
6710 right->exp_left->ref++;
6711 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_left,
6712 tmp, NULL, 0, 0);
6713 xmlExpFree(ctxt, right);
6714 return(tmp);
6715 }
6716 }
6717 /* we know both types are != XML_EXP_OR here */
6718 else if (left->key > right->key) {
6719 xmlExpNodePtr tmp = left;
6720 left = right;
6721 right = tmp;
6722 }
6723 kbase = xmlExpHashComputeKey(type, left, right);
6724 } else if (type == XML_EXP_SEQ) {
6725 /* Forbid reduction rules */
6726 if (left->type == XML_EXP_FORBID) {
6727 xmlExpFree(ctxt, right);
6728 return(left);
6729 }
6730 if (right->type == XML_EXP_FORBID) {
6731 xmlExpFree(ctxt, left);
6732 return(right);
6733 }
6734 /* Empty reduction rules */
6735 if (right->type == XML_EXP_EMPTY) {
6736 return(left);
6737 }
6738 if (left->type == XML_EXP_EMPTY) {
6739 return(right);
6740 }
6741 kbase = xmlExpHashComputeKey(type, left, right);
6742 } else
6743 return(NULL);
6744
6745 key = kbase % ctxt->size;
6746 if (ctxt->table[key] != NULL) {
6747 for (insert = ctxt->table[key]; insert != NULL;
6748 insert = insert->next) {
6749 if ((insert->key == kbase) &&
6750 (insert->type == type)) {
6751 if (type == XML_EXP_ATOM) {
6752 if (name == insert->exp_str) {
6753 insert->ref++;
6754 return(insert);
6755 }
6756 } else if (type == XML_EXP_COUNT) {
6757 if ((insert->exp_min == min) && (insert->exp_max == max) &&
6758 (insert->exp_left == left)) {
6759 insert->ref++;
6760 left->ref--;
6761 return(insert);
6762 }
6763 } else if ((insert->exp_left == left) &&
6764 (insert->exp_right == right)) {
6765 insert->ref++;
6766 left->ref--;
6767 right->ref--;
6768 return(insert);
6769 }
6770 }
6771 }
6772 }
6773
6774 entry = xmlExpNewNode(ctxt, type);
6775 if (entry == NULL)
6776 return(NULL);
6777 entry->key = kbase;
6778 if (type == XML_EXP_ATOM) {
6779 entry->exp_str = name;
6780 entry->c_max = 1;
6781 } else if (type == XML_EXP_COUNT) {
6782 entry->exp_min = min;
6783 entry->exp_max = max;
6784 entry->exp_left = left;
6785 if ((min == 0) || (IS_NILLABLE(left)))
6786 entry->info |= XML_EXP_NILABLE;
6787 if (max < 0)
6788 entry->c_max = -1;
6789 else
6790 entry->c_max = max * entry->exp_left->c_max;
6791 } else {
6792 entry->exp_left = left;
6793 entry->exp_right = right;
6794 if (type == XML_EXP_OR) {
6795 if ((IS_NILLABLE(left)) || (IS_NILLABLE(right)))
6796 entry->info |= XML_EXP_NILABLE;
6797 if ((entry->exp_left->c_max == -1) ||
6798 (entry->exp_right->c_max == -1))
6799 entry->c_max = -1;
6800 else if (entry->exp_left->c_max > entry->exp_right->c_max)
6801 entry->c_max = entry->exp_left->c_max;
6802 else
6803 entry->c_max = entry->exp_right->c_max;
6804 } else {
6805 if ((IS_NILLABLE(left)) && (IS_NILLABLE(right)))
6806 entry->info |= XML_EXP_NILABLE;
6807 if ((entry->exp_left->c_max == -1) ||
6808 (entry->exp_right->c_max == -1))
6809 entry->c_max = -1;
6810 else
6811 entry->c_max = entry->exp_left->c_max + entry->exp_right->c_max;
6812 }
6813 }
6814 entry->ref = 1;
6815 if (ctxt->table[key] != NULL)
6816 entry->next = ctxt->table[key];
6817
6818 ctxt->table[key] = entry;
6819 ctxt->nbElems++;
6820
6821 return(entry);
6822 }
6823
6824 /**
6825 * xmlExpFree:
6826 * @ctxt: the expression context
6827 * @exp: the expression
6828 *
6829 * Dereference the expression
6830 */
6831 void
xmlExpFree(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp)6832 xmlExpFree(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp) {
6833 if ((exp == NULL) || (exp == forbiddenExp) || (exp == emptyExp))
6834 return;
6835 exp->ref--;
6836 if (exp->ref == 0) {
6837 unsigned short key;
6838
6839 /* Unlink it first from the hash table */
6840 key = exp->key % ctxt->size;
6841 if (ctxt->table[key] == exp) {
6842 ctxt->table[key] = exp->next;
6843 } else {
6844 xmlExpNodePtr tmp;
6845
6846 tmp = ctxt->table[key];
6847 while (tmp != NULL) {
6848 if (tmp->next == exp) {
6849 tmp->next = exp->next;
6850 break;
6851 }
6852 tmp = tmp->next;
6853 }
6854 }
6855
6856 if ((exp->type == XML_EXP_SEQ) || (exp->type == XML_EXP_OR)) {
6857 xmlExpFree(ctxt, exp->exp_left);
6858 xmlExpFree(ctxt, exp->exp_right);
6859 } else if (exp->type == XML_EXP_COUNT) {
6860 xmlExpFree(ctxt, exp->exp_left);
6861 }
6862 xmlFree(exp);
6863 ctxt->nb_nodes--;
6864 }
6865 }
6866
6867 /**
6868 * xmlExpRef:
6869 * @exp: the expression
6870 *
6871 * Increase the reference count of the expression
6872 */
6873 void
xmlExpRef(xmlExpNodePtr exp)6874 xmlExpRef(xmlExpNodePtr exp) {
6875 if (exp != NULL)
6876 exp->ref++;
6877 }
6878
6879 /**
6880 * xmlExpNewAtom:
6881 * @ctxt: the expression context
6882 * @name: the atom name
6883 * @len: the atom name length in byte (or -1);
6884 *
6885 * Get the atom associated to this name from that context
6886 *
6887 * Returns the node or NULL in case of error
6888 */
6889 xmlExpNodePtr
xmlExpNewAtom(xmlExpCtxtPtr ctxt,const xmlChar * name,int len)6890 xmlExpNewAtom(xmlExpCtxtPtr ctxt, const xmlChar *name, int len) {
6891 if ((ctxt == NULL) || (name == NULL))
6892 return(NULL);
6893 name = xmlDictLookup(ctxt->dict, name, len);
6894 if (name == NULL)
6895 return(NULL);
6896 return(xmlExpHashGetEntry(ctxt, XML_EXP_ATOM, NULL, NULL, name, 0, 0));
6897 }
6898
6899 /**
6900 * xmlExpNewOr:
6901 * @ctxt: the expression context
6902 * @left: left expression
6903 * @right: right expression
6904 *
6905 * Get the atom associated to the choice @left | @right
6906 * Note that @left and @right are consumed in the operation, to keep
6907 * an handle on them use xmlExpRef() and use xmlExpFree() to release them,
6908 * this is true even in case of failure (unless ctxt == NULL).
6909 *
6910 * Returns the node or NULL in case of error
6911 */
6912 xmlExpNodePtr
xmlExpNewOr(xmlExpCtxtPtr ctxt,xmlExpNodePtr left,xmlExpNodePtr right)6913 xmlExpNewOr(xmlExpCtxtPtr ctxt, xmlExpNodePtr left, xmlExpNodePtr right) {
6914 if (ctxt == NULL)
6915 return(NULL);
6916 if ((left == NULL) || (right == NULL)) {
6917 xmlExpFree(ctxt, left);
6918 xmlExpFree(ctxt, right);
6919 return(NULL);
6920 }
6921 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, left, right, NULL, 0, 0));
6922 }
6923
6924 /**
6925 * xmlExpNewSeq:
6926 * @ctxt: the expression context
6927 * @left: left expression
6928 * @right: right expression
6929 *
6930 * Get the atom associated to the sequence @left , @right
6931 * Note that @left and @right are consumed in the operation, to keep
6932 * an handle on them use xmlExpRef() and use xmlExpFree() to release them,
6933 * this is true even in case of failure (unless ctxt == NULL).
6934 *
6935 * Returns the node or NULL in case of error
6936 */
6937 xmlExpNodePtr
xmlExpNewSeq(xmlExpCtxtPtr ctxt,xmlExpNodePtr left,xmlExpNodePtr right)6938 xmlExpNewSeq(xmlExpCtxtPtr ctxt, xmlExpNodePtr left, xmlExpNodePtr right) {
6939 if (ctxt == NULL)
6940 return(NULL);
6941 if ((left == NULL) || (right == NULL)) {
6942 xmlExpFree(ctxt, left);
6943 xmlExpFree(ctxt, right);
6944 return(NULL);
6945 }
6946 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, left, right, NULL, 0, 0));
6947 }
6948
6949 /**
6950 * xmlExpNewRange:
6951 * @ctxt: the expression context
6952 * @subset: the expression to be repeated
6953 * @min: the lower bound for the repetition
6954 * @max: the upper bound for the repetition, -1 means infinite
6955 *
6956 * Get the atom associated to the range (@subset){@min, @max}
6957 * Note that @subset is consumed in the operation, to keep
6958 * an handle on it use xmlExpRef() and use xmlExpFree() to release it,
6959 * this is true even in case of failure (unless ctxt == NULL).
6960 *
6961 * Returns the node or NULL in case of error
6962 */
6963 xmlExpNodePtr
xmlExpNewRange(xmlExpCtxtPtr ctxt,xmlExpNodePtr subset,int min,int max)6964 xmlExpNewRange(xmlExpCtxtPtr ctxt, xmlExpNodePtr subset, int min, int max) {
6965 if (ctxt == NULL)
6966 return(NULL);
6967 if ((subset == NULL) || (min < 0) || (max < -1) ||
6968 ((max >= 0) && (min > max))) {
6969 xmlExpFree(ctxt, subset);
6970 return(NULL);
6971 }
6972 return(xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, subset,
6973 NULL, NULL, min, max));
6974 }
6975
6976 /************************************************************************
6977 * *
6978 * Public API for operations on expressions *
6979 * *
6980 ************************************************************************/
6981
6982 static int
xmlExpGetLanguageInt(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar ** list,int len,int nb)6983 xmlExpGetLanguageInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
6984 const xmlChar**list, int len, int nb) {
6985 int tmp, tmp2;
6986 tail:
6987 switch (exp->type) {
6988 case XML_EXP_EMPTY:
6989 return(0);
6990 case XML_EXP_ATOM:
6991 for (tmp = 0;tmp < nb;tmp++)
6992 if (list[tmp] == exp->exp_str)
6993 return(0);
6994 if (nb >= len)
6995 return(-2);
6996 list[nb] = exp->exp_str;
6997 return(1);
6998 case XML_EXP_COUNT:
6999 exp = exp->exp_left;
7000 goto tail;
7001 case XML_EXP_SEQ:
7002 case XML_EXP_OR:
7003 tmp = xmlExpGetLanguageInt(ctxt, exp->exp_left, list, len, nb);
7004 if (tmp < 0)
7005 return(tmp);
7006 tmp2 = xmlExpGetLanguageInt(ctxt, exp->exp_right, list, len,
7007 nb + tmp);
7008 if (tmp2 < 0)
7009 return(tmp2);
7010 return(tmp + tmp2);
7011 }
7012 return(-1);
7013 }
7014
7015 /**
7016 * xmlExpGetLanguage:
7017 * @ctxt: the expression context
7018 * @exp: the expression
7019 * @langList: where to store the tokens
7020 * @len: the allocated length of @list
7021 *
7022 * Find all the strings used in @exp and store them in @list
7023 *
7024 * Returns the number of unique strings found, -1 in case of errors and
7025 * -2 if there is more than @len strings
7026 */
7027 int
xmlExpGetLanguage(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar ** langList,int len)7028 xmlExpGetLanguage(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7029 const xmlChar**langList, int len) {
7030 if ((ctxt == NULL) || (exp == NULL) || (langList == NULL) || (len <= 0))
7031 return(-1);
7032 return(xmlExpGetLanguageInt(ctxt, exp, langList, len, 0));
7033 }
7034
7035 static int
xmlExpGetStartInt(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar ** list,int len,int nb)7036 xmlExpGetStartInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7037 const xmlChar**list, int len, int nb) {
7038 int tmp, tmp2;
7039 tail:
7040 switch (exp->type) {
7041 case XML_EXP_FORBID:
7042 return(0);
7043 case XML_EXP_EMPTY:
7044 return(0);
7045 case XML_EXP_ATOM:
7046 for (tmp = 0;tmp < nb;tmp++)
7047 if (list[tmp] == exp->exp_str)
7048 return(0);
7049 if (nb >= len)
7050 return(-2);
7051 list[nb] = exp->exp_str;
7052 return(1);
7053 case XML_EXP_COUNT:
7054 exp = exp->exp_left;
7055 goto tail;
7056 case XML_EXP_SEQ:
7057 tmp = xmlExpGetStartInt(ctxt, exp->exp_left, list, len, nb);
7058 if (tmp < 0)
7059 return(tmp);
7060 if (IS_NILLABLE(exp->exp_left)) {
7061 tmp2 = xmlExpGetStartInt(ctxt, exp->exp_right, list, len,
7062 nb + tmp);
7063 if (tmp2 < 0)
7064 return(tmp2);
7065 tmp += tmp2;
7066 }
7067 return(tmp);
7068 case XML_EXP_OR:
7069 tmp = xmlExpGetStartInt(ctxt, exp->exp_left, list, len, nb);
7070 if (tmp < 0)
7071 return(tmp);
7072 tmp2 = xmlExpGetStartInt(ctxt, exp->exp_right, list, len,
7073 nb + tmp);
7074 if (tmp2 < 0)
7075 return(tmp2);
7076 return(tmp + tmp2);
7077 }
7078 return(-1);
7079 }
7080
7081 /**
7082 * xmlExpGetStart:
7083 * @ctxt: the expression context
7084 * @exp: the expression
7085 * @tokList: where to store the tokens
7086 * @len: the allocated length of @list
7087 *
7088 * Find all the strings that appears at the start of the languages
7089 * accepted by @exp and store them in @list. E.g. for (a, b) | c
7090 * it will return the list [a, c]
7091 *
7092 * Returns the number of unique strings found, -1 in case of errors and
7093 * -2 if there is more than @len strings
7094 */
7095 int
xmlExpGetStart(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar ** tokList,int len)7096 xmlExpGetStart(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7097 const xmlChar**tokList, int len) {
7098 if ((ctxt == NULL) || (exp == NULL) || (tokList == NULL) || (len <= 0))
7099 return(-1);
7100 return(xmlExpGetStartInt(ctxt, exp, tokList, len, 0));
7101 }
7102
7103 /**
7104 * xmlExpIsNillable:
7105 * @exp: the expression
7106 *
7107 * Finds if the expression is nillable, i.e. if it accepts the empty sequqnce
7108 *
7109 * Returns 1 if nillable, 0 if not and -1 in case of error
7110 */
7111 int
xmlExpIsNillable(xmlExpNodePtr exp)7112 xmlExpIsNillable(xmlExpNodePtr exp) {
7113 if (exp == NULL)
7114 return(-1);
7115 return(IS_NILLABLE(exp) != 0);
7116 }
7117
7118 static xmlExpNodePtr
xmlExpStringDeriveInt(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar * str)7119 xmlExpStringDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, const xmlChar *str)
7120 {
7121 xmlExpNodePtr ret;
7122
7123 switch (exp->type) {
7124 case XML_EXP_EMPTY:
7125 return(forbiddenExp);
7126 case XML_EXP_FORBID:
7127 return(forbiddenExp);
7128 case XML_EXP_ATOM:
7129 if (exp->exp_str == str) {
7130 #ifdef DEBUG_DERIV
7131 printf("deriv atom: equal => Empty\n");
7132 #endif
7133 ret = emptyExp;
7134 } else {
7135 #ifdef DEBUG_DERIV
7136 printf("deriv atom: mismatch => forbid\n");
7137 #endif
7138 /* TODO wildcards here */
7139 ret = forbiddenExp;
7140 }
7141 return(ret);
7142 case XML_EXP_OR: {
7143 xmlExpNodePtr tmp;
7144
7145 #ifdef DEBUG_DERIV
7146 printf("deriv or: => or(derivs)\n");
7147 #endif
7148 tmp = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7149 if (tmp == NULL) {
7150 return(NULL);
7151 }
7152 ret = xmlExpStringDeriveInt(ctxt, exp->exp_right, str);
7153 if (ret == NULL) {
7154 xmlExpFree(ctxt, tmp);
7155 return(NULL);
7156 }
7157 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, tmp, ret,
7158 NULL, 0, 0);
7159 return(ret);
7160 }
7161 case XML_EXP_SEQ:
7162 #ifdef DEBUG_DERIV
7163 printf("deriv seq: starting with left\n");
7164 #endif
7165 ret = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7166 if (ret == NULL) {
7167 return(NULL);
7168 } else if (ret == forbiddenExp) {
7169 if (IS_NILLABLE(exp->exp_left)) {
7170 #ifdef DEBUG_DERIV
7171 printf("deriv seq: left failed but nillable\n");
7172 #endif
7173 ret = xmlExpStringDeriveInt(ctxt, exp->exp_right, str);
7174 }
7175 } else {
7176 #ifdef DEBUG_DERIV
7177 printf("deriv seq: left match => sequence\n");
7178 #endif
7179 exp->exp_right->ref++;
7180 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, exp->exp_right,
7181 NULL, 0, 0);
7182 }
7183 return(ret);
7184 case XML_EXP_COUNT: {
7185 int min, max;
7186 xmlExpNodePtr tmp;
7187
7188 if (exp->exp_max == 0)
7189 return(forbiddenExp);
7190 ret = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7191 if (ret == NULL)
7192 return(NULL);
7193 if (ret == forbiddenExp) {
7194 #ifdef DEBUG_DERIV
7195 printf("deriv count: pattern mismatch => forbid\n");
7196 #endif
7197 return(ret);
7198 }
7199 if (exp->exp_max == 1)
7200 return(ret);
7201 if (exp->exp_max < 0) /* unbounded */
7202 max = -1;
7203 else
7204 max = exp->exp_max - 1;
7205 if (exp->exp_min > 0)
7206 min = exp->exp_min - 1;
7207 else
7208 min = 0;
7209 exp->exp_left->ref++;
7210 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left, NULL,
7211 NULL, min, max);
7212 if (ret == emptyExp) {
7213 #ifdef DEBUG_DERIV
7214 printf("deriv count: match to empty => new count\n");
7215 #endif
7216 return(tmp);
7217 }
7218 #ifdef DEBUG_DERIV
7219 printf("deriv count: match => sequence with new count\n");
7220 #endif
7221 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, tmp,
7222 NULL, 0, 0));
7223 }
7224 }
7225 return(NULL);
7226 }
7227
7228 /**
7229 * xmlExpStringDerive:
7230 * @ctxt: the expression context
7231 * @exp: the expression
7232 * @str: the string
7233 * @len: the string len in bytes if available
7234 *
7235 * Do one step of Brzozowski derivation of the expression @exp with
7236 * respect to the input string
7237 *
7238 * Returns the resulting expression or NULL in case of internal error
7239 */
7240 xmlExpNodePtr
xmlExpStringDerive(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,const xmlChar * str,int len)7241 xmlExpStringDerive(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7242 const xmlChar *str, int len) {
7243 const xmlChar *input;
7244
7245 if ((exp == NULL) || (ctxt == NULL) || (str == NULL)) {
7246 return(NULL);
7247 }
7248 /*
7249 * check the string is in the dictionary, if yes use an interned
7250 * copy, otherwise we know it's not an acceptable input
7251 */
7252 input = xmlDictExists(ctxt->dict, str, len);
7253 if (input == NULL) {
7254 return(forbiddenExp);
7255 }
7256 return(xmlExpStringDeriveInt(ctxt, exp, input));
7257 }
7258
7259 static int
xmlExpCheckCard(xmlExpNodePtr exp,xmlExpNodePtr sub)7260 xmlExpCheckCard(xmlExpNodePtr exp, xmlExpNodePtr sub) {
7261 int ret = 1;
7262
7263 if (sub->c_max == -1) {
7264 if (exp->c_max != -1)
7265 ret = 0;
7266 } else if ((exp->c_max >= 0) && (exp->c_max < sub->c_max)) {
7267 ret = 0;
7268 }
7269 #if 0
7270 if ((IS_NILLABLE(sub)) && (!IS_NILLABLE(exp)))
7271 ret = 0;
7272 #endif
7273 return(ret);
7274 }
7275
7276 static xmlExpNodePtr xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7277 xmlExpNodePtr sub);
7278 /**
7279 * xmlExpDivide:
7280 * @ctxt: the expressions context
7281 * @exp: the englobing expression
7282 * @sub: the subexpression
7283 * @mult: the multiple expression
7284 * @remain: the remain from the derivation of the multiple
7285 *
7286 * Check if exp is a multiple of sub, i.e. if there is a finite number n
7287 * so that sub{n} subsume exp
7288 *
7289 * Returns the multiple value if successful, 0 if it is not a multiple
7290 * and -1 in case of internel error.
7291 */
7292
7293 static int
xmlExpDivide(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,xmlExpNodePtr sub,xmlExpNodePtr * mult,xmlExpNodePtr * remain)7294 xmlExpDivide(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub,
7295 xmlExpNodePtr *mult, xmlExpNodePtr *remain) {
7296 int i;
7297 xmlExpNodePtr tmp, tmp2;
7298
7299 if (mult != NULL) *mult = NULL;
7300 if (remain != NULL) *remain = NULL;
7301 if (exp->c_max == -1) return(0);
7302 if (IS_NILLABLE(exp) && (!IS_NILLABLE(sub))) return(0);
7303
7304 for (i = 1;i <= exp->c_max;i++) {
7305 sub->ref++;
7306 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT,
7307 sub, NULL, NULL, i, i);
7308 if (tmp == NULL) {
7309 return(-1);
7310 }
7311 if (!xmlExpCheckCard(tmp, exp)) {
7312 xmlExpFree(ctxt, tmp);
7313 continue;
7314 }
7315 tmp2 = xmlExpExpDeriveInt(ctxt, tmp, exp);
7316 if (tmp2 == NULL) {
7317 xmlExpFree(ctxt, tmp);
7318 return(-1);
7319 }
7320 if ((tmp2 != forbiddenExp) && (IS_NILLABLE(tmp2))) {
7321 if (remain != NULL)
7322 *remain = tmp2;
7323 else
7324 xmlExpFree(ctxt, tmp2);
7325 if (mult != NULL)
7326 *mult = tmp;
7327 else
7328 xmlExpFree(ctxt, tmp);
7329 #ifdef DEBUG_DERIV
7330 printf("Divide succeeded %d\n", i);
7331 #endif
7332 return(i);
7333 }
7334 xmlExpFree(ctxt, tmp);
7335 xmlExpFree(ctxt, tmp2);
7336 }
7337 #ifdef DEBUG_DERIV
7338 printf("Divide failed\n");
7339 #endif
7340 return(0);
7341 }
7342
7343 /**
7344 * xmlExpExpDeriveInt:
7345 * @ctxt: the expressions context
7346 * @exp: the englobing expression
7347 * @sub: the subexpression
7348 *
7349 * Try to do a step of Brzozowski derivation but at a higher level
7350 * the input being a subexpression.
7351 *
7352 * Returns the resulting expression or NULL in case of internal error
7353 */
7354 static xmlExpNodePtr
xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,xmlExpNodePtr sub)7355 xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7356 xmlExpNodePtr ret, tmp, tmp2, tmp3;
7357 const xmlChar **tab;
7358 int len, i;
7359
7360 /*
7361 * In case of equality and if the expression can only consume a finite
7362 * amount, then the derivation is empty
7363 */
7364 if ((exp == sub) && (exp->c_max >= 0)) {
7365 #ifdef DEBUG_DERIV
7366 printf("Equal(exp, sub) and finite -> Empty\n");
7367 #endif
7368 return(emptyExp);
7369 }
7370 /*
7371 * decompose sub sequence first
7372 */
7373 if (sub->type == XML_EXP_EMPTY) {
7374 #ifdef DEBUG_DERIV
7375 printf("Empty(sub) -> Empty\n");
7376 #endif
7377 exp->ref++;
7378 return(exp);
7379 }
7380 if (sub->type == XML_EXP_SEQ) {
7381 #ifdef DEBUG_DERIV
7382 printf("Seq(sub) -> decompose\n");
7383 #endif
7384 tmp = xmlExpExpDeriveInt(ctxt, exp, sub->exp_left);
7385 if (tmp == NULL)
7386 return(NULL);
7387 if (tmp == forbiddenExp)
7388 return(tmp);
7389 ret = xmlExpExpDeriveInt(ctxt, tmp, sub->exp_right);
7390 xmlExpFree(ctxt, tmp);
7391 return(ret);
7392 }
7393 if (sub->type == XML_EXP_OR) {
7394 #ifdef DEBUG_DERIV
7395 printf("Or(sub) -> decompose\n");
7396 #endif
7397 tmp = xmlExpExpDeriveInt(ctxt, exp, sub->exp_left);
7398 if (tmp == forbiddenExp)
7399 return(tmp);
7400 if (tmp == NULL)
7401 return(NULL);
7402 ret = xmlExpExpDeriveInt(ctxt, exp, sub->exp_right);
7403 if ((ret == NULL) || (ret == forbiddenExp)) {
7404 xmlExpFree(ctxt, tmp);
7405 return(ret);
7406 }
7407 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, tmp, ret, NULL, 0, 0));
7408 }
7409 if (!xmlExpCheckCard(exp, sub)) {
7410 #ifdef DEBUG_DERIV
7411 printf("CheckCard(exp, sub) failed -> Forbid\n");
7412 #endif
7413 return(forbiddenExp);
7414 }
7415 switch (exp->type) {
7416 case XML_EXP_EMPTY:
7417 if (sub == emptyExp)
7418 return(emptyExp);
7419 #ifdef DEBUG_DERIV
7420 printf("Empty(exp) -> Forbid\n");
7421 #endif
7422 return(forbiddenExp);
7423 case XML_EXP_FORBID:
7424 #ifdef DEBUG_DERIV
7425 printf("Forbid(exp) -> Forbid\n");
7426 #endif
7427 return(forbiddenExp);
7428 case XML_EXP_ATOM:
7429 if (sub->type == XML_EXP_ATOM) {
7430 /* TODO: handle wildcards */
7431 if (exp->exp_str == sub->exp_str) {
7432 #ifdef DEBUG_DERIV
7433 printf("Atom match -> Empty\n");
7434 #endif
7435 return(emptyExp);
7436 }
7437 #ifdef DEBUG_DERIV
7438 printf("Atom mismatch -> Forbid\n");
7439 #endif
7440 return(forbiddenExp);
7441 }
7442 if ((sub->type == XML_EXP_COUNT) &&
7443 (sub->exp_max == 1) &&
7444 (sub->exp_left->type == XML_EXP_ATOM)) {
7445 /* TODO: handle wildcards */
7446 if (exp->exp_str == sub->exp_left->exp_str) {
7447 #ifdef DEBUG_DERIV
7448 printf("Atom match -> Empty\n");
7449 #endif
7450 return(emptyExp);
7451 }
7452 #ifdef DEBUG_DERIV
7453 printf("Atom mismatch -> Forbid\n");
7454 #endif
7455 return(forbiddenExp);
7456 }
7457 #ifdef DEBUG_DERIV
7458 printf("Compex exp vs Atom -> Forbid\n");
7459 #endif
7460 return(forbiddenExp);
7461 case XML_EXP_SEQ:
7462 /* try to get the sequence consumed only if possible */
7463 if (xmlExpCheckCard(exp->exp_left, sub)) {
7464 /* See if the sequence can be consumed directly */
7465 #ifdef DEBUG_DERIV
7466 printf("Seq trying left only\n");
7467 #endif
7468 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7469 if ((ret != forbiddenExp) && (ret != NULL)) {
7470 #ifdef DEBUG_DERIV
7471 printf("Seq trying left only worked\n");
7472 #endif
7473 /*
7474 * TODO: assumption here that we are determinist
7475 * i.e. we won't get to a nillable exp left
7476 * subset which could be matched by the right
7477 * part too.
7478 * e.g.: (a | b)+,(a | c) and 'a+,a'
7479 */
7480 exp->exp_right->ref++;
7481 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret,
7482 exp->exp_right, NULL, 0, 0));
7483 }
7484 #ifdef DEBUG_DERIV
7485 } else {
7486 printf("Seq: left too short\n");
7487 #endif
7488 }
7489 /* Try instead to decompose */
7490 if (sub->type == XML_EXP_COUNT) {
7491 int min, max;
7492
7493 #ifdef DEBUG_DERIV
7494 printf("Seq: sub is a count\n");
7495 #endif
7496 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub->exp_left);
7497 if (ret == NULL)
7498 return(NULL);
7499 if (ret != forbiddenExp) {
7500 #ifdef DEBUG_DERIV
7501 printf("Seq , Count match on left\n");
7502 #endif
7503 if (sub->exp_max < 0)
7504 max = -1;
7505 else
7506 max = sub->exp_max -1;
7507 if (sub->exp_min > 0)
7508 min = sub->exp_min -1;
7509 else
7510 min = 0;
7511 exp->exp_right->ref++;
7512 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret,
7513 exp->exp_right, NULL, 0, 0);
7514 if (tmp == NULL)
7515 return(NULL);
7516
7517 sub->exp_left->ref++;
7518 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT,
7519 sub->exp_left, NULL, NULL, min, max);
7520 if (tmp2 == NULL) {
7521 xmlExpFree(ctxt, tmp);
7522 return(NULL);
7523 }
7524 ret = xmlExpExpDeriveInt(ctxt, tmp, tmp2);
7525 xmlExpFree(ctxt, tmp);
7526 xmlExpFree(ctxt, tmp2);
7527 return(ret);
7528 }
7529 }
7530 /* we made no progress on structured operations */
7531 break;
7532 case XML_EXP_OR:
7533 #ifdef DEBUG_DERIV
7534 printf("Or , trying both side\n");
7535 #endif
7536 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7537 if (ret == NULL)
7538 return(NULL);
7539 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_right, sub);
7540 if (tmp == NULL) {
7541 xmlExpFree(ctxt, ret);
7542 return(NULL);
7543 }
7544 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, tmp, NULL, 0, 0));
7545 case XML_EXP_COUNT: {
7546 int min, max;
7547
7548 if (sub->type == XML_EXP_COUNT) {
7549 /*
7550 * Try to see if the loop is completely subsumed
7551 */
7552 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub->exp_left);
7553 if (tmp == NULL)
7554 return(NULL);
7555 if (tmp == forbiddenExp) {
7556 int mult;
7557
7558 #ifdef DEBUG_DERIV
7559 printf("Count, Count inner don't subsume\n");
7560 #endif
7561 mult = xmlExpDivide(ctxt, sub->exp_left, exp->exp_left,
7562 NULL, &tmp);
7563 if (mult <= 0) {
7564 #ifdef DEBUG_DERIV
7565 printf("Count, Count not multiple => forbidden\n");
7566 #endif
7567 return(forbiddenExp);
7568 }
7569 if (sub->exp_max == -1) {
7570 max = -1;
7571 if (exp->exp_max == -1) {
7572 if (exp->exp_min <= sub->exp_min * mult)
7573 min = 0;
7574 else
7575 min = exp->exp_min - sub->exp_min * mult;
7576 } else {
7577 #ifdef DEBUG_DERIV
7578 printf("Count, Count finite can't subsume infinite\n");
7579 #endif
7580 xmlExpFree(ctxt, tmp);
7581 return(forbiddenExp);
7582 }
7583 } else {
7584 if (exp->exp_max == -1) {
7585 #ifdef DEBUG_DERIV
7586 printf("Infinite loop consume mult finite loop\n");
7587 #endif
7588 if (exp->exp_min > sub->exp_min * mult) {
7589 max = -1;
7590 min = exp->exp_min - sub->exp_min * mult;
7591 } else {
7592 max = -1;
7593 min = 0;
7594 }
7595 } else {
7596 if (exp->exp_max < sub->exp_max * mult) {
7597 #ifdef DEBUG_DERIV
7598 printf("loops max mult mismatch => forbidden\n");
7599 #endif
7600 xmlExpFree(ctxt, tmp);
7601 return(forbiddenExp);
7602 }
7603 if (sub->exp_max * mult > exp->exp_min)
7604 min = 0;
7605 else
7606 min = exp->exp_min - sub->exp_max * mult;
7607 max = exp->exp_max - sub->exp_max * mult;
7608 }
7609 }
7610 } else if (!IS_NILLABLE(tmp)) {
7611 /*
7612 * TODO: loop here to try to grow if working on finite
7613 * blocks.
7614 */
7615 #ifdef DEBUG_DERIV
7616 printf("Count, Count remain not nillable => forbidden\n");
7617 #endif
7618 xmlExpFree(ctxt, tmp);
7619 return(forbiddenExp);
7620 } else if (sub->exp_max == -1) {
7621 if (exp->exp_max == -1) {
7622 if (exp->exp_min <= sub->exp_min) {
7623 #ifdef DEBUG_DERIV
7624 printf("Infinite loops Okay => COUNT(0,Inf)\n");
7625 #endif
7626 max = -1;
7627 min = 0;
7628 } else {
7629 #ifdef DEBUG_DERIV
7630 printf("Infinite loops min => Count(X,Inf)\n");
7631 #endif
7632 max = -1;
7633 min = exp->exp_min - sub->exp_min;
7634 }
7635 } else if (exp->exp_min > sub->exp_min) {
7636 #ifdef DEBUG_DERIV
7637 printf("loops min mismatch 1 => forbidden ???\n");
7638 #endif
7639 xmlExpFree(ctxt, tmp);
7640 return(forbiddenExp);
7641 } else {
7642 max = -1;
7643 min = 0;
7644 }
7645 } else {
7646 if (exp->exp_max == -1) {
7647 #ifdef DEBUG_DERIV
7648 printf("Infinite loop consume finite loop\n");
7649 #endif
7650 if (exp->exp_min > sub->exp_min) {
7651 max = -1;
7652 min = exp->exp_min - sub->exp_min;
7653 } else {
7654 max = -1;
7655 min = 0;
7656 }
7657 } else {
7658 if (exp->exp_max < sub->exp_max) {
7659 #ifdef DEBUG_DERIV
7660 printf("loops max mismatch => forbidden\n");
7661 #endif
7662 xmlExpFree(ctxt, tmp);
7663 return(forbiddenExp);
7664 }
7665 if (sub->exp_max > exp->exp_min)
7666 min = 0;
7667 else
7668 min = exp->exp_min - sub->exp_max;
7669 max = exp->exp_max - sub->exp_max;
7670 }
7671 }
7672 #ifdef DEBUG_DERIV
7673 printf("loops match => SEQ(COUNT())\n");
7674 #endif
7675 exp->exp_left->ref++;
7676 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left,
7677 NULL, NULL, min, max);
7678 if (tmp2 == NULL) {
7679 return(NULL);
7680 }
7681 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, tmp, tmp2,
7682 NULL, 0, 0);
7683 return(ret);
7684 }
7685 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7686 if (tmp == NULL)
7687 return(NULL);
7688 if (tmp == forbiddenExp) {
7689 #ifdef DEBUG_DERIV
7690 printf("loop mismatch => forbidden\n");
7691 #endif
7692 return(forbiddenExp);
7693 }
7694 if (exp->exp_min > 0)
7695 min = exp->exp_min - 1;
7696 else
7697 min = 0;
7698 if (exp->exp_max < 0)
7699 max = -1;
7700 else
7701 max = exp->exp_max - 1;
7702
7703 #ifdef DEBUG_DERIV
7704 printf("loop match => SEQ(COUNT())\n");
7705 #endif
7706 exp->exp_left->ref++;
7707 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left,
7708 NULL, NULL, min, max);
7709 if (tmp2 == NULL)
7710 return(NULL);
7711 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, tmp, tmp2,
7712 NULL, 0, 0);
7713 return(ret);
7714 }
7715 }
7716
7717 #ifdef DEBUG_DERIV
7718 printf("Fallback to derivative\n");
7719 #endif
7720 if (IS_NILLABLE(sub)) {
7721 if (!(IS_NILLABLE(exp)))
7722 return(forbiddenExp);
7723 else
7724 ret = emptyExp;
7725 } else
7726 ret = NULL;
7727 /*
7728 * here the structured derivation made no progress so
7729 * we use the default token based derivation to force one more step
7730 */
7731 if (ctxt->tabSize == 0)
7732 ctxt->tabSize = 40;
7733
7734 tab = (const xmlChar **) xmlMalloc(ctxt->tabSize *
7735 sizeof(const xmlChar *));
7736 if (tab == NULL) {
7737 return(NULL);
7738 }
7739
7740 /*
7741 * collect all the strings accepted by the subexpression on input
7742 */
7743 len = xmlExpGetStartInt(ctxt, sub, tab, ctxt->tabSize, 0);
7744 while (len < 0) {
7745 const xmlChar **temp;
7746 temp = (const xmlChar **) xmlRealloc((xmlChar **) tab, ctxt->tabSize * 2 *
7747 sizeof(const xmlChar *));
7748 if (temp == NULL) {
7749 xmlFree((xmlChar **) tab);
7750 return(NULL);
7751 }
7752 tab = temp;
7753 ctxt->tabSize *= 2;
7754 len = xmlExpGetStartInt(ctxt, sub, tab, ctxt->tabSize, 0);
7755 }
7756 for (i = 0;i < len;i++) {
7757 tmp = xmlExpStringDeriveInt(ctxt, exp, tab[i]);
7758 if ((tmp == NULL) || (tmp == forbiddenExp)) {
7759 xmlExpFree(ctxt, ret);
7760 xmlFree((xmlChar **) tab);
7761 return(tmp);
7762 }
7763 tmp2 = xmlExpStringDeriveInt(ctxt, sub, tab[i]);
7764 if ((tmp2 == NULL) || (tmp2 == forbiddenExp)) {
7765 xmlExpFree(ctxt, tmp);
7766 xmlExpFree(ctxt, ret);
7767 xmlFree((xmlChar **) tab);
7768 return(tmp);
7769 }
7770 tmp3 = xmlExpExpDeriveInt(ctxt, tmp, tmp2);
7771 xmlExpFree(ctxt, tmp);
7772 xmlExpFree(ctxt, tmp2);
7773
7774 if ((tmp3 == NULL) || (tmp3 == forbiddenExp)) {
7775 xmlExpFree(ctxt, ret);
7776 xmlFree((xmlChar **) tab);
7777 return(tmp3);
7778 }
7779
7780 if (ret == NULL)
7781 ret = tmp3;
7782 else {
7783 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, tmp3, NULL, 0, 0);
7784 if (ret == NULL) {
7785 xmlFree((xmlChar **) tab);
7786 return(NULL);
7787 }
7788 }
7789 }
7790 xmlFree((xmlChar **) tab);
7791 return(ret);
7792 }
7793
7794 /**
7795 * xmlExpExpDerive:
7796 * @ctxt: the expressions context
7797 * @exp: the englobing expression
7798 * @sub: the subexpression
7799 *
7800 * Evaluates the expression resulting from @exp consuming a sub expression @sub
7801 * Based on algebraic derivation and sometimes direct Brzozowski derivation
7802 * it usually tatkes less than linear time and can handle expressions generating
7803 * infinite languages.
7804 *
7805 * Returns the resulting expression or NULL in case of internal error, the
7806 * result must be freed
7807 */
7808 xmlExpNodePtr
xmlExpExpDerive(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,xmlExpNodePtr sub)7809 xmlExpExpDerive(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7810 if ((exp == NULL) || (ctxt == NULL) || (sub == NULL))
7811 return(NULL);
7812
7813 /*
7814 * O(1) speedups
7815 */
7816 if (IS_NILLABLE(sub) && (!IS_NILLABLE(exp))) {
7817 #ifdef DEBUG_DERIV
7818 printf("Sub nillable and not exp : can't subsume\n");
7819 #endif
7820 return(forbiddenExp);
7821 }
7822 if (xmlExpCheckCard(exp, sub) == 0) {
7823 #ifdef DEBUG_DERIV
7824 printf("sub generate longuer sequances than exp : can't subsume\n");
7825 #endif
7826 return(forbiddenExp);
7827 }
7828 return(xmlExpExpDeriveInt(ctxt, exp, sub));
7829 }
7830
7831 /**
7832 * xmlExpSubsume:
7833 * @ctxt: the expressions context
7834 * @exp: the englobing expression
7835 * @sub: the subexpression
7836 *
7837 * Check whether @exp accepts all the languages accexpted by @sub
7838 * the input being a subexpression.
7839 *
7840 * Returns 1 if true 0 if false and -1 in case of failure.
7841 */
7842 int
xmlExpSubsume(xmlExpCtxtPtr ctxt,xmlExpNodePtr exp,xmlExpNodePtr sub)7843 xmlExpSubsume(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7844 xmlExpNodePtr tmp;
7845
7846 if ((exp == NULL) || (ctxt == NULL) || (sub == NULL))
7847 return(-1);
7848
7849 /*
7850 * TODO: speedup by checking the language of sub is a subset of the
7851 * language of exp
7852 */
7853 /*
7854 * O(1) speedups
7855 */
7856 if (IS_NILLABLE(sub) && (!IS_NILLABLE(exp))) {
7857 #ifdef DEBUG_DERIV
7858 printf("Sub nillable and not exp : can't subsume\n");
7859 #endif
7860 return(0);
7861 }
7862 if (xmlExpCheckCard(exp, sub) == 0) {
7863 #ifdef DEBUG_DERIV
7864 printf("sub generate longuer sequances than exp : can't subsume\n");
7865 #endif
7866 return(0);
7867 }
7868 tmp = xmlExpExpDeriveInt(ctxt, exp, sub);
7869 #ifdef DEBUG_DERIV
7870 printf("Result derivation :\n");
7871 PRINT_EXP(tmp);
7872 #endif
7873 if (tmp == NULL)
7874 return(-1);
7875 if (tmp == forbiddenExp)
7876 return(0);
7877 if (tmp == emptyExp)
7878 return(1);
7879 if ((tmp != NULL) && (IS_NILLABLE(tmp))) {
7880 xmlExpFree(ctxt, tmp);
7881 return(1);
7882 }
7883 xmlExpFree(ctxt, tmp);
7884 return(0);
7885 }
7886
7887 /************************************************************************
7888 * *
7889 * Parsing expression *
7890 * *
7891 ************************************************************************/
7892
7893 static xmlExpNodePtr xmlExpParseExpr(xmlExpCtxtPtr ctxt);
7894
7895 #undef CUR
7896 #define CUR (*ctxt->cur)
7897 #undef NEXT
7898 #define NEXT ctxt->cur++;
7899 #undef IS_BLANK
7900 #define IS_BLANK(c) ((c == ' ') || (c == '\n') || (c == '\r') || (c == '\t'))
7901 #define SKIP_BLANKS while (IS_BLANK(*ctxt->cur)) ctxt->cur++;
7902
7903 static int
xmlExpParseNumber(xmlExpCtxtPtr ctxt)7904 xmlExpParseNumber(xmlExpCtxtPtr ctxt) {
7905 int ret = 0;
7906
7907 SKIP_BLANKS
7908 if (CUR == '*') {
7909 NEXT
7910 return(-1);
7911 }
7912 if ((CUR < '0') || (CUR > '9'))
7913 return(-1);
7914 while ((CUR >= '0') && (CUR <= '9')) {
7915 ret = ret * 10 + (CUR - '0');
7916 NEXT
7917 }
7918 return(ret);
7919 }
7920
7921 static xmlExpNodePtr
xmlExpParseOr(xmlExpCtxtPtr ctxt)7922 xmlExpParseOr(xmlExpCtxtPtr ctxt) {
7923 const char *base;
7924 xmlExpNodePtr ret;
7925 const xmlChar *val;
7926
7927 SKIP_BLANKS
7928 base = ctxt->cur;
7929 if (*ctxt->cur == '(') {
7930 NEXT
7931 ret = xmlExpParseExpr(ctxt);
7932 SKIP_BLANKS
7933 if (*ctxt->cur != ')') {
7934 fprintf(stderr, "unbalanced '(' : %s\n", base);
7935 xmlExpFree(ctxt, ret);
7936 return(NULL);
7937 }
7938 NEXT;
7939 SKIP_BLANKS
7940 goto parse_quantifier;
7941 }
7942 while ((CUR != 0) && (!(IS_BLANK(CUR))) && (CUR != '(') &&
7943 (CUR != ')') && (CUR != '|') && (CUR != ',') && (CUR != '{') &&
7944 (CUR != '*') && (CUR != '+') && (CUR != '?') && (CUR != '}'))
7945 NEXT;
7946 val = xmlDictLookup(ctxt->dict, BAD_CAST base, ctxt->cur - base);
7947 if (val == NULL)
7948 return(NULL);
7949 ret = xmlExpHashGetEntry(ctxt, XML_EXP_ATOM, NULL, NULL, val, 0, 0);
7950 if (ret == NULL)
7951 return(NULL);
7952 SKIP_BLANKS
7953 parse_quantifier:
7954 if (CUR == '{') {
7955 int min, max;
7956
7957 NEXT
7958 min = xmlExpParseNumber(ctxt);
7959 if (min < 0) {
7960 xmlExpFree(ctxt, ret);
7961 return(NULL);
7962 }
7963 SKIP_BLANKS
7964 if (CUR == ',') {
7965 NEXT
7966 max = xmlExpParseNumber(ctxt);
7967 SKIP_BLANKS
7968 } else
7969 max = min;
7970 if (CUR != '}') {
7971 xmlExpFree(ctxt, ret);
7972 return(NULL);
7973 }
7974 NEXT
7975 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7976 min, max);
7977 SKIP_BLANKS
7978 } else if (CUR == '?') {
7979 NEXT
7980 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7981 0, 1);
7982 SKIP_BLANKS
7983 } else if (CUR == '+') {
7984 NEXT
7985 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7986 1, -1);
7987 SKIP_BLANKS
7988 } else if (CUR == '*') {
7989 NEXT
7990 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7991 0, -1);
7992 SKIP_BLANKS
7993 }
7994 return(ret);
7995 }
7996
7997
7998 static xmlExpNodePtr
xmlExpParseSeq(xmlExpCtxtPtr ctxt)7999 xmlExpParseSeq(xmlExpCtxtPtr ctxt) {
8000 xmlExpNodePtr ret, right;
8001
8002 ret = xmlExpParseOr(ctxt);
8003 SKIP_BLANKS
8004 while (CUR == '|') {
8005 NEXT
8006 right = xmlExpParseOr(ctxt);
8007 if (right == NULL) {
8008 xmlExpFree(ctxt, ret);
8009 return(NULL);
8010 }
8011 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, right, NULL, 0, 0);
8012 if (ret == NULL)
8013 return(NULL);
8014 }
8015 return(ret);
8016 }
8017
8018 static xmlExpNodePtr
xmlExpParseExpr(xmlExpCtxtPtr ctxt)8019 xmlExpParseExpr(xmlExpCtxtPtr ctxt) {
8020 xmlExpNodePtr ret, right;
8021
8022 ret = xmlExpParseSeq(ctxt);
8023 SKIP_BLANKS
8024 while (CUR == ',') {
8025 NEXT
8026 right = xmlExpParseSeq(ctxt);
8027 if (right == NULL) {
8028 xmlExpFree(ctxt, ret);
8029 return(NULL);
8030 }
8031 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, right, NULL, 0, 0);
8032 if (ret == NULL)
8033 return(NULL);
8034 }
8035 return(ret);
8036 }
8037
8038 /**
8039 * xmlExpParse:
8040 * @ctxt: the expressions context
8041 * @expr: the 0 terminated string
8042 *
8043 * Minimal parser for regexps, it understand the following constructs
8044 * - string terminals
8045 * - choice operator |
8046 * - sequence operator ,
8047 * - subexpressions (...)
8048 * - usual cardinality operators + * and ?
8049 * - finite sequences { min, max }
8050 * - infinite sequences { min, * }
8051 * There is minimal checkings made especially no checking on strings values
8052 *
8053 * Returns a new expression or NULL in case of failure
8054 */
8055 xmlExpNodePtr
xmlExpParse(xmlExpCtxtPtr ctxt,const char * expr)8056 xmlExpParse(xmlExpCtxtPtr ctxt, const char *expr) {
8057 xmlExpNodePtr ret;
8058
8059 ctxt->expr = expr;
8060 ctxt->cur = expr;
8061
8062 ret = xmlExpParseExpr(ctxt);
8063 SKIP_BLANKS
8064 if (*ctxt->cur != 0) {
8065 xmlExpFree(ctxt, ret);
8066 return(NULL);
8067 }
8068 return(ret);
8069 }
8070
8071 static void
xmlExpDumpInt(xmlBufferPtr buf,xmlExpNodePtr expr,int glob)8072 xmlExpDumpInt(xmlBufferPtr buf, xmlExpNodePtr expr, int glob) {
8073 xmlExpNodePtr c;
8074
8075 if (expr == NULL) return;
8076 if (glob) xmlBufferWriteChar(buf, "(");
8077 switch (expr->type) {
8078 case XML_EXP_EMPTY:
8079 xmlBufferWriteChar(buf, "empty");
8080 break;
8081 case XML_EXP_FORBID:
8082 xmlBufferWriteChar(buf, "forbidden");
8083 break;
8084 case XML_EXP_ATOM:
8085 xmlBufferWriteCHAR(buf, expr->exp_str);
8086 break;
8087 case XML_EXP_SEQ:
8088 c = expr->exp_left;
8089 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8090 xmlExpDumpInt(buf, c, 1);
8091 else
8092 xmlExpDumpInt(buf, c, 0);
8093 xmlBufferWriteChar(buf, " , ");
8094 c = expr->exp_right;
8095 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8096 xmlExpDumpInt(buf, c, 1);
8097 else
8098 xmlExpDumpInt(buf, c, 0);
8099 break;
8100 case XML_EXP_OR:
8101 c = expr->exp_left;
8102 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8103 xmlExpDumpInt(buf, c, 1);
8104 else
8105 xmlExpDumpInt(buf, c, 0);
8106 xmlBufferWriteChar(buf, " | ");
8107 c = expr->exp_right;
8108 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8109 xmlExpDumpInt(buf, c, 1);
8110 else
8111 xmlExpDumpInt(buf, c, 0);
8112 break;
8113 case XML_EXP_COUNT: {
8114 char rep[40];
8115
8116 c = expr->exp_left;
8117 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8118 xmlExpDumpInt(buf, c, 1);
8119 else
8120 xmlExpDumpInt(buf, c, 0);
8121 if ((expr->exp_min == 0) && (expr->exp_max == 1)) {
8122 rep[0] = '?';
8123 rep[1] = 0;
8124 } else if ((expr->exp_min == 0) && (expr->exp_max == -1)) {
8125 rep[0] = '*';
8126 rep[1] = 0;
8127 } else if ((expr->exp_min == 1) && (expr->exp_max == -1)) {
8128 rep[0] = '+';
8129 rep[1] = 0;
8130 } else if (expr->exp_max == expr->exp_min) {
8131 snprintf(rep, 39, "{%d}", expr->exp_min);
8132 } else if (expr->exp_max < 0) {
8133 snprintf(rep, 39, "{%d,inf}", expr->exp_min);
8134 } else {
8135 snprintf(rep, 39, "{%d,%d}", expr->exp_min, expr->exp_max);
8136 }
8137 rep[39] = 0;
8138 xmlBufferWriteChar(buf, rep);
8139 break;
8140 }
8141 default:
8142 fprintf(stderr, "Error in tree\n");
8143 }
8144 if (glob)
8145 xmlBufferWriteChar(buf, ")");
8146 }
8147 /**
8148 * xmlExpDump:
8149 * @buf: a buffer to receive the output
8150 * @expr: the compiled expression
8151 *
8152 * Serialize the expression as compiled to the buffer
8153 */
8154 void
xmlExpDump(xmlBufferPtr buf,xmlExpNodePtr expr)8155 xmlExpDump(xmlBufferPtr buf, xmlExpNodePtr expr) {
8156 if ((buf == NULL) || (expr == NULL))
8157 return;
8158 xmlExpDumpInt(buf, expr, 0);
8159 }
8160
8161 /**
8162 * xmlExpMaxToken:
8163 * @expr: a compiled expression
8164 *
8165 * Indicate the maximum number of input a expression can accept
8166 *
8167 * Returns the maximum length or -1 in case of error
8168 */
8169 int
xmlExpMaxToken(xmlExpNodePtr expr)8170 xmlExpMaxToken(xmlExpNodePtr expr) {
8171 if (expr == NULL)
8172 return(-1);
8173 return(expr->c_max);
8174 }
8175
8176 /**
8177 * xmlExpCtxtNbNodes:
8178 * @ctxt: an expression context
8179 *
8180 * Debugging facility provides the number of allocated nodes at a that point
8181 *
8182 * Returns the number of nodes in use or -1 in case of error
8183 */
8184 int
xmlExpCtxtNbNodes(xmlExpCtxtPtr ctxt)8185 xmlExpCtxtNbNodes(xmlExpCtxtPtr ctxt) {
8186 if (ctxt == NULL)
8187 return(-1);
8188 return(ctxt->nb_nodes);
8189 }
8190
8191 /**
8192 * xmlExpCtxtNbCons:
8193 * @ctxt: an expression context
8194 *
8195 * Debugging facility provides the number of allocated nodes over lifetime
8196 *
8197 * Returns the number of nodes ever allocated or -1 in case of error
8198 */
8199 int
xmlExpCtxtNbCons(xmlExpCtxtPtr ctxt)8200 xmlExpCtxtNbCons(xmlExpCtxtPtr ctxt) {
8201 if (ctxt == NULL)
8202 return(-1);
8203 return(ctxt->nb_cons);
8204 }
8205
8206 #endif /* LIBXML_EXPR_ENABLED */
8207 #define bottom_xmlregexp
8208 #include "elfgcchack.h"
8209 #endif /* LIBXML_REGEXP_ENABLED */
8210