2 * regexp.c: generic and extensible Regular Expression engine
4 * Basically designed with the purpose of compiling regexps for
5 * the variety of validation/shemas mechanisms now available in
6 * XML related specifications these include:
7 * - XML-1.0 DTD validation
8 * - XML Schemas structure part 1
9 * - XML Schemas Datatypes part 2 especially Appendix F
10 * - RELAX-NG/TREX i.e. the counter proposal
12 * See Copyright for the status of this software.
14 * Daniel Veillard <veillard@redhat.com>
20 #ifdef LIBXML_REGEXP_ENABLED
22 /* #define DEBUG_ERR */
30 #include <libxml/tree.h>
31 #include <libxml/parserInternals.h>
32 #include <libxml/xmlregexp.h>
33 #include <libxml/xmlautomata.h>
34 #include <libxml/xmlunicode.h>
37 #define INT_MAX 123456789 /* easy to flag and big enough for our needs */
40 /* #define DEBUG_REGEXP_GRAPH */
41 /* #define DEBUG_REGEXP_EXEC */
42 /* #define DEBUG_PUSH */
43 /* #define DEBUG_COMPACTION */
45 #define MAX_PUSH 10000000
51 ctxt->error = XML_REGEXP_COMPILE_ERROR; \
52 xmlRegexpErrCompile(ctxt, str);
53 #define NEXT ctxt->cur++
54 #define CUR (*(ctxt->cur))
55 #define NXT(index) (ctxt->cur[index])
57 #define CUR_SCHAR(s, l) xmlStringCurrentChar(NULL, s, &l)
58 #define NEXTL(l) ctxt->cur += l;
59 #define XML_REG_STRING_SEPARATOR '|'
61 * Need PREV to check on a '-' within a Character Group. May only be used
62 * when it's guaranteed that cur is not at the beginning of ctxt->string!
64 #define PREV (ctxt->cur[-1])
69 * macro to flag unimplemented blocks
72 xmlGenericError(xmlGenericErrorContext, \
73 "Unimplemented block at %s:%d\n", \
76 /************************************************************************
78 * Datatypes and structures *
80 ************************************************************************/
83 * Note: the order of the enums below is significant, do not shuffle
86 XML_REGEXP_EPSILON = 1,
89 XML_REGEXP_SUBREG, /* used for () sub regexps */
91 XML_REGEXP_ANYCHAR, /* . */
92 XML_REGEXP_ANYSPACE, /* \s */
93 XML_REGEXP_NOTSPACE, /* \S */
94 XML_REGEXP_INITNAME, /* \l */
95 XML_REGEXP_NOTINITNAME, /* \L */
96 XML_REGEXP_NAMECHAR, /* \c */
97 XML_REGEXP_NOTNAMECHAR, /* \C */
98 XML_REGEXP_DECIMAL, /* \d */
99 XML_REGEXP_NOTDECIMAL, /* \D */
100 XML_REGEXP_REALCHAR, /* \w */
101 XML_REGEXP_NOTREALCHAR, /* \W */
102 XML_REGEXP_LETTER = 100,
103 XML_REGEXP_LETTER_UPPERCASE,
104 XML_REGEXP_LETTER_LOWERCASE,
105 XML_REGEXP_LETTER_TITLECASE,
106 XML_REGEXP_LETTER_MODIFIER,
107 XML_REGEXP_LETTER_OTHERS,
109 XML_REGEXP_MARK_NONSPACING,
110 XML_REGEXP_MARK_SPACECOMBINING,
111 XML_REGEXP_MARK_ENCLOSING,
113 XML_REGEXP_NUMBER_DECIMAL,
114 XML_REGEXP_NUMBER_LETTER,
115 XML_REGEXP_NUMBER_OTHERS,
117 XML_REGEXP_PUNCT_CONNECTOR,
118 XML_REGEXP_PUNCT_DASH,
119 XML_REGEXP_PUNCT_OPEN,
120 XML_REGEXP_PUNCT_CLOSE,
121 XML_REGEXP_PUNCT_INITQUOTE,
122 XML_REGEXP_PUNCT_FINQUOTE,
123 XML_REGEXP_PUNCT_OTHERS,
125 XML_REGEXP_SEPAR_SPACE,
126 XML_REGEXP_SEPAR_LINE,
127 XML_REGEXP_SEPAR_PARA,
129 XML_REGEXP_SYMBOL_MATH,
130 XML_REGEXP_SYMBOL_CURRENCY,
131 XML_REGEXP_SYMBOL_MODIFIER,
132 XML_REGEXP_SYMBOL_OTHERS,
134 XML_REGEXP_OTHER_CONTROL,
135 XML_REGEXP_OTHER_FORMAT,
136 XML_REGEXP_OTHER_PRIVATE,
138 XML_REGEXP_BLOCK_NAME
142 XML_REGEXP_QUANT_EPSILON = 1,
143 XML_REGEXP_QUANT_ONCE,
144 XML_REGEXP_QUANT_OPT,
145 XML_REGEXP_QUANT_MULT,
146 XML_REGEXP_QUANT_PLUS,
147 XML_REGEXP_QUANT_ONCEONLY,
148 XML_REGEXP_QUANT_ALL,
149 XML_REGEXP_QUANT_RANGE
153 XML_REGEXP_START_STATE = 1,
154 XML_REGEXP_FINAL_STATE,
155 XML_REGEXP_TRANS_STATE,
156 XML_REGEXP_SINK_STATE,
157 XML_REGEXP_UNREACH_STATE
161 XML_REGEXP_MARK_NORMAL = 0,
162 XML_REGEXP_MARK_START,
163 XML_REGEXP_MARK_VISITED
166 typedef struct _xmlRegRange xmlRegRange;
167 typedef xmlRegRange *xmlRegRangePtr;
169 struct _xmlRegRange {
170 int neg; /* 0 normal, 1 not, 2 exclude */
177 typedef struct _xmlRegAtom xmlRegAtom;
178 typedef xmlRegAtom *xmlRegAtomPtr;
180 typedef struct _xmlAutomataState xmlRegState;
181 typedef xmlRegState *xmlRegStatePtr;
186 xmlRegQuantType quant;
194 xmlRegStatePtr start;
195 xmlRegStatePtr start0;
199 xmlRegRangePtr *ranges;
203 typedef struct _xmlRegCounter xmlRegCounter;
204 typedef xmlRegCounter *xmlRegCounterPtr;
206 struct _xmlRegCounter {
211 typedef struct _xmlRegTrans xmlRegTrans;
212 typedef xmlRegTrans *xmlRegTransPtr;
214 struct _xmlRegTrans {
222 struct _xmlAutomataState {
223 xmlRegStateType type;
224 xmlRegMarkedType mark;
225 xmlRegMarkedType markd;
226 xmlRegMarkedType reached;
231 /* knowing states ponting to us can speed things up */
237 typedef struct _xmlAutomata xmlRegParserCtxt;
238 typedef xmlRegParserCtxt *xmlRegParserCtxtPtr;
240 #define AM_AUTOMATA_RNG 1
242 struct _xmlAutomata {
249 xmlRegStatePtr start;
251 xmlRegStatePtr state;
257 xmlRegAtomPtr *atoms;
261 xmlRegStatePtr *states;
265 xmlRegCounter *counters;
275 xmlRegStatePtr *states;
277 xmlRegAtomPtr *atoms;
279 xmlRegCounter *counters;
283 * That's the compact form for determinists automatas
292 typedef struct _xmlRegExecRollback xmlRegExecRollback;
293 typedef xmlRegExecRollback *xmlRegExecRollbackPtr;
295 struct _xmlRegExecRollback {
296 xmlRegStatePtr state;/* the current state */
297 int index; /* the index in the input stack */
298 int nextbranch; /* the next transition to explore in that state */
299 int *counts; /* save the automata state if it has some */
302 typedef struct _xmlRegInputToken xmlRegInputToken;
303 typedef xmlRegInputToken *xmlRegInputTokenPtr;
305 struct _xmlRegInputToken {
310 struct _xmlRegExecCtxt {
311 int status; /* execution status != 0 indicate an error */
312 int determinist; /* did we find an indeterministic behaviour */
313 xmlRegexpPtr comp; /* the compiled regexp */
314 xmlRegExecCallbacks callback;
317 xmlRegStatePtr state;/* the current state */
318 int transno; /* the current transition on that state */
319 int transcount; /* the number of chars in char counted transitions */
322 * A stack of rollback states
326 xmlRegExecRollback *rollbacks;
329 * The state of the automata if any
340 const xmlChar *inputString; /* when operating on characters */
341 xmlRegInputTokenPtr inputStack;/* when operating on strings */
346 int errStateNo; /* the error state number */
347 xmlRegStatePtr errState; /* the error state */
348 xmlChar *errString; /* the string raising the error */
349 int *errCounts; /* counters at the error state */
353 #define REGEXP_ALL_COUNTER 0x123456
354 #define REGEXP_ALL_LAX_COUNTER 0x123457
356 static void xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top);
357 static void xmlRegFreeState(xmlRegStatePtr state);
358 static void xmlRegFreeAtom(xmlRegAtomPtr atom);
359 static int xmlRegStrEqualWildcard(const xmlChar *expStr, const xmlChar *valStr);
360 static int xmlRegCheckCharacter(xmlRegAtomPtr atom, int codepoint);
361 static int xmlRegCheckCharacterRange(xmlRegAtomType type, int codepoint,
362 int neg, int start, int end, const xmlChar *blockName);
364 void xmlAutomataSetFlags(xmlAutomataPtr am, int flags);
366 /************************************************************************
368 * Regexp memory error handler *
370 ************************************************************************/
372 * xmlRegexpErrMemory:
373 * @extra: extra information
375 * Handle an out of memory condition
378 xmlRegexpErrMemory(xmlRegParserCtxtPtr ctxt, const char *extra)
380 const char *regexp = NULL;
382 regexp = (const char *) ctxt->string;
383 ctxt->error = XML_ERR_NO_MEMORY;
385 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_REGEXP,
386 XML_ERR_NO_MEMORY, XML_ERR_FATAL, NULL, 0, extra,
388 "Memory allocation failed : %s\n", extra);
392 * xmlRegexpErrCompile:
393 * @extra: extra information
395 * Handle a compilation failure
398 xmlRegexpErrCompile(xmlRegParserCtxtPtr ctxt, const char *extra)
400 const char *regexp = NULL;
404 regexp = (const char *) ctxt->string;
405 idx = ctxt->cur - ctxt->string;
406 ctxt->error = XML_REGEXP_COMPILE_ERROR;
408 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_REGEXP,
409 XML_REGEXP_COMPILE_ERROR, XML_ERR_FATAL, NULL, 0, extra,
410 regexp, NULL, idx, 0,
411 "failed to compile: %s\n", extra);
414 /************************************************************************
416 * Allocation/Deallocation *
418 ************************************************************************/
420 static int xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt);
422 * xmlRegEpxFromParse:
423 * @ctxt: the parser context used to build it
425 * Allocate a new regexp and fill it with the result from the parser
427 * Returns the new regexp or NULL in case of error
430 xmlRegEpxFromParse(xmlRegParserCtxtPtr ctxt) {
433 ret = (xmlRegexpPtr) xmlMalloc(sizeof(xmlRegexp));
435 xmlRegexpErrMemory(ctxt, "compiling regexp");
438 memset(ret, 0, sizeof(xmlRegexp));
439 ret->string = ctxt->string;
440 ret->nbStates = ctxt->nbStates;
441 ret->states = ctxt->states;
442 ret->nbAtoms = ctxt->nbAtoms;
443 ret->atoms = ctxt->atoms;
444 ret->nbCounters = ctxt->nbCounters;
445 ret->counters = ctxt->counters;
446 ret->determinist = ctxt->determinist;
447 ret->flags = ctxt->flags;
448 if (ret->determinist == -1) {
449 xmlRegexpIsDeterminist(ret);
452 if ((ret->determinist != 0) &&
453 (ret->nbCounters == 0) &&
455 (ret->atoms != NULL) &&
456 (ret->atoms[0] != NULL) &&
457 (ret->atoms[0]->type == XML_REGEXP_STRING)) {
458 int i, j, nbstates = 0, nbatoms = 0;
467 * Switch to a compact representation
468 * 1/ counting the effective number of states left
469 * 2/ counting the unique number of atoms, and check that
470 * they are all of the string type
471 * 3/ build a table state x atom for the transitions
474 stateRemap = xmlMalloc(ret->nbStates * sizeof(int));
475 if (stateRemap == NULL) {
476 xmlRegexpErrMemory(ctxt, "compiling regexp");
480 for (i = 0;i < ret->nbStates;i++) {
481 if (ret->states[i] != NULL) {
482 stateRemap[i] = nbstates;
488 #ifdef DEBUG_COMPACTION
489 printf("Final: %d states\n", nbstates);
491 stringMap = xmlMalloc(ret->nbAtoms * sizeof(char *));
492 if (stringMap == NULL) {
493 xmlRegexpErrMemory(ctxt, "compiling regexp");
498 stringRemap = xmlMalloc(ret->nbAtoms * sizeof(int));
499 if (stringRemap == NULL) {
500 xmlRegexpErrMemory(ctxt, "compiling regexp");
506 for (i = 0;i < ret->nbAtoms;i++) {
507 if ((ret->atoms[i]->type == XML_REGEXP_STRING) &&
508 (ret->atoms[i]->quant == XML_REGEXP_QUANT_ONCE)) {
509 value = ret->atoms[i]->valuep;
510 for (j = 0;j < nbatoms;j++) {
511 if (xmlStrEqual(stringMap[j], value)) {
517 stringRemap[i] = nbatoms;
518 stringMap[nbatoms] = xmlStrdup(value);
519 if (stringMap[nbatoms] == NULL) {
520 for (i = 0;i < nbatoms;i++)
521 xmlFree(stringMap[i]);
522 xmlFree(stringRemap);
532 xmlFree(stringRemap);
533 for (i = 0;i < nbatoms;i++)
534 xmlFree(stringMap[i]);
540 #ifdef DEBUG_COMPACTION
541 printf("Final: %d atoms\n", nbatoms);
543 transitions = (int *) xmlMalloc((nbstates + 1) *
544 (nbatoms + 1) * sizeof(int));
545 if (transitions == NULL) {
547 xmlFree(stringRemap);
552 memset(transitions, 0, (nbstates + 1) * (nbatoms + 1) * sizeof(int));
555 * Allocate the transition table. The first entry for each
556 * state corresponds to the state type.
560 for (i = 0;i < ret->nbStates;i++) {
561 int stateno, atomno, targetno, prev;
562 xmlRegStatePtr state;
563 xmlRegTransPtr trans;
565 stateno = stateRemap[i];
568 state = ret->states[i];
570 transitions[stateno * (nbatoms + 1)] = state->type;
572 for (j = 0;j < state->nbTrans;j++) {
573 trans = &(state->trans[j]);
574 if ((trans->to == -1) || (trans->atom == NULL))
576 atomno = stringRemap[trans->atom->no];
577 if ((trans->atom->data != NULL) && (transdata == NULL)) {
578 transdata = (void **) xmlMalloc(nbstates * nbatoms *
580 if (transdata != NULL)
582 nbstates * nbatoms * sizeof(void *));
584 xmlRegexpErrMemory(ctxt, "compiling regexp");
588 targetno = stateRemap[trans->to];
590 * if the same atom can generate transitions to 2 different
591 * states then it means the automata is not determinist and
592 * the compact form can't be used !
594 prev = transitions[stateno * (nbatoms + 1) + atomno + 1];
596 if (prev != targetno + 1) {
597 ret->determinist = 0;
598 #ifdef DEBUG_COMPACTION
599 printf("Indet: state %d trans %d, atom %d to %d : %d to %d\n",
600 i, j, trans->atom->no, trans->to, atomno, targetno);
601 printf(" previous to is %d\n", prev);
603 if (transdata != NULL)
605 xmlFree(transitions);
607 xmlFree(stringRemap);
608 for (i = 0;i < nbatoms;i++)
609 xmlFree(stringMap[i]);
615 printf("State %d trans %d: atom %d to %d : %d to %d\n",
616 i, j, trans->atom->no, trans->to, atomno, targetno);
618 transitions[stateno * (nbatoms + 1) + atomno + 1] =
619 targetno + 1; /* to avoid 0 */
620 if (transdata != NULL)
621 transdata[stateno * nbatoms + atomno] =
626 ret->determinist = 1;
627 #ifdef DEBUG_COMPACTION
631 for (i = 0;i < nbstates;i++) {
632 for (j = 0;j < nbatoms + 1;j++) {
633 printf("%02d ", transitions[i * (nbatoms + 1) + j]);
640 * Cleanup of the old data
642 if (ret->states != NULL) {
643 for (i = 0;i < ret->nbStates;i++)
644 xmlRegFreeState(ret->states[i]);
645 xmlFree(ret->states);
649 if (ret->atoms != NULL) {
650 for (i = 0;i < ret->nbAtoms;i++)
651 xmlRegFreeAtom(ret->atoms[i]);
657 ret->compact = transitions;
658 ret->transdata = transdata;
659 ret->stringMap = stringMap;
660 ret->nbstrings = nbatoms;
661 ret->nbstates = nbstates;
663 xmlFree(stringRemap);
671 ctxt->nbCounters = 0;
672 ctxt->counters = NULL;
677 * xmlRegNewParserCtxt:
678 * @string: the string to parse
680 * Allocate a new regexp parser context
682 * Returns the new context or NULL in case of error
684 static xmlRegParserCtxtPtr
685 xmlRegNewParserCtxt(const xmlChar *string) {
686 xmlRegParserCtxtPtr ret;
688 ret = (xmlRegParserCtxtPtr) xmlMalloc(sizeof(xmlRegParserCtxt));
691 memset(ret, 0, sizeof(xmlRegParserCtxt));
693 ret->string = xmlStrdup(string);
694 ret->cur = ret->string;
698 ret->determinist = -1;
704 * @ctxt: the regexp parser context
705 * @neg: is that negative
706 * @type: the type of range
707 * @start: the start codepoint
708 * @end: the end codepoint
710 * Allocate a new regexp range
712 * Returns the new range or NULL in case of error
714 static xmlRegRangePtr
715 xmlRegNewRange(xmlRegParserCtxtPtr ctxt,
716 int neg, xmlRegAtomType type, int start, int end) {
719 ret = (xmlRegRangePtr) xmlMalloc(sizeof(xmlRegRange));
721 xmlRegexpErrMemory(ctxt, "allocating range");
733 * @range: the regexp range
735 * Free a regexp range
738 xmlRegFreeRange(xmlRegRangePtr range) {
742 if (range->blockName != NULL)
743 xmlFree(range->blockName);
749 * @range: the regexp range
751 * Copy a regexp range
753 * Returns the new copy or NULL in case of error.
755 static xmlRegRangePtr
756 xmlRegCopyRange(xmlRegParserCtxtPtr ctxt, xmlRegRangePtr range) {
762 ret = xmlRegNewRange(ctxt, range->neg, range->type, range->start,
766 if (range->blockName != NULL) {
767 ret->blockName = xmlStrdup(range->blockName);
768 if (ret->blockName == NULL) {
769 xmlRegexpErrMemory(ctxt, "allocating range");
770 xmlRegFreeRange(ret);
779 * @ctxt: the regexp parser context
780 * @type: the type of atom
782 * Allocate a new atom
784 * Returns the new atom or NULL in case of error
787 xmlRegNewAtom(xmlRegParserCtxtPtr ctxt, xmlRegAtomType type) {
790 ret = (xmlRegAtomPtr) xmlMalloc(sizeof(xmlRegAtom));
792 xmlRegexpErrMemory(ctxt, "allocating atom");
795 memset(ret, 0, sizeof(xmlRegAtom));
797 ret->quant = XML_REGEXP_QUANT_ONCE;
805 * @atom: the regexp atom
810 xmlRegFreeAtom(xmlRegAtomPtr atom) {
816 for (i = 0;i < atom->nbRanges;i++)
817 xmlRegFreeRange(atom->ranges[i]);
818 if (atom->ranges != NULL)
819 xmlFree(atom->ranges);
820 if ((atom->type == XML_REGEXP_STRING) && (atom->valuep != NULL))
821 xmlFree(atom->valuep);
822 if ((atom->type == XML_REGEXP_STRING) && (atom->valuep2 != NULL))
823 xmlFree(atom->valuep2);
824 if ((atom->type == XML_REGEXP_BLOCK_NAME) && (atom->valuep != NULL))
825 xmlFree(atom->valuep);
831 * @ctxt: the regexp parser context
832 * @atom: the oiginal atom
834 * Allocate a new regexp range
836 * Returns the new atom or NULL in case of error
839 xmlRegCopyAtom(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom) {
842 ret = (xmlRegAtomPtr) xmlMalloc(sizeof(xmlRegAtom));
844 xmlRegexpErrMemory(ctxt, "copying atom");
847 memset(ret, 0, sizeof(xmlRegAtom));
848 ret->type = atom->type;
849 ret->quant = atom->quant;
850 ret->min = atom->min;
851 ret->max = atom->max;
852 if (atom->nbRanges > 0) {
855 ret->ranges = (xmlRegRangePtr *) xmlMalloc(sizeof(xmlRegRangePtr) *
857 if (ret->ranges == NULL) {
858 xmlRegexpErrMemory(ctxt, "copying atom");
861 for (i = 0;i < atom->nbRanges;i++) {
862 ret->ranges[i] = xmlRegCopyRange(ctxt, atom->ranges[i]);
863 if (ret->ranges[i] == NULL)
865 ret->nbRanges = i + 1;
875 static xmlRegStatePtr
876 xmlRegNewState(xmlRegParserCtxtPtr ctxt) {
879 ret = (xmlRegStatePtr) xmlMalloc(sizeof(xmlRegState));
881 xmlRegexpErrMemory(ctxt, "allocating state");
884 memset(ret, 0, sizeof(xmlRegState));
885 ret->type = XML_REGEXP_TRANS_STATE;
886 ret->mark = XML_REGEXP_MARK_NORMAL;
892 * @state: the regexp state
894 * Free a regexp state
897 xmlRegFreeState(xmlRegStatePtr state) {
901 if (state->trans != NULL)
902 xmlFree(state->trans);
903 if (state->transTo != NULL)
904 xmlFree(state->transTo);
909 * xmlRegFreeParserCtxt:
910 * @ctxt: the regexp parser context
912 * Free a regexp parser context
915 xmlRegFreeParserCtxt(xmlRegParserCtxtPtr ctxt) {
920 if (ctxt->string != NULL)
921 xmlFree(ctxt->string);
922 if (ctxt->states != NULL) {
923 for (i = 0;i < ctxt->nbStates;i++)
924 xmlRegFreeState(ctxt->states[i]);
925 xmlFree(ctxt->states);
927 if (ctxt->atoms != NULL) {
928 for (i = 0;i < ctxt->nbAtoms;i++)
929 xmlRegFreeAtom(ctxt->atoms[i]);
930 xmlFree(ctxt->atoms);
932 if (ctxt->counters != NULL)
933 xmlFree(ctxt->counters);
937 /************************************************************************
939 * Display of Data structures *
941 ************************************************************************/
944 xmlRegPrintAtomType(FILE *output, xmlRegAtomType type) {
946 case XML_REGEXP_EPSILON:
947 fprintf(output, "epsilon "); break;
948 case XML_REGEXP_CHARVAL:
949 fprintf(output, "charval "); break;
950 case XML_REGEXP_RANGES:
951 fprintf(output, "ranges "); break;
952 case XML_REGEXP_SUBREG:
953 fprintf(output, "subexpr "); break;
954 case XML_REGEXP_STRING:
955 fprintf(output, "string "); break;
956 case XML_REGEXP_ANYCHAR:
957 fprintf(output, "anychar "); break;
958 case XML_REGEXP_ANYSPACE:
959 fprintf(output, "anyspace "); break;
960 case XML_REGEXP_NOTSPACE:
961 fprintf(output, "notspace "); break;
962 case XML_REGEXP_INITNAME:
963 fprintf(output, "initname "); break;
964 case XML_REGEXP_NOTINITNAME:
965 fprintf(output, "notinitname "); break;
966 case XML_REGEXP_NAMECHAR:
967 fprintf(output, "namechar "); break;
968 case XML_REGEXP_NOTNAMECHAR:
969 fprintf(output, "notnamechar "); break;
970 case XML_REGEXP_DECIMAL:
971 fprintf(output, "decimal "); break;
972 case XML_REGEXP_NOTDECIMAL:
973 fprintf(output, "notdecimal "); break;
974 case XML_REGEXP_REALCHAR:
975 fprintf(output, "realchar "); break;
976 case XML_REGEXP_NOTREALCHAR:
977 fprintf(output, "notrealchar "); break;
978 case XML_REGEXP_LETTER:
979 fprintf(output, "LETTER "); break;
980 case XML_REGEXP_LETTER_UPPERCASE:
981 fprintf(output, "LETTER_UPPERCASE "); break;
982 case XML_REGEXP_LETTER_LOWERCASE:
983 fprintf(output, "LETTER_LOWERCASE "); break;
984 case XML_REGEXP_LETTER_TITLECASE:
985 fprintf(output, "LETTER_TITLECASE "); break;
986 case XML_REGEXP_LETTER_MODIFIER:
987 fprintf(output, "LETTER_MODIFIER "); break;
988 case XML_REGEXP_LETTER_OTHERS:
989 fprintf(output, "LETTER_OTHERS "); break;
990 case XML_REGEXP_MARK:
991 fprintf(output, "MARK "); break;
992 case XML_REGEXP_MARK_NONSPACING:
993 fprintf(output, "MARK_NONSPACING "); break;
994 case XML_REGEXP_MARK_SPACECOMBINING:
995 fprintf(output, "MARK_SPACECOMBINING "); break;
996 case XML_REGEXP_MARK_ENCLOSING:
997 fprintf(output, "MARK_ENCLOSING "); break;
998 case XML_REGEXP_NUMBER:
999 fprintf(output, "NUMBER "); break;
1000 case XML_REGEXP_NUMBER_DECIMAL:
1001 fprintf(output, "NUMBER_DECIMAL "); break;
1002 case XML_REGEXP_NUMBER_LETTER:
1003 fprintf(output, "NUMBER_LETTER "); break;
1004 case XML_REGEXP_NUMBER_OTHERS:
1005 fprintf(output, "NUMBER_OTHERS "); break;
1006 case XML_REGEXP_PUNCT:
1007 fprintf(output, "PUNCT "); break;
1008 case XML_REGEXP_PUNCT_CONNECTOR:
1009 fprintf(output, "PUNCT_CONNECTOR "); break;
1010 case XML_REGEXP_PUNCT_DASH:
1011 fprintf(output, "PUNCT_DASH "); break;
1012 case XML_REGEXP_PUNCT_OPEN:
1013 fprintf(output, "PUNCT_OPEN "); break;
1014 case XML_REGEXP_PUNCT_CLOSE:
1015 fprintf(output, "PUNCT_CLOSE "); break;
1016 case XML_REGEXP_PUNCT_INITQUOTE:
1017 fprintf(output, "PUNCT_INITQUOTE "); break;
1018 case XML_REGEXP_PUNCT_FINQUOTE:
1019 fprintf(output, "PUNCT_FINQUOTE "); break;
1020 case XML_REGEXP_PUNCT_OTHERS:
1021 fprintf(output, "PUNCT_OTHERS "); break;
1022 case XML_REGEXP_SEPAR:
1023 fprintf(output, "SEPAR "); break;
1024 case XML_REGEXP_SEPAR_SPACE:
1025 fprintf(output, "SEPAR_SPACE "); break;
1026 case XML_REGEXP_SEPAR_LINE:
1027 fprintf(output, "SEPAR_LINE "); break;
1028 case XML_REGEXP_SEPAR_PARA:
1029 fprintf(output, "SEPAR_PARA "); break;
1030 case XML_REGEXP_SYMBOL:
1031 fprintf(output, "SYMBOL "); break;
1032 case XML_REGEXP_SYMBOL_MATH:
1033 fprintf(output, "SYMBOL_MATH "); break;
1034 case XML_REGEXP_SYMBOL_CURRENCY:
1035 fprintf(output, "SYMBOL_CURRENCY "); break;
1036 case XML_REGEXP_SYMBOL_MODIFIER:
1037 fprintf(output, "SYMBOL_MODIFIER "); break;
1038 case XML_REGEXP_SYMBOL_OTHERS:
1039 fprintf(output, "SYMBOL_OTHERS "); break;
1040 case XML_REGEXP_OTHER:
1041 fprintf(output, "OTHER "); break;
1042 case XML_REGEXP_OTHER_CONTROL:
1043 fprintf(output, "OTHER_CONTROL "); break;
1044 case XML_REGEXP_OTHER_FORMAT:
1045 fprintf(output, "OTHER_FORMAT "); break;
1046 case XML_REGEXP_OTHER_PRIVATE:
1047 fprintf(output, "OTHER_PRIVATE "); break;
1048 case XML_REGEXP_OTHER_NA:
1049 fprintf(output, "OTHER_NA "); break;
1050 case XML_REGEXP_BLOCK_NAME:
1051 fprintf(output, "BLOCK "); break;
1056 xmlRegPrintQuantType(FILE *output, xmlRegQuantType type) {
1058 case XML_REGEXP_QUANT_EPSILON:
1059 fprintf(output, "epsilon "); break;
1060 case XML_REGEXP_QUANT_ONCE:
1061 fprintf(output, "once "); break;
1062 case XML_REGEXP_QUANT_OPT:
1063 fprintf(output, "? "); break;
1064 case XML_REGEXP_QUANT_MULT:
1065 fprintf(output, "* "); break;
1066 case XML_REGEXP_QUANT_PLUS:
1067 fprintf(output, "+ "); break;
1068 case XML_REGEXP_QUANT_RANGE:
1069 fprintf(output, "range "); break;
1070 case XML_REGEXP_QUANT_ONCEONLY:
1071 fprintf(output, "onceonly "); break;
1072 case XML_REGEXP_QUANT_ALL:
1073 fprintf(output, "all "); break;
1077 xmlRegPrintRange(FILE *output, xmlRegRangePtr range) {
1078 fprintf(output, " range: ");
1080 fprintf(output, "negative ");
1081 xmlRegPrintAtomType(output, range->type);
1082 fprintf(output, "%c - %c\n", range->start, range->end);
1086 xmlRegPrintAtom(FILE *output, xmlRegAtomPtr atom) {
1087 fprintf(output, " atom: ");
1089 fprintf(output, "NULL\n");
1093 fprintf(output, "not ");
1094 xmlRegPrintAtomType(output, atom->type);
1095 xmlRegPrintQuantType(output, atom->quant);
1096 if (atom->quant == XML_REGEXP_QUANT_RANGE)
1097 fprintf(output, "%d-%d ", atom->min, atom->max);
1098 if (atom->type == XML_REGEXP_STRING)
1099 fprintf(output, "'%s' ", (char *) atom->valuep);
1100 if (atom->type == XML_REGEXP_CHARVAL)
1101 fprintf(output, "char %c\n", atom->codepoint);
1102 else if (atom->type == XML_REGEXP_RANGES) {
1104 fprintf(output, "%d entries\n", atom->nbRanges);
1105 for (i = 0; i < atom->nbRanges;i++)
1106 xmlRegPrintRange(output, atom->ranges[i]);
1107 } else if (atom->type == XML_REGEXP_SUBREG) {
1108 fprintf(output, "start %d end %d\n", atom->start->no, atom->stop->no);
1110 fprintf(output, "\n");
1115 xmlRegPrintTrans(FILE *output, xmlRegTransPtr trans) {
1116 fprintf(output, " trans: ");
1117 if (trans == NULL) {
1118 fprintf(output, "NULL\n");
1121 if (trans->to < 0) {
1122 fprintf(output, "removed\n");
1125 if (trans->nd != 0) {
1127 fprintf(output, "last not determinist, ");
1129 fprintf(output, "not determinist, ");
1131 if (trans->counter >= 0) {
1132 fprintf(output, "counted %d, ", trans->counter);
1134 if (trans->count == REGEXP_ALL_COUNTER) {
1135 fprintf(output, "all transition, ");
1136 } else if (trans->count >= 0) {
1137 fprintf(output, "count based %d, ", trans->count);
1139 if (trans->atom == NULL) {
1140 fprintf(output, "epsilon to %d\n", trans->to);
1143 if (trans->atom->type == XML_REGEXP_CHARVAL)
1144 fprintf(output, "char %c ", trans->atom->codepoint);
1145 fprintf(output, "atom %d, to %d\n", trans->atom->no, trans->to);
1149 xmlRegPrintState(FILE *output, xmlRegStatePtr state) {
1152 fprintf(output, " state: ");
1153 if (state == NULL) {
1154 fprintf(output, "NULL\n");
1157 if (state->type == XML_REGEXP_START_STATE)
1158 fprintf(output, "START ");
1159 if (state->type == XML_REGEXP_FINAL_STATE)
1160 fprintf(output, "FINAL ");
1162 fprintf(output, "%d, %d transitions:\n", state->no, state->nbTrans);
1163 for (i = 0;i < state->nbTrans; i++) {
1164 xmlRegPrintTrans(output, &(state->trans[i]));
1168 #ifdef DEBUG_REGEXP_GRAPH
1170 xmlRegPrintCtxt(FILE *output, xmlRegParserCtxtPtr ctxt) {
1173 fprintf(output, " ctxt: ");
1175 fprintf(output, "NULL\n");
1178 fprintf(output, "'%s' ", ctxt->string);
1180 fprintf(output, "error ");
1182 fprintf(output, "neg ");
1183 fprintf(output, "\n");
1184 fprintf(output, "%d atoms:\n", ctxt->nbAtoms);
1185 for (i = 0;i < ctxt->nbAtoms; i++) {
1186 fprintf(output, " %02d ", i);
1187 xmlRegPrintAtom(output, ctxt->atoms[i]);
1189 if (ctxt->atom != NULL) {
1190 fprintf(output, "current atom:\n");
1191 xmlRegPrintAtom(output, ctxt->atom);
1193 fprintf(output, "%d states:", ctxt->nbStates);
1194 if (ctxt->start != NULL)
1195 fprintf(output, " start: %d", ctxt->start->no);
1196 if (ctxt->end != NULL)
1197 fprintf(output, " end: %d", ctxt->end->no);
1198 fprintf(output, "\n");
1199 for (i = 0;i < ctxt->nbStates; i++) {
1200 xmlRegPrintState(output, ctxt->states[i]);
1202 fprintf(output, "%d counters:\n", ctxt->nbCounters);
1203 for (i = 0;i < ctxt->nbCounters; i++) {
1204 fprintf(output, " %d: min %d max %d\n", i, ctxt->counters[i].min,
1205 ctxt->counters[i].max);
1210 /************************************************************************
1212 * Finite Automata structures manipulations *
1214 ************************************************************************/
1217 xmlRegAtomAddRange(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom,
1218 int neg, xmlRegAtomType type, int start, int end,
1219 xmlChar *blockName) {
1220 xmlRegRangePtr range;
1223 ERROR("add range: atom is NULL");
1226 if (atom->type != XML_REGEXP_RANGES) {
1227 ERROR("add range: atom is not ranges");
1230 if (atom->maxRanges == 0) {
1231 atom->maxRanges = 4;
1232 atom->ranges = (xmlRegRangePtr *) xmlMalloc(atom->maxRanges *
1233 sizeof(xmlRegRangePtr));
1234 if (atom->ranges == NULL) {
1235 xmlRegexpErrMemory(ctxt, "adding ranges");
1236 atom->maxRanges = 0;
1239 } else if (atom->nbRanges >= atom->maxRanges) {
1240 xmlRegRangePtr *tmp;
1241 atom->maxRanges *= 2;
1242 tmp = (xmlRegRangePtr *) xmlRealloc(atom->ranges, atom->maxRanges *
1243 sizeof(xmlRegRangePtr));
1245 xmlRegexpErrMemory(ctxt, "adding ranges");
1246 atom->maxRanges /= 2;
1251 range = xmlRegNewRange(ctxt, neg, type, start, end);
1254 range->blockName = blockName;
1255 atom->ranges[atom->nbRanges++] = range;
1260 xmlRegGetCounter(xmlRegParserCtxtPtr ctxt) {
1261 if (ctxt->maxCounters == 0) {
1262 ctxt->maxCounters = 4;
1263 ctxt->counters = (xmlRegCounter *) xmlMalloc(ctxt->maxCounters *
1264 sizeof(xmlRegCounter));
1265 if (ctxt->counters == NULL) {
1266 xmlRegexpErrMemory(ctxt, "allocating counter");
1267 ctxt->maxCounters = 0;
1270 } else if (ctxt->nbCounters >= ctxt->maxCounters) {
1272 ctxt->maxCounters *= 2;
1273 tmp = (xmlRegCounter *) xmlRealloc(ctxt->counters, ctxt->maxCounters *
1274 sizeof(xmlRegCounter));
1276 xmlRegexpErrMemory(ctxt, "allocating counter");
1277 ctxt->maxCounters /= 2;
1280 ctxt->counters = tmp;
1282 ctxt->counters[ctxt->nbCounters].min = -1;
1283 ctxt->counters[ctxt->nbCounters].max = -1;
1284 return(ctxt->nbCounters++);
1288 xmlRegAtomPush(xmlRegParserCtxtPtr ctxt, xmlRegAtomPtr atom) {
1290 ERROR("atom push: atom is NULL");
1293 if (ctxt->maxAtoms == 0) {
1295 ctxt->atoms = (xmlRegAtomPtr *) xmlMalloc(ctxt->maxAtoms *
1296 sizeof(xmlRegAtomPtr));
1297 if (ctxt->atoms == NULL) {
1298 xmlRegexpErrMemory(ctxt, "pushing atom");
1302 } else if (ctxt->nbAtoms >= ctxt->maxAtoms) {
1304 ctxt->maxAtoms *= 2;
1305 tmp = (xmlRegAtomPtr *) xmlRealloc(ctxt->atoms, ctxt->maxAtoms *
1306 sizeof(xmlRegAtomPtr));
1308 xmlRegexpErrMemory(ctxt, "allocating counter");
1309 ctxt->maxAtoms /= 2;
1314 atom->no = ctxt->nbAtoms;
1315 ctxt->atoms[ctxt->nbAtoms++] = atom;
1320 xmlRegStateAddTransTo(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr target,
1322 if (target->maxTransTo == 0) {
1323 target->maxTransTo = 8;
1324 target->transTo = (int *) xmlMalloc(target->maxTransTo *
1326 if (target->transTo == NULL) {
1327 xmlRegexpErrMemory(ctxt, "adding transition");
1328 target->maxTransTo = 0;
1331 } else if (target->nbTransTo >= target->maxTransTo) {
1333 target->maxTransTo *= 2;
1334 tmp = (int *) xmlRealloc(target->transTo, target->maxTransTo *
1337 xmlRegexpErrMemory(ctxt, "adding transition");
1338 target->maxTransTo /= 2;
1341 target->transTo = tmp;
1343 target->transTo[target->nbTransTo] = from;
1344 target->nbTransTo++;
1348 xmlRegStateAddTrans(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
1349 xmlRegAtomPtr atom, xmlRegStatePtr target,
1350 int counter, int count) {
1354 if (state == NULL) {
1355 ERROR("add state: state is NULL");
1358 if (target == NULL) {
1359 ERROR("add state: target is NULL");
1363 * Other routines follow the philosophy 'When in doubt, add a transition'
1364 * so we check here whether such a transition is already present and, if
1365 * so, silently ignore this request.
1368 for (nrtrans = state->nbTrans - 1; nrtrans >= 0; nrtrans--) {
1369 xmlRegTransPtr trans = &(state->trans[nrtrans]);
1370 if ((trans->atom == atom) &&
1371 (trans->to == target->no) &&
1372 (trans->counter == counter) &&
1373 (trans->count == count)) {
1374 #ifdef DEBUG_REGEXP_GRAPH
1375 printf("Ignoring duplicate transition from %d to %d\n",
1376 state->no, target->no);
1382 if (state->maxTrans == 0) {
1383 state->maxTrans = 8;
1384 state->trans = (xmlRegTrans *) xmlMalloc(state->maxTrans *
1385 sizeof(xmlRegTrans));
1386 if (state->trans == NULL) {
1387 xmlRegexpErrMemory(ctxt, "adding transition");
1388 state->maxTrans = 0;
1391 } else if (state->nbTrans >= state->maxTrans) {
1393 state->maxTrans *= 2;
1394 tmp = (xmlRegTrans *) xmlRealloc(state->trans, state->maxTrans *
1395 sizeof(xmlRegTrans));
1397 xmlRegexpErrMemory(ctxt, "adding transition");
1398 state->maxTrans /= 2;
1403 #ifdef DEBUG_REGEXP_GRAPH
1404 printf("Add trans from %d to %d ", state->no, target->no);
1405 if (count == REGEXP_ALL_COUNTER)
1406 printf("all transition\n");
1407 else if (count >= 0)
1408 printf("count based %d\n", count);
1409 else if (counter >= 0)
1410 printf("counted %d\n", counter);
1411 else if (atom == NULL)
1412 printf("epsilon transition\n");
1413 else if (atom != NULL)
1414 xmlRegPrintAtom(stdout, atom);
1417 state->trans[state->nbTrans].atom = atom;
1418 state->trans[state->nbTrans].to = target->no;
1419 state->trans[state->nbTrans].counter = counter;
1420 state->trans[state->nbTrans].count = count;
1421 state->trans[state->nbTrans].nd = 0;
1423 xmlRegStateAddTransTo(ctxt, target, state->no);
1427 xmlRegStatePush(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state) {
1428 if (state == NULL) return(-1);
1429 if (ctxt->maxStates == 0) {
1430 ctxt->maxStates = 4;
1431 ctxt->states = (xmlRegStatePtr *) xmlMalloc(ctxt->maxStates *
1432 sizeof(xmlRegStatePtr));
1433 if (ctxt->states == NULL) {
1434 xmlRegexpErrMemory(ctxt, "adding state");
1435 ctxt->maxStates = 0;
1438 } else if (ctxt->nbStates >= ctxt->maxStates) {
1439 xmlRegStatePtr *tmp;
1440 ctxt->maxStates *= 2;
1441 tmp = (xmlRegStatePtr *) xmlRealloc(ctxt->states, ctxt->maxStates *
1442 sizeof(xmlRegStatePtr));
1444 xmlRegexpErrMemory(ctxt, "adding state");
1445 ctxt->maxStates /= 2;
1450 state->no = ctxt->nbStates;
1451 ctxt->states[ctxt->nbStates++] = state;
1456 * xmlFAGenerateAllTransition:
1457 * @ctxt: a regexp parser context
1458 * @from: the from state
1459 * @to: the target state or NULL for building a new one
1464 xmlFAGenerateAllTransition(xmlRegParserCtxtPtr ctxt,
1465 xmlRegStatePtr from, xmlRegStatePtr to,
1468 to = xmlRegNewState(ctxt);
1469 xmlRegStatePush(ctxt, to);
1473 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_LAX_COUNTER);
1475 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, REGEXP_ALL_COUNTER);
1479 * xmlFAGenerateEpsilonTransition:
1480 * @ctxt: a regexp parser context
1481 * @from: the from state
1482 * @to: the target state or NULL for building a new one
1486 xmlFAGenerateEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1487 xmlRegStatePtr from, xmlRegStatePtr to) {
1489 to = xmlRegNewState(ctxt);
1490 xmlRegStatePush(ctxt, to);
1493 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, -1);
1497 * xmlFAGenerateCountedEpsilonTransition:
1498 * @ctxt: a regexp parser context
1499 * @from: the from state
1500 * @to: the target state or NULL for building a new one
1501 * counter: the counter for that transition
1505 xmlFAGenerateCountedEpsilonTransition(xmlRegParserCtxtPtr ctxt,
1506 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1508 to = xmlRegNewState(ctxt);
1509 xmlRegStatePush(ctxt, to);
1512 xmlRegStateAddTrans(ctxt, from, NULL, to, counter, -1);
1516 * xmlFAGenerateCountedTransition:
1517 * @ctxt: a regexp parser context
1518 * @from: the from state
1519 * @to: the target state or NULL for building a new one
1520 * counter: the counter for that transition
1524 xmlFAGenerateCountedTransition(xmlRegParserCtxtPtr ctxt,
1525 xmlRegStatePtr from, xmlRegStatePtr to, int counter) {
1527 to = xmlRegNewState(ctxt);
1528 xmlRegStatePush(ctxt, to);
1531 xmlRegStateAddTrans(ctxt, from, NULL, to, -1, counter);
1535 * xmlFAGenerateTransitions:
1536 * @ctxt: a regexp parser context
1537 * @from: the from state
1538 * @to: the target state or NULL for building a new one
1539 * @atom: the atom generating the transition
1541 * Returns 0 if success and -1 in case of error.
1544 xmlFAGenerateTransitions(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr from,
1545 xmlRegStatePtr to, xmlRegAtomPtr atom) {
1549 ERROR("genrate transition: atom == NULL");
1552 if (atom->type == XML_REGEXP_SUBREG) {
1554 * this is a subexpression handling one should not need to
1555 * create a new node except for XML_REGEXP_QUANT_RANGE.
1557 if (xmlRegAtomPush(ctxt, atom) < 0) {
1560 if ((to != NULL) && (atom->stop != to) &&
1561 (atom->quant != XML_REGEXP_QUANT_RANGE)) {
1563 * Generate an epsilon transition to link to the target
1565 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, to);
1567 } else if ((to == NULL) && (atom->quant != XML_REGEXP_QUANT_RANGE) &&
1568 (atom->quant != XML_REGEXP_QUANT_ONCE)) {
1569 to = xmlRegNewState(ctxt);
1570 xmlRegStatePush(ctxt, to);
1572 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, to);
1575 switch (atom->quant) {
1576 case XML_REGEXP_QUANT_OPT:
1577 atom->quant = XML_REGEXP_QUANT_ONCE;
1579 * transition done to the state after end of atom.
1580 * 1. set transition from atom start to new state
1581 * 2. set transition from atom end to this state.
1584 xmlFAGenerateEpsilonTransition(ctxt, atom->start, 0);
1585 xmlFAGenerateEpsilonTransition(ctxt, atom->stop,
1588 xmlFAGenerateEpsilonTransition(ctxt, atom->start, to);
1591 case XML_REGEXP_QUANT_MULT:
1592 atom->quant = XML_REGEXP_QUANT_ONCE;
1593 xmlFAGenerateEpsilonTransition(ctxt, atom->start, atom->stop);
1594 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1596 case XML_REGEXP_QUANT_PLUS:
1597 atom->quant = XML_REGEXP_QUANT_ONCE;
1598 xmlFAGenerateEpsilonTransition(ctxt, atom->stop, atom->start);
1600 case XML_REGEXP_QUANT_RANGE: {
1602 xmlRegStatePtr inter, newstate;
1605 * create the final state now if needed
1610 newstate = xmlRegNewState(ctxt);
1611 xmlRegStatePush(ctxt, newstate);
1615 * The principle here is to use counted transition
1616 * to avoid explosion in the number of states in the
1617 * graph. This is clearly more complex but should not
1618 * be exploitable at runtime.
1620 if ((atom->min == 0) && (atom->start0 == NULL)) {
1623 * duplicate a transition based on atom to count next
1624 * occurences after 1. We cannot loop to atom->start
1625 * directly because we need an epsilon transition to
1628 /* ???? For some reason it seems we never reach that
1629 case, I suppose this got optimized out before when
1630 building the automata */
1631 copy = xmlRegCopyAtom(ctxt, atom);
1634 copy->quant = XML_REGEXP_QUANT_ONCE;
1638 if (xmlFAGenerateTransitions(ctxt, atom->start, NULL, copy)
1641 inter = ctxt->state;
1642 counter = xmlRegGetCounter(ctxt);
1643 ctxt->counters[counter].min = atom->min - 1;
1644 ctxt->counters[counter].max = atom->max - 1;
1645 /* count the number of times we see it again */
1646 xmlFAGenerateCountedEpsilonTransition(ctxt, inter,
1647 atom->stop, counter);
1648 /* allow a way out based on the count */
1649 xmlFAGenerateCountedTransition(ctxt, inter,
1651 /* and also allow a direct exit for 0 */
1652 xmlFAGenerateEpsilonTransition(ctxt, atom->start,
1656 * either we need the atom at least once or there
1657 * is an atom->start0 allowing to easilly plug the
1658 * epsilon transition.
1660 counter = xmlRegGetCounter(ctxt);
1661 ctxt->counters[counter].min = atom->min - 1;
1662 ctxt->counters[counter].max = atom->max - 1;
1663 /* count the number of times we see it again */
1664 xmlFAGenerateCountedEpsilonTransition(ctxt, atom->stop,
1665 atom->start, counter);
1666 /* allow a way out based on the count */
1667 xmlFAGenerateCountedTransition(ctxt, atom->stop,
1669 /* and if needed allow a direct exit for 0 */
1671 xmlFAGenerateEpsilonTransition(ctxt, atom->start0,
1677 atom->quant = XML_REGEXP_QUANT_ONCE;
1678 ctxt->state = newstate;
1685 if ((atom->min == 0) && (atom->max == 0) &&
1686 (atom->quant == XML_REGEXP_QUANT_RANGE)) {
1688 * we can discard the atom and generate an epsilon transition instead
1691 to = xmlRegNewState(ctxt);
1693 xmlRegStatePush(ctxt, to);
1698 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1700 xmlRegFreeAtom(atom);
1704 to = xmlRegNewState(ctxt);
1706 xmlRegStatePush(ctxt, to);
1712 if ((atom->quant == XML_REGEXP_QUANT_MULT) ||
1713 (atom->quant == XML_REGEXP_QUANT_PLUS)) {
1715 * Do not pollute the target state by adding transitions from
1716 * it as it is likely to be the shared target of multiple branches.
1717 * So isolate with an epsilon transition.
1721 tmp = xmlRegNewState(ctxt);
1723 xmlRegStatePush(ctxt, tmp);
1727 xmlFAGenerateEpsilonTransition(ctxt, tmp, to);
1730 if (xmlRegAtomPush(ctxt, atom) < 0) {
1733 xmlRegStateAddTrans(ctxt, from, atom, to, -1, -1);
1735 switch (atom->quant) {
1736 case XML_REGEXP_QUANT_OPT:
1737 atom->quant = XML_REGEXP_QUANT_ONCE;
1738 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1740 case XML_REGEXP_QUANT_MULT:
1741 atom->quant = XML_REGEXP_QUANT_ONCE;
1742 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1743 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1745 case XML_REGEXP_QUANT_PLUS:
1746 atom->quant = XML_REGEXP_QUANT_ONCE;
1747 xmlRegStateAddTrans(ctxt, to, atom, to, -1, -1);
1749 case XML_REGEXP_QUANT_RANGE:
1751 if (atom->min == 0) {
1752 xmlFAGenerateEpsilonTransition(ctxt, from, to);
1763 * xmlFAReduceEpsilonTransitions:
1764 * @ctxt: a regexp parser context
1765 * @fromnr: the from state
1766 * @tonr: the to state
1767 * @counter: should that transition be associated to a counted
1771 xmlFAReduceEpsilonTransitions(xmlRegParserCtxtPtr ctxt, int fromnr,
1772 int tonr, int counter) {
1774 xmlRegStatePtr from;
1777 #ifdef DEBUG_REGEXP_GRAPH
1778 printf("xmlFAReduceEpsilonTransitions(%d, %d)\n", fromnr, tonr);
1780 from = ctxt->states[fromnr];
1783 to = ctxt->states[tonr];
1786 if ((to->mark == XML_REGEXP_MARK_START) ||
1787 (to->mark == XML_REGEXP_MARK_VISITED))
1790 to->mark = XML_REGEXP_MARK_VISITED;
1791 if (to->type == XML_REGEXP_FINAL_STATE) {
1792 #ifdef DEBUG_REGEXP_GRAPH
1793 printf("State %d is final, so %d becomes final\n", tonr, fromnr);
1795 from->type = XML_REGEXP_FINAL_STATE;
1797 for (transnr = 0;transnr < to->nbTrans;transnr++) {
1798 if (to->trans[transnr].to < 0)
1800 if (to->trans[transnr].atom == NULL) {
1802 * Don't remove counted transitions
1805 if (to->trans[transnr].to != fromnr) {
1806 if (to->trans[transnr].count >= 0) {
1807 int newto = to->trans[transnr].to;
1809 xmlRegStateAddTrans(ctxt, from, NULL,
1810 ctxt->states[newto],
1811 -1, to->trans[transnr].count);
1813 #ifdef DEBUG_REGEXP_GRAPH
1814 printf("Found epsilon trans %d from %d to %d\n",
1815 transnr, tonr, to->trans[transnr].to);
1817 if (to->trans[transnr].counter >= 0) {
1818 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1819 to->trans[transnr].to,
1820 to->trans[transnr].counter);
1822 xmlFAReduceEpsilonTransitions(ctxt, fromnr,
1823 to->trans[transnr].to,
1829 int newto = to->trans[transnr].to;
1831 if (to->trans[transnr].counter >= 0) {
1832 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1833 ctxt->states[newto],
1834 to->trans[transnr].counter, -1);
1836 xmlRegStateAddTrans(ctxt, from, to->trans[transnr].atom,
1837 ctxt->states[newto], counter, -1);
1841 to->mark = XML_REGEXP_MARK_NORMAL;
1845 * xmlFAEliminateSimpleEpsilonTransitions:
1846 * @ctxt: a regexp parser context
1848 * Eliminating general epsilon transitions can get costly in the general
1849 * algorithm due to the large amount of generated new transitions and
1850 * associated comparisons. However for simple epsilon transition used just
1851 * to separate building blocks when generating the automata this can be
1852 * reduced to state elimination:
1853 * - if there exists an epsilon from X to Y
1854 * - if there is no other transition from X
1855 * then X and Y are semantically equivalent and X can be eliminated
1856 * If X is the start state then make Y the start state, else replace the
1857 * target of all transitions to X by transitions to Y.
1860 xmlFAEliminateSimpleEpsilonTransitions(xmlRegParserCtxtPtr ctxt) {
1861 int statenr, i, j, newto;
1862 xmlRegStatePtr state, tmp;
1864 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1865 state = ctxt->states[statenr];
1868 if (state->nbTrans != 1)
1870 if (state->type == XML_REGEXP_UNREACH_STATE)
1872 /* is the only transition out a basic transition */
1873 if ((state->trans[0].atom == NULL) &&
1874 (state->trans[0].to >= 0) &&
1875 (state->trans[0].to != statenr) &&
1876 (state->trans[0].counter < 0) &&
1877 (state->trans[0].count < 0)) {
1878 newto = state->trans[0].to;
1880 if (state->type == XML_REGEXP_START_STATE) {
1881 #ifdef DEBUG_REGEXP_GRAPH
1882 printf("Found simple epsilon trans from start %d to %d\n",
1886 #ifdef DEBUG_REGEXP_GRAPH
1887 printf("Found simple epsilon trans from %d to %d\n",
1890 for (i = 0;i < state->nbTransTo;i++) {
1891 tmp = ctxt->states[state->transTo[i]];
1892 for (j = 0;j < tmp->nbTrans;j++) {
1893 if (tmp->trans[j].to == statenr) {
1894 #ifdef DEBUG_REGEXP_GRAPH
1895 printf("Changed transition %d on %d to go to %d\n",
1898 tmp->trans[j].to = -1;
1899 xmlRegStateAddTrans(ctxt, tmp, tmp->trans[j].atom,
1900 ctxt->states[newto],
1901 tmp->trans[j].counter,
1902 tmp->trans[j].count);
1906 if (state->type == XML_REGEXP_FINAL_STATE)
1907 ctxt->states[newto]->type = XML_REGEXP_FINAL_STATE;
1908 /* eliminate the transition completely */
1911 state->type = XML_REGEXP_UNREACH_STATE;
1919 * xmlFAEliminateEpsilonTransitions:
1920 * @ctxt: a regexp parser context
1924 xmlFAEliminateEpsilonTransitions(xmlRegParserCtxtPtr ctxt) {
1925 int statenr, transnr;
1926 xmlRegStatePtr state;
1929 if (ctxt->states == NULL) return;
1932 * Eliminate simple epsilon transition and the associated unreachable
1935 xmlFAEliminateSimpleEpsilonTransitions(ctxt);
1936 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
1937 state = ctxt->states[statenr];
1938 if ((state != NULL) && (state->type == XML_REGEXP_UNREACH_STATE)) {
1939 #ifdef DEBUG_REGEXP_GRAPH
1940 printf("Removed unreachable state %d\n", statenr);
1942 xmlRegFreeState(state);
1943 ctxt->states[statenr] = NULL;
1950 * Build the completed transitions bypassing the epsilons
1951 * Use a marking algorithm to avoid loops
1952 * Mark sink states too.
1953 * Process from the latests states backward to the start when
1954 * there is long cascading epsilon chains this minimize the
1955 * recursions and transition compares when adding the new ones
1957 for (statenr = ctxt->nbStates - 1;statenr >= 0;statenr--) {
1958 state = ctxt->states[statenr];
1961 if ((state->nbTrans == 0) &&
1962 (state->type != XML_REGEXP_FINAL_STATE)) {
1963 state->type = XML_REGEXP_SINK_STATE;
1965 for (transnr = 0;transnr < state->nbTrans;transnr++) {
1966 if ((state->trans[transnr].atom == NULL) &&
1967 (state->trans[transnr].to >= 0)) {
1968 if (state->trans[transnr].to == statenr) {
1969 state->trans[transnr].to = -1;
1970 #ifdef DEBUG_REGEXP_GRAPH
1971 printf("Removed loopback epsilon trans %d on %d\n",
1974 } else if (state->trans[transnr].count < 0) {
1975 int newto = state->trans[transnr].to;
1977 #ifdef DEBUG_REGEXP_GRAPH
1978 printf("Found epsilon trans %d from %d to %d\n",
1979 transnr, statenr, newto);
1982 state->trans[transnr].to = -2;
1983 state->mark = XML_REGEXP_MARK_START;
1984 xmlFAReduceEpsilonTransitions(ctxt, statenr,
1985 newto, state->trans[transnr].counter);
1986 state->mark = XML_REGEXP_MARK_NORMAL;
1987 #ifdef DEBUG_REGEXP_GRAPH
1989 printf("Found counted transition %d on %d\n",
1997 * Eliminate the epsilon transitions
2000 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2001 state = ctxt->states[statenr];
2004 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2005 xmlRegTransPtr trans = &(state->trans[transnr]);
2006 if ((trans->atom == NULL) &&
2007 (trans->count < 0) &&
2016 * Use this pass to detect unreachable states too
2018 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2019 state = ctxt->states[statenr];
2021 state->reached = XML_REGEXP_MARK_NORMAL;
2023 state = ctxt->states[0];
2025 state->reached = XML_REGEXP_MARK_START;
2026 while (state != NULL) {
2027 xmlRegStatePtr target = NULL;
2028 state->reached = XML_REGEXP_MARK_VISITED;
2030 * Mark all states reachable from the current reachable state
2032 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2033 if ((state->trans[transnr].to >= 0) &&
2034 ((state->trans[transnr].atom != NULL) ||
2035 (state->trans[transnr].count >= 0))) {
2036 int newto = state->trans[transnr].to;
2038 if (ctxt->states[newto] == NULL)
2040 if (ctxt->states[newto]->reached == XML_REGEXP_MARK_NORMAL) {
2041 ctxt->states[newto]->reached = XML_REGEXP_MARK_START;
2042 target = ctxt->states[newto];
2048 * find the next accessible state not explored
2050 if (target == NULL) {
2051 for (statenr = 1;statenr < ctxt->nbStates;statenr++) {
2052 state = ctxt->states[statenr];
2053 if ((state != NULL) && (state->reached ==
2054 XML_REGEXP_MARK_START)) {
2062 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2063 state = ctxt->states[statenr];
2064 if ((state != NULL) && (state->reached == XML_REGEXP_MARK_NORMAL)) {
2065 #ifdef DEBUG_REGEXP_GRAPH
2066 printf("Removed unreachable state %d\n", statenr);
2068 xmlRegFreeState(state);
2069 ctxt->states[statenr] = NULL;
2076 xmlFACompareRanges(xmlRegRangePtr range1, xmlRegRangePtr range2) {
2079 if ((range1->type == XML_REGEXP_RANGES) ||
2080 (range2->type == XML_REGEXP_RANGES) ||
2081 (range2->type == XML_REGEXP_SUBREG) ||
2082 (range1->type == XML_REGEXP_SUBREG) ||
2083 (range1->type == XML_REGEXP_STRING) ||
2084 (range2->type == XML_REGEXP_STRING))
2087 /* put them in order */
2088 if (range1->type > range2->type) {
2095 if ((range1->type == XML_REGEXP_ANYCHAR) ||
2096 (range2->type == XML_REGEXP_ANYCHAR)) {
2098 } else if ((range1->type == XML_REGEXP_EPSILON) ||
2099 (range2->type == XML_REGEXP_EPSILON)) {
2101 } else if (range1->type == range2->type) {
2102 if (range1->type != XML_REGEXP_CHARVAL)
2104 else if ((range1->end < range2->start) ||
2105 (range2->end < range1->start))
2109 } else if (range1->type == XML_REGEXP_CHARVAL) {
2114 * just check all codepoints in the range for acceptance,
2115 * this is usually way cheaper since done only once at
2116 * compilation than testing over and over at runtime or
2117 * pushing too many states when evaluating.
2119 if (((range1->neg == 0) && (range2->neg != 0)) ||
2120 ((range1->neg != 0) && (range2->neg == 0)))
2123 for (codepoint = range1->start;codepoint <= range1->end ;codepoint++) {
2124 ret = xmlRegCheckCharacterRange(range2->type, codepoint,
2125 0, range2->start, range2->end,
2129 if (((neg == 1) && (ret == 0)) ||
2130 ((neg == 0) && (ret == 1)))
2134 } else if ((range1->type == XML_REGEXP_BLOCK_NAME) ||
2135 (range2->type == XML_REGEXP_BLOCK_NAME)) {
2136 if (range1->type == range2->type) {
2137 ret = xmlStrEqual(range1->blockName, range2->blockName);
2140 * comparing a block range with anything else is way
2141 * too costly, and maintining the table is like too much
2142 * memory too, so let's force the automata to save state
2147 } else if ((range1->type < XML_REGEXP_LETTER) ||
2148 (range2->type < XML_REGEXP_LETTER)) {
2149 if ((range1->type == XML_REGEXP_ANYSPACE) &&
2150 (range2->type == XML_REGEXP_NOTSPACE))
2152 else if ((range1->type == XML_REGEXP_INITNAME) &&
2153 (range2->type == XML_REGEXP_NOTINITNAME))
2155 else if ((range1->type == XML_REGEXP_NAMECHAR) &&
2156 (range2->type == XML_REGEXP_NOTNAMECHAR))
2158 else if ((range1->type == XML_REGEXP_DECIMAL) &&
2159 (range2->type == XML_REGEXP_NOTDECIMAL))
2161 else if ((range1->type == XML_REGEXP_REALCHAR) &&
2162 (range2->type == XML_REGEXP_NOTREALCHAR))
2165 /* same thing to limit complexity */
2170 /* range1->type < range2->type here */
2171 switch (range1->type) {
2172 case XML_REGEXP_LETTER:
2173 /* all disjoint except in the subgroups */
2174 if ((range2->type == XML_REGEXP_LETTER_UPPERCASE) ||
2175 (range2->type == XML_REGEXP_LETTER_LOWERCASE) ||
2176 (range2->type == XML_REGEXP_LETTER_TITLECASE) ||
2177 (range2->type == XML_REGEXP_LETTER_MODIFIER) ||
2178 (range2->type == XML_REGEXP_LETTER_OTHERS))
2181 case XML_REGEXP_MARK:
2182 if ((range2->type == XML_REGEXP_MARK_NONSPACING) ||
2183 (range2->type == XML_REGEXP_MARK_SPACECOMBINING) ||
2184 (range2->type == XML_REGEXP_MARK_ENCLOSING))
2187 case XML_REGEXP_NUMBER:
2188 if ((range2->type == XML_REGEXP_NUMBER_DECIMAL) ||
2189 (range2->type == XML_REGEXP_NUMBER_LETTER) ||
2190 (range2->type == XML_REGEXP_NUMBER_OTHERS))
2193 case XML_REGEXP_PUNCT:
2194 if ((range2->type == XML_REGEXP_PUNCT_CONNECTOR) ||
2195 (range2->type == XML_REGEXP_PUNCT_DASH) ||
2196 (range2->type == XML_REGEXP_PUNCT_OPEN) ||
2197 (range2->type == XML_REGEXP_PUNCT_CLOSE) ||
2198 (range2->type == XML_REGEXP_PUNCT_INITQUOTE) ||
2199 (range2->type == XML_REGEXP_PUNCT_FINQUOTE) ||
2200 (range2->type == XML_REGEXP_PUNCT_OTHERS))
2203 case XML_REGEXP_SEPAR:
2204 if ((range2->type == XML_REGEXP_SEPAR_SPACE) ||
2205 (range2->type == XML_REGEXP_SEPAR_LINE) ||
2206 (range2->type == XML_REGEXP_SEPAR_PARA))
2209 case XML_REGEXP_SYMBOL:
2210 if ((range2->type == XML_REGEXP_SYMBOL_MATH) ||
2211 (range2->type == XML_REGEXP_SYMBOL_CURRENCY) ||
2212 (range2->type == XML_REGEXP_SYMBOL_MODIFIER) ||
2213 (range2->type == XML_REGEXP_SYMBOL_OTHERS))
2216 case XML_REGEXP_OTHER:
2217 if ((range2->type == XML_REGEXP_OTHER_CONTROL) ||
2218 (range2->type == XML_REGEXP_OTHER_FORMAT) ||
2219 (range2->type == XML_REGEXP_OTHER_PRIVATE))
2223 if ((range2->type >= XML_REGEXP_LETTER) &&
2224 (range2->type < XML_REGEXP_BLOCK_NAME))
2232 if (((range1->neg == 0) && (range2->neg != 0)) ||
2233 ((range1->neg != 0) && (range2->neg == 0)))
2239 * xmlFACompareAtomTypes:
2240 * @type1: an atom type
2241 * @type2: an atom type
2243 * Compares two atoms type to check whether they intersect in some ways,
2244 * this is used by xmlFACompareAtoms only
2246 * Returns 1 if they may intersect and 0 otherwise
2249 xmlFACompareAtomTypes(xmlRegAtomType type1, xmlRegAtomType type2) {
2250 if ((type1 == XML_REGEXP_EPSILON) ||
2251 (type1 == XML_REGEXP_CHARVAL) ||
2252 (type1 == XML_REGEXP_RANGES) ||
2253 (type1 == XML_REGEXP_SUBREG) ||
2254 (type1 == XML_REGEXP_STRING) ||
2255 (type1 == XML_REGEXP_ANYCHAR))
2257 if ((type2 == XML_REGEXP_EPSILON) ||
2258 (type2 == XML_REGEXP_CHARVAL) ||
2259 (type2 == XML_REGEXP_RANGES) ||
2260 (type2 == XML_REGEXP_SUBREG) ||
2261 (type2 == XML_REGEXP_STRING) ||
2262 (type2 == XML_REGEXP_ANYCHAR))
2265 if (type1 == type2) return(1);
2267 /* simplify subsequent compares by making sure type1 < type2 */
2268 if (type1 > type2) {
2269 xmlRegAtomType tmp = type1;
2274 case XML_REGEXP_ANYSPACE: /* \s */
2275 /* can't be a letter, number, mark, pontuation, symbol */
2276 if ((type2 == XML_REGEXP_NOTSPACE) ||
2277 ((type2 >= XML_REGEXP_LETTER) &&
2278 (type2 <= XML_REGEXP_LETTER_OTHERS)) ||
2279 ((type2 >= XML_REGEXP_NUMBER) &&
2280 (type2 <= XML_REGEXP_NUMBER_OTHERS)) ||
2281 ((type2 >= XML_REGEXP_MARK) &&
2282 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2283 ((type2 >= XML_REGEXP_PUNCT) &&
2284 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2285 ((type2 >= XML_REGEXP_SYMBOL) &&
2286 (type2 <= XML_REGEXP_SYMBOL_OTHERS))
2289 case XML_REGEXP_NOTSPACE: /* \S */
2291 case XML_REGEXP_INITNAME: /* \l */
2292 /* can't be a number, mark, separator, pontuation, symbol or other */
2293 if ((type2 == XML_REGEXP_NOTINITNAME) ||
2294 ((type2 >= XML_REGEXP_NUMBER) &&
2295 (type2 <= XML_REGEXP_NUMBER_OTHERS)) ||
2296 ((type2 >= XML_REGEXP_MARK) &&
2297 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2298 ((type2 >= XML_REGEXP_SEPAR) &&
2299 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2300 ((type2 >= XML_REGEXP_PUNCT) &&
2301 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2302 ((type2 >= XML_REGEXP_SYMBOL) &&
2303 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2304 ((type2 >= XML_REGEXP_OTHER) &&
2305 (type2 <= XML_REGEXP_OTHER_NA))
2308 case XML_REGEXP_NOTINITNAME: /* \L */
2310 case XML_REGEXP_NAMECHAR: /* \c */
2311 /* can't be a mark, separator, pontuation, symbol or other */
2312 if ((type2 == XML_REGEXP_NOTNAMECHAR) ||
2313 ((type2 >= XML_REGEXP_MARK) &&
2314 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2315 ((type2 >= XML_REGEXP_PUNCT) &&
2316 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2317 ((type2 >= XML_REGEXP_SEPAR) &&
2318 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2319 ((type2 >= XML_REGEXP_SYMBOL) &&
2320 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2321 ((type2 >= XML_REGEXP_OTHER) &&
2322 (type2 <= XML_REGEXP_OTHER_NA))
2325 case XML_REGEXP_NOTNAMECHAR: /* \C */
2327 case XML_REGEXP_DECIMAL: /* \d */
2328 /* can't be a letter, mark, separator, pontuation, symbol or other */
2329 if ((type2 == XML_REGEXP_NOTDECIMAL) ||
2330 (type2 == XML_REGEXP_REALCHAR) ||
2331 ((type2 >= XML_REGEXP_LETTER) &&
2332 (type2 <= XML_REGEXP_LETTER_OTHERS)) ||
2333 ((type2 >= XML_REGEXP_MARK) &&
2334 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2335 ((type2 >= XML_REGEXP_PUNCT) &&
2336 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2337 ((type2 >= XML_REGEXP_SEPAR) &&
2338 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2339 ((type2 >= XML_REGEXP_SYMBOL) &&
2340 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2341 ((type2 >= XML_REGEXP_OTHER) &&
2342 (type2 <= XML_REGEXP_OTHER_NA))
2345 case XML_REGEXP_NOTDECIMAL: /* \D */
2347 case XML_REGEXP_REALCHAR: /* \w */
2348 /* can't be a mark, separator, pontuation, symbol or other */
2349 if ((type2 == XML_REGEXP_NOTDECIMAL) ||
2350 ((type2 >= XML_REGEXP_MARK) &&
2351 (type2 <= XML_REGEXP_MARK_ENCLOSING)) ||
2352 ((type2 >= XML_REGEXP_PUNCT) &&
2353 (type2 <= XML_REGEXP_PUNCT_OTHERS)) ||
2354 ((type2 >= XML_REGEXP_SEPAR) &&
2355 (type2 <= XML_REGEXP_SEPAR_PARA)) ||
2356 ((type2 >= XML_REGEXP_SYMBOL) &&
2357 (type2 <= XML_REGEXP_SYMBOL_OTHERS)) ||
2358 ((type2 >= XML_REGEXP_OTHER) &&
2359 (type2 <= XML_REGEXP_OTHER_NA))
2362 case XML_REGEXP_NOTREALCHAR: /* \W */
2365 * at that point we know both type 1 and type2 are from
2366 * character categories are ordered and are different,
2367 * it becomes simple because this is a partition
2369 case XML_REGEXP_LETTER:
2370 if (type2 <= XML_REGEXP_LETTER_OTHERS)
2373 case XML_REGEXP_LETTER_UPPERCASE:
2374 case XML_REGEXP_LETTER_LOWERCASE:
2375 case XML_REGEXP_LETTER_TITLECASE:
2376 case XML_REGEXP_LETTER_MODIFIER:
2377 case XML_REGEXP_LETTER_OTHERS:
2379 case XML_REGEXP_MARK:
2380 if (type2 <= XML_REGEXP_MARK_ENCLOSING)
2383 case XML_REGEXP_MARK_NONSPACING:
2384 case XML_REGEXP_MARK_SPACECOMBINING:
2385 case XML_REGEXP_MARK_ENCLOSING:
2387 case XML_REGEXP_NUMBER:
2388 if (type2 <= XML_REGEXP_NUMBER_OTHERS)
2391 case XML_REGEXP_NUMBER_DECIMAL:
2392 case XML_REGEXP_NUMBER_LETTER:
2393 case XML_REGEXP_NUMBER_OTHERS:
2395 case XML_REGEXP_PUNCT:
2396 if (type2 <= XML_REGEXP_PUNCT_OTHERS)
2399 case XML_REGEXP_PUNCT_CONNECTOR:
2400 case XML_REGEXP_PUNCT_DASH:
2401 case XML_REGEXP_PUNCT_OPEN:
2402 case XML_REGEXP_PUNCT_CLOSE:
2403 case XML_REGEXP_PUNCT_INITQUOTE:
2404 case XML_REGEXP_PUNCT_FINQUOTE:
2405 case XML_REGEXP_PUNCT_OTHERS:
2407 case XML_REGEXP_SEPAR:
2408 if (type2 <= XML_REGEXP_SEPAR_PARA)
2411 case XML_REGEXP_SEPAR_SPACE:
2412 case XML_REGEXP_SEPAR_LINE:
2413 case XML_REGEXP_SEPAR_PARA:
2415 case XML_REGEXP_SYMBOL:
2416 if (type2 <= XML_REGEXP_SYMBOL_OTHERS)
2419 case XML_REGEXP_SYMBOL_MATH:
2420 case XML_REGEXP_SYMBOL_CURRENCY:
2421 case XML_REGEXP_SYMBOL_MODIFIER:
2422 case XML_REGEXP_SYMBOL_OTHERS:
2424 case XML_REGEXP_OTHER:
2425 if (type2 <= XML_REGEXP_OTHER_NA)
2428 case XML_REGEXP_OTHER_CONTROL:
2429 case XML_REGEXP_OTHER_FORMAT:
2430 case XML_REGEXP_OTHER_PRIVATE:
2431 case XML_REGEXP_OTHER_NA:
2443 * @deep: if not set only compare string pointers
2445 * Compares two atoms to check whether they are the same exactly
2446 * this is used to remove equivalent transitions
2448 * Returns 1 if same and 0 otherwise
2451 xmlFAEqualAtoms(xmlRegAtomPtr atom1, xmlRegAtomPtr atom2, int deep) {
2456 if ((atom1 == NULL) || (atom2 == NULL))
2459 if (atom1->type != atom2->type)
2461 switch (atom1->type) {
2462 case XML_REGEXP_EPSILON:
2465 case XML_REGEXP_STRING:
2467 ret = (atom1->valuep == atom2->valuep);
2469 ret = xmlStrEqual((xmlChar *)atom1->valuep,
2470 (xmlChar *)atom2->valuep);
2472 case XML_REGEXP_CHARVAL:
2473 ret = (atom1->codepoint == atom2->codepoint);
2475 case XML_REGEXP_RANGES:
2476 /* too hard to do in the general case */
2485 * xmlFACompareAtoms:
2488 * @deep: if not set only compare string pointers
2490 * Compares two atoms to check whether they intersect in some ways,
2491 * this is used by xmlFAComputesDeterminism and xmlFARecurseDeterminism only
2493 * Returns 1 if yes and 0 otherwise
2496 xmlFACompareAtoms(xmlRegAtomPtr atom1, xmlRegAtomPtr atom2, int deep) {
2501 if ((atom1 == NULL) || (atom2 == NULL))
2504 if ((atom1->type == XML_REGEXP_ANYCHAR) ||
2505 (atom2->type == XML_REGEXP_ANYCHAR))
2508 if (atom1->type > atom2->type) {
2514 if (atom1->type != atom2->type) {
2515 ret = xmlFACompareAtomTypes(atom1->type, atom2->type);
2516 /* if they can't intersect at the type level break now */
2520 switch (atom1->type) {
2521 case XML_REGEXP_STRING:
2523 ret = (atom1->valuep != atom2->valuep);
2525 ret = xmlRegStrEqualWildcard((xmlChar *)atom1->valuep,
2526 (xmlChar *)atom2->valuep);
2528 case XML_REGEXP_EPSILON:
2529 goto not_determinist;
2530 case XML_REGEXP_CHARVAL:
2531 if (atom2->type == XML_REGEXP_CHARVAL) {
2532 ret = (atom1->codepoint == atom2->codepoint);
2534 ret = xmlRegCheckCharacter(atom2, atom1->codepoint);
2539 case XML_REGEXP_RANGES:
2540 if (atom2->type == XML_REGEXP_RANGES) {
2542 xmlRegRangePtr r1, r2;
2545 * need to check that none of the ranges eventually matches
2547 for (i = 0;i < atom1->nbRanges;i++) {
2548 for (j = 0;j < atom2->nbRanges;j++) {
2549 r1 = atom1->ranges[i];
2550 r2 = atom2->ranges[j];
2551 res = xmlFACompareRanges(r1, r2);
2562 goto not_determinist;
2565 if (atom1->neg != atom2->neg) {
2575 * xmlFARecurseDeterminism:
2576 * @ctxt: a regexp parser context
2578 * Check whether the associated regexp is determinist,
2579 * should be called after xmlFAEliminateEpsilonTransitions()
2583 xmlFARecurseDeterminism(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
2584 int to, xmlRegAtomPtr atom) {
2587 int transnr, nbTrans;
2593 if (state->markd == XML_REGEXP_MARK_VISITED)
2596 if (ctxt->flags & AM_AUTOMATA_RNG)
2600 * don't recurse on transitions potentially added in the course of
2603 nbTrans = state->nbTrans;
2604 for (transnr = 0;transnr < nbTrans;transnr++) {
2605 t1 = &(state->trans[transnr]);
2607 * check transitions conflicting with the one looked at
2609 if (t1->atom == NULL) {
2612 state->markd = XML_REGEXP_MARK_VISITED;
2613 res = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
2624 if (xmlFACompareAtoms(t1->atom, atom, deep)) {
2626 /* mark the transition as non-deterministic */
2634 * xmlFAComputesDeterminism:
2635 * @ctxt: a regexp parser context
2637 * Check whether the associated regexp is determinist,
2638 * should be called after xmlFAEliminateEpsilonTransitions()
2642 xmlFAComputesDeterminism(xmlRegParserCtxtPtr ctxt) {
2643 int statenr, transnr;
2644 xmlRegStatePtr state;
2645 xmlRegTransPtr t1, t2, last;
2650 #ifdef DEBUG_REGEXP_GRAPH
2651 printf("xmlFAComputesDeterminism\n");
2652 xmlRegPrintCtxt(stdout, ctxt);
2654 if (ctxt->determinist != -1)
2655 return(ctxt->determinist);
2657 if (ctxt->flags & AM_AUTOMATA_RNG)
2661 * First cleanup the automata removing cancelled transitions
2663 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2664 state = ctxt->states[statenr];
2667 if (state->nbTrans < 2)
2669 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2670 t1 = &(state->trans[transnr]);
2672 * Determinism checks in case of counted or all transitions
2673 * will have to be handled separately
2675 if (t1->atom == NULL) {
2679 if (t1->to == -1) /* eliminated */
2681 for (i = 0;i < transnr;i++) {
2682 t2 = &(state->trans[i]);
2683 if (t2->to == -1) /* eliminated */
2685 if (t2->atom != NULL) {
2686 if (t1->to == t2->to) {
2688 * Here we use deep because we want to keep the
2689 * transitions which indicate a conflict
2691 if (xmlFAEqualAtoms(t1->atom, t2->atom, deep) &&
2692 (t1->counter == t2->counter) &&
2693 (t1->count == t2->count))
2694 t2->to = -1; /* eliminated */
2702 * Check for all states that there aren't 2 transitions
2703 * with the same atom and a different target.
2705 for (statenr = 0;statenr < ctxt->nbStates;statenr++) {
2706 state = ctxt->states[statenr];
2709 if (state->nbTrans < 2)
2712 for (transnr = 0;transnr < state->nbTrans;transnr++) {
2713 t1 = &(state->trans[transnr]);
2715 * Determinism checks in case of counted or all transitions
2716 * will have to be handled separately
2718 if (t1->atom == NULL) {
2721 if (t1->to == -1) /* eliminated */
2723 for (i = 0;i < transnr;i++) {
2724 t2 = &(state->trans[i]);
2725 if (t2->to == -1) /* eliminated */
2727 if (t2->atom != NULL) {
2729 * But here we don't use deep because we want to
2730 * find transitions which indicate a conflict
2732 if (xmlFACompareAtoms(t1->atom, t2->atom, 1)) {
2734 /* mark the transitions as non-deterministic ones */
2739 } else if (t1->to != -1) {
2741 * do the closure in case of remaining specific
2742 * epsilon transitions like choices or all
2744 ret = xmlFARecurseDeterminism(ctxt, ctxt->states[t1->to],
2746 /* don't shortcut the computation so all non deterministic
2747 transition get marked down
2758 /* don't shortcut the computation so all non deterministic
2759 transition get marked down
2765 * mark specifically the last non-deterministic transition
2766 * from a state since there is no need to set-up rollback
2773 /* don't shortcut the computation so all non deterministic
2774 transition get marked down
2779 ctxt->determinist = ret;
2783 /************************************************************************
2785 * Routines to check input against transition atoms *
2787 ************************************************************************/
2790 xmlRegCheckCharacterRange(xmlRegAtomType type, int codepoint, int neg,
2791 int start, int end, const xmlChar *blockName) {
2795 case XML_REGEXP_STRING:
2796 case XML_REGEXP_SUBREG:
2797 case XML_REGEXP_RANGES:
2798 case XML_REGEXP_EPSILON:
2800 case XML_REGEXP_ANYCHAR:
2801 ret = ((codepoint != '\n') && (codepoint != '\r'));
2803 case XML_REGEXP_CHARVAL:
2804 ret = ((codepoint >= start) && (codepoint <= end));
2806 case XML_REGEXP_NOTSPACE:
2808 case XML_REGEXP_ANYSPACE:
2809 ret = ((codepoint == '\n') || (codepoint == '\r') ||
2810 (codepoint == '\t') || (codepoint == ' '));
2812 case XML_REGEXP_NOTINITNAME:
2814 case XML_REGEXP_INITNAME:
2815 ret = (IS_LETTER(codepoint) ||
2816 (codepoint == '_') || (codepoint == ':'));
2818 case XML_REGEXP_NOTNAMECHAR:
2820 case XML_REGEXP_NAMECHAR:
2821 ret = (IS_LETTER(codepoint) || IS_DIGIT(codepoint) ||
2822 (codepoint == '.') || (codepoint == '-') ||
2823 (codepoint == '_') || (codepoint == ':') ||
2824 IS_COMBINING(codepoint) || IS_EXTENDER(codepoint));
2826 case XML_REGEXP_NOTDECIMAL:
2828 case XML_REGEXP_DECIMAL:
2829 ret = xmlUCSIsCatNd(codepoint);
2831 case XML_REGEXP_REALCHAR:
2833 case XML_REGEXP_NOTREALCHAR:
2834 ret = xmlUCSIsCatP(codepoint);
2836 ret = xmlUCSIsCatZ(codepoint);
2838 ret = xmlUCSIsCatC(codepoint);
2840 case XML_REGEXP_LETTER:
2841 ret = xmlUCSIsCatL(codepoint);
2843 case XML_REGEXP_LETTER_UPPERCASE:
2844 ret = xmlUCSIsCatLu(codepoint);
2846 case XML_REGEXP_LETTER_LOWERCASE:
2847 ret = xmlUCSIsCatLl(codepoint);
2849 case XML_REGEXP_LETTER_TITLECASE:
2850 ret = xmlUCSIsCatLt(codepoint);
2852 case XML_REGEXP_LETTER_MODIFIER:
2853 ret = xmlUCSIsCatLm(codepoint);
2855 case XML_REGEXP_LETTER_OTHERS:
2856 ret = xmlUCSIsCatLo(codepoint);
2858 case XML_REGEXP_MARK:
2859 ret = xmlUCSIsCatM(codepoint);
2861 case XML_REGEXP_MARK_NONSPACING:
2862 ret = xmlUCSIsCatMn(codepoint);
2864 case XML_REGEXP_MARK_SPACECOMBINING:
2865 ret = xmlUCSIsCatMc(codepoint);
2867 case XML_REGEXP_MARK_ENCLOSING:
2868 ret = xmlUCSIsCatMe(codepoint);
2870 case XML_REGEXP_NUMBER:
2871 ret = xmlUCSIsCatN(codepoint);
2873 case XML_REGEXP_NUMBER_DECIMAL:
2874 ret = xmlUCSIsCatNd(codepoint);
2876 case XML_REGEXP_NUMBER_LETTER:
2877 ret = xmlUCSIsCatNl(codepoint);
2879 case XML_REGEXP_NUMBER_OTHERS:
2880 ret = xmlUCSIsCatNo(codepoint);
2882 case XML_REGEXP_PUNCT:
2883 ret = xmlUCSIsCatP(codepoint);
2885 case XML_REGEXP_PUNCT_CONNECTOR:
2886 ret = xmlUCSIsCatPc(codepoint);
2888 case XML_REGEXP_PUNCT_DASH:
2889 ret = xmlUCSIsCatPd(codepoint);
2891 case XML_REGEXP_PUNCT_OPEN:
2892 ret = xmlUCSIsCatPs(codepoint);
2894 case XML_REGEXP_PUNCT_CLOSE:
2895 ret = xmlUCSIsCatPe(codepoint);
2897 case XML_REGEXP_PUNCT_INITQUOTE:
2898 ret = xmlUCSIsCatPi(codepoint);
2900 case XML_REGEXP_PUNCT_FINQUOTE:
2901 ret = xmlUCSIsCatPf(codepoint);
2903 case XML_REGEXP_PUNCT_OTHERS:
2904 ret = xmlUCSIsCatPo(codepoint);
2906 case XML_REGEXP_SEPAR:
2907 ret = xmlUCSIsCatZ(codepoint);
2909 case XML_REGEXP_SEPAR_SPACE:
2910 ret = xmlUCSIsCatZs(codepoint);
2912 case XML_REGEXP_SEPAR_LINE:
2913 ret = xmlUCSIsCatZl(codepoint);
2915 case XML_REGEXP_SEPAR_PARA:
2916 ret = xmlUCSIsCatZp(codepoint);
2918 case XML_REGEXP_SYMBOL:
2919 ret = xmlUCSIsCatS(codepoint);
2921 case XML_REGEXP_SYMBOL_MATH:
2922 ret = xmlUCSIsCatSm(codepoint);
2924 case XML_REGEXP_SYMBOL_CURRENCY:
2925 ret = xmlUCSIsCatSc(codepoint);
2927 case XML_REGEXP_SYMBOL_MODIFIER:
2928 ret = xmlUCSIsCatSk(codepoint);
2930 case XML_REGEXP_SYMBOL_OTHERS:
2931 ret = xmlUCSIsCatSo(codepoint);
2933 case XML_REGEXP_OTHER:
2934 ret = xmlUCSIsCatC(codepoint);
2936 case XML_REGEXP_OTHER_CONTROL:
2937 ret = xmlUCSIsCatCc(codepoint);
2939 case XML_REGEXP_OTHER_FORMAT:
2940 ret = xmlUCSIsCatCf(codepoint);
2942 case XML_REGEXP_OTHER_PRIVATE:
2943 ret = xmlUCSIsCatCo(codepoint);
2945 case XML_REGEXP_OTHER_NA:
2946 /* ret = xmlUCSIsCatCn(codepoint); */
2947 /* Seems it doesn't exist anymore in recent Unicode releases */
2950 case XML_REGEXP_BLOCK_NAME:
2951 ret = xmlUCSIsBlock(codepoint, (const char *) blockName);
2960 xmlRegCheckCharacter(xmlRegAtomPtr atom, int codepoint) {
2962 xmlRegRangePtr range;
2964 if ((atom == NULL) || (!IS_CHAR(codepoint)))
2967 switch (atom->type) {
2968 case XML_REGEXP_SUBREG:
2969 case XML_REGEXP_EPSILON:
2971 case XML_REGEXP_CHARVAL:
2972 return(codepoint == atom->codepoint);
2973 case XML_REGEXP_RANGES: {
2976 for (i = 0;i < atom->nbRanges;i++) {
2977 range = atom->ranges[i];
2978 if (range->neg == 2) {
2979 ret = xmlRegCheckCharacterRange(range->type, codepoint,
2980 0, range->start, range->end,
2983 return(0); /* excluded char */
2984 } else if (range->neg) {
2985 ret = xmlRegCheckCharacterRange(range->type, codepoint,
2986 0, range->start, range->end,
2993 ret = xmlRegCheckCharacterRange(range->type, codepoint,
2994 0, range->start, range->end,
2997 accept = 1; /* might still be excluded */
3002 case XML_REGEXP_STRING:
3003 printf("TODO: XML_REGEXP_STRING\n");
3005 case XML_REGEXP_ANYCHAR:
3006 case XML_REGEXP_ANYSPACE:
3007 case XML_REGEXP_NOTSPACE:
3008 case XML_REGEXP_INITNAME:
3009 case XML_REGEXP_NOTINITNAME:
3010 case XML_REGEXP_NAMECHAR:
3011 case XML_REGEXP_NOTNAMECHAR:
3012 case XML_REGEXP_DECIMAL:
3013 case XML_REGEXP_NOTDECIMAL:
3014 case XML_REGEXP_REALCHAR:
3015 case XML_REGEXP_NOTREALCHAR:
3016 case XML_REGEXP_LETTER:
3017 case XML_REGEXP_LETTER_UPPERCASE:
3018 case XML_REGEXP_LETTER_LOWERCASE:
3019 case XML_REGEXP_LETTER_TITLECASE:
3020 case XML_REGEXP_LETTER_MODIFIER:
3021 case XML_REGEXP_LETTER_OTHERS:
3022 case XML_REGEXP_MARK:
3023 case XML_REGEXP_MARK_NONSPACING:
3024 case XML_REGEXP_MARK_SPACECOMBINING:
3025 case XML_REGEXP_MARK_ENCLOSING:
3026 case XML_REGEXP_NUMBER:
3027 case XML_REGEXP_NUMBER_DECIMAL:
3028 case XML_REGEXP_NUMBER_LETTER:
3029 case XML_REGEXP_NUMBER_OTHERS:
3030 case XML_REGEXP_PUNCT:
3031 case XML_REGEXP_PUNCT_CONNECTOR:
3032 case XML_REGEXP_PUNCT_DASH:
3033 case XML_REGEXP_PUNCT_OPEN:
3034 case XML_REGEXP_PUNCT_CLOSE:
3035 case XML_REGEXP_PUNCT_INITQUOTE:
3036 case XML_REGEXP_PUNCT_FINQUOTE:
3037 case XML_REGEXP_PUNCT_OTHERS:
3038 case XML_REGEXP_SEPAR:
3039 case XML_REGEXP_SEPAR_SPACE:
3040 case XML_REGEXP_SEPAR_LINE:
3041 case XML_REGEXP_SEPAR_PARA:
3042 case XML_REGEXP_SYMBOL:
3043 case XML_REGEXP_SYMBOL_MATH:
3044 case XML_REGEXP_SYMBOL_CURRENCY:
3045 case XML_REGEXP_SYMBOL_MODIFIER:
3046 case XML_REGEXP_SYMBOL_OTHERS:
3047 case XML_REGEXP_OTHER:
3048 case XML_REGEXP_OTHER_CONTROL:
3049 case XML_REGEXP_OTHER_FORMAT:
3050 case XML_REGEXP_OTHER_PRIVATE:
3051 case XML_REGEXP_OTHER_NA:
3052 case XML_REGEXP_BLOCK_NAME:
3053 ret = xmlRegCheckCharacterRange(atom->type, codepoint, 0, 0, 0,
3054 (const xmlChar *)atom->valuep);
3062 /************************************************************************
3064 * Saving and restoring state of an execution context *
3066 ************************************************************************/
3068 #ifdef DEBUG_REGEXP_EXEC
3070 xmlFARegDebugExec(xmlRegExecCtxtPtr exec) {
3071 printf("state: %d:%d:idx %d", exec->state->no, exec->transno, exec->index);
3072 if (exec->inputStack != NULL) {
3075 for (i = 0;(i < 3) && (i < exec->inputStackNr);i++)
3076 printf("%s ", (const char *)
3077 exec->inputStack[exec->inputStackNr - (i + 1)].value);
3079 printf(": %s", &(exec->inputString[exec->index]));
3086 xmlFARegExecSave(xmlRegExecCtxtPtr exec) {
3087 #ifdef DEBUG_REGEXP_EXEC
3090 xmlFARegDebugExec(exec);
3094 if (exec->nbPush > MAX_PUSH) {
3100 if (exec->maxRollbacks == 0) {
3101 exec->maxRollbacks = 4;
3102 exec->rollbacks = (xmlRegExecRollback *) xmlMalloc(exec->maxRollbacks *
3103 sizeof(xmlRegExecRollback));
3104 if (exec->rollbacks == NULL) {
3105 xmlRegexpErrMemory(NULL, "saving regexp");
3106 exec->maxRollbacks = 0;
3109 memset(exec->rollbacks, 0,
3110 exec->maxRollbacks * sizeof(xmlRegExecRollback));
3111 } else if (exec->nbRollbacks >= exec->maxRollbacks) {
3112 xmlRegExecRollback *tmp;
3113 int len = exec->maxRollbacks;
3115 exec->maxRollbacks *= 2;
3116 tmp = (xmlRegExecRollback *) xmlRealloc(exec->rollbacks,
3117 exec->maxRollbacks * sizeof(xmlRegExecRollback));
3119 xmlRegexpErrMemory(NULL, "saving regexp");
3120 exec->maxRollbacks /= 2;
3123 exec->rollbacks = tmp;
3124 tmp = &exec->rollbacks[len];
3125 memset(tmp, 0, (exec->maxRollbacks - len) * sizeof(xmlRegExecRollback));
3127 exec->rollbacks[exec->nbRollbacks].state = exec->state;
3128 exec->rollbacks[exec->nbRollbacks].index = exec->index;
3129 exec->rollbacks[exec->nbRollbacks].nextbranch = exec->transno + 1;
3130 if (exec->comp->nbCounters > 0) {
3131 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3132 exec->rollbacks[exec->nbRollbacks].counts = (int *)
3133 xmlMalloc(exec->comp->nbCounters * sizeof(int));
3134 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3135 xmlRegexpErrMemory(NULL, "saving regexp");
3140 memcpy(exec->rollbacks[exec->nbRollbacks].counts, exec->counts,
3141 exec->comp->nbCounters * sizeof(int));
3143 exec->nbRollbacks++;
3147 xmlFARegExecRollBack(xmlRegExecCtxtPtr exec) {
3148 if (exec->nbRollbacks <= 0) {
3150 #ifdef DEBUG_REGEXP_EXEC
3151 printf("rollback failed on empty stack\n");
3155 exec->nbRollbacks--;
3156 exec->state = exec->rollbacks[exec->nbRollbacks].state;
3157 exec->index = exec->rollbacks[exec->nbRollbacks].index;
3158 exec->transno = exec->rollbacks[exec->nbRollbacks].nextbranch;
3159 if (exec->comp->nbCounters > 0) {
3160 if (exec->rollbacks[exec->nbRollbacks].counts == NULL) {
3161 fprintf(stderr, "exec save: allocation failed");
3165 memcpy(exec->counts, exec->rollbacks[exec->nbRollbacks].counts,
3166 exec->comp->nbCounters * sizeof(int));
3169 #ifdef DEBUG_REGEXP_EXEC
3170 printf("restored ");
3171 xmlFARegDebugExec(exec);
3175 /************************************************************************
3177 * Verifier, running an input against a compiled regexp *
3179 ************************************************************************/
3182 xmlFARegExec(xmlRegexpPtr comp, const xmlChar *content) {
3183 xmlRegExecCtxt execval;
3184 xmlRegExecCtxtPtr exec = &execval;
3185 int ret, codepoint = 0, len, deter;
3187 exec->inputString = content;
3190 exec->determinist = 1;
3191 exec->maxRollbacks = 0;
3192 exec->nbRollbacks = 0;
3193 exec->rollbacks = NULL;
3196 exec->state = comp->states[0];
3198 exec->transcount = 0;
3199 exec->inputStack = NULL;
3200 exec->inputStackMax = 0;
3201 if (comp->nbCounters > 0) {
3202 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int));
3203 if (exec->counts == NULL) {
3204 xmlRegexpErrMemory(NULL, "running regexp");
3207 memset(exec->counts, 0, comp->nbCounters * sizeof(int));
3209 exec->counts = NULL;
3210 while ((exec->status == 0) && (exec->state != NULL) &&
3211 ((exec->inputString[exec->index] != 0) ||
3212 ((exec->state != NULL) &&
3213 (exec->state->type != XML_REGEXP_FINAL_STATE)))) {
3214 xmlRegTransPtr trans;
3218 * If end of input on non-terminal state, rollback, however we may
3219 * still have epsilon like transition for counted transitions
3220 * on counters, in that case don't break too early. Additionally,
3221 * if we are working on a range like "AB{0,2}", where B is not present,
3222 * we don't want to break.
3225 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL)) {
3227 * if there is a transition, we must check if
3228 * atom allows minOccurs of 0
3230 if (exec->transno < exec->state->nbTrans) {
3231 trans = &exec->state->trans[exec->transno];
3232 if (trans->to >=0) {
3234 if (!((atom->min == 0) && (atom->max > 0)))
3241 exec->transcount = 0;
3242 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
3243 trans = &exec->state->trans[exec->transno];
3249 if (trans->count >= 0) {
3251 xmlRegCounterPtr counter;
3253 if (exec->counts == NULL) {
3258 * A counted transition.
3261 count = exec->counts[trans->count];
3262 counter = &exec->comp->counters[trans->count];
3263 #ifdef DEBUG_REGEXP_EXEC
3264 printf("testing count %d: val %d, min %d, max %d\n",
3265 trans->count, count, counter->min, counter->max);
3267 ret = ((count >= counter->min) && (count <= counter->max));
3268 if ((ret) && (counter->min != counter->max))
3270 } else if (atom == NULL) {
3271 fprintf(stderr, "epsilon transition left at runtime\n");
3274 } else if (exec->inputString[exec->index] != 0) {
3275 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
3276 ret = xmlRegCheckCharacter(atom, codepoint);
3277 if ((ret == 1) && (atom->min >= 0) && (atom->max > 0)) {
3278 xmlRegStatePtr to = comp->states[trans->to];
3281 * this is a multiple input sequence
3282 * If there is a counter associated increment it now.
3283 * before potentially saving and rollback
3284 * do not increment if the counter is already over the
3285 * maximum limit in which case get to next transition
3287 if (trans->counter >= 0) {
3288 xmlRegCounterPtr counter;
3290 if ((exec->counts == NULL) ||
3291 (exec->comp == NULL) ||
3292 (exec->comp->counters == NULL)) {
3296 counter = &exec->comp->counters[trans->counter];
3297 if (exec->counts[trans->counter] >= counter->max)
3298 continue; /* for loop on transitions */
3300 #ifdef DEBUG_REGEXP_EXEC
3301 printf("Increasing count %d\n", trans->counter);
3303 exec->counts[trans->counter]++;
3305 if (exec->state->nbTrans > exec->transno + 1) {
3306 xmlFARegExecSave(exec);
3308 exec->transcount = 1;
3311 * Try to progress as much as possible on the input
3313 if (exec->transcount == atom->max) {
3318 * End of input: stop here
3320 if (exec->inputString[exec->index] == 0) {
3324 if (exec->transcount >= atom->min) {
3325 int transno = exec->transno;
3326 xmlRegStatePtr state = exec->state;
3329 * The transition is acceptable save it
3331 exec->transno = -1; /* trick */
3333 xmlFARegExecSave(exec);
3334 exec->transno = transno;
3335 exec->state = state;
3337 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
3339 ret = xmlRegCheckCharacter(atom, codepoint);
3342 if (exec->transcount < atom->min)
3346 * If the last check failed but one transition was found
3347 * possible, rollback
3354 if (trans->counter >= 0) {
3355 if (exec->counts == NULL) {
3359 #ifdef DEBUG_REGEXP_EXEC
3360 printf("Decreasing count %d\n", trans->counter);
3362 exec->counts[trans->counter]--;
3364 } else if ((ret == 0) && (atom->min == 0) && (atom->max > 0)) {
3366 * we don't match on the codepoint, but minOccurs of 0
3367 * says that's ok. Setting len to 0 inhibits stepping
3368 * over the codepoint.
3370 exec->transcount = 1;
3374 } else if ((atom->min == 0) && (atom->max > 0)) {
3375 /* another spot to match when minOccurs is 0 */
3376 exec->transcount = 1;
3381 if ((trans->nd == 1) ||
3382 ((trans->count >= 0) && (deter == 0) &&
3383 (exec->state->nbTrans > exec->transno + 1))) {
3384 #ifdef DEBUG_REGEXP_EXEC
3386 printf("Saving on nd transition atom %d for %c at %d\n",
3387 trans->atom->no, codepoint, exec->index);
3389 printf("Saving on counted transition count %d for %c at %d\n",
3390 trans->count, codepoint, exec->index);
3392 xmlFARegExecSave(exec);
3394 if (trans->counter >= 0) {
3395 xmlRegCounterPtr counter;
3397 /* make sure we don't go over the counter maximum value */
3398 if ((exec->counts == NULL) ||
3399 (exec->comp == NULL) ||
3400 (exec->comp->counters == NULL)) {
3404 counter = &exec->comp->counters[trans->counter];
3405 if (exec->counts[trans->counter] >= counter->max)
3406 continue; /* for loop on transitions */
3407 #ifdef DEBUG_REGEXP_EXEC
3408 printf("Increasing count %d\n", trans->counter);
3410 exec->counts[trans->counter]++;
3412 if ((trans->count >= 0) &&
3413 (trans->count < REGEXP_ALL_COUNTER)) {
3414 if (exec->counts == NULL) {
3418 #ifdef DEBUG_REGEXP_EXEC
3419 printf("resetting count %d on transition\n",
3422 exec->counts[trans->count] = 0;
3424 #ifdef DEBUG_REGEXP_EXEC
3425 printf("entering state %d\n", trans->to);
3427 exec->state = comp->states[trans->to];
3429 if (trans->atom != NULL) {
3433 } else if (ret < 0) {
3438 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
3441 * Failed to find a way out
3443 exec->determinist = 0;
3444 #ifdef DEBUG_REGEXP_EXEC
3445 printf("rollback from state %d on %d:%c\n", exec->state->no,
3446 codepoint,codepoint);
3448 xmlFARegExecRollBack(exec);
3454 if (exec->rollbacks != NULL) {
3455 if (exec->counts != NULL) {
3458 for (i = 0;i < exec->maxRollbacks;i++)
3459 if (exec->rollbacks[i].counts != NULL)
3460 xmlFree(exec->rollbacks[i].counts);
3462 xmlFree(exec->rollbacks);
3464 if (exec->state == NULL)
3466 if (exec->counts != NULL)
3467 xmlFree(exec->counts);
3468 if (exec->status == 0)
3470 if (exec->status == -1) {
3471 if (exec->nbPush > MAX_PUSH)
3475 return(exec->status);
3478 /************************************************************************
3480 * Progressive interface to the verifier one atom at a time *
3482 ************************************************************************/
3484 static void testerr(xmlRegExecCtxtPtr exec);
3488 * xmlRegNewExecCtxt:
3489 * @comp: a precompiled regular expression
3490 * @callback: a callback function used for handling progresses in the
3491 * automata matching phase
3492 * @data: the context data associated to the callback in this context
3494 * Build a context used for progressive evaluation of a regexp.
3496 * Returns the new context
3499 xmlRegNewExecCtxt(xmlRegexpPtr comp, xmlRegExecCallbacks callback, void *data) {
3500 xmlRegExecCtxtPtr exec;
3504 if ((comp->compact == NULL) && (comp->states == NULL))
3506 exec = (xmlRegExecCtxtPtr) xmlMalloc(sizeof(xmlRegExecCtxt));
3508 xmlRegexpErrMemory(NULL, "creating execution context");
3511 memset(exec, 0, sizeof(xmlRegExecCtxt));
3512 exec->inputString = NULL;
3514 exec->determinist = 1;
3515 exec->maxRollbacks = 0;
3516 exec->nbRollbacks = 0;
3517 exec->rollbacks = NULL;
3520 if (comp->compact == NULL)
3521 exec->state = comp->states[0];
3523 exec->transcount = 0;
3524 exec->callback = callback;
3526 if (comp->nbCounters > 0) {
3528 * For error handling, exec->counts is allocated twice the size
3529 * the second half is used to store the data in case of rollback
3531 exec->counts = (int *) xmlMalloc(comp->nbCounters * sizeof(int)
3533 if (exec->counts == NULL) {
3534 xmlRegexpErrMemory(NULL, "creating execution context");
3538 memset(exec->counts, 0, comp->nbCounters * sizeof(int) * 2);
3539 exec->errCounts = &exec->counts[comp->nbCounters];
3541 exec->counts = NULL;
3542 exec->errCounts = NULL;
3544 exec->inputStackMax = 0;
3545 exec->inputStackNr = 0;
3546 exec->inputStack = NULL;
3547 exec->errStateNo = -1;
3548 exec->errString = NULL;
3554 * xmlRegFreeExecCtxt:
3555 * @exec: a regular expression evaulation context
3557 * Free the structures associated to a regular expression evaulation context.
3560 xmlRegFreeExecCtxt(xmlRegExecCtxtPtr exec) {
3564 if (exec->rollbacks != NULL) {
3565 if (exec->counts != NULL) {
3568 for (i = 0;i < exec->maxRollbacks;i++)
3569 if (exec->rollbacks[i].counts != NULL)
3570 xmlFree(exec->rollbacks[i].counts);
3572 xmlFree(exec->rollbacks);
3574 if (exec->counts != NULL)
3575 xmlFree(exec->counts);
3576 if (exec->inputStack != NULL) {
3579 for (i = 0;i < exec->inputStackNr;i++) {
3580 if (exec->inputStack[i].value != NULL)
3581 xmlFree(exec->inputStack[i].value);
3583 xmlFree(exec->inputStack);
3585 if (exec->errString != NULL)
3586 xmlFree(exec->errString);
3591 xmlFARegExecSaveInputString(xmlRegExecCtxtPtr exec, const xmlChar *value,
3594 printf("saving value: %d:%s\n", exec->inputStackNr, value);
3596 if (exec->inputStackMax == 0) {
3597 exec->inputStackMax = 4;
3598 exec->inputStack = (xmlRegInputTokenPtr)
3599 xmlMalloc(exec->inputStackMax * sizeof(xmlRegInputToken));
3600 if (exec->inputStack == NULL) {
3601 xmlRegexpErrMemory(NULL, "pushing input string");
3602 exec->inputStackMax = 0;
3605 } else if (exec->inputStackNr + 1 >= exec->inputStackMax) {
3606 xmlRegInputTokenPtr tmp;
3608 exec->inputStackMax *= 2;
3609 tmp = (xmlRegInputTokenPtr) xmlRealloc(exec->inputStack,
3610 exec->inputStackMax * sizeof(xmlRegInputToken));
3612 xmlRegexpErrMemory(NULL, "pushing input string");
3613 exec->inputStackMax /= 2;
3616 exec->inputStack = tmp;
3618 exec->inputStack[exec->inputStackNr].value = xmlStrdup(value);
3619 exec->inputStack[exec->inputStackNr].data = data;
3620 exec->inputStackNr++;
3621 exec->inputStack[exec->inputStackNr].value = NULL;
3622 exec->inputStack[exec->inputStackNr].data = NULL;
3626 * xmlRegStrEqualWildcard:
3627 * @expStr: the string to be evaluated
3628 * @valStr: the validation string
3630 * Checks if both strings are equal or have the same content. "*"
3631 * can be used as a wildcard in @valStr; "|" is used as a seperator of
3632 * substrings in both @expStr and @valStr.
3634 * Returns 1 if the comparison is satisfied and the number of substrings
3635 * is equal, 0 otherwise.
3639 xmlRegStrEqualWildcard(const xmlChar *expStr, const xmlChar *valStr) {
3640 if (expStr == valStr) return(1);
3641 if (expStr == NULL) return(0);
3642 if (valStr == NULL) return(0);
3645 * Eval if we have a wildcard for the current item.
3647 if (*expStr != *valStr) {
3648 /* if one of them starts with a wildcard make valStr be it */
3649 if (*valStr == '*') {
3656 if ((*valStr != 0) && (*expStr != 0) && (*expStr++ == '*')) {
3658 if (*valStr == XML_REG_STRING_SEPARATOR)
3661 } while (*valStr != 0);
3668 } while (*valStr != 0);
3676 * xmlRegCompactPushString:
3677 * @exec: a regexp execution context
3678 * @comp: the precompiled exec with a compact table
3679 * @value: a string token input
3680 * @data: data associated to the token to reuse in callbacks
3682 * Push one input token in the execution context
3684 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
3685 * a negative value in case of error.
3688 xmlRegCompactPushString(xmlRegExecCtxtPtr exec,
3690 const xmlChar *value,
3692 int state = exec->index;
3695 if ((comp == NULL) || (comp->compact == NULL) || (comp->stringMap == NULL))
3698 if (value == NULL) {
3700 * are we at a final state ?
3702 if (comp->compact[state * (comp->nbstrings + 1)] ==
3703 XML_REGEXP_FINAL_STATE)
3709 printf("value pushed: %s\n", value);
3713 * Examine all outside transitions from current state
3715 for (i = 0;i < comp->nbstrings;i++) {
3716 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
3717 if ((target > 0) && (target <= comp->nbstates)) {
3718 target--; /* to avoid 0 */
3719 if (xmlRegStrEqualWildcard(comp->stringMap[i], value)) {
3720 exec->index = target;
3721 if ((exec->callback != NULL) && (comp->transdata != NULL)) {
3722 exec->callback(exec->data, value,
3723 comp->transdata[state * comp->nbstrings + i], data);
3726 printf("entering state %d\n", target);
3728 if (comp->compact[target * (comp->nbstrings + 1)] ==
3729 XML_REGEXP_SINK_STATE)
3732 if (comp->compact[target * (comp->nbstrings + 1)] ==
3733 XML_REGEXP_FINAL_STATE)
3740 * Failed to find an exit transition out from current state for the
3744 printf("failed to find a transition for %s on state %d\n", value, state);
3747 if (exec->errString != NULL)
3748 xmlFree(exec->errString);
3749 exec->errString = xmlStrdup(value);
3750 exec->errStateNo = state;
3759 * xmlRegExecPushStringInternal:
3760 * @exec: a regexp execution context or NULL to indicate the end
3761 * @value: a string token input
3762 * @data: data associated to the token to reuse in callbacks
3763 * @compound: value was assembled from 2 strings
3765 * Push one input token in the execution context
3767 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
3768 * a negative value in case of error.
3771 xmlRegExecPushStringInternal(xmlRegExecCtxtPtr exec, const xmlChar *value,
3772 void *data, int compound) {
3773 xmlRegTransPtr trans;
3781 if (exec->comp == NULL)
3783 if (exec->status != 0)
3784 return(exec->status);
3786 if (exec->comp->compact != NULL)
3787 return(xmlRegCompactPushString(exec, exec->comp, value, data));
3789 if (value == NULL) {
3790 if (exec->state->type == XML_REGEXP_FINAL_STATE)
3796 printf("value pushed: %s\n", value);
3799 * If we have an active rollback stack push the new value there
3800 * and get back to where we were left
3802 if ((value != NULL) && (exec->inputStackNr > 0)) {
3803 xmlFARegExecSaveInputString(exec, value, data);
3804 value = exec->inputStack[exec->index].value;
3805 data = exec->inputStack[exec->index].data;
3807 printf("value loaded: %s\n", value);
3811 while ((exec->status == 0) &&
3814 (exec->state->type != XML_REGEXP_FINAL_STATE)))) {
3817 * End of input on non-terminal state, rollback, however we may
3818 * still have epsilon like transition for counted transitions
3819 * on counters, in that case don't break too early.
3821 if ((value == NULL) && (exec->counts == NULL))
3824 exec->transcount = 0;
3825 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
3826 trans = &exec->state->trans[exec->transno];
3831 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
3835 xmlRegCounterPtr counter;
3840 printf("testing all lax %d\n", trans->count);
3843 * Check all counted transitions from the current state
3845 if ((value == NULL) && (final)) {
3847 } else if (value != NULL) {
3848 for (i = 0;i < exec->state->nbTrans;i++) {
3849 t = &exec->state->trans[i];
3850 if ((t->counter < 0) || (t == trans))
3852 counter = &exec->comp->counters[t->counter];
3853 count = exec->counts[t->counter];
3854 if ((count < counter->max) &&
3855 (t->atom != NULL) &&
3856 (xmlStrEqual(value, t->atom->valuep))) {
3860 if ((count >= counter->min) &&
3861 (count < counter->max) &&
3862 (t->atom != NULL) &&
3863 (xmlStrEqual(value, t->atom->valuep))) {
3869 } else if (trans->count == REGEXP_ALL_COUNTER) {
3873 xmlRegCounterPtr counter;
3878 printf("testing all %d\n", trans->count);
3881 * Check all counted transitions from the current state
3883 for (i = 0;i < exec->state->nbTrans;i++) {
3884 t = &exec->state->trans[i];
3885 if ((t->counter < 0) || (t == trans))
3887 counter = &exec->comp->counters[t->counter];
3888 count = exec->counts[t->counter];
3889 if ((count < counter->min) || (count > counter->max)) {
3894 } else if (trans->count >= 0) {
3896 xmlRegCounterPtr counter;
3899 * A counted transition.
3902 count = exec->counts[trans->count];
3903 counter = &exec->comp->counters[trans->count];
3905 printf("testing count %d: val %d, min %d, max %d\n",
3906 trans->count, count, counter->min, counter->max);
3908 ret = ((count >= counter->min) && (count <= counter->max));
3909 } else if (atom == NULL) {
3910 fprintf(stderr, "epsilon transition left at runtime\n");
3913 } else if (value != NULL) {
3914 ret = xmlRegStrEqualWildcard(atom->valuep, value);
3920 if ((ret == 1) && (trans->counter >= 0)) {
3921 xmlRegCounterPtr counter;
3924 count = exec->counts[trans->counter];
3925 counter = &exec->comp->counters[trans->counter];
3926 if (count >= counter->max)
3930 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
3931 xmlRegStatePtr to = exec->comp->states[trans->to];
3934 * this is a multiple input sequence
3936 if (exec->state->nbTrans > exec->transno + 1) {
3937 if (exec->inputStackNr <= 0) {
3938 xmlFARegExecSaveInputString(exec, value, data);
3940 xmlFARegExecSave(exec);
3942 exec->transcount = 1;
3945 * Try to progress as much as possible on the input
3947 if (exec->transcount == atom->max) {
3951 value = exec->inputStack[exec->index].value;
3952 data = exec->inputStack[exec->index].data;
3954 printf("value loaded: %s\n", value);
3958 * End of input: stop here
3960 if (value == NULL) {
3964 if (exec->transcount >= atom->min) {
3965 int transno = exec->transno;
3966 xmlRegStatePtr state = exec->state;
3969 * The transition is acceptable save it
3971 exec->transno = -1; /* trick */
3973 if (exec->inputStackNr <= 0) {
3974 xmlFARegExecSaveInputString(exec, value, data);
3976 xmlFARegExecSave(exec);
3977 exec->transno = transno;
3978 exec->state = state;
3980 ret = xmlStrEqual(value, atom->valuep);
3983 if (exec->transcount < atom->min)
3987 * If the last check failed but one transition was found
3988 * possible, rollback
3998 if ((exec->callback != NULL) && (atom != NULL) &&
4000 exec->callback(exec->data, atom->valuep,
4003 if (exec->state->nbTrans > exec->transno + 1) {
4004 if (exec->inputStackNr <= 0) {
4005 xmlFARegExecSaveInputString(exec, value, data);
4007 xmlFARegExecSave(exec);
4009 if (trans->counter >= 0) {
4011 printf("Increasing count %d\n", trans->counter);
4013 exec->counts[trans->counter]++;
4015 if ((trans->count >= 0) &&
4016 (trans->count < REGEXP_ALL_COUNTER)) {
4017 #ifdef DEBUG_REGEXP_EXEC
4018 printf("resetting count %d on transition\n",
4021 exec->counts[trans->count] = 0;
4024 printf("entering state %d\n", trans->to);
4026 if ((exec->comp->states[trans->to] != NULL) &&
4027 (exec->comp->states[trans->to]->type ==
4028 XML_REGEXP_SINK_STATE)) {
4030 * entering a sink state, save the current state as error
4033 if (exec->errString != NULL)
4034 xmlFree(exec->errString);
4035 exec->errString = xmlStrdup(value);
4036 exec->errState = exec->state;
4037 memcpy(exec->errCounts, exec->counts,
4038 exec->comp->nbCounters * sizeof(int));
4040 exec->state = exec->comp->states[trans->to];
4042 if (trans->atom != NULL) {
4043 if (exec->inputStack != NULL) {
4045 if (exec->index < exec->inputStackNr) {
4046 value = exec->inputStack[exec->index].value;
4047 data = exec->inputStack[exec->index].data;
4049 printf("value loaded: %s\n", value);
4055 printf("end of input\n");
4062 printf("end of input\n");
4067 } else if (ret < 0) {
4072 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
4075 * if we didn't yet rollback on the current input
4076 * store the current state as the error state.
4078 if ((progress) && (exec->state != NULL) &&
4079 (exec->state->type != XML_REGEXP_SINK_STATE)) {
4081 if (exec->errString != NULL)
4082 xmlFree(exec->errString);
4083 exec->errString = xmlStrdup(value);
4084 exec->errState = exec->state;
4085 memcpy(exec->errCounts, exec->counts,
4086 exec->comp->nbCounters * sizeof(int));
4090 * Failed to find a way out
4092 exec->determinist = 0;
4093 xmlFARegExecRollBack(exec);
4094 if (exec->status == 0) {
4095 value = exec->inputStack[exec->index].value;
4096 data = exec->inputStack[exec->index].data;
4098 printf("value loaded: %s\n", value);
4107 if (exec->status == 0) {
4108 return(exec->state->type == XML_REGEXP_FINAL_STATE);
4111 if (exec->status < 0) {
4115 return(exec->status);
4119 * xmlRegExecPushString:
4120 * @exec: a regexp execution context or NULL to indicate the end
4121 * @value: a string token input
4122 * @data: data associated to the token to reuse in callbacks
4124 * Push one input token in the execution context
4126 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
4127 * a negative value in case of error.
4130 xmlRegExecPushString(xmlRegExecCtxtPtr exec, const xmlChar *value,
4132 return(xmlRegExecPushStringInternal(exec, value, data, 0));
4136 * xmlRegExecPushString2:
4137 * @exec: a regexp execution context or NULL to indicate the end
4138 * @value: the first string token input
4139 * @value2: the second string token input
4140 * @data: data associated to the token to reuse in callbacks
4142 * Push one input token in the execution context
4144 * Returns: 1 if the regexp reached a final state, 0 if non-final, and
4145 * a negative value in case of error.
4148 xmlRegExecPushString2(xmlRegExecCtxtPtr exec, const xmlChar *value,
4149 const xmlChar *value2, void *data) {
4151 int lenn, lenp, ret;
4156 if (exec->comp == NULL)
4158 if (exec->status != 0)
4159 return(exec->status);
4162 return(xmlRegExecPushString(exec, value, data));
4164 lenn = strlen((char *) value2);
4165 lenp = strlen((char *) value);
4167 if (150 < lenn + lenp + 2) {
4168 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
4176 memcpy(&str[0], value, lenp);
4177 str[lenp] = XML_REG_STRING_SEPARATOR;
4178 memcpy(&str[lenp + 1], value2, lenn);
4179 str[lenn + lenp + 1] = 0;
4181 if (exec->comp->compact != NULL)
4182 ret = xmlRegCompactPushString(exec, exec->comp, str, data);
4184 ret = xmlRegExecPushStringInternal(exec, str, data, 1);
4192 * xmlRegExecGetValues:
4193 * @exec: a regexp execution context
4194 * @err: error extraction or normal one
4195 * @nbval: pointer to the number of accepted values IN/OUT
4196 * @nbneg: return number of negative transitions
4197 * @values: pointer to the array of acceptable values
4198 * @terminal: return value if this was a terminal state
4200 * Extract informations from the regexp execution, internal routine to
4201 * implement xmlRegExecNextValues() and xmlRegExecErrInfo()
4203 * Returns: 0 in case of success or -1 in case of error.
4206 xmlRegExecGetValues(xmlRegExecCtxtPtr exec, int err,
4207 int *nbval, int *nbneg,
4208 xmlChar **values, int *terminal) {
4212 if ((exec == NULL) || (nbval == NULL) || (nbneg == NULL) ||
4213 (values == NULL) || (*nbval <= 0))
4219 if ((exec->comp != NULL) && (exec->comp->compact != NULL)) {
4221 int target, i, state;
4226 if (exec->errStateNo == -1) return(-1);
4227 state = exec->errStateNo;
4229 state = exec->index;
4231 if (terminal != NULL) {
4232 if (comp->compact[state * (comp->nbstrings + 1)] ==
4233 XML_REGEXP_FINAL_STATE)
4238 for (i = 0;(i < comp->nbstrings) && (nb < maxval);i++) {
4239 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
4240 if ((target > 0) && (target <= comp->nbstates) &&
4241 (comp->compact[(target - 1) * (comp->nbstrings + 1)] !=
4242 XML_REGEXP_SINK_STATE)) {
4243 values[nb++] = comp->stringMap[i];
4247 for (i = 0;(i < comp->nbstrings) && (nb < maxval);i++) {
4248 target = comp->compact[state * (comp->nbstrings + 1) + i + 1];
4249 if ((target > 0) && (target <= comp->nbstates) &&
4250 (comp->compact[(target - 1) * (comp->nbstrings + 1)] ==
4251 XML_REGEXP_SINK_STATE)) {
4252 values[nb++] = comp->stringMap[i];
4258 xmlRegTransPtr trans;
4260 xmlRegStatePtr state;
4262 if (terminal != NULL) {
4263 if (exec->state->type == XML_REGEXP_FINAL_STATE)
4270 if (exec->errState == NULL) return(-1);
4271 state = exec->errState;
4273 if (exec->state == NULL) return(-1);
4274 state = exec->state;
4277 (transno < state->nbTrans) && (nb < maxval);
4279 trans = &state->trans[transno];
4283 if ((atom == NULL) || (atom->valuep == NULL))
4285 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
4286 /* this should not be reached but ... */
4288 } else if (trans->count == REGEXP_ALL_COUNTER) {
4289 /* this should not be reached but ... */
4291 } else if (trans->counter >= 0) {
4292 xmlRegCounterPtr counter = NULL;
4296 count = exec->errCounts[trans->counter];
4298 count = exec->counts[trans->counter];
4299 if (exec->comp != NULL)
4300 counter = &exec->comp->counters[trans->counter];
4301 if ((counter == NULL) || (count < counter->max)) {
4303 values[nb++] = (xmlChar *) atom->valuep2;
4305 values[nb++] = (xmlChar *) atom->valuep;
4309 if ((exec->comp->states[trans->to] != NULL) &&
4310 (exec->comp->states[trans->to]->type !=
4311 XML_REGEXP_SINK_STATE)) {
4313 values[nb++] = (xmlChar *) atom->valuep2;
4315 values[nb++] = (xmlChar *) atom->valuep;
4321 (transno < state->nbTrans) && (nb < maxval);
4323 trans = &state->trans[transno];
4327 if ((atom == NULL) || (atom->valuep == NULL))
4329 if (trans->count == REGEXP_ALL_LAX_COUNTER) {
4331 } else if (trans->count == REGEXP_ALL_COUNTER) {
4333 } else if (trans->counter >= 0) {
4336 if ((exec->comp->states[trans->to] != NULL) &&
4337 (exec->comp->states[trans->to]->type ==
4338 XML_REGEXP_SINK_STATE)) {
4340 values[nb++] = (xmlChar *) atom->valuep2;
4342 values[nb++] = (xmlChar *) atom->valuep;
4352 * xmlRegExecNextValues:
4353 * @exec: a regexp execution context
4354 * @nbval: pointer to the number of accepted values IN/OUT
4355 * @nbneg: return number of negative transitions
4356 * @values: pointer to the array of acceptable values
4357 * @terminal: return value if this was a terminal state
4359 * Extract informations from the regexp execution,
4360 * the parameter @values must point to an array of @nbval string pointers
4361 * on return nbval will contain the number of possible strings in that
4362 * state and the @values array will be updated with them. The string values
4363 * returned will be freed with the @exec context and don't need to be
4366 * Returns: 0 in case of success or -1 in case of error.
4369 xmlRegExecNextValues(xmlRegExecCtxtPtr exec, int *nbval, int *nbneg,
4370 xmlChar **values, int *terminal) {
4371 return(xmlRegExecGetValues(exec, 0, nbval, nbneg, values, terminal));
4375 * xmlRegExecErrInfo:
4376 * @exec: a regexp execution context generating an error
4377 * @string: return value for the error string
4378 * @nbval: pointer to the number of accepted values IN/OUT
4379 * @nbneg: return number of negative transitions
4380 * @values: pointer to the array of acceptable values
4381 * @terminal: return value if this was a terminal state
4383 * Extract error informations from the regexp execution, the parameter
4384 * @string will be updated with the value pushed and not accepted,
4385 * the parameter @values must point to an array of @nbval string pointers
4386 * on return nbval will contain the number of possible strings in that
4387 * state and the @values array will be updated with them. The string values
4388 * returned will be freed with the @exec context and don't need to be
4391 * Returns: 0 in case of success or -1 in case of error.
4394 xmlRegExecErrInfo(xmlRegExecCtxtPtr exec, const xmlChar **string,
4395 int *nbval, int *nbneg, xmlChar **values, int *terminal) {
4398 if (string != NULL) {
4399 if (exec->status != 0)
4400 *string = exec->errString;
4404 return(xmlRegExecGetValues(exec, 1, nbval, nbneg, values, terminal));
4408 static void testerr(xmlRegExecCtxtPtr exec) {
4409 const xmlChar *string;
4414 xmlRegExecErrInfo(exec, &string, &nb, &nbneg, &values[0], &terminal);
4420 xmlRegExecPushChar(xmlRegExecCtxtPtr exec, int UCS) {
4421 xmlRegTransPtr trans;
4428 if (exec->status != 0)
4429 return(exec->status);
4431 while ((exec->status == 0) &&
4432 ((exec->inputString[exec->index] != 0) ||
4433 (exec->state->type != XML_REGEXP_FINAL_STATE))) {
4436 * End of input on non-terminal state, rollback, however we may
4437 * still have epsilon like transition for counted transitions
4438 * on counters, in that case don't break too early.
4440 if ((exec->inputString[exec->index] == 0) && (exec->counts == NULL))
4443 exec->transcount = 0;
4444 for (;exec->transno < exec->state->nbTrans;exec->transno++) {
4445 trans = &exec->state->trans[exec->transno];
4450 if (trans->count >= 0) {
4452 xmlRegCounterPtr counter;
4455 * A counted transition.
4458 count = exec->counts[trans->count];
4459 counter = &exec->comp->counters[trans->count];
4460 #ifdef DEBUG_REGEXP_EXEC
4461 printf("testing count %d: val %d, min %d, max %d\n",
4462 trans->count, count, counter->min, counter->max);
4464 ret = ((count >= counter->min) && (count <= counter->max));
4465 } else if (atom == NULL) {
4466 fprintf(stderr, "epsilon transition left at runtime\n");
4469 } else if (exec->inputString[exec->index] != 0) {
4470 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]), len);
4471 ret = xmlRegCheckCharacter(atom, codepoint);
4472 if ((ret == 1) && (atom->min > 0) && (atom->max > 0)) {
4473 xmlRegStatePtr to = exec->comp->states[trans->to];
4476 * this is a multiple input sequence
4478 if (exec->state->nbTrans > exec->transno + 1) {
4479 xmlFARegExecSave(exec);
4481 exec->transcount = 1;
4484 * Try to progress as much as possible on the input
4486 if (exec->transcount == atom->max) {
4491 * End of input: stop here
4493 if (exec->inputString[exec->index] == 0) {
4497 if (exec->transcount >= atom->min) {
4498 int transno = exec->transno;
4499 xmlRegStatePtr state = exec->state;
4502 * The transition is acceptable save it
4504 exec->transno = -1; /* trick */
4506 xmlFARegExecSave(exec);
4507 exec->transno = transno;
4508 exec->state = state;
4510 codepoint = CUR_SCHAR(&(exec->inputString[exec->index]),
4512 ret = xmlRegCheckCharacter(atom, codepoint);
4515 if (exec->transcount < atom->min)
4519 * If the last check failed but one transition was found
4520 * possible, rollback
4530 if (exec->state->nbTrans > exec->transno + 1) {
4531 xmlFARegExecSave(exec);
4534 * restart count for expressions like this ((abc){2})*
4536 if (trans->count >= 0) {
4537 #ifdef DEBUG_REGEXP_EXEC
4538 printf("Reset count %d\n", trans->count);
4540 exec->counts[trans->count] = 0;
4542 if (trans->counter >= 0) {
4543 #ifdef DEBUG_REGEXP_EXEC
4544 printf("Increasing count %d\n", trans->counter);
4546 exec->counts[trans->counter]++;
4548 #ifdef DEBUG_REGEXP_EXEC
4549 printf("entering state %d\n", trans->to);
4551 exec->state = exec->comp->states[trans->to];
4553 if (trans->atom != NULL) {
4557 } else if (ret < 0) {
4562 if ((exec->transno != 0) || (exec->state->nbTrans == 0)) {
4565 * Failed to find a way out
4567 exec->determinist = 0;
4568 xmlFARegExecRollBack(exec);
4575 /************************************************************************
4577 * Parser for the Schemas Datatype Regular Expressions *
4578 * http://www.w3.org/TR/2001/REC-xmlschema-2-20010502/#regexs *
4580 ************************************************************************/
4584 * @ctxt: a regexp parser context
4586 * [10] Char ::= [^.\?*+()|#x5B#x5D]
4589 xmlFAIsChar(xmlRegParserCtxtPtr ctxt) {
4593 cur = CUR_SCHAR(ctxt->cur, len);
4594 if ((cur == '.') || (cur == '\\') || (cur == '?') ||
4595 (cur == '*') || (cur == '+') || (cur == '(') ||
4596 (cur == ')') || (cur == '|') || (cur == 0x5B) ||
4597 (cur == 0x5D) || (cur == 0))
4603 * xmlFAParseCharProp:
4604 * @ctxt: a regexp parser context
4606 * [27] charProp ::= IsCategory | IsBlock
4607 * [28] IsCategory ::= Letters | Marks | Numbers | Punctuation |
4608 * Separators | Symbols | Others
4609 * [29] Letters ::= 'L' [ultmo]?
4610 * [30] Marks ::= 'M' [nce]?
4611 * [31] Numbers ::= 'N' [dlo]?
4612 * [32] Punctuation ::= 'P' [cdseifo]?
4613 * [33] Separators ::= 'Z' [slp]?
4614 * [34] Symbols ::= 'S' [mcko]?
4615 * [35] Others ::= 'C' [cfon]?
4616 * [36] IsBlock ::= 'Is' [a-zA-Z0-9#x2D]+
4619 xmlFAParseCharProp(xmlRegParserCtxtPtr ctxt) {
4621 xmlRegAtomType type = (xmlRegAtomType) 0;
4622 xmlChar *blockName = NULL;
4630 type = XML_REGEXP_LETTER_UPPERCASE;
4631 } else if (cur == 'l') {
4633 type = XML_REGEXP_LETTER_LOWERCASE;
4634 } else if (cur == 't') {
4636 type = XML_REGEXP_LETTER_TITLECASE;
4637 } else if (cur == 'm') {
4639 type = XML_REGEXP_LETTER_MODIFIER;
4640 } else if (cur == 'o') {
4642 type = XML_REGEXP_LETTER_OTHERS;
4644 type = XML_REGEXP_LETTER;
4646 } else if (cur == 'M') {
4652 type = XML_REGEXP_MARK_NONSPACING;
4653 } else if (cur == 'c') {
4655 /* spacing combining */
4656 type = XML_REGEXP_MARK_SPACECOMBINING;
4657 } else if (cur == 'e') {
4660 type = XML_REGEXP_MARK_ENCLOSING;
4663 type = XML_REGEXP_MARK;
4665 } else if (cur == 'N') {
4671 type = XML_REGEXP_NUMBER_DECIMAL;
4672 } else if (cur == 'l') {
4675 type = XML_REGEXP_NUMBER_LETTER;
4676 } else if (cur == 'o') {
4679 type = XML_REGEXP_NUMBER_OTHERS;
4682 type = XML_REGEXP_NUMBER;
4684 } else if (cur == 'P') {
4690 type = XML_REGEXP_PUNCT_CONNECTOR;
4691 } else if (cur == 'd') {
4694 type = XML_REGEXP_PUNCT_DASH;
4695 } else if (cur == 's') {
4698 type = XML_REGEXP_PUNCT_OPEN;
4699 } else if (cur == 'e') {
4702 type = XML_REGEXP_PUNCT_CLOSE;
4703 } else if (cur == 'i') {
4706 type = XML_REGEXP_PUNCT_INITQUOTE;
4707 } else if (cur == 'f') {
4710 type = XML_REGEXP_PUNCT_FINQUOTE;
4711 } else if (cur == 'o') {
4714 type = XML_REGEXP_PUNCT_OTHERS;
4716 /* all punctuation */
4717 type = XML_REGEXP_PUNCT;
4719 } else if (cur == 'Z') {
4725 type = XML_REGEXP_SEPAR_SPACE;
4726 } else if (cur == 'l') {
4729 type = XML_REGEXP_SEPAR_LINE;
4730 } else if (cur == 'p') {
4733 type = XML_REGEXP_SEPAR_PARA;
4735 /* all separators */
4736 type = XML_REGEXP_SEPAR;
4738 } else if (cur == 'S') {
4743 type = XML_REGEXP_SYMBOL_MATH;
4745 } else if (cur == 'c') {
4747 type = XML_REGEXP_SYMBOL_CURRENCY;
4749 } else if (cur == 'k') {
4751 type = XML_REGEXP_SYMBOL_MODIFIER;
4753 } else if (cur == 'o') {
4755 type = XML_REGEXP_SYMBOL_OTHERS;
4759 type = XML_REGEXP_SYMBOL;
4761 } else if (cur == 'C') {
4767 type = XML_REGEXP_OTHER_CONTROL;
4768 } else if (cur == 'f') {
4771 type = XML_REGEXP_OTHER_FORMAT;
4772 } else if (cur == 'o') {
4775 type = XML_REGEXP_OTHER_PRIVATE;
4776 } else if (cur == 'n') {
4779 type = XML_REGEXP_OTHER_NA;
4782 type = XML_REGEXP_OTHER;
4784 } else if (cur == 'I') {
4785 const xmlChar *start;
4789 ERROR("IsXXXX expected");
4795 if (((cur >= 'a') && (cur <= 'z')) ||
4796 ((cur >= 'A') && (cur <= 'Z')) ||
4797 ((cur >= '0') && (cur <= '9')) ||
4801 while (((cur >= 'a') && (cur <= 'z')) ||
4802 ((cur >= 'A') && (cur <= 'Z')) ||
4803 ((cur >= '0') && (cur <= '9')) ||
4809 type = XML_REGEXP_BLOCK_NAME;
4810 blockName = xmlStrndup(start, ctxt->cur - start);
4812 ERROR("Unknown char property");
4815 if (ctxt->atom == NULL) {
4816 ctxt->atom = xmlRegNewAtom(ctxt, type);
4817 if (ctxt->atom != NULL)
4818 ctxt->atom->valuep = blockName;
4819 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4820 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4821 type, 0, 0, blockName);
4826 * xmlFAParseCharClassEsc:
4827 * @ctxt: a regexp parser context
4829 * [23] charClassEsc ::= ( SingleCharEsc | MultiCharEsc | catEsc | complEsc )
4830 * [24] SingleCharEsc ::= '\' [nrt\|.?*+(){}#x2D#x5B#x5D#x5E]
4831 * [25] catEsc ::= '\p{' charProp '}'
4832 * [26] complEsc ::= '\P{' charProp '}'
4833 * [37] MultiCharEsc ::= '.' | ('\' [sSiIcCdDwW])
4836 xmlFAParseCharClassEsc(xmlRegParserCtxtPtr ctxt) {
4840 if (ctxt->atom == NULL) {
4841 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_ANYCHAR);
4842 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4843 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4844 XML_REGEXP_ANYCHAR, 0, 0, NULL);
4850 ERROR("Escaped sequence: expecting \\");
4858 ERROR("Expecting '{'");
4862 xmlFAParseCharProp(ctxt);
4864 ERROR("Expecting '}'");
4868 } else if (cur == 'P') {
4871 ERROR("Expecting '{'");
4875 xmlFAParseCharProp(ctxt);
4876 ctxt->atom->neg = 1;
4878 ERROR("Expecting '}'");
4882 } else if ((cur == 'n') || (cur == 'r') || (cur == 't') || (cur == '\\') ||
4883 (cur == '|') || (cur == '.') || (cur == '?') || (cur == '*') ||
4884 (cur == '+') || (cur == '(') || (cur == ')') || (cur == '{') ||
4885 (cur == '}') || (cur == 0x2D) || (cur == 0x5B) || (cur == 0x5D) ||
4887 if (ctxt->atom == NULL) {
4888 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
4889 if (ctxt->atom != NULL) {
4892 ctxt->atom->codepoint = '\n';
4895 ctxt->atom->codepoint = '\r';
4898 ctxt->atom->codepoint = '\t';
4901 ctxt->atom->codepoint = cur;
4904 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4916 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4917 XML_REGEXP_CHARVAL, cur, cur, NULL);
4920 } else if ((cur == 's') || (cur == 'S') || (cur == 'i') || (cur == 'I') ||
4921 (cur == 'c') || (cur == 'C') || (cur == 'd') || (cur == 'D') ||
4922 (cur == 'w') || (cur == 'W')) {
4923 xmlRegAtomType type = XML_REGEXP_ANYSPACE;
4927 type = XML_REGEXP_ANYSPACE;
4930 type = XML_REGEXP_NOTSPACE;
4933 type = XML_REGEXP_INITNAME;
4936 type = XML_REGEXP_NOTINITNAME;
4939 type = XML_REGEXP_NAMECHAR;
4942 type = XML_REGEXP_NOTNAMECHAR;
4945 type = XML_REGEXP_DECIMAL;
4948 type = XML_REGEXP_NOTDECIMAL;
4951 type = XML_REGEXP_REALCHAR;
4954 type = XML_REGEXP_NOTREALCHAR;
4958 if (ctxt->atom == NULL) {
4959 ctxt->atom = xmlRegNewAtom(ctxt, type);
4960 } else if (ctxt->atom->type == XML_REGEXP_RANGES) {
4961 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
4965 ERROR("Wrong escape sequence, misuse of character '\\'");
4970 * xmlFAParseCharRange:
4971 * @ctxt: a regexp parser context
4973 * [17] charRange ::= seRange | XmlCharRef | XmlCharIncDash
4974 * [18] seRange ::= charOrEsc '-' charOrEsc
4975 * [20] charOrEsc ::= XmlChar | SingleCharEsc
4976 * [21] XmlChar ::= [^\#x2D#x5B#x5D]
4977 * [22] XmlCharIncDash ::= [^\#x5B#x5D]
4980 xmlFAParseCharRange(xmlRegParserCtxtPtr ctxt) {
4986 ERROR("Expecting ']'");
4995 case 'n': start = 0xA; break;
4996 case 'r': start = 0xD; break;
4997 case 't': start = 0x9; break;
4998 case '\\': case '|': case '.': case '-': case '^': case '?':
4999 case '*': case '+': case '{': case '}': case '(': case ')':
5003 ERROR("Invalid escape value");
5008 } else if ((cur != 0x5B) && (cur != 0x5D)) {
5009 end = start = CUR_SCHAR(ctxt->cur, len);
5011 ERROR("Expecting a char range");
5015 * Since we are "inside" a range, we can assume ctxt->cur is past
5016 * the start of ctxt->string, and PREV should be safe
5018 if ((start == '-') && (NXT(1) != ']') && (PREV != '[') && (PREV != '^')) {
5024 if ((cur != '-') || (NXT(1) == ']')) {
5025 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5026 XML_REGEXP_CHARVAL, start, end, NULL);
5035 case 'n': end = 0xA; break;
5036 case 'r': end = 0xD; break;
5037 case 't': end = 0x9; break;
5038 case '\\': case '|': case '.': case '-': case '^': case '?':
5039 case '*': case '+': case '{': case '}': case '(': case ')':
5043 ERROR("Invalid escape value");
5047 } else if ((cur != 0x5B) && (cur != 0x5D)) {
5048 end = CUR_SCHAR(ctxt->cur, len);
5050 ERROR("Expecting the end of a char range");
5054 /* TODO check that the values are acceptable character ranges for XML */
5056 ERROR("End of range is before start of range");
5058 xmlRegAtomAddRange(ctxt, ctxt->atom, ctxt->neg,
5059 XML_REGEXP_CHARVAL, start, end, NULL);
5065 * xmlFAParsePosCharGroup:
5066 * @ctxt: a regexp parser context
5068 * [14] posCharGroup ::= ( charRange | charClassEsc )+
5071 xmlFAParsePosCharGroup(xmlRegParserCtxtPtr ctxt) {
5074 xmlFAParseCharClassEsc(ctxt);
5076 xmlFAParseCharRange(ctxt);
5078 } while ((CUR != ']') && (CUR != '^') && (CUR != '-') &&
5079 (CUR != 0) && (ctxt->error == 0));
5083 * xmlFAParseCharGroup:
5084 * @ctxt: a regexp parser context
5086 * [13] charGroup ::= posCharGroup | negCharGroup | charClassSub
5087 * [15] negCharGroup ::= '^' posCharGroup
5088 * [16] charClassSub ::= ( posCharGroup | negCharGroup ) '-' charClassExpr
5089 * [12] charClassExpr ::= '[' charGroup ']'
5092 xmlFAParseCharGroup(xmlRegParserCtxtPtr ctxt) {
5094 while ((CUR != ']') && (ctxt->error == 0)) {
5096 int neg = ctxt->neg;
5099 ctxt->neg = !ctxt->neg;
5100 xmlFAParsePosCharGroup(ctxt);
5102 } else if ((CUR == '-') && (NXT(1) == '[')) {
5103 int neg = ctxt->neg;
5105 NEXT; /* eat the '-' */
5106 NEXT; /* eat the '[' */
5107 xmlFAParseCharGroup(ctxt);
5111 ERROR("charClassExpr: ']' expected");
5116 } else if (CUR != ']') {
5117 xmlFAParsePosCharGroup(ctxt);
5124 * xmlFAParseCharClass:
5125 * @ctxt: a regexp parser context
5127 * [11] charClass ::= charClassEsc | charClassExpr
5128 * [12] charClassExpr ::= '[' charGroup ']'
5131 xmlFAParseCharClass(xmlRegParserCtxtPtr ctxt) {
5134 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_RANGES);
5135 if (ctxt->atom == NULL)
5137 xmlFAParseCharGroup(ctxt);
5141 ERROR("xmlFAParseCharClass: ']' expected");
5144 xmlFAParseCharClassEsc(ctxt);
5149 * xmlFAParseQuantExact:
5150 * @ctxt: a regexp parser context
5152 * [8] QuantExact ::= [0-9]+
5154 * Returns 0 if success or -1 in case of error
5157 xmlFAParseQuantExact(xmlRegParserCtxtPtr ctxt) {
5161 while ((CUR >= '0') && (CUR <= '9')) {
5162 ret = ret * 10 + (CUR - '0');
5173 * xmlFAParseQuantifier:
5174 * @ctxt: a regexp parser context
5176 * [4] quantifier ::= [?*+] | ( '{' quantity '}' )
5177 * [5] quantity ::= quantRange | quantMin | QuantExact
5178 * [6] quantRange ::= QuantExact ',' QuantExact
5179 * [7] quantMin ::= QuantExact ','
5180 * [8] QuantExact ::= [0-9]+
5183 xmlFAParseQuantifier(xmlRegParserCtxtPtr ctxt) {
5187 if ((cur == '?') || (cur == '*') || (cur == '+')) {
5188 if (ctxt->atom != NULL) {
5190 ctxt->atom->quant = XML_REGEXP_QUANT_OPT;
5191 else if (cur == '*')
5192 ctxt->atom->quant = XML_REGEXP_QUANT_MULT;
5193 else if (cur == '+')
5194 ctxt->atom->quant = XML_REGEXP_QUANT_PLUS;
5200 int min = 0, max = 0;
5203 cur = xmlFAParseQuantExact(ctxt);
5211 cur = xmlFAParseQuantExact(ctxt);
5215 ERROR("Improper quantifier");
5222 ERROR("Unterminated quantifier");
5226 if (ctxt->atom != NULL) {
5227 ctxt->atom->quant = XML_REGEXP_QUANT_RANGE;
5228 ctxt->atom->min = min;
5229 ctxt->atom->max = max;
5238 * @ctxt: a regexp parser context
5240 * [9] atom ::= Char | charClass | ( '(' regExp ')' )
5243 xmlFAParseAtom(xmlRegParserCtxtPtr ctxt) {
5246 codepoint = xmlFAIsChar(ctxt);
5247 if (codepoint > 0) {
5248 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_CHARVAL);
5249 if (ctxt->atom == NULL)
5251 codepoint = CUR_SCHAR(ctxt->cur, len);
5252 ctxt->atom->codepoint = codepoint;
5255 } else if (CUR == '|') {
5257 } else if (CUR == 0) {
5259 } else if (CUR == ')') {
5261 } else if (CUR == '(') {
5262 xmlRegStatePtr start, oldend, start0;
5266 * this extra Epsilon transition is needed if we count with 0 allowed
5267 * unfortunately this can't be known at that point
5269 xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, NULL);
5270 start0 = ctxt->state;
5271 xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, NULL);
5272 start = ctxt->state;
5276 xmlFAParseRegExp(ctxt, 0);
5280 ERROR("xmlFAParseAtom: expecting ')'");
5282 ctxt->atom = xmlRegNewAtom(ctxt, XML_REGEXP_SUBREG);
5283 if (ctxt->atom == NULL)
5285 ctxt->atom->start = start;
5286 ctxt->atom->start0 = start0;
5287 ctxt->atom->stop = ctxt->state;
5290 } else if ((CUR == '[') || (CUR == '\\') || (CUR == '.')) {
5291 xmlFAParseCharClass(ctxt);
5299 * @ctxt: a regexp parser context
5301 * [3] piece ::= atom quantifier?
5304 xmlFAParsePiece(xmlRegParserCtxtPtr ctxt) {
5308 ret = xmlFAParseAtom(ctxt);
5311 if (ctxt->atom == NULL) {
5312 ERROR("internal: no atom generated");
5314 xmlFAParseQuantifier(ctxt);
5320 * @ctxt: a regexp parser context
5321 * @to: optional target to the end of the branch
5323 * @to is used to optimize by removing duplicate path in automata
5324 * in expressions like (a|b)(c|d)
5326 * [2] branch ::= piece*
5329 xmlFAParseBranch(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr to) {
5330 xmlRegStatePtr previous;
5333 previous = ctxt->state;
5334 ret = xmlFAParsePiece(ctxt);
5336 if (xmlFAGenerateTransitions(ctxt, previous,
5337 (CUR=='|' || CUR==')') ? to : NULL, ctxt->atom) < 0)
5339 previous = ctxt->state;
5342 while ((ret != 0) && (ctxt->error == 0)) {
5343 ret = xmlFAParsePiece(ctxt);
5345 if (xmlFAGenerateTransitions(ctxt, previous,
5346 (CUR=='|' || CUR==')') ? to : NULL, ctxt->atom) < 0)
5348 previous = ctxt->state;
5357 * @ctxt: a regexp parser context
5358 * @top: is this the top-level expression ?
5360 * [1] regExp ::= branch ( '|' branch )*
5363 xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top) {
5364 xmlRegStatePtr start, end;
5366 /* if not top start should have been generated by an epsilon trans */
5367 start = ctxt->state;
5369 xmlFAParseBranch(ctxt, NULL);
5371 #ifdef DEBUG_REGEXP_GRAPH
5372 printf("State %d is final\n", ctxt->state->no);
5374 ctxt->state->type = XML_REGEXP_FINAL_STATE;
5377 ctxt->end = ctxt->state;
5381 while ((CUR == '|') && (ctxt->error == 0)) {
5384 ERROR("expecting a branch after |")
5387 ctxt->state = start;
5389 xmlFAParseBranch(ctxt, end);
5397 /************************************************************************
5401 ************************************************************************/
5405 * @output: the file for the output debug
5406 * @regexp: the compiled regexp
5408 * Print the content of the compiled regular expression
5411 xmlRegexpPrint(FILE *output, xmlRegexpPtr regexp) {
5416 fprintf(output, " regexp: ");
5417 if (regexp == NULL) {
5418 fprintf(output, "NULL\n");
5421 fprintf(output, "'%s' ", regexp->string);
5422 fprintf(output, "\n");
5423 fprintf(output, "%d atoms:\n", regexp->nbAtoms);
5424 for (i = 0;i < regexp->nbAtoms; i++) {
5425 fprintf(output, " %02d ", i);
5426 xmlRegPrintAtom(output, regexp->atoms[i]);
5428 fprintf(output, "%d states:", regexp->nbStates);
5429 fprintf(output, "\n");
5430 for (i = 0;i < regexp->nbStates; i++) {
5431 xmlRegPrintState(output, regexp->states[i]);
5433 fprintf(output, "%d counters:\n", regexp->nbCounters);
5434 for (i = 0;i < regexp->nbCounters; i++) {
5435 fprintf(output, " %d: min %d max %d\n", i, regexp->counters[i].min,
5436 regexp->counters[i].max);
5442 * @regexp: a regular expression string
5444 * Parses a regular expression conforming to XML Schemas Part 2 Datatype
5445 * Appendix F and builds an automata suitable for testing strings against
5446 * that regular expression
5448 * Returns the compiled expression or NULL in case of error
5451 xmlRegexpCompile(const xmlChar *regexp) {
5453 xmlRegParserCtxtPtr ctxt;
5455 ctxt = xmlRegNewParserCtxt(regexp);
5459 /* initialize the parser */
5461 ctxt->start = ctxt->state = xmlRegNewState(ctxt);
5462 xmlRegStatePush(ctxt, ctxt->start);
5464 /* parse the expression building an automata */
5465 xmlFAParseRegExp(ctxt, 1);
5467 ERROR("xmlFAParseRegExp: extra characters");
5469 if (ctxt->error != 0) {
5470 xmlRegFreeParserCtxt(ctxt);
5473 ctxt->end = ctxt->state;
5474 ctxt->start->type = XML_REGEXP_START_STATE;
5475 ctxt->end->type = XML_REGEXP_FINAL_STATE;
5477 /* remove the Epsilon except for counted transitions */
5478 xmlFAEliminateEpsilonTransitions(ctxt);
5481 if (ctxt->error != 0) {
5482 xmlRegFreeParserCtxt(ctxt);
5485 ret = xmlRegEpxFromParse(ctxt);
5486 xmlRegFreeParserCtxt(ctxt);
5492 * @comp: the compiled regular expression
5493 * @content: the value to check against the regular expression
5495 * Check if the regular expression generates the value
5497 * Returns 1 if it matches, 0 if not and a negative value in case of error
5500 xmlRegexpExec(xmlRegexpPtr comp, const xmlChar *content) {
5501 if ((comp == NULL) || (content == NULL))
5503 return(xmlFARegExec(comp, content));
5507 * xmlRegexpIsDeterminist:
5508 * @comp: the compiled regular expression
5510 * Check if the regular expression is determinist
5512 * Returns 1 if it yes, 0 if not and a negative value in case of error
5515 xmlRegexpIsDeterminist(xmlRegexpPtr comp) {
5521 if (comp->determinist != -1)
5522 return(comp->determinist);
5524 am = xmlNewAutomata();
5525 if (am->states != NULL) {
5528 for (i = 0;i < am->nbStates;i++)
5529 xmlRegFreeState(am->states[i]);
5530 xmlFree(am->states);
5532 am->nbAtoms = comp->nbAtoms;
5533 am->atoms = comp->atoms;
5534 am->nbStates = comp->nbStates;
5535 am->states = comp->states;
5536 am->determinist = -1;
5537 am->flags = comp->flags;
5538 ret = xmlFAComputesDeterminism(am);
5541 xmlFreeAutomata(am);
5542 comp->determinist = ret;
5548 * @regexp: the regexp
5553 xmlRegFreeRegexp(xmlRegexpPtr regexp) {
5558 if (regexp->string != NULL)
5559 xmlFree(regexp->string);
5560 if (regexp->states != NULL) {
5561 for (i = 0;i < regexp->nbStates;i++)
5562 xmlRegFreeState(regexp->states[i]);
5563 xmlFree(regexp->states);
5565 if (regexp->atoms != NULL) {
5566 for (i = 0;i < regexp->nbAtoms;i++)
5567 xmlRegFreeAtom(regexp->atoms[i]);
5568 xmlFree(regexp->atoms);
5570 if (regexp->counters != NULL)
5571 xmlFree(regexp->counters);
5572 if (regexp->compact != NULL)
5573 xmlFree(regexp->compact);
5574 if (regexp->transdata != NULL)
5575 xmlFree(regexp->transdata);
5576 if (regexp->stringMap != NULL) {
5577 for (i = 0; i < regexp->nbstrings;i++)
5578 xmlFree(regexp->stringMap[i]);
5579 xmlFree(regexp->stringMap);
5585 #ifdef LIBXML_AUTOMATA_ENABLED
5586 /************************************************************************
5588 * The Automata interface *
5590 ************************************************************************/
5595 * Create a new automata
5597 * Returns the new object or NULL in case of failure
5600 xmlNewAutomata(void) {
5601 xmlAutomataPtr ctxt;
5603 ctxt = xmlRegNewParserCtxt(NULL);
5607 /* initialize the parser */
5609 ctxt->start = ctxt->state = xmlRegNewState(ctxt);
5610 if (ctxt->start == NULL) {
5611 xmlFreeAutomata(ctxt);
5614 ctxt->start->type = XML_REGEXP_START_STATE;
5615 if (xmlRegStatePush(ctxt, ctxt->start) < 0) {
5616 xmlRegFreeState(ctxt->start);
5617 xmlFreeAutomata(ctxt);
5632 xmlFreeAutomata(xmlAutomataPtr am) {
5635 xmlRegFreeParserCtxt(am);
5639 * xmlAutomataSetFlags:
5641 * @flags: a set of internal flags
5643 * Set some flags on the automata
5646 xmlAutomataSetFlags(xmlAutomataPtr am, int flags) {
5653 * xmlAutomataGetInitState:
5656 * Initial state lookup
5658 * Returns the initial state of the automata
5661 xmlAutomataGetInitState(xmlAutomataPtr am) {
5668 * xmlAutomataSetFinalState:
5670 * @state: a state in this automata
5672 * Makes that state a final state
5674 * Returns 0 or -1 in case of error
5677 xmlAutomataSetFinalState(xmlAutomataPtr am, xmlAutomataStatePtr state) {
5678 if ((am == NULL) || (state == NULL))
5680 state->type = XML_REGEXP_FINAL_STATE;
5685 * xmlAutomataNewTransition:
5687 * @from: the starting point of the transition
5688 * @to: the target point of the transition or NULL
5689 * @token: the input string associated to that transition
5690 * @data: data passed to the callback function if the transition is activated
5692 * If @to is NULL, this creates first a new target state in the automata
5693 * and then adds a transition from the @from state to the target state
5694 * activated by the value of @token
5696 * Returns the target state or NULL in case of error
5699 xmlAutomataNewTransition(xmlAutomataPtr am, xmlAutomataStatePtr from,
5700 xmlAutomataStatePtr to, const xmlChar *token,
5704 if ((am == NULL) || (from == NULL) || (token == NULL))
5706 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5712 atom->valuep = xmlStrdup(token);
5714 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5715 xmlRegFreeAtom(atom);
5724 * xmlAutomataNewTransition2:
5726 * @from: the starting point of the transition
5727 * @to: the target point of the transition or NULL
5728 * @token: the first input string associated to that transition
5729 * @token2: the second input string associated to that transition
5730 * @data: data passed to the callback function if the transition is activated
5732 * If @to is NULL, this creates first a new target state in the automata
5733 * and then adds a transition from the @from state to the target state
5734 * activated by the value of @token
5736 * Returns the target state or NULL in case of error
5739 xmlAutomataNewTransition2(xmlAutomataPtr am, xmlAutomataStatePtr from,
5740 xmlAutomataStatePtr to, const xmlChar *token,
5741 const xmlChar *token2, void *data) {
5744 if ((am == NULL) || (from == NULL) || (token == NULL))
5746 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5750 if ((token2 == NULL) || (*token2 == 0)) {
5751 atom->valuep = xmlStrdup(token);
5756 lenn = strlen((char *) token2);
5757 lenp = strlen((char *) token);
5759 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5761 xmlRegFreeAtom(atom);
5764 memcpy(&str[0], token, lenp);
5766 memcpy(&str[lenp + 1], token2, lenn);
5767 str[lenn + lenp + 1] = 0;
5772 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5773 xmlRegFreeAtom(atom);
5782 * xmlAutomataNewNegTrans:
5784 * @from: the starting point of the transition
5785 * @to: the target point of the transition or NULL
5786 * @token: the first input string associated to that transition
5787 * @token2: the second input string associated to that transition
5788 * @data: data passed to the callback function if the transition is activated
5790 * If @to is NULL, this creates first a new target state in the automata
5791 * and then adds a transition from the @from state to the target state
5792 * activated by any value except (@token,@token2)
5793 * Note that if @token2 is not NULL, then (X, NULL) won't match to follow
5794 # the semantic of XSD ##other
5796 * Returns the target state or NULL in case of error
5799 xmlAutomataNewNegTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
5800 xmlAutomataStatePtr to, const xmlChar *token,
5801 const xmlChar *token2, void *data) {
5803 xmlChar err_msg[200];
5805 if ((am == NULL) || (from == NULL) || (token == NULL))
5807 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5812 if ((token2 == NULL) || (*token2 == 0)) {
5813 atom->valuep = xmlStrdup(token);
5818 lenn = strlen((char *) token2);
5819 lenp = strlen((char *) token);
5821 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5823 xmlRegFreeAtom(atom);
5826 memcpy(&str[0], token, lenp);
5828 memcpy(&str[lenp + 1], token2, lenn);
5829 str[lenn + lenp + 1] = 0;
5833 snprintf((char *) err_msg, 199, "not %s", (const char *) atom->valuep);
5835 atom->valuep2 = xmlStrdup(err_msg);
5837 if (xmlFAGenerateTransitions(am, from, to, atom) < 0) {
5838 xmlRegFreeAtom(atom);
5848 * xmlAutomataNewCountTrans2:
5850 * @from: the starting point of the transition
5851 * @to: the target point of the transition or NULL
5852 * @token: the input string associated to that transition
5853 * @token2: the second input string associated to that transition
5854 * @min: the minimum successive occurences of token
5855 * @max: the maximum successive occurences of token
5856 * @data: data associated to the transition
5858 * If @to is NULL, this creates first a new target state in the automata
5859 * and then adds a transition from the @from state to the target state
5860 * activated by a succession of input of value @token and @token2 and
5861 * whose number is between @min and @max
5863 * Returns the target state or NULL in case of error
5866 xmlAutomataNewCountTrans2(xmlAutomataPtr am, xmlAutomataStatePtr from,
5867 xmlAutomataStatePtr to, const xmlChar *token,
5868 const xmlChar *token2,
5869 int min, int max, void *data) {
5873 if ((am == NULL) || (from == NULL) || (token == NULL))
5877 if ((max < min) || (max < 1))
5879 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5882 if ((token2 == NULL) || (*token2 == 0)) {
5883 atom->valuep = xmlStrdup(token);
5888 lenn = strlen((char *) token2);
5889 lenp = strlen((char *) token);
5891 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
5893 xmlRegFreeAtom(atom);
5896 memcpy(&str[0], token, lenp);
5898 memcpy(&str[lenp + 1], token2, lenn);
5899 str[lenn + lenp + 1] = 0;
5911 * associate a counter to the transition.
5913 counter = xmlRegGetCounter(am);
5914 am->counters[counter].min = min;
5915 am->counters[counter].max = max;
5917 /* xmlFAGenerateTransitions(am, from, to, atom); */
5919 to = xmlRegNewState(am);
5920 xmlRegStatePush(am, to);
5922 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
5923 xmlRegAtomPush(am, atom);
5931 xmlFAGenerateEpsilonTransition(am, from, to);
5936 * xmlAutomataNewCountTrans:
5938 * @from: the starting point of the transition
5939 * @to: the target point of the transition or NULL
5940 * @token: the input string associated to that transition
5941 * @min: the minimum successive occurences of token
5942 * @max: the maximum successive occurences of token
5943 * @data: data associated to the transition
5945 * If @to is NULL, this creates first a new target state in the automata
5946 * and then adds a transition from the @from state to the target state
5947 * activated by a succession of input of value @token and whose number
5948 * is between @min and @max
5950 * Returns the target state or NULL in case of error
5953 xmlAutomataNewCountTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
5954 xmlAutomataStatePtr to, const xmlChar *token,
5955 int min, int max, void *data) {
5959 if ((am == NULL) || (from == NULL) || (token == NULL))
5963 if ((max < min) || (max < 1))
5965 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
5968 atom->valuep = xmlStrdup(token);
5977 * associate a counter to the transition.
5979 counter = xmlRegGetCounter(am);
5980 am->counters[counter].min = min;
5981 am->counters[counter].max = max;
5983 /* xmlFAGenerateTransitions(am, from, to, atom); */
5985 to = xmlRegNewState(am);
5986 xmlRegStatePush(am, to);
5988 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
5989 xmlRegAtomPush(am, atom);
5997 xmlFAGenerateEpsilonTransition(am, from, to);
6002 * xmlAutomataNewOnceTrans2:
6004 * @from: the starting point of the transition
6005 * @to: the target point of the transition or NULL
6006 * @token: the input string associated to that transition
6007 * @token2: the second input string associated to that transition
6008 * @min: the minimum successive occurences of token
6009 * @max: the maximum successive occurences of token
6010 * @data: data associated to the transition
6012 * If @to is NULL, this creates first a new target state in the automata
6013 * and then adds a transition from the @from state to the target state
6014 * activated by a succession of input of value @token and @token2 and whose
6015 * number is between @min and @max, moreover that transition can only be
6018 * Returns the target state or NULL in case of error
6021 xmlAutomataNewOnceTrans2(xmlAutomataPtr am, xmlAutomataStatePtr from,
6022 xmlAutomataStatePtr to, const xmlChar *token,
6023 const xmlChar *token2,
6024 int min, int max, void *data) {
6028 if ((am == NULL) || (from == NULL) || (token == NULL))
6032 if ((max < min) || (max < 1))
6034 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6037 if ((token2 == NULL) || (*token2 == 0)) {
6038 atom->valuep = xmlStrdup(token);
6043 lenn = strlen((char *) token2);
6044 lenp = strlen((char *) token);
6046 str = (xmlChar *) xmlMallocAtomic(lenn + lenp + 2);
6048 xmlRegFreeAtom(atom);
6051 memcpy(&str[0], token, lenp);
6053 memcpy(&str[lenp + 1], token2, lenn);
6054 str[lenn + lenp + 1] = 0;
6059 atom->quant = XML_REGEXP_QUANT_ONCEONLY;
6063 * associate a counter to the transition.
6065 counter = xmlRegGetCounter(am);
6066 am->counters[counter].min = 1;
6067 am->counters[counter].max = 1;
6069 /* xmlFAGenerateTransitions(am, from, to, atom); */
6071 to = xmlRegNewState(am);
6072 xmlRegStatePush(am, to);
6074 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6075 xmlRegAtomPush(am, atom);
6083 * xmlAutomataNewOnceTrans:
6085 * @from: the starting point of the transition
6086 * @to: the target point of the transition or NULL
6087 * @token: the input string associated to that transition
6088 * @min: the minimum successive occurences of token
6089 * @max: the maximum successive occurences of token
6090 * @data: data associated to the transition
6092 * If @to is NULL, this creates first a new target state in the automata
6093 * and then adds a transition from the @from state to the target state
6094 * activated by a succession of input of value @token and whose number
6095 * is between @min and @max, moreover that transition can only be crossed
6098 * Returns the target state or NULL in case of error
6101 xmlAutomataNewOnceTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6102 xmlAutomataStatePtr to, const xmlChar *token,
6103 int min, int max, void *data) {
6107 if ((am == NULL) || (from == NULL) || (token == NULL))
6111 if ((max < min) || (max < 1))
6113 atom = xmlRegNewAtom(am, XML_REGEXP_STRING);
6116 atom->valuep = xmlStrdup(token);
6118 atom->quant = XML_REGEXP_QUANT_ONCEONLY;
6122 * associate a counter to the transition.
6124 counter = xmlRegGetCounter(am);
6125 am->counters[counter].min = 1;
6126 am->counters[counter].max = 1;
6128 /* xmlFAGenerateTransitions(am, from, to, atom); */
6130 to = xmlRegNewState(am);
6131 xmlRegStatePush(am, to);
6133 xmlRegStateAddTrans(am, from, atom, to, counter, -1);
6134 xmlRegAtomPush(am, atom);
6140 * xmlAutomataNewState:
6143 * Create a new disconnected state in the automata
6145 * Returns the new state or NULL in case of error
6148 xmlAutomataNewState(xmlAutomataPtr am) {
6149 xmlAutomataStatePtr to;
6153 to = xmlRegNewState(am);
6154 xmlRegStatePush(am, to);
6159 * xmlAutomataNewEpsilon:
6161 * @from: the starting point of the transition
6162 * @to: the target point of the transition or NULL
6164 * If @to is NULL, this creates first a new target state in the automata
6165 * and then adds an epsilon transition from the @from state to the
6168 * Returns the target state or NULL in case of error
6171 xmlAutomataNewEpsilon(xmlAutomataPtr am, xmlAutomataStatePtr from,
6172 xmlAutomataStatePtr to) {
6173 if ((am == NULL) || (from == NULL))
6175 xmlFAGenerateEpsilonTransition(am, from, to);
6182 * xmlAutomataNewAllTrans:
6184 * @from: the starting point of the transition
6185 * @to: the target point of the transition or NULL
6186 * @lax: allow to transition if not all all transitions have been activated
6188 * If @to is NULL, this creates first a new target state in the automata
6189 * and then adds a an ALL transition from the @from state to the
6190 * target state. That transition is an epsilon transition allowed only when
6191 * all transitions from the @from node have been activated.
6193 * Returns the target state or NULL in case of error
6196 xmlAutomataNewAllTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6197 xmlAutomataStatePtr to, int lax) {
6198 if ((am == NULL) || (from == NULL))
6200 xmlFAGenerateAllTransition(am, from, to, lax);
6207 * xmlAutomataNewCounter:
6209 * @min: the minimal value on the counter
6210 * @max: the maximal value on the counter
6212 * Create a new counter
6214 * Returns the counter number or -1 in case of error
6217 xmlAutomataNewCounter(xmlAutomataPtr am, int min, int max) {
6223 ret = xmlRegGetCounter(am);
6226 am->counters[ret].min = min;
6227 am->counters[ret].max = max;
6232 * xmlAutomataNewCountedTrans:
6234 * @from: the starting point of the transition
6235 * @to: the target point of the transition or NULL
6236 * @counter: the counter associated to that transition
6238 * If @to is NULL, this creates first a new target state in the automata
6239 * and then adds an epsilon transition from the @from state to the target state
6240 * which will increment the counter provided
6242 * Returns the target state or NULL in case of error
6245 xmlAutomataNewCountedTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6246 xmlAutomataStatePtr to, int counter) {
6247 if ((am == NULL) || (from == NULL) || (counter < 0))
6249 xmlFAGenerateCountedEpsilonTransition(am, from, to, counter);
6256 * xmlAutomataNewCounterTrans:
6258 * @from: the starting point of the transition
6259 * @to: the target point of the transition or NULL
6260 * @counter: the counter associated to that transition
6262 * If @to is NULL, this creates first a new target state in the automata
6263 * and then adds an epsilon transition from the @from state to the target state
6264 * which will be allowed only if the counter is within the right range.
6266 * Returns the target state or NULL in case of error
6269 xmlAutomataNewCounterTrans(xmlAutomataPtr am, xmlAutomataStatePtr from,
6270 xmlAutomataStatePtr to, int counter) {
6271 if ((am == NULL) || (from == NULL) || (counter < 0))
6273 xmlFAGenerateCountedTransition(am, from, to, counter);
6280 * xmlAutomataCompile:
6283 * Compile the automata into a Reg Exp ready for being executed.
6284 * The automata should be free after this point.
6286 * Returns the compiled regexp or NULL in case of error
6289 xmlAutomataCompile(xmlAutomataPtr am) {
6292 if ((am == NULL) || (am->error != 0)) return(NULL);
6293 xmlFAEliminateEpsilonTransitions(am);
6294 /* xmlFAComputesDeterminism(am); */
6295 ret = xmlRegEpxFromParse(am);
6301 * xmlAutomataIsDeterminist:
6304 * Checks if an automata is determinist.
6306 * Returns 1 if true, 0 if not, and -1 in case of error
6309 xmlAutomataIsDeterminist(xmlAutomataPtr am) {
6315 ret = xmlFAComputesDeterminism(am);
6318 #endif /* LIBXML_AUTOMATA_ENABLED */
6320 #ifdef LIBXML_EXPR_ENABLED
6321 /************************************************************************
6323 * Formal Expression handling code *
6325 ************************************************************************/
6326 /************************************************************************
6328 * Expression handling context *
6330 ************************************************************************/
6332 struct _xmlExpCtxt {
6334 xmlExpNodePtr *table;
6347 * @maxNodes: the maximum number of nodes
6348 * @dict: optional dictionnary to use internally
6350 * Creates a new context for manipulating expressions
6352 * Returns the context or NULL in case of error
6355 xmlExpNewCtxt(int maxNodes, xmlDictPtr dict) {
6359 if (maxNodes <= 4096)
6362 ret = (xmlExpCtxtPtr) xmlMalloc(sizeof(xmlExpCtxt));
6365 memset(ret, 0, sizeof(xmlExpCtxt));
6368 ret->maxNodes = maxNodes;
6369 ret->table = xmlMalloc(size * sizeof(xmlExpNodePtr));
6370 if (ret->table == NULL) {
6374 memset(ret->table, 0, size * sizeof(xmlExpNodePtr));
6376 ret->dict = xmlDictCreate();
6377 if (ret->dict == NULL) {
6378 xmlFree(ret->table);
6384 xmlDictReference(ret->dict);
6391 * @ctxt: an expression context
6393 * Free an expression context
6396 xmlExpFreeCtxt(xmlExpCtxtPtr ctxt) {
6399 xmlDictFree(ctxt->dict);
6400 if (ctxt->table != NULL)
6401 xmlFree(ctxt->table);
6405 /************************************************************************
6407 * Structure associated to an expression node *
6409 ************************************************************************/
6410 #define MAX_NODES 10000
6412 /* #define DEBUG_DERIV */
6417 * - public API for creation
6420 * - regression testing
6423 * - split into module and test tool
6428 XML_EXP_NILABLE = (1 << 0)
6431 #define IS_NILLABLE(node) ((node)->info & XML_EXP_NILABLE)
6433 struct _xmlExpNode {
6434 unsigned char type;/* xmlExpNodeType */
6435 unsigned char info;/* OR of xmlExpNodeInfo */
6436 unsigned short key; /* the hash key */
6437 unsigned int ref; /* The number of references */
6438 int c_max; /* the maximum length it can consume */
6439 xmlExpNodePtr exp_left;
6440 xmlExpNodePtr next;/* the next node in the hash table or free list */
6447 xmlExpNodePtr f_right;
6449 const xmlChar *f_str;
6453 #define exp_min field.count.f_min
6454 #define exp_max field.count.f_max
6455 /* #define exp_left field.children.f_left */
6456 #define exp_right field.children.f_right
6457 #define exp_str field.f_str
6459 static xmlExpNodePtr xmlExpNewNode(xmlExpCtxtPtr ctxt, xmlExpNodeType type);
6460 static xmlExpNode forbiddenExpNode = {
6461 XML_EXP_FORBID, 0, 0, 0, 0, NULL, NULL, {{ 0, 0}}
6463 xmlExpNodePtr forbiddenExp = &forbiddenExpNode;
6464 static xmlExpNode emptyExpNode = {
6465 XML_EXP_EMPTY, 1, 0, 0, 0, NULL, NULL, {{ 0, 0}}
6467 xmlExpNodePtr emptyExp = &emptyExpNode;
6469 /************************************************************************
6471 * The custom hash table for unicity and canonicalization *
6472 * of sub-expressions pointers *
6474 ************************************************************************/
6476 * xmlExpHashNameComputeKey:
6477 * Calculate the hash key for a token
6479 static unsigned short
6480 xmlExpHashNameComputeKey(const xmlChar *name) {
6481 unsigned short value = 0L;
6485 value += 30 * (*name);
6486 while ((ch = *name++) != 0) {
6487 value = value ^ ((value << 5) + (value >> 3) + (unsigned long)ch);
6494 * xmlExpHashComputeKey:
6495 * Calculate the hash key for a compound expression
6497 static unsigned short
6498 xmlExpHashComputeKey(xmlExpNodeType type, xmlExpNodePtr left,
6499 xmlExpNodePtr right) {
6500 unsigned long value;
6506 value += right->key;
6508 ret = (unsigned short) value;
6512 value += right->key;
6514 ret = (unsigned short) value;
6518 value += right->key;
6519 ret = (unsigned short) value;
6528 static xmlExpNodePtr
6529 xmlExpNewNode(xmlExpCtxtPtr ctxt, xmlExpNodeType type) {
6532 if (ctxt->nb_nodes >= MAX_NODES)
6534 ret = (xmlExpNodePtr) xmlMalloc(sizeof(xmlExpNode));
6537 memset(ret, 0, sizeof(xmlExpNode));
6546 * xmlExpHashGetEntry:
6547 * @table: the hash table
6549 * Get the unique entry from the hash table. The entry is created if
6550 * needed. @left and @right are consumed, i.e. their ref count will
6551 * be decremented by the operation.
6553 * Returns the pointer or NULL in case of error
6555 static xmlExpNodePtr
6556 xmlExpHashGetEntry(xmlExpCtxtPtr ctxt, xmlExpNodeType type,
6557 xmlExpNodePtr left, xmlExpNodePtr right,
6558 const xmlChar *name, int min, int max) {
6559 unsigned short kbase, key;
6560 xmlExpNodePtr entry;
6561 xmlExpNodePtr insert;
6567 * Check for duplicate and insertion location.
6569 if (type == XML_EXP_ATOM) {
6570 kbase = xmlExpHashNameComputeKey(name);
6571 } else if (type == XML_EXP_COUNT) {
6572 /* COUNT reduction rule 1 */
6579 xmlExpFree(ctxt, left);
6584 xmlExpFree(ctxt, left);
6585 return(forbiddenExp);
6592 } else if (type == XML_EXP_OR) {
6593 /* Forbid reduction rules */
6594 if (left->type == XML_EXP_FORBID) {
6595 xmlExpFree(ctxt, left);
6598 if (right->type == XML_EXP_FORBID) {
6599 xmlExpFree(ctxt, right);
6603 /* OR reduction rule 1 */
6604 /* a | a reduced to a */
6605 if (left == right) {
6609 /* OR canonicalization rule 1 */
6610 /* linearize (a | b) | c into a | (b | c) */
6611 if ((left->type == XML_EXP_OR) && (right->type != XML_EXP_OR)) {
6612 xmlExpNodePtr tmp = left;
6616 /* OR reduction rule 2 */
6617 /* a | (a | b) and b | (a | b) are reduced to a | b */
6618 if (right->type == XML_EXP_OR) {
6619 if ((left == right->exp_left) ||
6620 (left == right->exp_right)) {
6621 xmlExpFree(ctxt, left);
6625 /* OR canonicalization rule 2 */
6626 /* linearize (a | b) | c into a | (b | c) */
6627 if (left->type == XML_EXP_OR) {
6630 /* OR canonicalization rule 2 */
6631 if ((left->exp_right->type != XML_EXP_OR) &&
6632 (left->exp_right->key < left->exp_left->key)) {
6633 tmp = left->exp_right;
6634 left->exp_right = left->exp_left;
6635 left->exp_left = tmp;
6637 left->exp_right->ref++;
6638 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left->exp_right, right,
6640 left->exp_left->ref++;
6641 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left->exp_left, tmp,
6644 xmlExpFree(ctxt, left);
6647 if (right->type == XML_EXP_OR) {
6648 /* Ordering in the tree */
6649 /* C | (A | B) -> A | (B | C) */
6650 if (left->key > right->exp_right->key) {
6652 right->exp_right->ref++;
6653 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_right,
6655 right->exp_left->ref++;
6656 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_left,
6658 xmlExpFree(ctxt, right);
6661 /* Ordering in the tree */
6662 /* B | (A | C) -> A | (B | C) */
6663 if (left->key > right->exp_left->key) {
6665 right->exp_right->ref++;
6666 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, left,
6667 right->exp_right, NULL, 0, 0);
6668 right->exp_left->ref++;
6669 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_OR, right->exp_left,
6671 xmlExpFree(ctxt, right);
6675 /* we know both types are != XML_EXP_OR here */
6676 else if (left->key > right->key) {
6677 xmlExpNodePtr tmp = left;
6681 kbase = xmlExpHashComputeKey(type, left, right);
6682 } else if (type == XML_EXP_SEQ) {
6683 /* Forbid reduction rules */
6684 if (left->type == XML_EXP_FORBID) {
6685 xmlExpFree(ctxt, right);
6688 if (right->type == XML_EXP_FORBID) {
6689 xmlExpFree(ctxt, left);
6692 /* Empty reduction rules */
6693 if (right->type == XML_EXP_EMPTY) {
6696 if (left->type == XML_EXP_EMPTY) {
6699 kbase = xmlExpHashComputeKey(type, left, right);
6703 key = kbase % ctxt->size;
6704 if (ctxt->table[key] != NULL) {
6705 for (insert = ctxt->table[key]; insert != NULL;
6706 insert = insert->next) {
6707 if ((insert->key == kbase) &&
6708 (insert->type == type)) {
6709 if (type == XML_EXP_ATOM) {
6710 if (name == insert->exp_str) {
6714 } else if (type == XML_EXP_COUNT) {
6715 if ((insert->exp_min == min) && (insert->exp_max == max) &&
6716 (insert->exp_left == left)) {
6721 } else if ((insert->exp_left == left) &&
6722 (insert->exp_right == right)) {
6732 entry = xmlExpNewNode(ctxt, type);
6736 if (type == XML_EXP_ATOM) {
6737 entry->exp_str = name;
6739 } else if (type == XML_EXP_COUNT) {
6740 entry->exp_min = min;
6741 entry->exp_max = max;
6742 entry->exp_left = left;
6743 if ((min == 0) || (IS_NILLABLE(left)))
6744 entry->info |= XML_EXP_NILABLE;
6748 entry->c_max = max * entry->exp_left->c_max;
6750 entry->exp_left = left;
6751 entry->exp_right = right;
6752 if (type == XML_EXP_OR) {
6753 if ((IS_NILLABLE(left)) || (IS_NILLABLE(right)))
6754 entry->info |= XML_EXP_NILABLE;
6755 if ((entry->exp_left->c_max == -1) ||
6756 (entry->exp_right->c_max == -1))
6758 else if (entry->exp_left->c_max > entry->exp_right->c_max)
6759 entry->c_max = entry->exp_left->c_max;
6761 entry->c_max = entry->exp_right->c_max;
6763 if ((IS_NILLABLE(left)) && (IS_NILLABLE(right)))
6764 entry->info |= XML_EXP_NILABLE;
6765 if ((entry->exp_left->c_max == -1) ||
6766 (entry->exp_right->c_max == -1))
6769 entry->c_max = entry->exp_left->c_max + entry->exp_right->c_max;
6773 if (ctxt->table[key] != NULL)
6774 entry->next = ctxt->table[key];
6776 ctxt->table[key] = entry;
6784 * @ctxt: the expression context
6785 * @exp: the expression
6787 * Dereference the expression
6790 xmlExpFree(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp) {
6791 if ((exp == NULL) || (exp == forbiddenExp) || (exp == emptyExp))
6794 if (exp->ref == 0) {
6797 /* Unlink it first from the hash table */
6798 key = exp->key % ctxt->size;
6799 if (ctxt->table[key] == exp) {
6800 ctxt->table[key] = exp->next;
6804 tmp = ctxt->table[key];
6805 while (tmp != NULL) {
6806 if (tmp->next == exp) {
6807 tmp->next = exp->next;
6814 if ((exp->type == XML_EXP_SEQ) || (exp->type == XML_EXP_OR)) {
6815 xmlExpFree(ctxt, exp->exp_left);
6816 xmlExpFree(ctxt, exp->exp_right);
6817 } else if (exp->type == XML_EXP_COUNT) {
6818 xmlExpFree(ctxt, exp->exp_left);
6827 * @exp: the expression
6829 * Increase the reference count of the expression
6832 xmlExpRef(xmlExpNodePtr exp) {
6839 * @ctxt: the expression context
6840 * @name: the atom name
6841 * @len: the atom name length in byte (or -1);
6843 * Get the atom associated to this name from that context
6845 * Returns the node or NULL in case of error
6848 xmlExpNewAtom(xmlExpCtxtPtr ctxt, const xmlChar *name, int len) {
6849 if ((ctxt == NULL) || (name == NULL))
6851 name = xmlDictLookup(ctxt->dict, name, len);
6854 return(xmlExpHashGetEntry(ctxt, XML_EXP_ATOM, NULL, NULL, name, 0, 0));
6859 * @ctxt: the expression context
6860 * @left: left expression
6861 * @right: right expression
6863 * Get the atom associated to the choice @left | @right
6864 * Note that @left and @right are consumed in the operation, to keep
6865 * an handle on them use xmlExpRef() and use xmlExpFree() to release them,
6866 * this is true even in case of failure (unless ctxt == NULL).
6868 * Returns the node or NULL in case of error
6871 xmlExpNewOr(xmlExpCtxtPtr ctxt, xmlExpNodePtr left, xmlExpNodePtr right) {
6874 if ((left == NULL) || (right == NULL)) {
6875 xmlExpFree(ctxt, left);
6876 xmlExpFree(ctxt, right);
6879 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, left, right, NULL, 0, 0));
6884 * @ctxt: the expression context
6885 * @left: left expression
6886 * @right: right expression
6888 * Get the atom associated to the sequence @left , @right
6889 * Note that @left and @right are consumed in the operation, to keep
6890 * an handle on them use xmlExpRef() and use xmlExpFree() to release them,
6891 * this is true even in case of failure (unless ctxt == NULL).
6893 * Returns the node or NULL in case of error
6896 xmlExpNewSeq(xmlExpCtxtPtr ctxt, xmlExpNodePtr left, xmlExpNodePtr right) {
6899 if ((left == NULL) || (right == NULL)) {
6900 xmlExpFree(ctxt, left);
6901 xmlExpFree(ctxt, right);
6904 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, left, right, NULL, 0, 0));
6909 * @ctxt: the expression context
6910 * @subset: the expression to be repeated
6911 * @min: the lower bound for the repetition
6912 * @max: the upper bound for the repetition, -1 means infinite
6914 * Get the atom associated to the range (@subset){@min, @max}
6915 * Note that @subset is consumed in the operation, to keep
6916 * an handle on it use xmlExpRef() and use xmlExpFree() to release it,
6917 * this is true even in case of failure (unless ctxt == NULL).
6919 * Returns the node or NULL in case of error
6922 xmlExpNewRange(xmlExpCtxtPtr ctxt, xmlExpNodePtr subset, int min, int max) {
6925 if ((subset == NULL) || (min < 0) || (max < -1) ||
6926 ((max >= 0) && (min > max))) {
6927 xmlExpFree(ctxt, subset);
6930 return(xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, subset,
6931 NULL, NULL, min, max));
6934 /************************************************************************
6936 * Public API for operations on expressions *
6938 ************************************************************************/
6941 xmlExpGetLanguageInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
6942 const xmlChar**list, int len, int nb) {
6945 switch (exp->type) {
6949 for (tmp = 0;tmp < nb;tmp++)
6950 if (list[tmp] == exp->exp_str)
6954 list[nb] = exp->exp_str;
6957 exp = exp->exp_left;
6961 tmp = xmlExpGetLanguageInt(ctxt, exp->exp_left, list, len, nb);
6964 tmp2 = xmlExpGetLanguageInt(ctxt, exp->exp_right, list, len,
6974 * xmlExpGetLanguage:
6975 * @ctxt: the expression context
6976 * @exp: the expression
6977 * @langList: where to store the tokens
6978 * @len: the allocated length of @list
6980 * Find all the strings used in @exp and store them in @list
6982 * Returns the number of unique strings found, -1 in case of errors and
6983 * -2 if there is more than @len strings
6986 xmlExpGetLanguage(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
6987 const xmlChar**langList, int len) {
6988 if ((ctxt == NULL) || (exp == NULL) || (langList == NULL) || (len <= 0))
6990 return(xmlExpGetLanguageInt(ctxt, exp, langList, len, 0));
6994 xmlExpGetStartInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
6995 const xmlChar**list, int len, int nb) {
6998 switch (exp->type) {
6999 case XML_EXP_FORBID:
7004 for (tmp = 0;tmp < nb;tmp++)
7005 if (list[tmp] == exp->exp_str)
7009 list[nb] = exp->exp_str;
7012 exp = exp->exp_left;
7015 tmp = xmlExpGetStartInt(ctxt, exp->exp_left, list, len, nb);
7018 if (IS_NILLABLE(exp->exp_left)) {
7019 tmp2 = xmlExpGetStartInt(ctxt, exp->exp_right, list, len,
7027 tmp = xmlExpGetStartInt(ctxt, exp->exp_left, list, len, nb);
7030 tmp2 = xmlExpGetStartInt(ctxt, exp->exp_right, list, len,
7041 * @ctxt: the expression context
7042 * @exp: the expression
7043 * @tokList: where to store the tokens
7044 * @len: the allocated length of @list
7046 * Find all the strings that appears at the start of the languages
7047 * accepted by @exp and store them in @list. E.g. for (a, b) | c
7048 * it will return the list [a, c]
7050 * Returns the number of unique strings found, -1 in case of errors and
7051 * -2 if there is more than @len strings
7054 xmlExpGetStart(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7055 const xmlChar**tokList, int len) {
7056 if ((ctxt == NULL) || (exp == NULL) || (tokList == NULL) || (len <= 0))
7058 return(xmlExpGetStartInt(ctxt, exp, tokList, len, 0));
7063 * @exp: the expression
7065 * Finds if the expression is nillable, i.e. if it accepts the empty sequqnce
7067 * Returns 1 if nillable, 0 if not and -1 in case of error
7070 xmlExpIsNillable(xmlExpNodePtr exp) {
7073 return(IS_NILLABLE(exp) != 0);
7076 static xmlExpNodePtr
7077 xmlExpStringDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, const xmlChar *str)
7081 switch (exp->type) {
7083 return(forbiddenExp);
7084 case XML_EXP_FORBID:
7085 return(forbiddenExp);
7087 if (exp->exp_str == str) {
7089 printf("deriv atom: equal => Empty\n");
7094 printf("deriv atom: mismatch => forbid\n");
7096 /* TODO wildcards here */
7104 printf("deriv or: => or(derivs)\n");
7106 tmp = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7110 ret = xmlExpStringDeriveInt(ctxt, exp->exp_right, str);
7112 xmlExpFree(ctxt, tmp);
7115 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, tmp, ret,
7121 printf("deriv seq: starting with left\n");
7123 ret = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7126 } else if (ret == forbiddenExp) {
7127 if (IS_NILLABLE(exp->exp_left)) {
7129 printf("deriv seq: left failed but nillable\n");
7131 ret = xmlExpStringDeriveInt(ctxt, exp->exp_right, str);
7135 printf("deriv seq: left match => sequence\n");
7137 exp->exp_right->ref++;
7138 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, exp->exp_right,
7142 case XML_EXP_COUNT: {
7146 if (exp->exp_max == 0)
7147 return(forbiddenExp);
7148 ret = xmlExpStringDeriveInt(ctxt, exp->exp_left, str);
7151 if (ret == forbiddenExp) {
7153 printf("deriv count: pattern mismatch => forbid\n");
7157 if (exp->exp_max == 1)
7159 if (exp->exp_max < 0) /* unbounded */
7162 max = exp->exp_max - 1;
7163 if (exp->exp_min > 0)
7164 min = exp->exp_min - 1;
7167 exp->exp_left->ref++;
7168 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left, NULL,
7170 if (ret == emptyExp) {
7172 printf("deriv count: match to empty => new count\n");
7177 printf("deriv count: match => sequence with new count\n");
7179 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, tmp,
7187 * xmlExpStringDerive:
7188 * @ctxt: the expression context
7189 * @exp: the expression
7191 * @len: the string len in bytes if available
7193 * Do one step of Brzozowski derivation of the expression @exp with
7194 * respect to the input string
7196 * Returns the resulting expression or NULL in case of internal error
7199 xmlExpStringDerive(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7200 const xmlChar *str, int len) {
7201 const xmlChar *input;
7203 if ((exp == NULL) || (ctxt == NULL) || (str == NULL)) {
7207 * check the string is in the dictionnary, if yes use an interned
7208 * copy, otherwise we know it's not an acceptable input
7210 input = xmlDictExists(ctxt->dict, str, len);
7211 if (input == NULL) {
7212 return(forbiddenExp);
7214 return(xmlExpStringDeriveInt(ctxt, exp, input));
7218 xmlExpCheckCard(xmlExpNodePtr exp, xmlExpNodePtr sub) {
7221 if (sub->c_max == -1) {
7222 if (exp->c_max != -1)
7224 } else if ((exp->c_max >= 0) && (exp->c_max < sub->c_max)) {
7228 if ((IS_NILLABLE(sub)) && (!IS_NILLABLE(exp)))
7234 static xmlExpNodePtr xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp,
7238 * @ctxt: the expressions context
7239 * @exp: the englobing expression
7240 * @sub: the subexpression
7241 * @mult: the multiple expression
7242 * @remain: the remain from the derivation of the multiple
7244 * Check if exp is a multiple of sub, i.e. if there is a finite number n
7245 * so that sub{n} subsume exp
7247 * Returns the multiple value if successful, 0 if it is not a multiple
7248 * and -1 in case of internel error.
7252 xmlExpDivide(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub,
7253 xmlExpNodePtr *mult, xmlExpNodePtr *remain) {
7255 xmlExpNodePtr tmp, tmp2;
7257 if (mult != NULL) *mult = NULL;
7258 if (remain != NULL) *remain = NULL;
7259 if (exp->c_max == -1) return(0);
7260 if (IS_NILLABLE(exp) && (!IS_NILLABLE(sub))) return(0);
7262 for (i = 1;i <= exp->c_max;i++) {
7264 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT,
7265 sub, NULL, NULL, i, i);
7269 if (!xmlExpCheckCard(tmp, exp)) {
7270 xmlExpFree(ctxt, tmp);
7273 tmp2 = xmlExpExpDeriveInt(ctxt, tmp, exp);
7275 xmlExpFree(ctxt, tmp);
7278 if ((tmp2 != forbiddenExp) && (IS_NILLABLE(tmp2))) {
7282 xmlExpFree(ctxt, tmp2);
7286 xmlExpFree(ctxt, tmp);
7288 printf("Divide succeeded %d\n", i);
7292 xmlExpFree(ctxt, tmp);
7293 xmlExpFree(ctxt, tmp2);
7296 printf("Divide failed\n");
7302 * xmlExpExpDeriveInt:
7303 * @ctxt: the expressions context
7304 * @exp: the englobing expression
7305 * @sub: the subexpression
7307 * Try to do a step of Brzozowski derivation but at a higher level
7308 * the input being a subexpression.
7310 * Returns the resulting expression or NULL in case of internal error
7312 static xmlExpNodePtr
7313 xmlExpExpDeriveInt(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7314 xmlExpNodePtr ret, tmp, tmp2, tmp3;
7315 const xmlChar **tab;
7319 * In case of equality and if the expression can only consume a finite
7320 * amount, then the derivation is empty
7322 if ((exp == sub) && (exp->c_max >= 0)) {
7324 printf("Equal(exp, sub) and finite -> Empty\n");
7329 * decompose sub sequence first
7331 if (sub->type == XML_EXP_EMPTY) {
7333 printf("Empty(sub) -> Empty\n");
7338 if (sub->type == XML_EXP_SEQ) {
7340 printf("Seq(sub) -> decompose\n");
7342 tmp = xmlExpExpDeriveInt(ctxt, exp, sub->exp_left);
7345 if (tmp == forbiddenExp)
7347 ret = xmlExpExpDeriveInt(ctxt, tmp, sub->exp_right);
7348 xmlExpFree(ctxt, tmp);
7351 if (sub->type == XML_EXP_OR) {
7353 printf("Or(sub) -> decompose\n");
7355 tmp = xmlExpExpDeriveInt(ctxt, exp, sub->exp_left);
7356 if (tmp == forbiddenExp)
7360 ret = xmlExpExpDeriveInt(ctxt, exp, sub->exp_right);
7361 if ((ret == NULL) || (ret == forbiddenExp)) {
7362 xmlExpFree(ctxt, tmp);
7365 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, tmp, ret, NULL, 0, 0));
7367 if (!xmlExpCheckCard(exp, sub)) {
7369 printf("CheckCard(exp, sub) failed -> Forbid\n");
7371 return(forbiddenExp);
7373 switch (exp->type) {
7375 if (sub == emptyExp)
7378 printf("Empty(exp) -> Forbid\n");
7380 return(forbiddenExp);
7381 case XML_EXP_FORBID:
7383 printf("Forbid(exp) -> Forbid\n");
7385 return(forbiddenExp);
7387 if (sub->type == XML_EXP_ATOM) {
7388 /* TODO: handle wildcards */
7389 if (exp->exp_str == sub->exp_str) {
7391 printf("Atom match -> Empty\n");
7396 printf("Atom mismatch -> Forbid\n");
7398 return(forbiddenExp);
7400 if ((sub->type == XML_EXP_COUNT) &&
7401 (sub->exp_max == 1) &&
7402 (sub->exp_left->type == XML_EXP_ATOM)) {
7403 /* TODO: handle wildcards */
7404 if (exp->exp_str == sub->exp_left->exp_str) {
7406 printf("Atom match -> Empty\n");
7411 printf("Atom mismatch -> Forbid\n");
7413 return(forbiddenExp);
7416 printf("Compex exp vs Atom -> Forbid\n");
7418 return(forbiddenExp);
7420 /* try to get the sequence consumed only if possible */
7421 if (xmlExpCheckCard(exp->exp_left, sub)) {
7422 /* See if the sequence can be consumed directly */
7424 printf("Seq trying left only\n");
7426 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7427 if ((ret != forbiddenExp) && (ret != NULL)) {
7429 printf("Seq trying left only worked\n");
7432 * TODO: assumption here that we are determinist
7433 * i.e. we won't get to a nillable exp left
7434 * subset which could be matched by the right
7436 * e.g.: (a | b)+,(a | c) and 'a+,a'
7438 exp->exp_right->ref++;
7439 return(xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret,
7440 exp->exp_right, NULL, 0, 0));
7444 printf("Seq: left too short\n");
7447 /* Try instead to decompose */
7448 if (sub->type == XML_EXP_COUNT) {
7452 printf("Seq: sub is a count\n");
7454 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub->exp_left);
7457 if (ret != forbiddenExp) {
7459 printf("Seq , Count match on left\n");
7461 if (sub->exp_max < 0)
7464 max = sub->exp_max -1;
7465 if (sub->exp_min > 0)
7466 min = sub->exp_min -1;
7469 exp->exp_right->ref++;
7470 tmp = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret,
7471 exp->exp_right, NULL, 0, 0);
7475 sub->exp_left->ref++;
7476 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT,
7477 sub->exp_left, NULL, NULL, min, max);
7479 xmlExpFree(ctxt, tmp);
7482 ret = xmlExpExpDeriveInt(ctxt, tmp, tmp2);
7483 xmlExpFree(ctxt, tmp);
7484 xmlExpFree(ctxt, tmp2);
7488 /* we made no progress on structured operations */
7492 printf("Or , trying both side\n");
7494 ret = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7497 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_right, sub);
7499 xmlExpFree(ctxt, ret);
7502 return(xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, tmp, NULL, 0, 0));
7503 case XML_EXP_COUNT: {
7506 if (sub->type == XML_EXP_COUNT) {
7508 * Try to see if the loop is completely subsumed
7510 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub->exp_left);
7513 if (tmp == forbiddenExp) {
7517 printf("Count, Count inner don't subsume\n");
7519 mult = xmlExpDivide(ctxt, sub->exp_left, exp->exp_left,
7523 printf("Count, Count not multiple => forbidden\n");
7525 return(forbiddenExp);
7527 if (sub->exp_max == -1) {
7529 if (exp->exp_max == -1) {
7530 if (exp->exp_min <= sub->exp_min * mult)
7533 min = exp->exp_min - sub->exp_min * mult;
7536 printf("Count, Count finite can't subsume infinite\n");
7538 xmlExpFree(ctxt, tmp);
7539 return(forbiddenExp);
7542 if (exp->exp_max == -1) {
7544 printf("Infinite loop consume mult finite loop\n");
7546 if (exp->exp_min > sub->exp_min * mult) {
7548 min = exp->exp_min - sub->exp_min * mult;
7554 if (exp->exp_max < sub->exp_max * mult) {
7556 printf("loops max mult mismatch => forbidden\n");
7558 xmlExpFree(ctxt, tmp);
7559 return(forbiddenExp);
7561 if (sub->exp_max * mult > exp->exp_min)
7564 min = exp->exp_min - sub->exp_max * mult;
7565 max = exp->exp_max - sub->exp_max * mult;
7568 } else if (!IS_NILLABLE(tmp)) {
7570 * TODO: loop here to try to grow if working on finite
7574 printf("Count, Count remain not nillable => forbidden\n");
7576 xmlExpFree(ctxt, tmp);
7577 return(forbiddenExp);
7578 } else if (sub->exp_max == -1) {
7579 if (exp->exp_max == -1) {
7580 if (exp->exp_min <= sub->exp_min) {
7582 printf("Infinite loops Okay => COUNT(0,Inf)\n");
7588 printf("Infinite loops min => Count(X,Inf)\n");
7591 min = exp->exp_min - sub->exp_min;
7593 } else if (exp->exp_min > sub->exp_min) {
7595 printf("loops min mismatch 1 => forbidden ???\n");
7597 xmlExpFree(ctxt, tmp);
7598 return(forbiddenExp);
7604 if (exp->exp_max == -1) {
7606 printf("Infinite loop consume finite loop\n");
7608 if (exp->exp_min > sub->exp_min) {
7610 min = exp->exp_min - sub->exp_min;
7616 if (exp->exp_max < sub->exp_max) {
7618 printf("loops max mismatch => forbidden\n");
7620 xmlExpFree(ctxt, tmp);
7621 return(forbiddenExp);
7623 if (sub->exp_max > exp->exp_min)
7626 min = exp->exp_min - sub->exp_max;
7627 max = exp->exp_max - sub->exp_max;
7631 printf("loops match => SEQ(COUNT())\n");
7633 exp->exp_left->ref++;
7634 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left,
7635 NULL, NULL, min, max);
7639 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, tmp, tmp2,
7643 tmp = xmlExpExpDeriveInt(ctxt, exp->exp_left, sub);
7646 if (tmp == forbiddenExp) {
7648 printf("loop mismatch => forbidden\n");
7650 return(forbiddenExp);
7652 if (exp->exp_min > 0)
7653 min = exp->exp_min - 1;
7656 if (exp->exp_max < 0)
7659 max = exp->exp_max - 1;
7662 printf("loop match => SEQ(COUNT())\n");
7664 exp->exp_left->ref++;
7665 tmp2 = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, exp->exp_left,
7666 NULL, NULL, min, max);
7669 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, tmp, tmp2,
7676 printf("Fallback to derivative\n");
7678 if (IS_NILLABLE(sub)) {
7679 if (!(IS_NILLABLE(exp)))
7680 return(forbiddenExp);
7686 * here the structured derivation made no progress so
7687 * we use the default token based derivation to force one more step
7689 if (ctxt->tabSize == 0)
7692 tab = (const xmlChar **) xmlMalloc(ctxt->tabSize *
7693 sizeof(const xmlChar *));
7699 * collect all the strings accepted by the subexpression on input
7701 len = xmlExpGetStartInt(ctxt, sub, tab, ctxt->tabSize, 0);
7703 const xmlChar **temp;
7704 temp = (const xmlChar **) xmlRealloc((xmlChar **) tab, ctxt->tabSize * 2 *
7705 sizeof(const xmlChar *));
7707 xmlFree((xmlChar **) tab);
7712 len = xmlExpGetStartInt(ctxt, sub, tab, ctxt->tabSize, 0);
7714 for (i = 0;i < len;i++) {
7715 tmp = xmlExpStringDeriveInt(ctxt, exp, tab[i]);
7716 if ((tmp == NULL) || (tmp == forbiddenExp)) {
7717 xmlExpFree(ctxt, ret);
7718 xmlFree((xmlChar **) tab);
7721 tmp2 = xmlExpStringDeriveInt(ctxt, sub, tab[i]);
7722 if ((tmp2 == NULL) || (tmp2 == forbiddenExp)) {
7723 xmlExpFree(ctxt, tmp);
7724 xmlExpFree(ctxt, ret);
7725 xmlFree((xmlChar **) tab);
7728 tmp3 = xmlExpExpDeriveInt(ctxt, tmp, tmp2);
7729 xmlExpFree(ctxt, tmp);
7730 xmlExpFree(ctxt, tmp2);
7732 if ((tmp3 == NULL) || (tmp3 == forbiddenExp)) {
7733 xmlExpFree(ctxt, ret);
7734 xmlFree((xmlChar **) tab);
7741 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, tmp3, NULL, 0, 0);
7743 xmlFree((xmlChar **) tab);
7748 xmlFree((xmlChar **) tab);
7754 * @ctxt: the expressions context
7755 * @exp: the englobing expression
7756 * @sub: the subexpression
7758 * Evaluates the expression resulting from @exp consuming a sub expression @sub
7759 * Based on algebraic derivation and sometimes direct Brzozowski derivation
7760 * it usually tatkes less than linear time and can handle expressions generating
7761 * infinite languages.
7763 * Returns the resulting expression or NULL in case of internal error, the
7764 * result must be freed
7767 xmlExpExpDerive(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7768 if ((exp == NULL) || (ctxt == NULL) || (sub == NULL))
7774 if (IS_NILLABLE(sub) && (!IS_NILLABLE(exp))) {
7776 printf("Sub nillable and not exp : can't subsume\n");
7778 return(forbiddenExp);
7780 if (xmlExpCheckCard(exp, sub) == 0) {
7782 printf("sub generate longuer sequances than exp : can't subsume\n");
7784 return(forbiddenExp);
7786 return(xmlExpExpDeriveInt(ctxt, exp, sub));
7791 * @ctxt: the expressions context
7792 * @exp: the englobing expression
7793 * @sub: the subexpression
7795 * Check whether @exp accepts all the languages accexpted by @sub
7796 * the input being a subexpression.
7798 * Returns 1 if true 0 if false and -1 in case of failure.
7801 xmlExpSubsume(xmlExpCtxtPtr ctxt, xmlExpNodePtr exp, xmlExpNodePtr sub) {
7804 if ((exp == NULL) || (ctxt == NULL) || (sub == NULL))
7808 * TODO: speedup by checking the language of sub is a subset of the
7814 if (IS_NILLABLE(sub) && (!IS_NILLABLE(exp))) {
7816 printf("Sub nillable and not exp : can't subsume\n");
7820 if (xmlExpCheckCard(exp, sub) == 0) {
7822 printf("sub generate longuer sequances than exp : can't subsume\n");
7826 tmp = xmlExpExpDeriveInt(ctxt, exp, sub);
7828 printf("Result derivation :\n");
7833 if (tmp == forbiddenExp)
7835 if (tmp == emptyExp)
7837 if ((tmp != NULL) && (IS_NILLABLE(tmp))) {
7838 xmlExpFree(ctxt, tmp);
7841 xmlExpFree(ctxt, tmp);
7845 /************************************************************************
7847 * Parsing expression *
7849 ************************************************************************/
7851 static xmlExpNodePtr xmlExpParseExpr(xmlExpCtxtPtr ctxt);
7854 #define CUR (*ctxt->cur)
7856 #define NEXT ctxt->cur++;
7858 #define IS_BLANK(c) ((c == ' ') || (c == '\n') || (c == '\r') || (c == '\t'))
7859 #define SKIP_BLANKS while (IS_BLANK(*ctxt->cur)) ctxt->cur++;
7862 xmlExpParseNumber(xmlExpCtxtPtr ctxt) {
7870 if ((CUR < '0') || (CUR > '9'))
7872 while ((CUR >= '0') && (CUR <= '9')) {
7873 ret = ret * 10 + (CUR - '0');
7879 static xmlExpNodePtr
7880 xmlExpParseOr(xmlExpCtxtPtr ctxt) {
7887 if (*ctxt->cur == '(') {
7889 ret = xmlExpParseExpr(ctxt);
7891 if (*ctxt->cur != ')') {
7892 fprintf(stderr, "unbalanced '(' : %s\n", base);
7893 xmlExpFree(ctxt, ret);
7898 goto parse_quantifier;
7900 while ((CUR != 0) && (!(IS_BLANK(CUR))) && (CUR != '(') &&
7901 (CUR != ')') && (CUR != '|') && (CUR != ',') && (CUR != '{') &&
7902 (CUR != '*') && (CUR != '+') && (CUR != '?') && (CUR != '}'))
7904 val = xmlDictLookup(ctxt->dict, BAD_CAST base, ctxt->cur - base);
7907 ret = xmlExpHashGetEntry(ctxt, XML_EXP_ATOM, NULL, NULL, val, 0, 0);
7916 min = xmlExpParseNumber(ctxt);
7918 xmlExpFree(ctxt, ret);
7924 max = xmlExpParseNumber(ctxt);
7929 xmlExpFree(ctxt, ret);
7933 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7936 } else if (CUR == '?') {
7938 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7941 } else if (CUR == '+') {
7943 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7946 } else if (CUR == '*') {
7948 ret = xmlExpHashGetEntry(ctxt, XML_EXP_COUNT, ret, NULL, NULL,
7956 static xmlExpNodePtr
7957 xmlExpParseSeq(xmlExpCtxtPtr ctxt) {
7958 xmlExpNodePtr ret, right;
7960 ret = xmlExpParseOr(ctxt);
7962 while (CUR == '|') {
7964 right = xmlExpParseOr(ctxt);
7965 if (right == NULL) {
7966 xmlExpFree(ctxt, ret);
7969 ret = xmlExpHashGetEntry(ctxt, XML_EXP_OR, ret, right, NULL, 0, 0);
7976 static xmlExpNodePtr
7977 xmlExpParseExpr(xmlExpCtxtPtr ctxt) {
7978 xmlExpNodePtr ret, right;
7980 ret = xmlExpParseSeq(ctxt);
7982 while (CUR == ',') {
7984 right = xmlExpParseSeq(ctxt);
7985 if (right == NULL) {
7986 xmlExpFree(ctxt, ret);
7989 ret = xmlExpHashGetEntry(ctxt, XML_EXP_SEQ, ret, right, NULL, 0, 0);
7998 * @ctxt: the expressions context
7999 * @expr: the 0 terminated string
8001 * Minimal parser for regexps, it understand the following constructs
8002 * - string terminals
8003 * - choice operator |
8004 * - sequence operator ,
8005 * - subexpressions (...)
8006 * - usual cardinality operators + * and ?
8007 * - finite sequences { min, max }
8008 * - infinite sequences { min, * }
8009 * There is minimal checkings made especially no checking on strings values
8011 * Returns a new expression or NULL in case of failure
8014 xmlExpParse(xmlExpCtxtPtr ctxt, const char *expr) {
8020 ret = xmlExpParseExpr(ctxt);
8022 if (*ctxt->cur != 0) {
8023 xmlExpFree(ctxt, ret);
8030 xmlExpDumpInt(xmlBufferPtr buf, xmlExpNodePtr expr, int glob) {
8033 if (expr == NULL) return;
8034 if (glob) xmlBufferWriteChar(buf, "(");
8035 switch (expr->type) {
8037 xmlBufferWriteChar(buf, "empty");
8039 case XML_EXP_FORBID:
8040 xmlBufferWriteChar(buf, "forbidden");
8043 xmlBufferWriteCHAR(buf, expr->exp_str);
8047 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8048 xmlExpDumpInt(buf, c, 1);
8050 xmlExpDumpInt(buf, c, 0);
8051 xmlBufferWriteChar(buf, " , ");
8052 c = expr->exp_right;
8053 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8054 xmlExpDumpInt(buf, c, 1);
8056 xmlExpDumpInt(buf, c, 0);
8060 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8061 xmlExpDumpInt(buf, c, 1);
8063 xmlExpDumpInt(buf, c, 0);
8064 xmlBufferWriteChar(buf, " | ");
8065 c = expr->exp_right;
8066 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8067 xmlExpDumpInt(buf, c, 1);
8069 xmlExpDumpInt(buf, c, 0);
8071 case XML_EXP_COUNT: {
8075 if ((c->type == XML_EXP_SEQ) || (c->type == XML_EXP_OR))
8076 xmlExpDumpInt(buf, c, 1);
8078 xmlExpDumpInt(buf, c, 0);
8079 if ((expr->exp_min == 0) && (expr->exp_max == 1)) {
8082 } else if ((expr->exp_min == 0) && (expr->exp_max == -1)) {
8085 } else if ((expr->exp_min == 1) && (expr->exp_max == -1)) {
8088 } else if (expr->exp_max == expr->exp_min) {
8089 snprintf(rep, 39, "{%d}", expr->exp_min);
8090 } else if (expr->exp_max < 0) {
8091 snprintf(rep, 39, "{%d,inf}", expr->exp_min);
8093 snprintf(rep, 39, "{%d,%d}", expr->exp_min, expr->exp_max);
8096 xmlBufferWriteChar(buf, rep);
8100 fprintf(stderr, "Error in tree\n");
8103 xmlBufferWriteChar(buf, ")");
8107 * @buf: a buffer to receive the output
8108 * @expr: the compiled expression
8110 * Serialize the expression as compiled to the buffer
8113 xmlExpDump(xmlBufferPtr buf, xmlExpNodePtr expr) {
8114 if ((buf == NULL) || (expr == NULL))
8116 xmlExpDumpInt(buf, expr, 0);
8121 * @expr: a compiled expression
8123 * Indicate the maximum number of input a expression can accept
8125 * Returns the maximum length or -1 in case of error
8128 xmlExpMaxToken(xmlExpNodePtr expr) {
8131 return(expr->c_max);
8135 * xmlExpCtxtNbNodes:
8136 * @ctxt: an expression context
8138 * Debugging facility provides the number of allocated nodes at a that point
8140 * Returns the number of nodes in use or -1 in case of error
8143 xmlExpCtxtNbNodes(xmlExpCtxtPtr ctxt) {
8146 return(ctxt->nb_nodes);
8151 * @ctxt: an expression context
8153 * Debugging facility provides the number of allocated nodes over lifetime
8155 * Returns the number of nodes ever allocated or -1 in case of error
8158 xmlExpCtxtNbCons(xmlExpCtxtPtr ctxt) {
8161 return(ctxt->nb_cons);
8164 #endif /* LIBXML_EXPR_ENABLED */
8165 #define bottom_xmlregexp
8166 #include "elfgcchack.h"
8167 #endif /* LIBXML_REGEXP_ENABLED */