Coverage Report

Created: 2025-07-07 10:01

/work/workdir/UnpackedTarball/libxml2/parser.c
Line
Count
Source (jump to first uncovered line)
1
/*
2
 * parser.c : an XML 1.0 parser, namespaces and validity support are mostly
3
 *            implemented on top of the SAX interfaces
4
 *
5
 * References:
6
 *   The XML specification:
7
 *     http://www.w3.org/TR/REC-xml
8
 *   Original 1.0 version:
9
 *     http://www.w3.org/TR/1998/REC-xml-19980210
10
 *   XML second edition working draft
11
 *     http://www.w3.org/TR/2000/WD-xml-2e-20000814
12
 *
13
 * Okay this is a big file, the parser core is around 7000 lines, then it
14
 * is followed by the progressive parser top routines, then the various
15
 * high level APIs to call the parser and a few miscellaneous functions.
16
 * A number of helper functions and deprecated ones have been moved to
17
 * parserInternals.c to reduce this file size.
18
 * As much as possible the functions are associated with their relative
19
 * production in the XML specification. A few productions defining the
20
 * different ranges of character are actually implanted either in
21
 * parserInternals.h or parserInternals.c
22
 * The DOM tree build is realized from the default SAX callbacks in
23
 * the module SAX2.c.
24
 * The routines doing the validation checks are in valid.c and called either
25
 * from the SAX callbacks or as standalone functions using a preparsed
26
 * document.
27
 *
28
 * See Copyright for the status of this software.
29
 *
30
 * daniel@veillard.com
31
 */
32
33
/* To avoid EBCDIC trouble when parsing on zOS */
34
#if defined(__MVS__)
35
#pragma convert("ISO8859-1")
36
#endif
37
38
#define IN_LIBXML
39
#include "libxml.h"
40
41
#if defined(_WIN32)
42
#define XML_DIR_SEP '\\'
43
#else
44
#define XML_DIR_SEP '/'
45
#endif
46
47
#include <stdlib.h>
48
#include <limits.h>
49
#include <string.h>
50
#include <stdarg.h>
51
#include <stddef.h>
52
#include <ctype.h>
53
#include <stdlib.h>
54
#include <libxml/parser.h>
55
#include <libxml/xmlmemory.h>
56
#include <libxml/tree.h>
57
#include <libxml/parserInternals.h>
58
#include <libxml/valid.h>
59
#include <libxml/entities.h>
60
#include <libxml/xmlerror.h>
61
#include <libxml/encoding.h>
62
#include <libxml/xmlIO.h>
63
#include <libxml/uri.h>
64
#include <libxml/SAX2.h>
65
#include <libxml/HTMLparser.h>
66
#ifdef LIBXML_CATALOG_ENABLED
67
#include <libxml/catalog.h>
68
#endif
69
70
#include "private/buf.h"
71
#include "private/dict.h"
72
#include "private/entities.h"
73
#include "private/error.h"
74
#include "private/html.h"
75
#include "private/io.h"
76
#include "private/memory.h"
77
#include "private/parser.h"
78
79
38.1M
#define NS_INDEX_EMPTY  INT_MAX
80
7.67M
#define NS_INDEX_XML    (INT_MAX - 1)
81
13.2M
#define URI_HASH_EMPTY  0xD943A04E
82
65.2k
#define URI_HASH_XML    0xF0451F02
83
84
#ifndef STDIN_FILENO
85
0
  #define STDIN_FILENO 0
86
#endif
87
88
#ifndef SIZE_MAX
89
  #define SIZE_MAX ((size_t) -1)
90
#endif
91
92
1.11M
#define XML_MAX_ATTRS 100000000 /* 100 million */
93
94
struct _xmlStartTag {
95
    const xmlChar *prefix;
96
    const xmlChar *URI;
97
    int line;
98
    int nsNr;
99
};
100
101
typedef struct {
102
    void *saxData;
103
    unsigned prefixHashValue;
104
    unsigned uriHashValue;
105
    unsigned elementId;
106
    int oldIndex;
107
} xmlParserNsExtra;
108
109
typedef struct {
110
    unsigned hashValue;
111
    int index;
112
} xmlParserNsBucket;
113
114
struct _xmlParserNsData {
115
    xmlParserNsExtra *extra;
116
117
    unsigned hashSize;
118
    unsigned hashElems;
119
    xmlParserNsBucket *hash;
120
121
    unsigned elementId;
122
    int defaultNsIndex;
123
    int minNsIndex;
124
};
125
126
static int
127
xmlParseElementStart(xmlParserCtxtPtr ctxt);
128
129
static void
130
xmlParseElementEnd(xmlParserCtxtPtr ctxt);
131
132
static xmlEntityPtr
133
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr);
134
135
static const xmlChar *
136
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt);
137
138
/************************************************************************
139
 *                  *
140
 *  Arbitrary limits set in the parser. See XML_PARSE_HUGE    *
141
 *                  *
142
 ************************************************************************/
143
144
#define XML_PARSER_BIG_ENTITY 1000
145
#define XML_PARSER_LOT_ENTITY 5000
146
147
/*
148
 * Constants for protection against abusive entity expansion
149
 * ("billion laughs").
150
 */
151
152
/*
153
 * A certain amount of entity expansion which is always allowed.
154
 */
155
638k
#define XML_PARSER_ALLOWED_EXPANSION 1000000
156
157
/*
158
 * Fixed cost for each entity reference. This crudely models processing time
159
 * as well to protect, for example, against exponential expansion of empty
160
 * or very short entities.
161
 */
162
638k
#define XML_ENT_FIXED_COST 20
163
164
76.7M
#define XML_PARSER_BIG_BUFFER_SIZE 300
165
234k
#define XML_PARSER_BUFFER_SIZE 100
166
29.2k
#define SAX_COMPAT_MODE BAD_CAST "SAX compatibility mode document"
167
168
/**
169
 * XML_PARSER_CHUNK_SIZE
170
 *
171
 * When calling GROW that's the minimal amount of data
172
 * the parser expected to have received. It is not a hard
173
 * limit but an optimization when reading strings like Names
174
 * It is not strictly needed as long as inputs available characters
175
 * are followed by 0, which should be provided by the I/O level
176
 */
177
#define XML_PARSER_CHUNK_SIZE 100
178
179
/**
180
 * xmlParserVersion:
181
 *
182
 * Constant string describing the internal version of the library
183
 */
184
const char *const
185
xmlParserVersion = LIBXML_VERSION_STRING LIBXML_VERSION_EXTRA;
186
187
/*
188
 * List of XML prefixed PI allowed by W3C specs
189
 */
190
191
static const char* const xmlW3CPIs[] = {
192
    "xml-stylesheet",
193
    "xml-model",
194
    NULL
195
};
196
197
198
/* DEPR void xmlParserHandleReference(xmlParserCtxtPtr ctxt); */
199
static xmlEntityPtr xmlParseStringPEReference(xmlParserCtxtPtr ctxt,
200
                                              const xmlChar **str);
201
202
static void
203
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent);
204
205
static int
206
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity);
207
208
/************************************************************************
209
 *                  *
210
 *    Some factorized error routines        *
211
 *                  *
212
 ************************************************************************/
213
214
static void
215
0
xmlErrMemory(xmlParserCtxtPtr ctxt) {
216
0
    xmlCtxtErrMemory(ctxt);
217
0
}
218
219
/**
220
 * xmlErrAttributeDup:
221
 * @ctxt:  an XML parser context
222
 * @prefix:  the attribute prefix
223
 * @localname:  the attribute localname
224
 *
225
 * Handle a redefinition of attribute error
226
 */
227
static void
228
xmlErrAttributeDup(xmlParserCtxtPtr ctxt, const xmlChar * prefix,
229
                   const xmlChar * localname)
230
192k
{
231
192k
    if (prefix == NULL)
232
107k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
233
107k
                   XML_ERR_FATAL, localname, NULL, NULL, 0,
234
107k
                   "Attribute %s redefined\n", localname);
235
84.5k
    else
236
84.5k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
237
84.5k
                   XML_ERR_FATAL, prefix, localname, NULL, 0,
238
84.5k
                   "Attribute %s:%s redefined\n", prefix, localname);
239
192k
}
240
241
/**
242
 * xmlFatalErrMsg:
243
 * @ctxt:  an XML parser context
244
 * @error:  the error number
245
 * @msg:  the error message
246
 *
247
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
248
 */
249
static void LIBXML_ATTR_FORMAT(3,0)
250
xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
251
               const char *msg)
252
1.94M
{
253
1.94M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
254
1.94M
               NULL, NULL, NULL, 0, "%s", msg);
255
1.94M
}
256
257
/**
258
 * xmlWarningMsg:
259
 * @ctxt:  an XML parser context
260
 * @error:  the error number
261
 * @msg:  the error message
262
 * @str1:  extra data
263
 * @str2:  extra data
264
 *
265
 * Handle a warning.
266
 */
267
void LIBXML_ATTR_FORMAT(3,0)
268
xmlWarningMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
269
              const char *msg, const xmlChar *str1, const xmlChar *str2)
270
28.6k
{
271
28.6k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING,
272
28.6k
               str1, str2, NULL, 0, msg, str1, str2);
273
28.6k
}
274
275
/**
276
 * xmlValidityError:
277
 * @ctxt:  an XML parser context
278
 * @error:  the error number
279
 * @msg:  the error message
280
 * @str1:  extra data
281
 *
282
 * Handle a validity error.
283
 */
284
static void LIBXML_ATTR_FORMAT(3,0)
285
xmlValidityError(xmlParserCtxtPtr ctxt, xmlParserErrors error,
286
              const char *msg, const xmlChar *str1, const xmlChar *str2)
287
0
{
288
0
    ctxt->valid = 0;
289
290
0
    xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR,
291
0
               str1, str2, NULL, 0, msg, str1, str2);
292
0
}
293
294
/**
295
 * xmlFatalErrMsgInt:
296
 * @ctxt:  an XML parser context
297
 * @error:  the error number
298
 * @msg:  the error message
299
 * @val:  an integer value
300
 *
301
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
302
 */
303
static void LIBXML_ATTR_FORMAT(3,0)
304
xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error,
305
                  const char *msg, int val)
306
9.65k
{
307
9.65k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
308
9.65k
               NULL, NULL, NULL, val, msg, val);
309
9.65k
}
310
311
/**
312
 * xmlFatalErrMsgStrIntStr:
313
 * @ctxt:  an XML parser context
314
 * @error:  the error number
315
 * @msg:  the error message
316
 * @str1:  an string info
317
 * @val:  an integer value
318
 * @str2:  an string info
319
 *
320
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
321
 */
322
static void LIBXML_ATTR_FORMAT(3,0)
323
xmlFatalErrMsgStrIntStr(xmlParserCtxtPtr ctxt, xmlParserErrors error,
324
                  const char *msg, const xmlChar *str1, int val,
325
      const xmlChar *str2)
326
72.2k
{
327
72.2k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
328
72.2k
               str1, str2, NULL, val, msg, str1, val, str2);
329
72.2k
}
330
331
/**
332
 * xmlFatalErrMsgStr:
333
 * @ctxt:  an XML parser context
334
 * @error:  the error number
335
 * @msg:  the error message
336
 * @val:  a string value
337
 *
338
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
339
 */
340
static void LIBXML_ATTR_FORMAT(3,0)
341
xmlFatalErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error,
342
                  const char *msg, const xmlChar * val)
343
151k
{
344
151k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
345
151k
               val, NULL, NULL, 0, msg, val);
346
151k
}
347
348
/**
349
 * xmlErrMsgStr:
350
 * @ctxt:  an XML parser context
351
 * @error:  the error number
352
 * @msg:  the error message
353
 * @val:  a string value
354
 *
355
 * Handle a non fatal parser error
356
 */
357
static void LIBXML_ATTR_FORMAT(3,0)
358
xmlErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error,
359
                  const char *msg, const xmlChar * val)
360
13.8k
{
361
13.8k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR,
362
13.8k
               val, NULL, NULL, 0, msg, val);
363
13.8k
}
364
365
/**
366
 * xmlNsErr:
367
 * @ctxt:  an XML parser context
368
 * @error:  the error number
369
 * @msg:  the message
370
 * @info1:  extra information string
371
 * @info2:  extra information string
372
 *
373
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
374
 */
375
static void LIBXML_ATTR_FORMAT(3,0)
376
xmlNsErr(xmlParserCtxtPtr ctxt, xmlParserErrors error,
377
         const char *msg,
378
         const xmlChar * info1, const xmlChar * info2,
379
         const xmlChar * info3)
380
1.79M
{
381
1.79M
    ctxt->nsWellFormed = 0;
382
383
1.79M
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR,
384
1.79M
               info1, info2, info3, 0, msg, info1, info2, info3);
385
1.79M
}
386
387
/**
388
 * xmlNsWarn
389
 * @ctxt:  an XML parser context
390
 * @error:  the error number
391
 * @msg:  the message
392
 * @info1:  extra information string
393
 * @info2:  extra information string
394
 *
395
 * Handle a namespace warning error
396
 */
397
static void LIBXML_ATTR_FORMAT(3,0)
398
xmlNsWarn(xmlParserCtxtPtr ctxt, xmlParserErrors error,
399
         const char *msg,
400
         const xmlChar * info1, const xmlChar * info2,
401
         const xmlChar * info3)
402
71.3k
{
403
71.3k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING,
404
71.3k
               info1, info2, info3, 0, msg, info1, info2, info3);
405
71.3k
}
406
407
static void
408
1.91M
xmlSaturatedAdd(unsigned long *dst, unsigned long val) {
409
1.91M
    if (val > ULONG_MAX - *dst)
410
0
        *dst = ULONG_MAX;
411
1.91M
    else
412
1.91M
        *dst += val;
413
1.91M
}
414
415
static void
416
638k
xmlSaturatedAddSizeT(unsigned long *dst, unsigned long val) {
417
638k
    if (val > ULONG_MAX - *dst)
418
0
        *dst = ULONG_MAX;
419
638k
    else
420
638k
        *dst += val;
421
638k
}
422
423
/**
424
 * xmlParserEntityCheck:
425
 * @ctxt:  parser context
426
 * @extra:  sum of unexpanded entity sizes
427
 *
428
 * Check for non-linear entity expansion behaviour.
429
 *
430
 * In some cases like xmlExpandEntityInAttValue, this function is called
431
 * for each, possibly nested entity and its unexpanded content length.
432
 *
433
 * In other cases like xmlParseReference, it's only called for each
434
 * top-level entity with its unexpanded content length plus the sum of
435
 * the unexpanded content lengths (plus fixed cost) of all nested
436
 * entities.
437
 *
438
 * Summing the unexpanded lengths also adds the length of the reference.
439
 * This is by design. Taking the length of the entity name into account
440
 * discourages attacks that try to waste CPU time with abusively long
441
 * entity names. See test/recurse/lol6.xml for example. Each call also
442
 * adds some fixed cost XML_ENT_FIXED_COST to discourage attacks with
443
 * short entities.
444
 *
445
 * Returns 1 on error, 0 on success.
446
 */
447
static int
448
xmlParserEntityCheck(xmlParserCtxtPtr ctxt, unsigned long extra)
449
638k
{
450
638k
    unsigned long consumed;
451
638k
    unsigned long *expandedSize;
452
638k
    xmlParserInputPtr input = ctxt->input;
453
638k
    xmlEntityPtr entity = input->entity;
454
455
638k
    if ((entity) && (entity->flags & XML_ENT_CHECKED))
456
0
        return(0);
457
458
    /*
459
     * Compute total consumed bytes so far, including input streams of
460
     * external entities.
461
     */
462
638k
    consumed = input->consumed;
463
638k
    xmlSaturatedAddSizeT(&consumed, input->cur - input->base);
464
638k
    xmlSaturatedAdd(&consumed, ctxt->sizeentities);
465
466
638k
    if (entity)
467
0
        expandedSize = &entity->expandedSize;
468
638k
    else
469
638k
        expandedSize = &ctxt->sizeentcopy;
470
471
    /*
472
     * Add extra cost and some fixed cost.
473
     */
474
638k
    xmlSaturatedAdd(expandedSize, extra);
475
638k
    xmlSaturatedAdd(expandedSize, XML_ENT_FIXED_COST);
476
477
    /*
478
     * It's important to always use saturation arithmetic when tracking
479
     * entity sizes to make the size checks reliable. If "sizeentcopy"
480
     * overflows, we have to abort.
481
     */
482
638k
    if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) &&
483
638k
        ((*expandedSize >= ULONG_MAX) ||
484
109
         (*expandedSize / ctxt->maxAmpl > consumed))) {
485
109
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
486
109
                       "Maximum entity amplification factor exceeded, see "
487
109
                       "xmlCtxtSetMaxAmplification.\n");
488
109
        xmlHaltParser(ctxt);
489
109
        return(1);
490
109
    }
491
492
638k
    return(0);
493
638k
}
494
495
/************************************************************************
496
 *                  *
497
 *    Library wide options          *
498
 *                  *
499
 ************************************************************************/
500
501
/**
502
  * xmlHasFeature:
503
  * @feature: the feature to be examined
504
  *
505
  * Examines if the library has been compiled with a given feature.
506
  *
507
  * Returns a non-zero value if the feature exist, otherwise zero.
508
  * Returns zero (0) if the feature does not exist or an unknown
509
  * unknown feature is requested, non-zero otherwise.
510
  */
511
int
512
xmlHasFeature(xmlFeature feature)
513
0
{
514
0
    switch (feature) {
515
0
  case XML_WITH_THREAD:
516
0
#ifdef LIBXML_THREAD_ENABLED
517
0
      return(1);
518
#else
519
      return(0);
520
#endif
521
0
        case XML_WITH_TREE:
522
0
            return(1);
523
0
        case XML_WITH_OUTPUT:
524
0
#ifdef LIBXML_OUTPUT_ENABLED
525
0
            return(1);
526
#else
527
            return(0);
528
#endif
529
0
        case XML_WITH_PUSH:
530
0
#ifdef LIBXML_PUSH_ENABLED
531
0
            return(1);
532
#else
533
            return(0);
534
#endif
535
0
        case XML_WITH_READER:
536
0
#ifdef LIBXML_READER_ENABLED
537
0
            return(1);
538
#else
539
            return(0);
540
#endif
541
0
        case XML_WITH_PATTERN:
542
0
#ifdef LIBXML_PATTERN_ENABLED
543
0
            return(1);
544
#else
545
            return(0);
546
#endif
547
0
        case XML_WITH_WRITER:
548
0
#ifdef LIBXML_WRITER_ENABLED
549
0
            return(1);
550
#else
551
            return(0);
552
#endif
553
0
        case XML_WITH_SAX1:
554
0
#ifdef LIBXML_SAX1_ENABLED
555
0
            return(1);
556
#else
557
            return(0);
558
#endif
559
0
        case XML_WITH_HTTP:
560
#ifdef LIBXML_HTTP_ENABLED
561
            return(1);
562
#else
563
0
            return(0);
564
0
#endif
565
0
        case XML_WITH_VALID:
566
0
#ifdef LIBXML_VALID_ENABLED
567
0
            return(1);
568
#else
569
            return(0);
570
#endif
571
0
        case XML_WITH_HTML:
572
0
#ifdef LIBXML_HTML_ENABLED
573
0
            return(1);
574
#else
575
            return(0);
576
#endif
577
0
        case XML_WITH_LEGACY:
578
0
            return(0);
579
0
        case XML_WITH_C14N:
580
0
#ifdef LIBXML_C14N_ENABLED
581
0
            return(1);
582
#else
583
            return(0);
584
#endif
585
0
        case XML_WITH_CATALOG:
586
0
#ifdef LIBXML_CATALOG_ENABLED
587
0
            return(1);
588
#else
589
            return(0);
590
#endif
591
0
        case XML_WITH_XPATH:
592
0
#ifdef LIBXML_XPATH_ENABLED
593
0
            return(1);
594
#else
595
            return(0);
596
#endif
597
0
        case XML_WITH_XPTR:
598
0
#ifdef LIBXML_XPTR_ENABLED
599
0
            return(1);
600
#else
601
            return(0);
602
#endif
603
0
        case XML_WITH_XINCLUDE:
604
0
#ifdef LIBXML_XINCLUDE_ENABLED
605
0
            return(1);
606
#else
607
            return(0);
608
#endif
609
0
        case XML_WITH_ICONV:
610
#ifdef LIBXML_ICONV_ENABLED
611
            return(1);
612
#else
613
0
            return(0);
614
0
#endif
615
0
        case XML_WITH_ISO8859X:
616
0
#ifdef LIBXML_ISO8859X_ENABLED
617
0
            return(1);
618
#else
619
            return(0);
620
#endif
621
0
        case XML_WITH_UNICODE:
622
0
            return(0);
623
0
        case XML_WITH_REGEXP:
624
0
#ifdef LIBXML_REGEXP_ENABLED
625
0
            return(1);
626
#else
627
            return(0);
628
#endif
629
0
        case XML_WITH_AUTOMATA:
630
0
#ifdef LIBXML_REGEXP_ENABLED
631
0
            return(1);
632
#else
633
            return(0);
634
#endif
635
0
        case XML_WITH_EXPR:
636
#ifdef LIBXML_EXPR_ENABLED
637
            return(1);
638
#else
639
0
            return(0);
640
0
#endif
641
0
        case XML_WITH_RELAXNG:
642
0
#ifdef LIBXML_RELAXNG_ENABLED
643
0
            return(1);
644
#else
645
            return(0);
646
#endif
647
0
        case XML_WITH_SCHEMAS:
648
0
#ifdef LIBXML_SCHEMAS_ENABLED
649
0
            return(1);
650
#else
651
            return(0);
652
#endif
653
0
        case XML_WITH_SCHEMATRON:
654
0
#ifdef LIBXML_SCHEMATRON_ENABLED
655
0
            return(1);
656
#else
657
            return(0);
658
#endif
659
0
        case XML_WITH_MODULES:
660
0
#ifdef LIBXML_MODULES_ENABLED
661
0
            return(1);
662
#else
663
            return(0);
664
#endif
665
0
        case XML_WITH_DEBUG:
666
0
#ifdef LIBXML_DEBUG_ENABLED
667
0
            return(1);
668
#else
669
            return(0);
670
#endif
671
0
        case XML_WITH_DEBUG_MEM:
672
0
            return(0);
673
0
        case XML_WITH_ZLIB:
674
#ifdef LIBXML_ZLIB_ENABLED
675
            return(1);
676
#else
677
0
            return(0);
678
0
#endif
679
0
        case XML_WITH_LZMA:
680
#ifdef LIBXML_LZMA_ENABLED
681
            return(1);
682
#else
683
0
            return(0);
684
0
#endif
685
0
        case XML_WITH_ICU:
686
#ifdef LIBXML_ICU_ENABLED
687
            return(1);
688
#else
689
0
            return(0);
690
0
#endif
691
0
        default:
692
0
      break;
693
0
     }
694
0
     return(0);
695
0
}
696
697
/************************************************************************
698
 *                  *
699
 *      Simple string buffer        *
700
 *                  *
701
 ************************************************************************/
702
703
typedef struct {
704
    xmlChar *mem;
705
    unsigned size;
706
    unsigned cap; /* size < cap */
707
    unsigned max; /* size <= max */
708
    xmlParserErrors code;
709
} xmlSBuf;
710
711
static void
712
28.1M
xmlSBufInit(xmlSBuf *buf, unsigned max) {
713
28.1M
    buf->mem = NULL;
714
28.1M
    buf->size = 0;
715
28.1M
    buf->cap = 0;
716
28.1M
    buf->max = max;
717
28.1M
    buf->code = XML_ERR_OK;
718
28.1M
}
719
720
static int
721
305k
xmlSBufGrow(xmlSBuf *buf, unsigned len) {
722
305k
    xmlChar *mem;
723
305k
    unsigned cap;
724
725
305k
    if (len >= UINT_MAX / 2 - buf->size) {
726
0
        if (buf->code == XML_ERR_OK)
727
0
            buf->code = XML_ERR_RESOURCE_LIMIT;
728
0
        return(-1);
729
0
    }
730
731
305k
    cap = (buf->size + len) * 2;
732
305k
    if (cap < 240)
733
256k
        cap = 240;
734
735
305k
    mem = xmlRealloc(buf->mem, cap);
736
305k
    if (mem == NULL) {
737
0
        buf->code = XML_ERR_NO_MEMORY;
738
0
        return(-1);
739
0
    }
740
741
305k
    buf->mem = mem;
742
305k
    buf->cap = cap;
743
744
305k
    return(0);
745
305k
}
746
747
static void
748
6.65M
xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) {
749
6.65M
    if (buf->max - buf->size < len) {
750
0
        if (buf->code == XML_ERR_OK)
751
0
            buf->code = XML_ERR_RESOURCE_LIMIT;
752
0
        return;
753
0
    }
754
755
6.65M
    if (buf->cap - buf->size <= len) {
756
301k
        if (xmlSBufGrow(buf, len) < 0)
757
0
            return;
758
301k
    }
759
760
6.65M
    if (len > 0)
761
6.65M
        memcpy(buf->mem + buf->size, str, len);
762
6.65M
    buf->size += len;
763
6.65M
}
764
765
static void
766
5.43M
xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) {
767
5.43M
    xmlSBufAddString(buf, (const xmlChar *) str, len);
768
5.43M
}
769
770
static void
771
59.8k
xmlSBufAddChar(xmlSBuf *buf, int c) {
772
59.8k
    xmlChar *end;
773
774
59.8k
    if (buf->max - buf->size < 4) {
775
0
        if (buf->code == XML_ERR_OK)
776
0
            buf->code = XML_ERR_RESOURCE_LIMIT;
777
0
        return;
778
0
    }
779
780
59.8k
    if (buf->cap - buf->size <= 4) {
781
3.67k
        if (xmlSBufGrow(buf, 4) < 0)
782
0
            return;
783
3.67k
    }
784
785
59.8k
    end = buf->mem + buf->size;
786
787
59.8k
    if (c < 0x80) {
788
19.7k
        *end = (xmlChar) c;
789
19.7k
        buf->size += 1;
790
40.1k
    } else {
791
40.1k
        buf->size += xmlCopyCharMultiByte(end, c);
792
40.1k
    }
793
59.8k
}
794
795
static void
796
4.37M
xmlSBufAddReplChar(xmlSBuf *buf) {
797
4.37M
    xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3);
798
4.37M
}
799
800
static void
801
0
xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
802
0
    if (buf->code == XML_ERR_NO_MEMORY)
803
0
        xmlCtxtErrMemory(ctxt);
804
0
    else
805
0
        xmlFatalErr(ctxt, buf->code, errMsg);
806
0
}
807
808
static xmlChar *
809
xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt,
810
279k
              const char *errMsg) {
811
279k
    if (buf->mem == NULL) {
812
17.4k
        buf->mem = xmlMalloc(1);
813
17.4k
        if (buf->mem == NULL) {
814
0
            buf->code = XML_ERR_NO_MEMORY;
815
17.4k
        } else {
816
17.4k
            buf->mem[0] = 0;
817
17.4k
        }
818
261k
    } else {
819
261k
        buf->mem[buf->size] = 0;
820
261k
    }
821
822
279k
    if (buf->code == XML_ERR_OK) {
823
279k
        if (sizeOut != NULL)
824
222k
            *sizeOut = buf->size;
825
279k
        return(buf->mem);
826
279k
    }
827
828
0
    xmlSBufReportError(buf, ctxt, errMsg);
829
830
0
    xmlFree(buf->mem);
831
832
0
    if (sizeOut != NULL)
833
0
        *sizeOut = 0;
834
0
    return(NULL);
835
279k
}
836
837
static void
838
27.8M
xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
839
27.8M
    if (buf->code != XML_ERR_OK)
840
0
        xmlSBufReportError(buf, ctxt, errMsg);
841
842
27.8M
    xmlFree(buf->mem);
843
27.8M
}
844
845
static int
846
xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str,
847
6.46M
                    const char *errMsg) {
848
6.46M
    int c = str[0];
849
6.46M
    int c1 = str[1];
850
851
6.46M
    if ((c1 & 0xC0) != 0x80)
852
1.88M
        goto encoding_error;
853
854
4.58M
    if (c < 0xE0) {
855
        /* 2-byte sequence */
856
1.71M
        if (c < 0xC2)
857
936k
            goto encoding_error;
858
859
779k
        return(2);
860
2.86M
    } else {
861
2.86M
        int c2 = str[2];
862
863
2.86M
        if ((c2 & 0xC0) != 0x80)
864
21.2k
            goto encoding_error;
865
866
2.84M
        if (c < 0xF0) {
867
            /* 3-byte sequence */
868
2.72M
            if (c == 0xE0) {
869
                /* overlong */
870
154k
                if (c1 < 0xA0)
871
1.13k
                    goto encoding_error;
872
2.56M
            } else if (c == 0xED) {
873
                /* surrogate */
874
6.15k
                if (c1 >= 0xA0)
875
1.11k
                    goto encoding_error;
876
2.56M
            } else if (c == 0xEF) {
877
                /* U+FFFE and U+FFFF are invalid Chars */
878
1.15M
                if ((c1 == 0xBF) && (c2 >= 0xBE))
879
16.1k
                    xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg);
880
1.15M
            }
881
882
2.72M
            return(3);
883
2.72M
        } else {
884
            /* 4-byte sequence */
885
120k
            if ((str[3] & 0xC0) != 0x80)
886
8.57k
                goto encoding_error;
887
112k
            if (c == 0xF0) {
888
                /* overlong */
889
5.72k
                if (c1 < 0x90)
890
1.99k
                    goto encoding_error;
891
106k
            } else if (c >= 0xF4) {
892
                /* greater than 0x10FFFF */
893
14.4k
                if ((c > 0xF4) || (c1 >= 0x90))
894
4.40k
                    goto encoding_error;
895
14.4k
            }
896
897
105k
            return(4);
898
112k
        }
899
2.84M
    }
900
901
2.85M
encoding_error:
902
    /* Only report the first error */
903
2.85M
    if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) {
904
5.85k
        xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL);
905
5.85k
        ctxt->input->flags |= XML_INPUT_ENCODING_ERROR;
906
5.85k
    }
907
908
2.85M
    return(0);
909
4.58M
}
910
911
/************************************************************************
912
 *                  *
913
 *    SAX2 defaulted attributes handling      *
914
 *                  *
915
 ************************************************************************/
916
917
/**
918
 * xmlCtxtInitializeLate:
919
 * @ctxt:  an XML parser context
920
 *
921
 * Final initialization of the parser context before starting to parse.
922
 *
923
 * This accounts for users modifying struct members of parser context
924
 * directly.
925
 */
926
static void
927
272k
xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) {
928
272k
    xmlSAXHandlerPtr sax;
929
930
    /* Avoid unused variable warning if features are disabled. */
931
272k
    (void) sax;
932
933
    /*
934
     * Changing the SAX struct directly is still widespread practice
935
     * in internal and external code.
936
     */
937
272k
    if (ctxt == NULL) return;
938
272k
    sax = ctxt->sax;
939
272k
#ifdef LIBXML_SAX1_ENABLED
940
    /*
941
     * Only enable SAX2 if there SAX2 element handlers, except when there
942
     * are no element handlers at all.
943
     */
944
272k
    if (((ctxt->options & XML_PARSE_SAX1) == 0) &&
945
272k
        (sax) &&
946
272k
        (sax->initialized == XML_SAX2_MAGIC) &&
947
272k
        ((sax->startElementNs != NULL) ||
948
272k
         (sax->endElementNs != NULL) ||
949
272k
         ((sax->startElement == NULL) && (sax->endElement == NULL))))
950
272k
        ctxt->sax2 = 1;
951
#else
952
    ctxt->sax2 = 1;
953
#endif /* LIBXML_SAX1_ENABLED */
954
955
    /*
956
     * Some users replace the dictionary directly in the context struct.
957
     * We really need an API function to do that cleanly.
958
     */
959
272k
    ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3);
960
272k
    ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5);
961
272k
    ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36);
962
272k
    if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) ||
963
272k
    (ctxt->str_xml_ns == NULL)) {
964
0
        xmlErrMemory(ctxt);
965
0
    }
966
967
272k
    xmlDictSetLimit(ctxt->dict,
968
272k
                    (ctxt->options & XML_PARSE_HUGE) ?
969
265k
                        0 :
970
272k
                        XML_MAX_DICTIONARY_LIMIT);
971
272k
}
972
973
typedef struct {
974
    xmlHashedString prefix;
975
    xmlHashedString name;
976
    xmlHashedString value;
977
    const xmlChar *valueEnd;
978
    int external;
979
    int expandedSize;
980
} xmlDefAttr;
981
982
typedef struct _xmlDefAttrs xmlDefAttrs;
983
typedef xmlDefAttrs *xmlDefAttrsPtr;
984
struct _xmlDefAttrs {
985
    int nbAttrs;  /* number of defaulted attributes on that element */
986
    int maxAttrs;       /* the size of the array */
987
#if __STDC_VERSION__ >= 199901L
988
    /* Using a C99 flexible array member avoids UBSan errors. */
989
    xmlDefAttr attrs[] ATTRIBUTE_COUNTED_BY(maxAttrs);
990
#else
991
    xmlDefAttr attrs[1];
992
#endif
993
};
994
995
/**
996
 * xmlAttrNormalizeSpace:
997
 * @src: the source string
998
 * @dst: the target string
999
 *
1000
 * Normalize the space in non CDATA attribute values:
1001
 * If the attribute type is not CDATA, then the XML processor MUST further
1002
 * process the normalized attribute value by discarding any leading and
1003
 * trailing space (#x20) characters, and by replacing sequences of space
1004
 * (#x20) characters by a single space (#x20) character.
1005
 * Note that the size of dst need to be at least src, and if one doesn't need
1006
 * to preserve dst (and it doesn't come from a dictionary or read-only) then
1007
 * passing src as dst is just fine.
1008
 *
1009
 * Returns a pointer to the normalized value (dst) or NULL if no conversion
1010
 *         is needed.
1011
 */
1012
static xmlChar *
1013
xmlAttrNormalizeSpace(const xmlChar *src, xmlChar *dst)
1014
29.9k
{
1015
29.9k
    if ((src == NULL) || (dst == NULL))
1016
0
        return(NULL);
1017
1018
42.9k
    while (*src == 0x20) src++;
1019
647k
    while (*src != 0) {
1020
617k
  if (*src == 0x20) {
1021
139k
      while (*src == 0x20) src++;
1022
17.3k
      if (*src != 0)
1023
13.4k
    *dst++ = 0x20;
1024
599k
  } else {
1025
599k
      *dst++ = *src++;
1026
599k
  }
1027
617k
    }
1028
29.9k
    *dst = 0;
1029
29.9k
    if (dst == src)
1030
19.3k
       return(NULL);
1031
10.6k
    return(dst);
1032
29.9k
}
1033
1034
/**
1035
 * xmlAddDefAttrs:
1036
 * @ctxt:  an XML parser context
1037
 * @fullname:  the element fullname
1038
 * @fullattr:  the attribute fullname
1039
 * @value:  the attribute value
1040
 *
1041
 * Add a defaulted attribute for an element
1042
 */
1043
static void
1044
xmlAddDefAttrs(xmlParserCtxtPtr ctxt,
1045
               const xmlChar *fullname,
1046
               const xmlChar *fullattr,
1047
32.7k
               const xmlChar *value) {
1048
32.7k
    xmlDefAttrsPtr defaults;
1049
32.7k
    xmlDefAttr *attr;
1050
32.7k
    int len, expandedSize;
1051
32.7k
    xmlHashedString name;
1052
32.7k
    xmlHashedString prefix;
1053
32.7k
    xmlHashedString hvalue;
1054
32.7k
    const xmlChar *localname;
1055
1056
    /*
1057
     * Allows to detect attribute redefinitions
1058
     */
1059
32.7k
    if (ctxt->attsSpecial != NULL) {
1060
30.0k
        if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL)
1061
10.9k
      return;
1062
30.0k
    }
1063
1064
21.7k
    if (ctxt->attsDefault == NULL) {
1065
2.73k
        ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict);
1066
2.73k
  if (ctxt->attsDefault == NULL)
1067
0
      goto mem_error;
1068
2.73k
    }
1069
1070
    /*
1071
     * split the element name into prefix:localname , the string found
1072
     * are within the DTD and then not associated to namespace names.
1073
     */
1074
21.7k
    localname = xmlSplitQName3(fullname, &len);
1075
21.7k
    if (localname == NULL) {
1076
18.7k
        name = xmlDictLookupHashed(ctxt->dict, fullname, -1);
1077
18.7k
  prefix.name = NULL;
1078
18.7k
    } else {
1079
3.07k
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1080
3.07k
  prefix = xmlDictLookupHashed(ctxt->dict, fullname, len);
1081
3.07k
        if (prefix.name == NULL)
1082
0
            goto mem_error;
1083
3.07k
    }
1084
21.7k
    if (name.name == NULL)
1085
0
        goto mem_error;
1086
1087
    /*
1088
     * make sure there is some storage
1089
     */
1090
21.7k
    defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name);
1091
21.7k
    if ((defaults == NULL) ||
1092
21.7k
        (defaults->nbAttrs >= defaults->maxAttrs)) {
1093
8.26k
        xmlDefAttrsPtr temp;
1094
8.26k
        int newSize;
1095
1096
8.26k
        if (defaults == NULL) {
1097
5.71k
            newSize = 4;
1098
5.71k
        } else {
1099
2.55k
            if ((defaults->maxAttrs >= XML_MAX_ATTRS) ||
1100
2.55k
                ((size_t) defaults->maxAttrs >
1101
2.55k
                     SIZE_MAX / 2 / sizeof(temp[0]) - sizeof(*defaults)))
1102
0
                goto mem_error;
1103
1104
2.55k
            if (defaults->maxAttrs > XML_MAX_ATTRS / 2)
1105
0
                newSize = XML_MAX_ATTRS;
1106
2.55k
            else
1107
2.55k
                newSize = defaults->maxAttrs * 2;
1108
2.55k
        }
1109
8.26k
        temp = xmlRealloc(defaults,
1110
8.26k
                          sizeof(*defaults) + newSize * sizeof(xmlDefAttr));
1111
8.26k
  if (temp == NULL)
1112
0
      goto mem_error;
1113
8.26k
        if (defaults == NULL)
1114
5.71k
            temp->nbAttrs = 0;
1115
8.26k
  temp->maxAttrs = newSize;
1116
8.26k
        defaults = temp;
1117
8.26k
  if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name,
1118
8.26k
                          defaults, NULL) < 0) {
1119
0
      xmlFree(defaults);
1120
0
      goto mem_error;
1121
0
  }
1122
8.26k
    }
1123
1124
    /*
1125
     * Split the attribute name into prefix:localname , the string found
1126
     * are within the DTD and hen not associated to namespace names.
1127
     */
1128
21.7k
    localname = xmlSplitQName3(fullattr, &len);
1129
21.7k
    if (localname == NULL) {
1130
11.6k
        name = xmlDictLookupHashed(ctxt->dict, fullattr, -1);
1131
11.6k
  prefix.name = NULL;
1132
11.6k
    } else {
1133
10.1k
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1134
10.1k
  prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len);
1135
10.1k
        if (prefix.name == NULL)
1136
0
            goto mem_error;
1137
10.1k
    }
1138
21.7k
    if (name.name == NULL)
1139
0
        goto mem_error;
1140
1141
    /* intern the string and precompute the end */
1142
21.7k
    len = strlen((const char *) value);
1143
21.7k
    hvalue = xmlDictLookupHashed(ctxt->dict, value, len);
1144
21.7k
    if (hvalue.name == NULL)
1145
0
        goto mem_error;
1146
1147
21.7k
    expandedSize = strlen((const char *) name.name);
1148
21.7k
    if (prefix.name != NULL)
1149
10.1k
        expandedSize += strlen((const char *) prefix.name);
1150
21.7k
    expandedSize += len;
1151
1152
21.7k
    attr = &defaults->attrs[defaults->nbAttrs++];
1153
21.7k
    attr->name = name;
1154
21.7k
    attr->prefix = prefix;
1155
21.7k
    attr->value = hvalue;
1156
21.7k
    attr->valueEnd = hvalue.name + len;
1157
21.7k
    attr->external = PARSER_EXTERNAL(ctxt);
1158
21.7k
    attr->expandedSize = expandedSize;
1159
1160
21.7k
    return;
1161
1162
0
mem_error:
1163
0
    xmlErrMemory(ctxt);
1164
0
}
1165
1166
/**
1167
 * xmlAddSpecialAttr:
1168
 * @ctxt:  an XML parser context
1169
 * @fullname:  the element fullname
1170
 * @fullattr:  the attribute fullname
1171
 * @type:  the attribute type
1172
 *
1173
 * Register this attribute type
1174
 */
1175
static void
1176
xmlAddSpecialAttr(xmlParserCtxtPtr ctxt,
1177
      const xmlChar *fullname,
1178
      const xmlChar *fullattr,
1179
      int type)
1180
43.1k
{
1181
43.1k
    if (ctxt->attsSpecial == NULL) {
1182
3.14k
        ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict);
1183
3.14k
  if (ctxt->attsSpecial == NULL)
1184
0
      goto mem_error;
1185
3.14k
    }
1186
1187
43.1k
    if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr,
1188
43.1k
                    XML_INT_TO_PTR(type)) < 0)
1189
0
        goto mem_error;
1190
43.1k
    return;
1191
1192
43.1k
mem_error:
1193
0
    xmlErrMemory(ctxt);
1194
0
}
1195
1196
/**
1197
 * xmlCleanSpecialAttrCallback:
1198
 *
1199
 * Removes CDATA attributes from the special attribute table
1200
 */
1201
static void
1202
xmlCleanSpecialAttrCallback(void *payload, void *data,
1203
                            const xmlChar *fullname, const xmlChar *fullattr,
1204
25.9k
                            const xmlChar *unused ATTRIBUTE_UNUSED) {
1205
25.9k
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data;
1206
1207
25.9k
    if (XML_PTR_TO_INT(payload) == XML_ATTRIBUTE_CDATA) {
1208
2.01k
        xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL);
1209
2.01k
    }
1210
25.9k
}
1211
1212
/**
1213
 * xmlCleanSpecialAttr:
1214
 * @ctxt:  an XML parser context
1215
 *
1216
 * Trim the list of attributes defined to remove all those of type
1217
 * CDATA as they are not special. This call should be done when finishing
1218
 * to parse the DTD and before starting to parse the document root.
1219
 */
1220
static void
1221
xmlCleanSpecialAttr(xmlParserCtxtPtr ctxt)
1222
7.77k
{
1223
7.77k
    if (ctxt->attsSpecial == NULL)
1224
4.62k
        return;
1225
1226
3.14k
    xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt);
1227
1228
3.14k
    if (xmlHashSize(ctxt->attsSpecial) == 0) {
1229
258
        xmlHashFree(ctxt->attsSpecial, NULL);
1230
258
        ctxt->attsSpecial = NULL;
1231
258
    }
1232
3.14k
}
1233
1234
/**
1235
 * xmlCheckLanguageID:
1236
 * @lang:  pointer to the string value
1237
 *
1238
 * DEPRECATED: Internal function, do not use.
1239
 *
1240
 * Checks that the value conforms to the LanguageID production:
1241
 *
1242
 * NOTE: this is somewhat deprecated, those productions were removed from
1243
 *       the XML Second edition.
1244
 *
1245
 * [33] LanguageID ::= Langcode ('-' Subcode)*
1246
 * [34] Langcode ::= ISO639Code |  IanaCode |  UserCode
1247
 * [35] ISO639Code ::= ([a-z] | [A-Z]) ([a-z] | [A-Z])
1248
 * [36] IanaCode ::= ('i' | 'I') '-' ([a-z] | [A-Z])+
1249
 * [37] UserCode ::= ('x' | 'X') '-' ([a-z] | [A-Z])+
1250
 * [38] Subcode ::= ([a-z] | [A-Z])+
1251
 *
1252
 * The current REC reference the successors of RFC 1766, currently 5646
1253
 *
1254
 * http://www.rfc-editor.org/rfc/rfc5646.txt
1255
 * langtag       = language
1256
 *                 ["-" script]
1257
 *                 ["-" region]
1258
 *                 *("-" variant)
1259
 *                 *("-" extension)
1260
 *                 ["-" privateuse]
1261
 * language      = 2*3ALPHA            ; shortest ISO 639 code
1262
 *                 ["-" extlang]       ; sometimes followed by
1263
 *                                     ; extended language subtags
1264
 *               / 4ALPHA              ; or reserved for future use
1265
 *               / 5*8ALPHA            ; or registered language subtag
1266
 *
1267
 * extlang       = 3ALPHA              ; selected ISO 639 codes
1268
 *                 *2("-" 3ALPHA)      ; permanently reserved
1269
 *
1270
 * script        = 4ALPHA              ; ISO 15924 code
1271
 *
1272
 * region        = 2ALPHA              ; ISO 3166-1 code
1273
 *               / 3DIGIT              ; UN M.49 code
1274
 *
1275
 * variant       = 5*8alphanum         ; registered variants
1276
 *               / (DIGIT 3alphanum)
1277
 *
1278
 * extension     = singleton 1*("-" (2*8alphanum))
1279
 *
1280
 *                                     ; Single alphanumerics
1281
 *                                     ; "x" reserved for private use
1282
 * singleton     = DIGIT               ; 0 - 9
1283
 *               / %x41-57             ; A - W
1284
 *               / %x59-5A             ; Y - Z
1285
 *               / %x61-77             ; a - w
1286
 *               / %x79-7A             ; y - z
1287
 *
1288
 * it sounds right to still allow Irregular i-xxx IANA and user codes too
1289
 * The parser below doesn't try to cope with extension or privateuse
1290
 * that could be added but that's not interoperable anyway
1291
 *
1292
 * Returns 1 if correct 0 otherwise
1293
 **/
1294
int
1295
xmlCheckLanguageID(const xmlChar * lang)
1296
0
{
1297
0
    const xmlChar *cur = lang, *nxt;
1298
1299
0
    if (cur == NULL)
1300
0
        return (0);
1301
0
    if (((cur[0] == 'i') && (cur[1] == '-')) ||
1302
0
        ((cur[0] == 'I') && (cur[1] == '-')) ||
1303
0
        ((cur[0] == 'x') && (cur[1] == '-')) ||
1304
0
        ((cur[0] == 'X') && (cur[1] == '-'))) {
1305
        /*
1306
         * Still allow IANA code and user code which were coming
1307
         * from the previous version of the XML-1.0 specification
1308
         * it's deprecated but we should not fail
1309
         */
1310
0
        cur += 2;
1311
0
        while (((cur[0] >= 'A') && (cur[0] <= 'Z')) ||
1312
0
               ((cur[0] >= 'a') && (cur[0] <= 'z')))
1313
0
            cur++;
1314
0
        return(cur[0] == 0);
1315
0
    }
1316
0
    nxt = cur;
1317
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1318
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1319
0
           nxt++;
1320
0
    if (nxt - cur >= 4) {
1321
        /*
1322
         * Reserved
1323
         */
1324
0
        if ((nxt - cur > 8) || (nxt[0] != 0))
1325
0
            return(0);
1326
0
        return(1);
1327
0
    }
1328
0
    if (nxt - cur < 2)
1329
0
        return(0);
1330
    /* we got an ISO 639 code */
1331
0
    if (nxt[0] == 0)
1332
0
        return(1);
1333
0
    if (nxt[0] != '-')
1334
0
        return(0);
1335
1336
0
    nxt++;
1337
0
    cur = nxt;
1338
    /* now we can have extlang or script or region or variant */
1339
0
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1340
0
        goto region_m49;
1341
1342
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1343
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1344
0
           nxt++;
1345
0
    if (nxt - cur == 4)
1346
0
        goto script;
1347
0
    if (nxt - cur == 2)
1348
0
        goto region;
1349
0
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1350
0
        goto variant;
1351
0
    if (nxt - cur != 3)
1352
0
        return(0);
1353
    /* we parsed an extlang */
1354
0
    if (nxt[0] == 0)
1355
0
        return(1);
1356
0
    if (nxt[0] != '-')
1357
0
        return(0);
1358
1359
0
    nxt++;
1360
0
    cur = nxt;
1361
    /* now we can have script or region or variant */
1362
0
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1363
0
        goto region_m49;
1364
1365
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1366
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1367
0
           nxt++;
1368
0
    if (nxt - cur == 2)
1369
0
        goto region;
1370
0
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1371
0
        goto variant;
1372
0
    if (nxt - cur != 4)
1373
0
        return(0);
1374
    /* we parsed a script */
1375
0
script:
1376
0
    if (nxt[0] == 0)
1377
0
        return(1);
1378
0
    if (nxt[0] != '-')
1379
0
        return(0);
1380
1381
0
    nxt++;
1382
0
    cur = nxt;
1383
    /* now we can have region or variant */
1384
0
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1385
0
        goto region_m49;
1386
1387
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1388
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1389
0
           nxt++;
1390
1391
0
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1392
0
        goto variant;
1393
0
    if (nxt - cur != 2)
1394
0
        return(0);
1395
    /* we parsed a region */
1396
0
region:
1397
0
    if (nxt[0] == 0)
1398
0
        return(1);
1399
0
    if (nxt[0] != '-')
1400
0
        return(0);
1401
1402
0
    nxt++;
1403
0
    cur = nxt;
1404
    /* now we can just have a variant */
1405
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1406
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1407
0
           nxt++;
1408
1409
0
    if ((nxt - cur < 5) || (nxt - cur > 8))
1410
0
        return(0);
1411
1412
    /* we parsed a variant */
1413
0
variant:
1414
0
    if (nxt[0] == 0)
1415
0
        return(1);
1416
0
    if (nxt[0] != '-')
1417
0
        return(0);
1418
    /* extensions and private use subtags not checked */
1419
0
    return (1);
1420
1421
0
region_m49:
1422
0
    if (((nxt[1] >= '0') && (nxt[1] <= '9')) &&
1423
0
        ((nxt[2] >= '0') && (nxt[2] <= '9'))) {
1424
0
        nxt += 3;
1425
0
        goto region;
1426
0
    }
1427
0
    return(0);
1428
0
}
1429
1430
/************************************************************************
1431
 *                  *
1432
 *    Parser stacks related functions and macros    *
1433
 *                  *
1434
 ************************************************************************/
1435
1436
static xmlChar *
1437
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar **str);
1438
1439
/**
1440
 * xmlParserNsCreate:
1441
 *
1442
 * Create a new namespace database.
1443
 *
1444
 * Returns the new obejct.
1445
 */
1446
xmlParserNsData *
1447
272k
xmlParserNsCreate(void) {
1448
272k
    xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb));
1449
1450
272k
    if (nsdb == NULL)
1451
0
        return(NULL);
1452
272k
    memset(nsdb, 0, sizeof(*nsdb));
1453
272k
    nsdb->defaultNsIndex = INT_MAX;
1454
1455
272k
    return(nsdb);
1456
272k
}
1457
1458
/**
1459
 * xmlParserNsFree:
1460
 * @nsdb: namespace database
1461
 *
1462
 * Free a namespace database.
1463
 */
1464
void
1465
272k
xmlParserNsFree(xmlParserNsData *nsdb) {
1466
272k
    if (nsdb == NULL)
1467
0
        return;
1468
1469
272k
    xmlFree(nsdb->extra);
1470
272k
    xmlFree(nsdb->hash);
1471
272k
    xmlFree(nsdb);
1472
272k
}
1473
1474
/**
1475
 * xmlParserNsReset:
1476
 * @nsdb: namespace database
1477
 *
1478
 * Reset a namespace database.
1479
 */
1480
static void
1481
6.77k
xmlParserNsReset(xmlParserNsData *nsdb) {
1482
6.77k
    if (nsdb == NULL)
1483
0
        return;
1484
1485
6.77k
    nsdb->hashElems = 0;
1486
6.77k
    nsdb->elementId = 0;
1487
6.77k
    nsdb->defaultNsIndex = INT_MAX;
1488
1489
6.77k
    if (nsdb->hash)
1490
0
        memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0]));
1491
6.77k
}
1492
1493
/**
1494
 * xmlParserStartElement:
1495
 * @nsdb: namespace database
1496
 *
1497
 * Signal that a new element has started.
1498
 *
1499
 * Returns 0 on success, -1 if the element counter overflowed.
1500
 */
1501
static int
1502
35.9M
xmlParserNsStartElement(xmlParserNsData *nsdb) {
1503
35.9M
    if (nsdb->elementId == UINT_MAX)
1504
0
        return(-1);
1505
35.9M
    nsdb->elementId++;
1506
1507
35.9M
    return(0);
1508
35.9M
}
1509
1510
/**
1511
 * xmlParserNsLookup:
1512
 * @ctxt: parser context
1513
 * @prefix: namespace prefix
1514
 * @bucketPtr: optional bucket (return value)
1515
 *
1516
 * Lookup namespace with given prefix. If @bucketPtr is non-NULL, it will
1517
 * be set to the matching bucket, or the first empty bucket if no match
1518
 * was found.
1519
 *
1520
 * Returns the namespace index on success, INT_MAX if no namespace was
1521
 * found.
1522
 */
1523
static int
1524
xmlParserNsLookup(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix,
1525
49.4M
                  xmlParserNsBucket **bucketPtr) {
1526
49.4M
    xmlParserNsBucket *bucket, *tombstone;
1527
49.4M
    unsigned index, hashValue;
1528
1529
49.4M
    if (prefix->name == NULL)
1530
20.9M
        return(ctxt->nsdb->defaultNsIndex);
1531
1532
28.5M
    if (ctxt->nsdb->hashSize == 0)
1533
418k
        return(INT_MAX);
1534
1535
28.1M
    hashValue = prefix->hashValue;
1536
28.1M
    index = hashValue & (ctxt->nsdb->hashSize - 1);
1537
28.1M
    bucket = &ctxt->nsdb->hash[index];
1538
28.1M
    tombstone = NULL;
1539
1540
29.6M
    while (bucket->hashValue) {
1541
27.6M
        if (bucket->index == INT_MAX) {
1542
239k
            if (tombstone == NULL)
1543
232k
                tombstone = bucket;
1544
27.4M
        } else if (bucket->hashValue == hashValue) {
1545
26.1M
            if (ctxt->nsTab[bucket->index * 2] == prefix->name) {
1546
26.1M
                if (bucketPtr != NULL)
1547
1.12M
                    *bucketPtr = bucket;
1548
26.1M
                return(bucket->index);
1549
26.1M
            }
1550
26.1M
        }
1551
1552
1.55M
        index++;
1553
1.55M
        bucket++;
1554
1.55M
        if (index == ctxt->nsdb->hashSize) {
1555
21.9k
            index = 0;
1556
21.9k
            bucket = ctxt->nsdb->hash;
1557
21.9k
        }
1558
1.55M
    }
1559
1560
1.96M
    if (bucketPtr != NULL)
1561
1.16M
        *bucketPtr = tombstone ? tombstone : bucket;
1562
1.96M
    return(INT_MAX);
1563
28.1M
}
1564
1565
/**
1566
 * xmlParserNsLookupUri:
1567
 * @ctxt: parser context
1568
 * @prefix: namespace prefix
1569
 *
1570
 * Lookup namespace URI with given prefix.
1571
 *
1572
 * Returns the namespace URI on success, NULL if no namespace was found.
1573
 */
1574
static const xmlChar *
1575
35.9M
xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) {
1576
35.9M
    const xmlChar *ret;
1577
35.9M
    int nsIndex;
1578
1579
35.9M
    if (prefix->name == ctxt->str_xml)
1580
28.7k
        return(ctxt->str_xml_ns);
1581
1582
    /*
1583
     * minNsIndex is used when building an entity tree. We must
1584
     * ignore namespaces declared outside the entity.
1585
     */
1586
35.8M
    nsIndex = xmlParserNsLookup(ctxt, prefix, NULL);
1587
35.8M
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1588
15.6M
        return(NULL);
1589
1590
20.2M
    ret = ctxt->nsTab[nsIndex * 2 + 1];
1591
20.2M
    if (ret[0] == 0)
1592
18.5k
        ret = NULL;
1593
20.2M
    return(ret);
1594
35.8M
}
1595
1596
/**
1597
 * xmlParserNsLookupSax:
1598
 * @ctxt: parser context
1599
 * @prefix: namespace prefix
1600
 *
1601
 * Lookup extra data for the given prefix. This returns data stored
1602
 * with xmlParserNsUdpateSax.
1603
 *
1604
 * Returns the data on success, NULL if no namespace was found.
1605
 */
1606
void *
1607
1.08M
xmlParserNsLookupSax(xmlParserCtxtPtr ctxt, const xmlChar *prefix) {
1608
1.08M
    xmlHashedString hprefix;
1609
1.08M
    int nsIndex;
1610
1611
1.08M
    if (prefix == ctxt->str_xml)
1612
0
        return(NULL);
1613
1614
1.08M
    hprefix.name = prefix;
1615
1.08M
    if (prefix != NULL)
1616
1.08M
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1617
1.32k
    else
1618
1.32k
        hprefix.hashValue = 0;
1619
1.08M
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1620
1.08M
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1621
0
        return(NULL);
1622
1623
1.08M
    return(ctxt->nsdb->extra[nsIndex].saxData);
1624
1.08M
}
1625
1626
/**
1627
 * xmlParserNsUpdateSax:
1628
 * @ctxt: parser context
1629
 * @prefix: namespace prefix
1630
 * @saxData: extra data for SAX handler
1631
 *
1632
 * Sets or updates extra data for the given prefix. This value will be
1633
 * returned by xmlParserNsLookupSax as long as the namespace with the
1634
 * given prefix is in scope.
1635
 *
1636
 * Returns the data on success, NULL if no namespace was found.
1637
 */
1638
int
1639
xmlParserNsUpdateSax(xmlParserCtxtPtr ctxt, const xmlChar *prefix,
1640
13.1k
                     void *saxData) {
1641
13.1k
    xmlHashedString hprefix;
1642
13.1k
    int nsIndex;
1643
1644
13.1k
    if (prefix == ctxt->str_xml)
1645
0
        return(-1);
1646
1647
13.1k
    hprefix.name = prefix;
1648
13.1k
    if (prefix != NULL)
1649
12.2k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1650
870
    else
1651
870
        hprefix.hashValue = 0;
1652
13.1k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1653
13.1k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1654
0
        return(-1);
1655
1656
13.1k
    ctxt->nsdb->extra[nsIndex].saxData = saxData;
1657
13.1k
    return(0);
1658
13.1k
}
1659
1660
/**
1661
 * xmlParserNsGrow:
1662
 * @ctxt: parser context
1663
 *
1664
 * Grows the namespace tables.
1665
 *
1666
 * Returns 0 on success, -1 if a memory allocation failed.
1667
 */
1668
static int
1669
702k
xmlParserNsGrow(xmlParserCtxtPtr ctxt) {
1670
702k
    const xmlChar **table;
1671
702k
    xmlParserNsExtra *extra;
1672
702k
    int newSize;
1673
1674
702k
    newSize = xmlGrowCapacity(ctxt->nsMax,
1675
702k
                              sizeof(table[0]) + sizeof(extra[0]),
1676
702k
                              16, XML_MAX_ITEMS);
1677
702k
    if (newSize < 0)
1678
0
        goto error;
1679
1680
702k
    table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0]));
1681
702k
    if (table == NULL)
1682
0
        goto error;
1683
702k
    ctxt->nsTab = table;
1684
1685
702k
    extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0]));
1686
702k
    if (extra == NULL)
1687
0
        goto error;
1688
702k
    ctxt->nsdb->extra = extra;
1689
1690
702k
    ctxt->nsMax = newSize;
1691
702k
    return(0);
1692
1693
0
error:
1694
0
    xmlErrMemory(ctxt);
1695
0
    return(-1);
1696
702k
}
1697
1698
/**
1699
 * xmlParserNsPush:
1700
 * @ctxt: parser context
1701
 * @prefix: prefix with hash value
1702
 * @uri: uri with hash value
1703
 * @saxData: extra data for SAX handler
1704
 * @defAttr: whether the namespace comes from a default attribute
1705
 *
1706
 * Push a new namespace on the table.
1707
 *
1708
 * Returns 1 if the namespace was pushed, 0 if the namespace was ignored,
1709
 * -1 if a memory allocation failed.
1710
 */
1711
static int
1712
xmlParserNsPush(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix,
1713
1.91M
                const xmlHashedString *uri, void *saxData, int defAttr) {
1714
1.91M
    xmlParserNsBucket *bucket = NULL;
1715
1.91M
    xmlParserNsExtra *extra;
1716
1.91M
    const xmlChar **ns;
1717
1.91M
    unsigned hashValue, nsIndex, oldIndex;
1718
1719
1.91M
    if ((prefix != NULL) && (prefix->name == ctxt->str_xml))
1720
59
        return(0);
1721
1722
1.91M
    if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) {
1723
0
        xmlErrMemory(ctxt);
1724
0
        return(-1);
1725
0
    }
1726
1727
    /*
1728
     * Default namespace and 'xml' namespace
1729
     */
1730
1.91M
    if ((prefix == NULL) || (prefix->name == NULL)) {
1731
334k
        oldIndex = ctxt->nsdb->defaultNsIndex;
1732
1733
334k
        if (oldIndex != INT_MAX) {
1734
176k
            extra = &ctxt->nsdb->extra[oldIndex];
1735
1736
176k
            if (extra->elementId == ctxt->nsdb->elementId) {
1737
84.0k
                if (defAttr == 0)
1738
82.8k
                    xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns");
1739
84.0k
                return(0);
1740
84.0k
            }
1741
1742
91.9k
            if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1743
91.9k
                (uri->name == ctxt->nsTab[oldIndex * 2 + 1]))
1744
0
                return(0);
1745
91.9k
        }
1746
1747
250k
        ctxt->nsdb->defaultNsIndex = ctxt->nsNr;
1748
250k
        goto populate_entry;
1749
334k
    }
1750
1751
    /*
1752
     * Hash table lookup
1753
     */
1754
1.57M
    oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket);
1755
1.57M
    if (oldIndex != INT_MAX) {
1756
262k
        extra = &ctxt->nsdb->extra[oldIndex];
1757
1758
        /*
1759
         * Check for duplicate definitions on the same element.
1760
         */
1761
262k
        if (extra->elementId == ctxt->nsdb->elementId) {
1762
82.4k
            if (defAttr == 0)
1763
82.4k
                xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name);
1764
82.4k
            return(0);
1765
82.4k
        }
1766
1767
179k
        if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1768
179k
            (uri->name == ctxt->nsTab[bucket->index * 2 + 1]))
1769
0
            return(0);
1770
1771
179k
        bucket->index = ctxt->nsNr;
1772
179k
        goto populate_entry;
1773
179k
    }
1774
1775
    /*
1776
     * Insert new bucket
1777
     */
1778
1779
1.31M
    hashValue = prefix->hashValue;
1780
1781
    /*
1782
     * Grow hash table, 50% fill factor
1783
     */
1784
1.31M
    if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) {
1785
211k
        xmlParserNsBucket *newHash;
1786
211k
        unsigned newSize, i, index;
1787
1788
211k
        if (ctxt->nsdb->hashSize > UINT_MAX / 2) {
1789
0
            xmlErrMemory(ctxt);
1790
0
            return(-1);
1791
0
        }
1792
211k
        newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16;
1793
211k
        newHash = xmlMalloc(newSize * sizeof(newHash[0]));
1794
211k
        if (newHash == NULL) {
1795
0
            xmlErrMemory(ctxt);
1796
0
            return(-1);
1797
0
        }
1798
211k
        memset(newHash, 0, newSize * sizeof(newHash[0]));
1799
1800
2.44M
        for (i = 0; i < ctxt->nsdb->hashSize; i++) {
1801
2.23M
            unsigned hv = ctxt->nsdb->hash[i].hashValue;
1802
2.23M
            unsigned newIndex;
1803
1804
2.23M
            if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX))
1805
1.43M
                continue;
1806
806k
            newIndex = hv & (newSize - 1);
1807
1808
831k
            while (newHash[newIndex].hashValue != 0) {
1809
25.5k
                newIndex++;
1810
25.5k
                if (newIndex == newSize)
1811
87
                    newIndex = 0;
1812
25.5k
            }
1813
1814
806k
            newHash[newIndex] = ctxt->nsdb->hash[i];
1815
806k
        }
1816
1817
211k
        xmlFree(ctxt->nsdb->hash);
1818
211k
        ctxt->nsdb->hash = newHash;
1819
211k
        ctxt->nsdb->hashSize = newSize;
1820
1821
        /*
1822
         * Relookup
1823
         */
1824
211k
        index = hashValue & (newSize - 1);
1825
1826
216k
        while (newHash[index].hashValue != 0) {
1827
4.96k
            index++;
1828
4.96k
            if (index == newSize)
1829
47
                index = 0;
1830
4.96k
        }
1831
1832
211k
        bucket = &newHash[index];
1833
211k
    }
1834
1835
1.31M
    bucket->hashValue = hashValue;
1836
1.31M
    bucket->index = ctxt->nsNr;
1837
1.31M
    ctxt->nsdb->hashElems++;
1838
1.31M
    oldIndex = INT_MAX;
1839
1840
1.74M
populate_entry:
1841
1.74M
    nsIndex = ctxt->nsNr;
1842
1843
1.74M
    ns = &ctxt->nsTab[nsIndex * 2];
1844
1.74M
    ns[0] = prefix ? prefix->name : NULL;
1845
1.74M
    ns[1] = uri->name;
1846
1847
1.74M
    extra = &ctxt->nsdb->extra[nsIndex];
1848
1.74M
    extra->saxData = saxData;
1849
1.74M
    extra->prefixHashValue = prefix ? prefix->hashValue : 0;
1850
1.74M
    extra->uriHashValue = uri->hashValue;
1851
1.74M
    extra->elementId = ctxt->nsdb->elementId;
1852
1.74M
    extra->oldIndex = oldIndex;
1853
1854
1.74M
    ctxt->nsNr++;
1855
1856
1.74M
    return(1);
1857
1.31M
}
1858
1859
/**
1860
 * xmlParserNsPop:
1861
 * @ctxt: an XML parser context
1862
 * @nr:  the number to pop
1863
 *
1864
 * Pops the top @nr namespaces and restores the hash table.
1865
 *
1866
 * Returns the number of namespaces popped.
1867
 */
1868
static int
1869
xmlParserNsPop(xmlParserCtxtPtr ctxt, int nr)
1870
380k
{
1871
380k
    int i;
1872
1873
    /* assert(nr <= ctxt->nsNr); */
1874
1875
1.40M
    for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) {
1876
1.02M
        const xmlChar *prefix = ctxt->nsTab[i * 2];
1877
1.02M
        xmlParserNsExtra *extra = &ctxt->nsdb->extra[i];
1878
1879
1.02M
        if (prefix == NULL) {
1880
166k
            ctxt->nsdb->defaultNsIndex = extra->oldIndex;
1881
858k
        } else {
1882
858k
            xmlHashedString hprefix;
1883
858k
            xmlParserNsBucket *bucket = NULL;
1884
1885
858k
            hprefix.name = prefix;
1886
858k
            hprefix.hashValue = extra->prefixHashValue;
1887
858k
            xmlParserNsLookup(ctxt, &hprefix, &bucket);
1888
            /* assert(bucket && bucket->hashValue); */
1889
858k
            bucket->index = extra->oldIndex;
1890
858k
        }
1891
1.02M
    }
1892
1893
380k
    ctxt->nsNr -= nr;
1894
380k
    return(nr);
1895
380k
}
1896
1897
static int
1898
737k
xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt) {
1899
737k
    const xmlChar **atts;
1900
737k
    unsigned *attallocs;
1901
737k
    int newSize;
1902
1903
737k
    newSize = xmlGrowCapacity(ctxt->maxatts / 5,
1904
737k
                              sizeof(atts[0]) * 5 + sizeof(attallocs[0]),
1905
737k
                              10, XML_MAX_ATTRS);
1906
737k
    if (newSize < 0) {
1907
0
        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
1908
0
                    "Maximum number of attributes exceeded");
1909
0
        return(-1);
1910
0
    }
1911
1912
737k
    atts = xmlRealloc(ctxt->atts, newSize * sizeof(atts[0]) * 5);
1913
737k
    if (atts == NULL)
1914
0
        goto mem_error;
1915
737k
    ctxt->atts = atts;
1916
1917
737k
    attallocs = xmlRealloc(ctxt->attallocs,
1918
737k
                           newSize * sizeof(attallocs[0]));
1919
737k
    if (attallocs == NULL)
1920
0
        goto mem_error;
1921
737k
    ctxt->attallocs = attallocs;
1922
1923
737k
    ctxt->maxatts = newSize * 5;
1924
1925
737k
    return(0);
1926
1927
0
mem_error:
1928
0
    xmlErrMemory(ctxt);
1929
0
    return(-1);
1930
737k
}
1931
1932
/**
1933
 * xmlCtxtPushInput:
1934
 * @ctxt:  an XML parser context
1935
 * @value:  the parser input
1936
 *
1937
 * Pushes a new parser input on top of the input stack
1938
 *
1939
 * Returns -1 in case of error, the index in the stack otherwise
1940
 */
1941
int
1942
xmlCtxtPushInput(xmlParserCtxtPtr ctxt, xmlParserInputPtr value)
1943
272k
{
1944
272k
    char *directory = NULL;
1945
272k
    int maxDepth;
1946
1947
272k
    if ((ctxt == NULL) || (value == NULL))
1948
0
        return(-1);
1949
1950
272k
    maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
1951
1952
272k
    if (ctxt->inputNr >= ctxt->inputMax) {
1953
0
        xmlParserInputPtr *tmp;
1954
0
        int newSize;
1955
1956
0
        newSize = xmlGrowCapacity(ctxt->inputMax, sizeof(tmp[0]),
1957
0
                                  5, maxDepth);
1958
0
        if (newSize < 0) {
1959
0
            xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
1960
0
                           "Maximum entity nesting depth exceeded");
1961
0
            xmlHaltParser(ctxt);
1962
0
            return(-1);
1963
0
        }
1964
0
        tmp = xmlRealloc(ctxt->inputTab, newSize * sizeof(tmp[0]));
1965
0
        if (tmp == NULL) {
1966
0
            xmlErrMemory(ctxt);
1967
0
            return(-1);
1968
0
        }
1969
0
        ctxt->inputTab = tmp;
1970
0
        ctxt->inputMax = newSize;
1971
0
    }
1972
1973
272k
    if ((ctxt->inputNr == 0) && (value->filename != NULL)) {
1974
0
        directory = xmlParserGetDirectory(value->filename);
1975
0
        if (directory == NULL) {
1976
0
            xmlErrMemory(ctxt);
1977
0
            return(-1);
1978
0
        }
1979
0
    }
1980
1981
272k
    if (ctxt->input_id >= INT_MAX) {
1982
0
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, "Input ID overflow\n");
1983
0
        return(-1);
1984
0
    }
1985
1986
272k
    ctxt->inputTab[ctxt->inputNr] = value;
1987
272k
    ctxt->input = value;
1988
1989
272k
    if (ctxt->inputNr == 0) {
1990
272k
        xmlFree(ctxt->directory);
1991
272k
        ctxt->directory = directory;
1992
272k
    }
1993
1994
    /*
1995
     * Internally, the input ID is only used to detect parameter entity
1996
     * boundaries. But there are entity loaders in downstream code that
1997
     * detect the main document by checking for "input_id == 1".
1998
     */
1999
272k
    value->id = ctxt->input_id++;
2000
2001
272k
    return(ctxt->inputNr++);
2002
272k
}
2003
2004
/**
2005
 * xmlCtxtPopInput:
2006
 * @ctxt: an XML parser context
2007
 *
2008
 * Pops the top parser input from the input stack
2009
 *
2010
 * Returns the input just removed
2011
 */
2012
xmlParserInputPtr
2013
xmlCtxtPopInput(xmlParserCtxtPtr ctxt)
2014
824k
{
2015
824k
    xmlParserInputPtr ret;
2016
2017
824k
    if (ctxt == NULL)
2018
0
        return(NULL);
2019
824k
    if (ctxt->inputNr <= 0)
2020
551k
        return (NULL);
2021
272k
    ctxt->inputNr--;
2022
272k
    if (ctxt->inputNr > 0)
2023
0
        ctxt->input = ctxt->inputTab[ctxt->inputNr - 1];
2024
272k
    else
2025
272k
        ctxt->input = NULL;
2026
272k
    ret = ctxt->inputTab[ctxt->inputNr];
2027
272k
    ctxt->inputTab[ctxt->inputNr] = NULL;
2028
272k
    return (ret);
2029
824k
}
2030
2031
/**
2032
 * nodePush:
2033
 * @ctxt:  an XML parser context
2034
 * @value:  the element node
2035
 *
2036
 * DEPRECATED: Internal function, do not use.
2037
 *
2038
 * Pushes a new element node on top of the node stack
2039
 *
2040
 * Returns -1 in case of error, the index in the stack otherwise
2041
 */
2042
int
2043
nodePush(xmlParserCtxtPtr ctxt, xmlNodePtr value)
2044
1.08M
{
2045
1.08M
    if (ctxt == NULL)
2046
0
        return(0);
2047
2048
1.08M
    if (ctxt->nodeNr >= ctxt->nodeMax) {
2049
27.1k
        int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
2050
27.1k
        xmlNodePtr *tmp;
2051
27.1k
        int newSize;
2052
2053
27.1k
        newSize = xmlGrowCapacity(ctxt->nodeMax, sizeof(tmp[0]),
2054
27.1k
                                  10, maxDepth);
2055
27.1k
        if (newSize < 0) {
2056
0
            xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
2057
0
                    "Excessive depth in document: %d,"
2058
0
                    " use XML_PARSE_HUGE option\n",
2059
0
                    ctxt->nodeNr);
2060
0
            xmlHaltParser(ctxt);
2061
0
            return(-1);
2062
0
        }
2063
2064
27.1k
  tmp = xmlRealloc(ctxt->nodeTab, newSize * sizeof(tmp[0]));
2065
27.1k
        if (tmp == NULL) {
2066
0
            xmlErrMemory(ctxt);
2067
0
            return (-1);
2068
0
        }
2069
27.1k
        ctxt->nodeTab = tmp;
2070
27.1k
  ctxt->nodeMax = newSize;
2071
27.1k
    }
2072
2073
1.08M
    ctxt->nodeTab[ctxt->nodeNr] = value;
2074
1.08M
    ctxt->node = value;
2075
1.08M
    return (ctxt->nodeNr++);
2076
1.08M
}
2077
2078
/**
2079
 * nodePop:
2080
 * @ctxt: an XML parser context
2081
 *
2082
 * DEPRECATED: Internal function, do not use.
2083
 *
2084
 * Pops the top element node from the node stack
2085
 *
2086
 * Returns the node just removed
2087
 */
2088
xmlNodePtr
2089
nodePop(xmlParserCtxtPtr ctxt)
2090
1.19M
{
2091
1.19M
    xmlNodePtr ret;
2092
2093
1.19M
    if (ctxt == NULL) return(NULL);
2094
1.19M
    if (ctxt->nodeNr <= 0)
2095
103k
        return (NULL);
2096
1.08M
    ctxt->nodeNr--;
2097
1.08M
    if (ctxt->nodeNr > 0)
2098
1.08M
        ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1];
2099
6.65k
    else
2100
6.65k
        ctxt->node = NULL;
2101
1.08M
    ret = ctxt->nodeTab[ctxt->nodeNr];
2102
1.08M
    ctxt->nodeTab[ctxt->nodeNr] = NULL;
2103
1.08M
    return (ret);
2104
1.19M
}
2105
2106
/**
2107
 * nameNsPush:
2108
 * @ctxt:  an XML parser context
2109
 * @value:  the element name
2110
 * @prefix:  the element prefix
2111
 * @URI:  the element namespace name
2112
 * @line:  the current line number for error messages
2113
 * @nsNr:  the number of namespaces pushed on the namespace table
2114
 *
2115
 * Pushes a new element name/prefix/URL on top of the name stack
2116
 *
2117
 * Returns -1 in case of error, the index in the stack otherwise
2118
 */
2119
static int
2120
nameNsPush(xmlParserCtxtPtr ctxt, const xmlChar * value,
2121
           const xmlChar *prefix, const xmlChar *URI, int line, int nsNr)
2122
26.1M
{
2123
26.1M
    xmlStartTag *tag;
2124
2125
26.1M
    if (ctxt->nameNr >= ctxt->nameMax) {
2126
863k
        const xmlChar **tmp;
2127
863k
        xmlStartTag *tmp2;
2128
863k
        int newSize;
2129
2130
863k
        newSize = xmlGrowCapacity(ctxt->nameMax,
2131
863k
                                  sizeof(tmp[0]) + sizeof(tmp2[0]),
2132
863k
                                  10, XML_MAX_ITEMS);
2133
863k
        if (newSize < 0)
2134
0
            goto mem_error;
2135
2136
863k
        tmp = xmlRealloc(ctxt->nameTab, newSize * sizeof(tmp[0]));
2137
863k
        if (tmp == NULL)
2138
0
      goto mem_error;
2139
863k
  ctxt->nameTab = tmp;
2140
2141
863k
        tmp2 = xmlRealloc(ctxt->pushTab, newSize * sizeof(tmp2[0]));
2142
863k
        if (tmp2 == NULL)
2143
0
      goto mem_error;
2144
863k
  ctxt->pushTab = tmp2;
2145
2146
863k
        ctxt->nameMax = newSize;
2147
25.3M
    } else if (ctxt->pushTab == NULL) {
2148
245k
        ctxt->pushTab = xmlMalloc(ctxt->nameMax * sizeof(ctxt->pushTab[0]));
2149
245k
        if (ctxt->pushTab == NULL)
2150
0
            goto mem_error;
2151
245k
    }
2152
26.1M
    ctxt->nameTab[ctxt->nameNr] = value;
2153
26.1M
    ctxt->name = value;
2154
26.1M
    tag = &ctxt->pushTab[ctxt->nameNr];
2155
26.1M
    tag->prefix = prefix;
2156
26.1M
    tag->URI = URI;
2157
26.1M
    tag->line = line;
2158
26.1M
    tag->nsNr = nsNr;
2159
26.1M
    return (ctxt->nameNr++);
2160
0
mem_error:
2161
0
    xmlErrMemory(ctxt);
2162
0
    return (-1);
2163
26.1M
}
2164
#ifdef LIBXML_PUSH_ENABLED
2165
/**
2166
 * nameNsPop:
2167
 * @ctxt: an XML parser context
2168
 *
2169
 * Pops the top element/prefix/URI name from the name stack
2170
 *
2171
 * Returns the name just removed
2172
 */
2173
static const xmlChar *
2174
nameNsPop(xmlParserCtxtPtr ctxt)
2175
9.88M
{
2176
9.88M
    const xmlChar *ret;
2177
2178
9.88M
    if (ctxt->nameNr <= 0)
2179
0
        return (NULL);
2180
9.88M
    ctxt->nameNr--;
2181
9.88M
    if (ctxt->nameNr > 0)
2182
9.73M
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2183
156k
    else
2184
156k
        ctxt->name = NULL;
2185
9.88M
    ret = ctxt->nameTab[ctxt->nameNr];
2186
9.88M
    ctxt->nameTab[ctxt->nameNr] = NULL;
2187
9.88M
    return (ret);
2188
9.88M
}
2189
#endif /* LIBXML_PUSH_ENABLED */
2190
2191
/**
2192
 * namePop:
2193
 * @ctxt: an XML parser context
2194
 *
2195
 * DEPRECATED: Internal function, do not use.
2196
 *
2197
 * Pops the top element name from the name stack
2198
 *
2199
 * Returns the name just removed
2200
 */
2201
static const xmlChar *
2202
namePop(xmlParserCtxtPtr ctxt)
2203
1.17M
{
2204
1.17M
    const xmlChar *ret;
2205
2206
1.17M
    if ((ctxt == NULL) || (ctxt->nameNr <= 0))
2207
0
        return (NULL);
2208
1.17M
    ctxt->nameNr--;
2209
1.17M
    if (ctxt->nameNr > 0)
2210
1.16M
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2211
6.54k
    else
2212
6.54k
        ctxt->name = NULL;
2213
1.17M
    ret = ctxt->nameTab[ctxt->nameNr];
2214
1.17M
    ctxt->nameTab[ctxt->nameNr] = NULL;
2215
1.17M
    return (ret);
2216
1.17M
}
2217
2218
35.9M
static int spacePush(xmlParserCtxtPtr ctxt, int val) {
2219
35.9M
    if (ctxt->spaceNr >= ctxt->spaceMax) {
2220
1.11M
        int *tmp;
2221
1.11M
        int newSize;
2222
2223
1.11M
        newSize = xmlGrowCapacity(ctxt->spaceMax, sizeof(tmp[0]),
2224
1.11M
                                  10, XML_MAX_ITEMS);
2225
1.11M
        if (newSize < 0) {
2226
0
      xmlErrMemory(ctxt);
2227
0
      return(-1);
2228
0
        }
2229
2230
1.11M
        tmp = xmlRealloc(ctxt->spaceTab, newSize * sizeof(tmp[0]));
2231
1.11M
        if (tmp == NULL) {
2232
0
      xmlErrMemory(ctxt);
2233
0
      return(-1);
2234
0
  }
2235
1.11M
  ctxt->spaceTab = tmp;
2236
2237
1.11M
        ctxt->spaceMax = newSize;
2238
1.11M
    }
2239
35.9M
    ctxt->spaceTab[ctxt->spaceNr] = val;
2240
35.9M
    ctxt->space = &ctxt->spaceTab[ctxt->spaceNr];
2241
35.9M
    return(ctxt->spaceNr++);
2242
35.9M
}
2243
2244
20.7M
static int spacePop(xmlParserCtxtPtr ctxt) {
2245
20.7M
    int ret;
2246
20.7M
    if (ctxt->spaceNr <= 0) return(0);
2247
20.7M
    ctxt->spaceNr--;
2248
20.7M
    if (ctxt->spaceNr > 0)
2249
20.7M
  ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1];
2250
6.54k
    else
2251
6.54k
        ctxt->space = &ctxt->spaceTab[0];
2252
20.7M
    ret = ctxt->spaceTab[ctxt->spaceNr];
2253
20.7M
    ctxt->spaceTab[ctxt->spaceNr] = -1;
2254
20.7M
    return(ret);
2255
20.7M
}
2256
2257
/*
2258
 * Macros for accessing the content. Those should be used only by the parser,
2259
 * and not exported.
2260
 *
2261
 * Dirty macros, i.e. one often need to make assumption on the context to
2262
 * use them
2263
 *
2264
 *   CUR_PTR return the current pointer to the xmlChar to be parsed.
2265
 *           To be used with extreme caution since operations consuming
2266
 *           characters may move the input buffer to a different location !
2267
 *   CUR     returns the current xmlChar value, i.e. a 8 bit value if compiled
2268
 *           This should be used internally by the parser
2269
 *           only to compare to ASCII values otherwise it would break when
2270
 *           running with UTF-8 encoding.
2271
 *   RAW     same as CUR but in the input buffer, bypass any token
2272
 *           extraction that may have been done
2273
 *   NXT(n)  returns the n'th next xmlChar. Same as CUR is should be used only
2274
 *           to compare on ASCII based substring.
2275
 *   SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined
2276
 *           strings without newlines within the parser.
2277
 *   NEXT1(l) Skip 1 xmlChar, and must also be used only to skip 1 non-newline ASCII
2278
 *           defined char within the parser.
2279
 * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding
2280
 *
2281
 *   NEXT    Skip to the next character, this does the proper decoding
2282
 *           in UTF-8 mode. It also pop-up unfinished entities on the fly.
2283
 *   NEXTL(l) Skip the current unicode character of l xmlChars long.
2284
 *   CUR_SCHAR  same but operate on a string instead of the context
2285
 *   COPY_BUF  copy the current unicode char to the target buffer, increment
2286
 *            the index
2287
 *   GROW, SHRINK  handling of input buffers
2288
 */
2289
2290
287M
#define RAW (*ctxt->input->cur)
2291
472M
#define CUR (*ctxt->input->cur)
2292
37.4M
#define NXT(val) ctxt->input->cur[(val)]
2293
727M
#define CUR_PTR ctxt->input->cur
2294
173M
#define BASE_PTR ctxt->input->base
2295
2296
#define CMP4( s, c1, c2, c3, c4 ) \
2297
6.35M
  ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \
2298
3.31M
    ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 )
2299
#define CMP5( s, c1, c2, c3, c4, c5 ) \
2300
5.97M
  ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 )
2301
#define CMP6( s, c1, c2, c3, c4, c5, c6 ) \
2302
5.69M
  ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 )
2303
#define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \
2304
5.27M
  ( CMP6( s, c1, c2, c3, c4, c5, c6 ) && ((unsigned char *) s)[ 6 ] == c7 )
2305
#define CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) \
2306
4.74M
  ( CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) && ((unsigned char *) s)[ 7 ] == c8 )
2307
#define CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) \
2308
2.35M
  ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \
2309
2.35M
    ((unsigned char *) s)[ 8 ] == c9 )
2310
#define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \
2311
119k
  ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \
2312
119k
    ((unsigned char *) s)[ 9 ] == c10 )
2313
2314
22.4M
#define SKIP(val) do {             \
2315
22.4M
    ctxt->input->cur += (val),ctxt->input->col+=(val);      \
2316
22.4M
    if (*ctxt->input->cur == 0)           \
2317
22.4M
        xmlParserGrow(ctxt);           \
2318
22.4M
  } while (0)
2319
2320
#define SKIPL(val) do {             \
2321
    int skipl;                \
2322
    for(skipl=0; skipl<val; skipl++) {          \
2323
  if (*(ctxt->input->cur) == '\n') {        \
2324
  ctxt->input->line++; ctxt->input->col = 1;      \
2325
  } else ctxt->input->col++;          \
2326
  ctxt->input->cur++;           \
2327
    }                 \
2328
    if (*ctxt->input->cur == 0)           \
2329
        xmlParserGrow(ctxt);            \
2330
  } while (0)
2331
2332
#define SHRINK \
2333
3.37M
    if (!PARSER_PROGRESSIVE(ctxt)) \
2334
3.37M
  xmlParserShrink(ctxt);
2335
2336
#define GROW \
2337
251M
    if ((!PARSER_PROGRESSIVE(ctxt)) && \
2338
251M
        (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \
2339
633k
  xmlParserGrow(ctxt);
2340
2341
120M
#define SKIP_BLANKS xmlSkipBlankChars(ctxt)
2342
2343
852k
#define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt)
2344
2345
86.2M
#define NEXT xmlNextChar(ctxt)
2346
2347
46.7M
#define NEXT1 {               \
2348
46.7M
  ctxt->input->col++;           \
2349
46.7M
  ctxt->input->cur++;           \
2350
46.7M
  if (*ctxt->input->cur == 0)         \
2351
46.7M
      xmlParserGrow(ctxt);           \
2352
46.7M
    }
2353
2354
508M
#define NEXTL(l) do {             \
2355
508M
    if (*(ctxt->input->cur) == '\n') {         \
2356
1.74M
  ctxt->input->line++; ctxt->input->col = 1;      \
2357
507M
    } else ctxt->input->col++;           \
2358
508M
    ctxt->input->cur += l;        \
2359
508M
  } while (0)
2360
2361
160k
#define CUR_SCHAR(s, l) xmlStringCurrentChar(ctxt, s, &l)
2362
2363
#define COPY_BUF(b, i, v)           \
2364
141M
    if (v < 0x80) b[i++] = v;           \
2365
141M
    else i += xmlCopyCharMultiByte(&b[i],v)
2366
2367
static int
2368
137M
xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) {
2369
137M
    int c = xmlCurrentChar(ctxt, len);
2370
2371
137M
    if (c == XML_INVALID_CHAR)
2372
517k
        c = 0xFFFD; /* replacement character */
2373
2374
137M
    return(c);
2375
137M
}
2376
2377
/**
2378
 * xmlSkipBlankChars:
2379
 * @ctxt:  the XML parser context
2380
 *
2381
 * DEPRECATED: Internal function, do not use.
2382
 *
2383
 * Skip whitespace in the input stream.
2384
 *
2385
 * Returns the number of space chars skipped
2386
 */
2387
int
2388
121M
xmlSkipBlankChars(xmlParserCtxtPtr ctxt) {
2389
121M
    const xmlChar *cur;
2390
121M
    int res = 0;
2391
2392
121M
    cur = ctxt->input->cur;
2393
121M
    while (IS_BLANK_CH(*cur)) {
2394
39.6M
        if (*cur == '\n') {
2395
1.22M
            ctxt->input->line++; ctxt->input->col = 1;
2396
38.3M
        } else {
2397
38.3M
            ctxt->input->col++;
2398
38.3M
        }
2399
39.6M
        cur++;
2400
39.6M
        if (res < INT_MAX)
2401
39.6M
            res++;
2402
39.6M
        if (*cur == 0) {
2403
14.4k
            ctxt->input->cur = cur;
2404
14.4k
            xmlParserGrow(ctxt);
2405
14.4k
            cur = ctxt->input->cur;
2406
14.4k
        }
2407
39.6M
    }
2408
121M
    ctxt->input->cur = cur;
2409
2410
121M
    if (res > 4)
2411
572k
        GROW;
2412
2413
121M
    return(res);
2414
121M
}
2415
2416
static void
2417
0
xmlPopPE(xmlParserCtxtPtr ctxt) {
2418
0
    unsigned long consumed;
2419
0
    xmlEntityPtr ent;
2420
2421
0
    ent = ctxt->input->entity;
2422
2423
0
    ent->flags &= ~XML_ENT_EXPANDING;
2424
2425
0
    if ((ent->flags & XML_ENT_CHECKED) == 0) {
2426
0
        int result;
2427
2428
        /*
2429
         * Read the rest of the stream in case of errors. We want
2430
         * to account for the whole entity size.
2431
         */
2432
0
        do {
2433
0
            ctxt->input->cur = ctxt->input->end;
2434
0
            xmlParserShrink(ctxt);
2435
0
            result = xmlParserGrow(ctxt);
2436
0
        } while (result > 0);
2437
2438
0
        consumed = ctxt->input->consumed;
2439
0
        xmlSaturatedAddSizeT(&consumed,
2440
0
                             ctxt->input->end - ctxt->input->base);
2441
2442
0
        xmlSaturatedAdd(&ent->expandedSize, consumed);
2443
2444
        /*
2445
         * Add to sizeentities when parsing an external entity
2446
         * for the first time.
2447
         */
2448
0
        if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
2449
0
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
2450
0
        }
2451
2452
0
        ent->flags |= XML_ENT_CHECKED;
2453
0
    }
2454
2455
0
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
2456
2457
0
    xmlParserEntityCheck(ctxt, ent->expandedSize);
2458
2459
0
    GROW;
2460
0
}
2461
2462
/**
2463
 * xmlSkipBlankCharsPE:
2464
 * @ctxt:  the XML parser context
2465
 *
2466
 * Skip whitespace in the input stream, also handling parameter
2467
 * entities.
2468
 *
2469
 * Returns the number of space chars skipped
2470
 */
2471
static int
2472
852k
xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) {
2473
852k
    int res = 0;
2474
852k
    int inParam;
2475
852k
    int expandParam;
2476
2477
852k
    inParam = PARSER_IN_PE(ctxt);
2478
852k
    expandParam = PARSER_EXTERNAL(ctxt);
2479
2480
852k
    if (!inParam && !expandParam)
2481
852k
        return(xmlSkipBlankChars(ctxt));
2482
2483
    /*
2484
     * It's Okay to use CUR/NEXT here since all the blanks are on
2485
     * the ASCII range.
2486
     */
2487
0
    while (PARSER_STOPPED(ctxt) == 0) {
2488
0
        if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */
2489
0
            NEXT;
2490
0
        } else if (CUR == '%') {
2491
0
            if ((expandParam == 0) ||
2492
0
                (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0))
2493
0
                break;
2494
2495
            /*
2496
             * Expand parameter entity. We continue to consume
2497
             * whitespace at the start of the entity and possible
2498
             * even consume the whole entity and pop it. We might
2499
             * even pop multiple PEs in this loop.
2500
             */
2501
0
            xmlParsePEReference(ctxt);
2502
2503
0
            inParam = PARSER_IN_PE(ctxt);
2504
0
            expandParam = PARSER_EXTERNAL(ctxt);
2505
0
        } else if (CUR == 0) {
2506
0
            if (inParam == 0)
2507
0
                break;
2508
2509
0
            xmlPopPE(ctxt);
2510
2511
0
            inParam = PARSER_IN_PE(ctxt);
2512
0
            expandParam = PARSER_EXTERNAL(ctxt);
2513
0
        } else {
2514
0
            break;
2515
0
        }
2516
2517
        /*
2518
         * Also increase the counter when entering or exiting a PERef.
2519
         * The spec says: "When a parameter-entity reference is recognized
2520
         * in the DTD and included, its replacement text MUST be enlarged
2521
         * by the attachment of one leading and one following space (#x20)
2522
         * character."
2523
         */
2524
0
        if (res < INT_MAX)
2525
0
            res++;
2526
0
    }
2527
2528
0
    return(res);
2529
852k
}
2530
2531
/************************************************************************
2532
 *                  *
2533
 *    Commodity functions to handle entities      *
2534
 *                  *
2535
 ************************************************************************/
2536
2537
/**
2538
 * xmlPopInput:
2539
 * @ctxt:  an XML parser context
2540
 *
2541
 * DEPRECATED: Internal function, don't use.
2542
 *
2543
 * Returns the current xmlChar in the parser context
2544
 */
2545
xmlChar
2546
0
xmlPopInput(xmlParserCtxtPtr ctxt) {
2547
0
    xmlParserInputPtr input;
2548
2549
0
    if ((ctxt == NULL) || (ctxt->inputNr <= 1)) return(0);
2550
0
    input = xmlCtxtPopInput(ctxt);
2551
0
    xmlFreeInputStream(input);
2552
0
    if (*ctxt->input->cur == 0)
2553
0
        xmlParserGrow(ctxt);
2554
0
    return(CUR);
2555
0
}
2556
2557
/**
2558
 * xmlPushInput:
2559
 * @ctxt:  an XML parser context
2560
 * @input:  an XML parser input fragment (entity, XML fragment ...).
2561
 *
2562
 * DEPRECATED: Internal function, don't use.
2563
 *
2564
 * Push an input stream onto the stack.
2565
 *
2566
 * Returns -1 in case of error or the index in the input stack
2567
 */
2568
int
2569
0
xmlPushInput(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) {
2570
0
    int ret;
2571
2572
0
    if ((ctxt == NULL) || (input == NULL))
2573
0
        return(-1);
2574
2575
0
    ret = xmlCtxtPushInput(ctxt, input);
2576
0
    if (ret >= 0)
2577
0
        GROW;
2578
0
    return(ret);
2579
0
}
2580
2581
/**
2582
 * xmlParseCharRef:
2583
 * @ctxt:  an XML parser context
2584
 *
2585
 * DEPRECATED: Internal function, don't use.
2586
 *
2587
 * Parse a numeric character reference. Always consumes '&'.
2588
 *
2589
 * [66] CharRef ::= '&#' [0-9]+ ';' |
2590
 *                  '&#x' [0-9a-fA-F]+ ';'
2591
 *
2592
 * [ WFC: Legal Character ]
2593
 * Characters referred to using character references must match the
2594
 * production for Char.
2595
 *
2596
 * Returns the value parsed (as an int), 0 in case of error
2597
 */
2598
int
2599
126k
xmlParseCharRef(xmlParserCtxtPtr ctxt) {
2600
126k
    int val = 0;
2601
126k
    int count = 0;
2602
2603
    /*
2604
     * Using RAW/CUR/NEXT is okay since we are working on ASCII range here
2605
     */
2606
126k
    if ((RAW == '&') && (NXT(1) == '#') &&
2607
126k
        (NXT(2) == 'x')) {
2608
79.4k
  SKIP(3);
2609
79.4k
  GROW;
2610
379k
  while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) {
2611
300k
      if (count++ > 20) {
2612
445
    count = 0;
2613
445
    GROW;
2614
445
      }
2615
300k
      if ((RAW >= '0') && (RAW <= '9'))
2616
194k
          val = val * 16 + (CUR - '0');
2617
106k
      else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20))
2618
69.6k
          val = val * 16 + (CUR - 'a') + 10;
2619
36.5k
      else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20))
2620
36.1k
          val = val * 16 + (CUR - 'A') + 10;
2621
383
      else {
2622
383
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2623
383
    val = 0;
2624
383
    break;
2625
383
      }
2626
300k
      if (val > 0x110000)
2627
2.59k
          val = 0x110000;
2628
2629
300k
      NEXT;
2630
300k
      count++;
2631
300k
  }
2632
79.4k
  if (RAW == ';') {
2633
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2634
79.1k
      ctxt->input->col++;
2635
79.1k
      ctxt->input->cur++;
2636
79.1k
  }
2637
79.4k
    } else if  ((RAW == '&') && (NXT(1) == '#')) {
2638
46.6k
  SKIP(2);
2639
46.6k
  GROW;
2640
155k
  while (RAW != ';') { /* loop blocked by count */
2641
109k
      if (count++ > 20) {
2642
319
    count = 0;
2643
319
    GROW;
2644
319
      }
2645
109k
      if ((RAW >= '0') && (RAW <= '9'))
2646
108k
          val = val * 10 + (CUR - '0');
2647
500
      else {
2648
500
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2649
500
    val = 0;
2650
500
    break;
2651
500
      }
2652
108k
      if (val > 0x110000)
2653
2.63k
          val = 0x110000;
2654
2655
108k
      NEXT;
2656
108k
      count++;
2657
108k
  }
2658
46.6k
  if (RAW == ';') {
2659
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2660
46.1k
      ctxt->input->col++;
2661
46.1k
      ctxt->input->cur++;
2662
46.1k
  }
2663
46.6k
    } else {
2664
0
        if (RAW == '&')
2665
0
            SKIP(1);
2666
0
        xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2667
0
    }
2668
2669
    /*
2670
     * [ WFC: Legal Character ]
2671
     * Characters referred to using character references must match the
2672
     * production for Char.
2673
     */
2674
126k
    if (val >= 0x110000) {
2675
75
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2676
75
                "xmlParseCharRef: character reference out of bounds\n",
2677
75
          val);
2678
126k
    } else if (IS_CHAR(val)) {
2679
124k
        return(val);
2680
124k
    } else {
2681
1.12k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2682
1.12k
                          "xmlParseCharRef: invalid xmlChar value %d\n",
2683
1.12k
                    val);
2684
1.12k
    }
2685
1.19k
    return(0);
2686
126k
}
2687
2688
/**
2689
 * xmlParseStringCharRef:
2690
 * @ctxt:  an XML parser context
2691
 * @str:  a pointer to an index in the string
2692
 *
2693
 * parse Reference declarations, variant parsing from a string rather
2694
 * than an an input flow.
2695
 *
2696
 * [66] CharRef ::= '&#' [0-9]+ ';' |
2697
 *                  '&#x' [0-9a-fA-F]+ ';'
2698
 *
2699
 * [ WFC: Legal Character ]
2700
 * Characters referred to using character references must match the
2701
 * production for Char.
2702
 *
2703
 * Returns the value parsed (as an int), 0 in case of error, str will be
2704
 *         updated to the current value of the index
2705
 */
2706
static int
2707
35.6k
xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) {
2708
35.6k
    const xmlChar *ptr;
2709
35.6k
    xmlChar cur;
2710
35.6k
    int val = 0;
2711
2712
35.6k
    if ((str == NULL) || (*str == NULL)) return(0);
2713
35.6k
    ptr = *str;
2714
35.6k
    cur = *ptr;
2715
35.6k
    if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) {
2716
7.42k
  ptr += 3;
2717
7.42k
  cur = *ptr;
2718
25.9k
  while (cur != ';') { /* Non input consuming loop */
2719
20.6k
      if ((cur >= '0') && (cur <= '9'))
2720
10.0k
          val = val * 16 + (cur - '0');
2721
10.6k
      else if ((cur >= 'a') && (cur <= 'f'))
2722
4.46k
          val = val * 16 + (cur - 'a') + 10;
2723
6.15k
      else if ((cur >= 'A') && (cur <= 'F'))
2724
4.05k
          val = val * 16 + (cur - 'A') + 10;
2725
2.10k
      else {
2726
2.10k
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2727
2.10k
    val = 0;
2728
2.10k
    break;
2729
2.10k
      }
2730
18.5k
      if (val > 0x110000)
2731
619
          val = 0x110000;
2732
2733
18.5k
      ptr++;
2734
18.5k
      cur = *ptr;
2735
18.5k
  }
2736
7.42k
  if (cur == ';')
2737
5.31k
      ptr++;
2738
28.1k
    } else if  ((cur == '&') && (ptr[1] == '#')){
2739
28.1k
  ptr += 2;
2740
28.1k
  cur = *ptr;
2741
156k
  while (cur != ';') { /* Non input consuming loops */
2742
129k
      if ((cur >= '0') && (cur <= '9'))
2743
128k
          val = val * 10 + (cur - '0');
2744
1.07k
      else {
2745
1.07k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2746
1.07k
    val = 0;
2747
1.07k
    break;
2748
1.07k
      }
2749
128k
      if (val > 0x110000)
2750
1.24k
          val = 0x110000;
2751
2752
128k
      ptr++;
2753
128k
      cur = *ptr;
2754
128k
  }
2755
28.1k
  if (cur == ';')
2756
27.1k
      ptr++;
2757
28.1k
    } else {
2758
0
  xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2759
0
  return(0);
2760
0
    }
2761
35.6k
    *str = ptr;
2762
2763
    /*
2764
     * [ WFC: Legal Character ]
2765
     * Characters referred to using character references must match the
2766
     * production for Char.
2767
     */
2768
35.6k
    if (val >= 0x110000) {
2769
52
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2770
52
                "xmlParseStringCharRef: character reference out of bounds\n",
2771
52
                val);
2772
35.5k
    } else if (IS_CHAR(val)) {
2773
31.1k
        return(val);
2774
31.1k
    } else {
2775
4.41k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2776
4.41k
        "xmlParseStringCharRef: invalid xmlChar value %d\n",
2777
4.41k
        val);
2778
4.41k
    }
2779
4.46k
    return(0);
2780
35.6k
}
2781
2782
/**
2783
 * xmlParserHandlePEReference:
2784
 * @ctxt:  the parser context
2785
 *
2786
 * DEPRECATED: Internal function, do not use.
2787
 *
2788
 * [69] PEReference ::= '%' Name ';'
2789
 *
2790
 * [ WFC: No Recursion ]
2791
 * A parsed entity must not contain a recursive
2792
 * reference to itself, either directly or indirectly.
2793
 *
2794
 * [ WFC: Entity Declared ]
2795
 * In a document without any DTD, a document with only an internal DTD
2796
 * subset which contains no parameter entity references, or a document
2797
 * with "standalone='yes'", ...  ... The declaration of a parameter
2798
 * entity must precede any reference to it...
2799
 *
2800
 * [ VC: Entity Declared ]
2801
 * In a document with an external subset or external parameter entities
2802
 * with "standalone='no'", ...  ... The declaration of a parameter entity
2803
 * must precede any reference to it...
2804
 *
2805
 * [ WFC: In DTD ]
2806
 * Parameter-entity references may only appear in the DTD.
2807
 * NOTE: misleading but this is handled.
2808
 *
2809
 * A PEReference may have been detected in the current input stream
2810
 * the handling is done accordingly to
2811
 *      http://www.w3.org/TR/REC-xml#entproc
2812
 * i.e.
2813
 *   - Included in literal in entity values
2814
 *   - Included as Parameter Entity reference within DTDs
2815
 */
2816
void
2817
0
xmlParserHandlePEReference(xmlParserCtxtPtr ctxt) {
2818
0
    xmlParsePEReference(ctxt);
2819
0
}
2820
2821
/**
2822
 * xmlStringLenDecodeEntities:
2823
 * @ctxt:  the parser context
2824
 * @str:  the input string
2825
 * @len: the string length
2826
 * @what:  combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF
2827
 * @end:  an end marker xmlChar, 0 if none
2828
 * @end2:  an end marker xmlChar, 0 if none
2829
 * @end3:  an end marker xmlChar, 0 if none
2830
 *
2831
 * DEPRECATED: Internal function, don't use.
2832
 *
2833
 * Returns A newly allocated string with the substitution done. The caller
2834
 *      must deallocate it !
2835
 */
2836
xmlChar *
2837
xmlStringLenDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, int len,
2838
                           int what ATTRIBUTE_UNUSED,
2839
0
                           xmlChar end, xmlChar end2, xmlChar end3) {
2840
0
    if ((ctxt == NULL) || (str == NULL) || (len < 0))
2841
0
        return(NULL);
2842
2843
0
    if ((str[len] != 0) ||
2844
0
        (end != 0) || (end2 != 0) || (end3 != 0))
2845
0
        return(NULL);
2846
2847
0
    return(xmlExpandEntitiesInAttValue(ctxt, str, 0));
2848
0
}
2849
2850
/**
2851
 * xmlStringDecodeEntities:
2852
 * @ctxt:  the parser context
2853
 * @str:  the input string
2854
 * @what:  combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF
2855
 * @end:  an end marker xmlChar, 0 if none
2856
 * @end2:  an end marker xmlChar, 0 if none
2857
 * @end3:  an end marker xmlChar, 0 if none
2858
 *
2859
 * DEPRECATED: Internal function, don't use.
2860
 *
2861
 * Returns A newly allocated string with the substitution done. The caller
2862
 *      must deallocate it !
2863
 */
2864
xmlChar *
2865
xmlStringDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str,
2866
                        int what ATTRIBUTE_UNUSED,
2867
0
            xmlChar end, xmlChar  end2, xmlChar end3) {
2868
0
    if ((ctxt == NULL) || (str == NULL))
2869
0
        return(NULL);
2870
2871
0
    if ((end != 0) || (end2 != 0) || (end3 != 0))
2872
0
        return(NULL);
2873
2874
0
    return(xmlExpandEntitiesInAttValue(ctxt, str, 0));
2875
0
}
2876
2877
/************************************************************************
2878
 *                  *
2879
 *    Commodity functions, cleanup needed ?     *
2880
 *                  *
2881
 ************************************************************************/
2882
2883
/**
2884
 * areBlanks:
2885
 * @ctxt:  an XML parser context
2886
 * @str:  a xmlChar *
2887
 * @len:  the size of @str
2888
 * @blank_chars: we know the chars are blanks
2889
 *
2890
 * Is this a sequence of blank chars that one can ignore ?
2891
 *
2892
 * Returns 1 if ignorable 0 otherwise.
2893
 */
2894
2895
static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len,
2896
14.1M
                     int blank_chars) {
2897
14.1M
    int i;
2898
14.1M
    xmlNodePtr lastChild;
2899
2900
    /*
2901
     * Check for xml:space value.
2902
     */
2903
14.1M
    if ((ctxt->space == NULL) || (*(ctxt->space) == 1) ||
2904
14.1M
        (*(ctxt->space) == -2))
2905
6.08M
  return(0);
2906
2907
    /*
2908
     * Check that the string is made of blanks
2909
     */
2910
8.02M
    if (blank_chars == 0) {
2911
8.36M
  for (i = 0;i < len;i++)
2912
8.17M
      if (!(IS_BLANK_CH(str[i]))) return(0);
2913
6.78M
    }
2914
2915
    /*
2916
     * Look if the element is mixed content in the DTD if available
2917
     */
2918
1.43M
    if (ctxt->node == NULL) return(0);
2919
57
    if (ctxt->myDoc != NULL) {
2920
0
        xmlElementPtr elemDecl = NULL;
2921
0
        xmlDocPtr doc = ctxt->myDoc;
2922
0
        const xmlChar *prefix = NULL;
2923
2924
0
        if (ctxt->node->ns)
2925
0
            prefix = ctxt->node->ns->prefix;
2926
0
        if (doc->intSubset != NULL)
2927
0
            elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name,
2928
0
                                      prefix);
2929
0
        if ((elemDecl == NULL) && (doc->extSubset != NULL))
2930
0
            elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name,
2931
0
                                      prefix);
2932
0
        if (elemDecl != NULL) {
2933
0
            if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)
2934
0
                return(1);
2935
0
            if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) ||
2936
0
                (elemDecl->etype == XML_ELEMENT_TYPE_MIXED))
2937
0
                return(0);
2938
0
        }
2939
0
    }
2940
2941
    /*
2942
     * Otherwise, heuristic :-\
2943
     *
2944
     * When push parsing, we could be at the end of a chunk.
2945
     * This makes the look-ahead and consequently the NOBLANKS
2946
     * option unreliable.
2947
     */
2948
57
    if ((RAW != '<') && (RAW != 0xD)) return(0);
2949
57
    if ((ctxt->node->children == NULL) &&
2950
57
  (RAW == '<') && (NXT(1) == '/')) return(0);
2951
2952
57
    lastChild = xmlGetLastChild(ctxt->node);
2953
57
    if (lastChild == NULL) {
2954
0
        if ((ctxt->node->type != XML_ELEMENT_NODE) &&
2955
0
            (ctxt->node->content != NULL)) return(0);
2956
57
    } else if (xmlNodeIsText(lastChild))
2957
0
        return(0);
2958
57
    else if ((ctxt->node->children != NULL) &&
2959
57
             (xmlNodeIsText(ctxt->node->children)))
2960
0
        return(0);
2961
57
    return(1);
2962
57
}
2963
2964
/************************************************************************
2965
 *                  *
2966
 *    Extra stuff for namespace support     *
2967
 *  Relates to http://www.w3.org/TR/WD-xml-names      *
2968
 *                  *
2969
 ************************************************************************/
2970
2971
/**
2972
 * xmlSplitQName:
2973
 * @ctxt:  an XML parser context
2974
 * @name:  an XML parser context
2975
 * @prefixOut:  a xmlChar **
2976
 *
2977
 * DEPRECATED: Don't use.
2978
 *
2979
 * parse an UTF8 encoded XML qualified name string
2980
 *
2981
 * [NS 5] QName ::= (Prefix ':')? LocalPart
2982
 *
2983
 * [NS 6] Prefix ::= NCName
2984
 *
2985
 * [NS 7] LocalPart ::= NCName
2986
 *
2987
 * Returns the local part, and prefix is updated
2988
 *   to get the Prefix if any.
2989
 */
2990
2991
xmlChar *
2992
0
xmlSplitQName(xmlParserCtxtPtr ctxt, const xmlChar *name, xmlChar **prefixOut) {
2993
0
    xmlChar buf[XML_MAX_NAMELEN + 5];
2994
0
    xmlChar *buffer = NULL;
2995
0
    int len = 0;
2996
0
    int max = XML_MAX_NAMELEN;
2997
0
    xmlChar *ret = NULL;
2998
0
    xmlChar *prefix;
2999
0
    const xmlChar *cur = name;
3000
0
    int c;
3001
3002
0
    if (prefixOut == NULL) return(NULL);
3003
0
    *prefixOut = NULL;
3004
3005
0
    if (cur == NULL) return(NULL);
3006
3007
    /* nasty but well=formed */
3008
0
    if (cur[0] == ':')
3009
0
  return(xmlStrdup(name));
3010
3011
0
    c = *cur++;
3012
0
    while ((c != 0) && (c != ':') && (len < max)) { /* tested bigname.xml */
3013
0
  buf[len++] = c;
3014
0
  c = *cur++;
3015
0
    }
3016
0
    if (len >= max) {
3017
  /*
3018
   * Okay someone managed to make a huge name, so he's ready to pay
3019
   * for the processing speed.
3020
   */
3021
0
  max = len * 2;
3022
3023
0
  buffer = xmlMalloc(max);
3024
0
  if (buffer == NULL) {
3025
0
      xmlErrMemory(ctxt);
3026
0
      return(NULL);
3027
0
  }
3028
0
  memcpy(buffer, buf, len);
3029
0
  while ((c != 0) && (c != ':')) { /* tested bigname.xml */
3030
0
      if (len + 10 > max) {
3031
0
          xmlChar *tmp;
3032
0
                int newSize;
3033
3034
0
                newSize = xmlGrowCapacity(max, 1, 1, XML_MAX_ITEMS);
3035
0
                if (newSize < 0) {
3036
0
        xmlErrMemory(ctxt);
3037
0
        xmlFree(buffer);
3038
0
        return(NULL);
3039
0
                }
3040
0
    tmp = xmlRealloc(buffer, newSize);
3041
0
    if (tmp == NULL) {
3042
0
        xmlErrMemory(ctxt);
3043
0
        xmlFree(buffer);
3044
0
        return(NULL);
3045
0
    }
3046
0
    buffer = tmp;
3047
0
    max = newSize;
3048
0
      }
3049
0
      buffer[len++] = c;
3050
0
      c = *cur++;
3051
0
  }
3052
0
  buffer[len] = 0;
3053
0
    }
3054
3055
0
    if ((c == ':') && (*cur == 0)) {
3056
0
        if (buffer != NULL)
3057
0
      xmlFree(buffer);
3058
0
  return(xmlStrdup(name));
3059
0
    }
3060
3061
0
    if (buffer == NULL) {
3062
0
  ret = xmlStrndup(buf, len);
3063
0
        if (ret == NULL) {
3064
0
      xmlErrMemory(ctxt);
3065
0
      return(NULL);
3066
0
        }
3067
0
    } else {
3068
0
  ret = buffer;
3069
0
  buffer = NULL;
3070
0
  max = XML_MAX_NAMELEN;
3071
0
    }
3072
3073
3074
0
    if (c == ':') {
3075
0
  c = *cur;
3076
0
        prefix = ret;
3077
0
  if (c == 0) {
3078
0
      ret = xmlStrndup(BAD_CAST "", 0);
3079
0
            if (ret == NULL) {
3080
0
                xmlFree(prefix);
3081
0
                return(NULL);
3082
0
            }
3083
0
            *prefixOut = prefix;
3084
0
            return(ret);
3085
0
  }
3086
0
  len = 0;
3087
3088
  /*
3089
   * Check that the first character is proper to start
3090
   * a new name
3091
   */
3092
0
  if (!(((c >= 0x61) && (c <= 0x7A)) ||
3093
0
        ((c >= 0x41) && (c <= 0x5A)) ||
3094
0
        (c == '_') || (c == ':'))) {
3095
0
      int l;
3096
0
      int first = CUR_SCHAR(cur, l);
3097
3098
0
      if (!IS_LETTER(first) && (first != '_')) {
3099
0
    xmlFatalErrMsgStr(ctxt, XML_NS_ERR_QNAME,
3100
0
          "Name %s is not XML Namespace compliant\n",
3101
0
          name);
3102
0
      }
3103
0
  }
3104
0
  cur++;
3105
3106
0
  while ((c != 0) && (len < max)) { /* tested bigname2.xml */
3107
0
      buf[len++] = c;
3108
0
      c = *cur++;
3109
0
  }
3110
0
  if (len >= max) {
3111
      /*
3112
       * Okay someone managed to make a huge name, so he's ready to pay
3113
       * for the processing speed.
3114
       */
3115
0
      max = len * 2;
3116
3117
0
      buffer = xmlMalloc(max);
3118
0
      if (buffer == NULL) {
3119
0
          xmlErrMemory(ctxt);
3120
0
                xmlFree(prefix);
3121
0
    return(NULL);
3122
0
      }
3123
0
      memcpy(buffer, buf, len);
3124
0
      while (c != 0) { /* tested bigname2.xml */
3125
0
    if (len + 10 > max) {
3126
0
        xmlChar *tmp;
3127
0
                    int newSize;
3128
3129
0
                    newSize = xmlGrowCapacity(max, 1, 1, XML_MAX_ITEMS);
3130
0
                    if (newSize < 0) {
3131
0
                        xmlErrMemory(ctxt);
3132
0
                        xmlFree(buffer);
3133
0
                        return(NULL);
3134
0
                    }
3135
0
        tmp = xmlRealloc(buffer, newSize);
3136
0
        if (tmp == NULL) {
3137
0
      xmlErrMemory(ctxt);
3138
0
                        xmlFree(prefix);
3139
0
      xmlFree(buffer);
3140
0
      return(NULL);
3141
0
        }
3142
0
        buffer = tmp;
3143
0
                    max = newSize;
3144
0
    }
3145
0
    buffer[len++] = c;
3146
0
    c = *cur++;
3147
0
      }
3148
0
      buffer[len] = 0;
3149
0
  }
3150
3151
0
  if (buffer == NULL) {
3152
0
      ret = xmlStrndup(buf, len);
3153
0
            if (ret == NULL) {
3154
0
                xmlFree(prefix);
3155
0
                return(NULL);
3156
0
            }
3157
0
  } else {
3158
0
      ret = buffer;
3159
0
  }
3160
3161
0
        *prefixOut = prefix;
3162
0
    }
3163
3164
0
    return(ret);
3165
0
}
3166
3167
/************************************************************************
3168
 *                  *
3169
 *      The parser itself       *
3170
 *  Relates to http://www.w3.org/TR/REC-xml       *
3171
 *                  *
3172
 ************************************************************************/
3173
3174
/************************************************************************
3175
 *                  *
3176
 *  Routines to parse Name, NCName and NmToken      *
3177
 *                  *
3178
 ************************************************************************/
3179
3180
/*
3181
 * The two following functions are related to the change of accepted
3182
 * characters for Name and NmToken in the Revision 5 of XML-1.0
3183
 * They correspond to the modified production [4] and the new production [4a]
3184
 * changes in that revision. Also note that the macros used for the
3185
 * productions Letter, Digit, CombiningChar and Extender are not needed
3186
 * anymore.
3187
 * We still keep compatibility to pre-revision5 parsing semantic if the
3188
 * new XML_PARSE_OLD10 option is given to the parser.
3189
 */
3190
static int
3191
1.43M
xmlIsNameStartChar(xmlParserCtxtPtr ctxt, int c) {
3192
1.43M
    if ((ctxt->options & XML_PARSE_OLD10) == 0) {
3193
        /*
3194
   * Use the new checks of production [4] [4a] amd [5] of the
3195
   * Update 5 of XML-1.0
3196
   */
3197
1.43M
  if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3198
1.43M
      (((c >= 'a') && (c <= 'z')) ||
3199
1.42M
       ((c >= 'A') && (c <= 'Z')) ||
3200
1.42M
       (c == '_') || (c == ':') ||
3201
1.42M
       ((c >= 0xC0) && (c <= 0xD6)) ||
3202
1.42M
       ((c >= 0xD8) && (c <= 0xF6)) ||
3203
1.42M
       ((c >= 0xF8) && (c <= 0x2FF)) ||
3204
1.42M
       ((c >= 0x370) && (c <= 0x37D)) ||
3205
1.42M
       ((c >= 0x37F) && (c <= 0x1FFF)) ||
3206
1.42M
       ((c >= 0x200C) && (c <= 0x200D)) ||
3207
1.42M
       ((c >= 0x2070) && (c <= 0x218F)) ||
3208
1.42M
       ((c >= 0x2C00) && (c <= 0x2FEF)) ||
3209
1.42M
       ((c >= 0x3001) && (c <= 0xD7FF)) ||
3210
1.42M
       ((c >= 0xF900) && (c <= 0xFDCF)) ||
3211
1.42M
       ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
3212
1.42M
       ((c >= 0x10000) && (c <= 0xEFFFF))))
3213
1.31M
      return(1);
3214
1.43M
    } else {
3215
0
        if (IS_LETTER(c) || (c == '_') || (c == ':'))
3216
0
      return(1);
3217
0
    }
3218
115k
    return(0);
3219
1.43M
}
3220
3221
static int
3222
31.2M
xmlIsNameChar(xmlParserCtxtPtr ctxt, int c) {
3223
31.2M
    if ((ctxt->options & XML_PARSE_OLD10) == 0) {
3224
        /*
3225
   * Use the new checks of production [4] [4a] amd [5] of the
3226
   * Update 5 of XML-1.0
3227
   */
3228
31.2M
  if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3229
31.2M
      (((c >= 'a') && (c <= 'z')) ||
3230
30.8M
       ((c >= 'A') && (c <= 'Z')) ||
3231
30.8M
       ((c >= '0') && (c <= '9')) || /* !start */
3232
30.8M
       (c == '_') || (c == ':') ||
3233
30.8M
       (c == '-') || (c == '.') || (c == 0xB7) || /* !start */
3234
30.8M
       ((c >= 0xC0) && (c <= 0xD6)) ||
3235
30.8M
       ((c >= 0xD8) && (c <= 0xF6)) ||
3236
30.8M
       ((c >= 0xF8) && (c <= 0x2FF)) ||
3237
30.8M
       ((c >= 0x300) && (c <= 0x36F)) || /* !start */
3238
30.8M
       ((c >= 0x370) && (c <= 0x37D)) ||
3239
30.8M
       ((c >= 0x37F) && (c <= 0x1FFF)) ||
3240
30.8M
       ((c >= 0x200C) && (c <= 0x200D)) ||
3241
30.8M
       ((c >= 0x203F) && (c <= 0x2040)) || /* !start */
3242
30.8M
       ((c >= 0x2070) && (c <= 0x218F)) ||
3243
30.8M
       ((c >= 0x2C00) && (c <= 0x2FEF)) ||
3244
30.8M
       ((c >= 0x3001) && (c <= 0xD7FF)) ||
3245
30.8M
       ((c >= 0xF900) && (c <= 0xFDCF)) ||
3246
30.8M
       ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
3247
30.8M
       ((c >= 0x10000) && (c <= 0xEFFFF))))
3248
30.6M
       return(1);
3249
31.2M
    } else {
3250
0
        if ((IS_LETTER(c)) || (IS_DIGIT(c)) ||
3251
0
            (c == '.') || (c == '-') ||
3252
0
      (c == '_') || (c == ':') ||
3253
0
      (IS_COMBINING(c)) ||
3254
0
      (IS_EXTENDER(c)))
3255
0
      return(1);
3256
0
    }
3257
601k
    return(0);
3258
31.2M
}
3259
3260
static const xmlChar *
3261
460k
xmlParseNameComplex(xmlParserCtxtPtr ctxt) {
3262
460k
    const xmlChar *ret;
3263
460k
    int len = 0, l;
3264
460k
    int c;
3265
460k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3266
459k
                    XML_MAX_TEXT_LENGTH :
3267
460k
                    XML_MAX_NAME_LENGTH;
3268
3269
    /*
3270
     * Handler for more complex cases
3271
     */
3272
460k
    c = xmlCurrentChar(ctxt, &l);
3273
460k
    if ((ctxt->options & XML_PARSE_OLD10) == 0) {
3274
        /*
3275
   * Use the new checks of production [4] [4a] amd [5] of the
3276
   * Update 5 of XML-1.0
3277
   */
3278
460k
  if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */
3279
460k
      (!(((c >= 'a') && (c <= 'z')) ||
3280
449k
         ((c >= 'A') && (c <= 'Z')) ||
3281
449k
         (c == '_') || (c == ':') ||
3282
449k
         ((c >= 0xC0) && (c <= 0xD6)) ||
3283
449k
         ((c >= 0xD8) && (c <= 0xF6)) ||
3284
449k
         ((c >= 0xF8) && (c <= 0x2FF)) ||
3285
449k
         ((c >= 0x370) && (c <= 0x37D)) ||
3286
449k
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
3287
449k
         ((c >= 0x200C) && (c <= 0x200D)) ||
3288
449k
         ((c >= 0x2070) && (c <= 0x218F)) ||
3289
449k
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
3290
449k
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
3291
449k
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
3292
449k
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
3293
449k
         ((c >= 0x10000) && (c <= 0xEFFFF))))) {
3294
291k
      return(NULL);
3295
291k
  }
3296
168k
  len += l;
3297
168k
  NEXTL(l);
3298
168k
  c = xmlCurrentChar(ctxt, &l);
3299
1.08M
  while ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3300
1.08M
         (((c >= 'a') && (c <= 'z')) ||
3301
1.07M
          ((c >= 'A') && (c <= 'Z')) ||
3302
1.07M
          ((c >= '0') && (c <= '9')) || /* !start */
3303
1.07M
          (c == '_') || (c == ':') ||
3304
1.07M
          (c == '-') || (c == '.') || (c == 0xB7) || /* !start */
3305
1.07M
          ((c >= 0xC0) && (c <= 0xD6)) ||
3306
1.07M
          ((c >= 0xD8) && (c <= 0xF6)) ||
3307
1.07M
          ((c >= 0xF8) && (c <= 0x2FF)) ||
3308
1.07M
          ((c >= 0x300) && (c <= 0x36F)) || /* !start */
3309
1.07M
          ((c >= 0x370) && (c <= 0x37D)) ||
3310
1.07M
          ((c >= 0x37F) && (c <= 0x1FFF)) ||
3311
1.07M
          ((c >= 0x200C) && (c <= 0x200D)) ||
3312
1.07M
          ((c >= 0x203F) && (c <= 0x2040)) || /* !start */
3313
1.07M
          ((c >= 0x2070) && (c <= 0x218F)) ||
3314
1.07M
          ((c >= 0x2C00) && (c <= 0x2FEF)) ||
3315
1.07M
          ((c >= 0x3001) && (c <= 0xD7FF)) ||
3316
1.07M
          ((c >= 0xF900) && (c <= 0xFDCF)) ||
3317
1.07M
          ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
3318
1.07M
          ((c >= 0x10000) && (c <= 0xEFFFF))
3319
1.07M
    )) {
3320
920k
            if (len <= INT_MAX - l)
3321
920k
          len += l;
3322
920k
      NEXTL(l);
3323
920k
      c = xmlCurrentChar(ctxt, &l);
3324
920k
  }
3325
168k
    } else {
3326
0
  if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */
3327
0
      (!IS_LETTER(c) && (c != '_') &&
3328
0
       (c != ':'))) {
3329
0
      return(NULL);
3330
0
  }
3331
0
  len += l;
3332
0
  NEXTL(l);
3333
0
  c = xmlCurrentChar(ctxt, &l);
3334
3335
0
  while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */
3336
0
         ((IS_LETTER(c)) || (IS_DIGIT(c)) ||
3337
0
    (c == '.') || (c == '-') ||
3338
0
    (c == '_') || (c == ':') ||
3339
0
    (IS_COMBINING(c)) ||
3340
0
    (IS_EXTENDER(c)))) {
3341
0
            if (len <= INT_MAX - l)
3342
0
          len += l;
3343
0
      NEXTL(l);
3344
0
      c = xmlCurrentChar(ctxt, &l);
3345
0
  }
3346
0
    }
3347
168k
    if (len > maxLength) {
3348
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3349
0
        return(NULL);
3350
0
    }
3351
168k
    if (ctxt->input->cur - ctxt->input->base < len) {
3352
        /*
3353
         * There were a couple of bugs where PERefs lead to to a change
3354
         * of the buffer. Check the buffer size to avoid passing an invalid
3355
         * pointer to xmlDictLookup.
3356
         */
3357
0
        xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR,
3358
0
                    "unexpected change of input buffer");
3359
0
        return (NULL);
3360
0
    }
3361
168k
    if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r'))
3362
629
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len);
3363
167k
    else
3364
167k
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len);
3365
168k
    if (ret == NULL)
3366
0
        xmlErrMemory(ctxt);
3367
168k
    return(ret);
3368
168k
}
3369
3370
/**
3371
 * xmlParseName:
3372
 * @ctxt:  an XML parser context
3373
 *
3374
 * DEPRECATED: Internal function, don't use.
3375
 *
3376
 * parse an XML name.
3377
 *
3378
 * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' |
3379
 *                  CombiningChar | Extender
3380
 *
3381
 * [5] Name ::= (Letter | '_' | ':') (NameChar)*
3382
 *
3383
 * [6] Names ::= Name (#x20 Name)*
3384
 *
3385
 * Returns the Name parsed or NULL
3386
 */
3387
3388
const xmlChar *
3389
1.27M
xmlParseName(xmlParserCtxtPtr ctxt) {
3390
1.27M
    const xmlChar *in;
3391
1.27M
    const xmlChar *ret;
3392
1.27M
    size_t count = 0;
3393
1.27M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3394
1.27M
                       XML_MAX_TEXT_LENGTH :
3395
1.27M
                       XML_MAX_NAME_LENGTH;
3396
3397
1.27M
    GROW;
3398
3399
    /*
3400
     * Accelerator for simple ASCII names
3401
     */
3402
1.27M
    in = ctxt->input->cur;
3403
1.27M
    if (((*in >= 0x61) && (*in <= 0x7A)) ||
3404
1.27M
  ((*in >= 0x41) && (*in <= 0x5A)) ||
3405
1.27M
  (*in == '_') || (*in == ':')) {
3406
889k
  in++;
3407
4.28M
  while (((*in >= 0x61) && (*in <= 0x7A)) ||
3408
4.28M
         ((*in >= 0x41) && (*in <= 0x5A)) ||
3409
4.28M
         ((*in >= 0x30) && (*in <= 0x39)) ||
3410
4.28M
         (*in == '_') || (*in == '-') ||
3411
4.28M
         (*in == ':') || (*in == '.'))
3412
3.39M
      in++;
3413
889k
  if ((*in > 0) && (*in < 0x80)) {
3414
816k
      count = in - ctxt->input->cur;
3415
816k
            if (count > maxLength) {
3416
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3417
0
                return(NULL);
3418
0
            }
3419
816k
      ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count);
3420
816k
      ctxt->input->cur = in;
3421
816k
      ctxt->input->col += count;
3422
816k
      if (ret == NULL)
3423
0
          xmlErrMemory(ctxt);
3424
816k
      return(ret);
3425
816k
  }
3426
889k
    }
3427
    /* accelerator for special cases */
3428
460k
    return(xmlParseNameComplex(ctxt));
3429
1.27M
}
3430
3431
static xmlHashedString
3432
1.48M
xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) {
3433
1.48M
    xmlHashedString ret;
3434
1.48M
    int len = 0, l;
3435
1.48M
    int c;
3436
1.48M
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3437
1.44M
                    XML_MAX_TEXT_LENGTH :
3438
1.48M
                    XML_MAX_NAME_LENGTH;
3439
1.48M
    size_t startPosition = 0;
3440
3441
1.48M
    ret.name = NULL;
3442
1.48M
    ret.hashValue = 0;
3443
3444
    /*
3445
     * Handler for more complex cases
3446
     */
3447
1.48M
    startPosition = CUR_PTR - BASE_PTR;
3448
1.48M
    c = xmlCurrentChar(ctxt, &l);
3449
1.48M
    if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */
3450
1.48M
  (!xmlIsNameStartChar(ctxt, c) || (c == ':'))) {
3451
389k
  return(ret);
3452
389k
    }
3453
3454
27.1M
    while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */
3455
27.1M
     (xmlIsNameChar(ctxt, c) && (c != ':'))) {
3456
26.0M
        if (len <= INT_MAX - l)
3457
26.0M
      len += l;
3458
26.0M
  NEXTL(l);
3459
26.0M
  c = xmlCurrentChar(ctxt, &l);
3460
26.0M
    }
3461
1.09M
    if (len > maxLength) {
3462
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3463
0
        return(ret);
3464
0
    }
3465
1.09M
    ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len);
3466
1.09M
    if (ret.name == NULL)
3467
0
        xmlErrMemory(ctxt);
3468
1.09M
    return(ret);
3469
1.09M
}
3470
3471
/**
3472
 * xmlParseNCName:
3473
 * @ctxt:  an XML parser context
3474
 * @len:  length of the string parsed
3475
 *
3476
 * parse an XML name.
3477
 *
3478
 * [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' |
3479
 *                      CombiningChar | Extender
3480
 *
3481
 * [5NS] NCName ::= (Letter | '_') (NCNameChar)*
3482
 *
3483
 * Returns the Name parsed or NULL
3484
 */
3485
3486
static xmlHashedString
3487
90.8M
xmlParseNCName(xmlParserCtxtPtr ctxt) {
3488
90.8M
    const xmlChar *in, *e;
3489
90.8M
    xmlHashedString ret;
3490
90.8M
    size_t count = 0;
3491
90.8M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3492
87.0M
                       XML_MAX_TEXT_LENGTH :
3493
90.8M
                       XML_MAX_NAME_LENGTH;
3494
3495
90.8M
    ret.name = NULL;
3496
3497
    /*
3498
     * Accelerator for simple ASCII names
3499
     */
3500
90.8M
    in = ctxt->input->cur;
3501
90.8M
    e = ctxt->input->end;
3502
90.8M
    if ((((*in >= 0x61) && (*in <= 0x7A)) ||
3503
90.8M
   ((*in >= 0x41) && (*in <= 0x5A)) ||
3504
90.8M
   (*in == '_')) && (in < e)) {
3505
90.2M
  in++;
3506
498M
  while ((((*in >= 0x61) && (*in <= 0x7A)) ||
3507
498M
          ((*in >= 0x41) && (*in <= 0x5A)) ||
3508
498M
          ((*in >= 0x30) && (*in <= 0x39)) ||
3509
498M
          (*in == '_') || (*in == '-') ||
3510
498M
          (*in == '.')) && (in < e))
3511
408M
      in++;
3512
90.2M
  if (in >= e)
3513
15.0k
      goto complex;
3514
90.2M
  if ((*in > 0) && (*in < 0x80)) {
3515
89.3M
      count = in - ctxt->input->cur;
3516
89.3M
            if (count > maxLength) {
3517
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3518
0
                return(ret);
3519
0
            }
3520
89.3M
      ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count);
3521
89.3M
      ctxt->input->cur = in;
3522
89.3M
      ctxt->input->col += count;
3523
89.3M
      if (ret.name == NULL) {
3524
0
          xmlErrMemory(ctxt);
3525
0
      }
3526
89.3M
      return(ret);
3527
89.3M
  }
3528
90.2M
    }
3529
1.48M
complex:
3530
1.48M
    return(xmlParseNCNameComplex(ctxt));
3531
90.8M
}
3532
3533
/**
3534
 * xmlParseNameAndCompare:
3535
 * @ctxt:  an XML parser context
3536
 *
3537
 * parse an XML name and compares for match
3538
 * (specialized for endtag parsing)
3539
 *
3540
 * Returns NULL for an illegal name, (xmlChar*) 1 for success
3541
 * and the name for mismatch
3542
 */
3543
3544
static const xmlChar *
3545
3.27M
xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) {
3546
3.27M
    register const xmlChar *cmp = other;
3547
3.27M
    register const xmlChar *in;
3548
3.27M
    const xmlChar *ret;
3549
3550
3.27M
    GROW;
3551
3552
3.27M
    in = ctxt->input->cur;
3553
10.7M
    while (*in != 0 && *in == *cmp) {
3554
7.46M
  ++in;
3555
7.46M
  ++cmp;
3556
7.46M
    }
3557
3.27M
    if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
3558
  /* success */
3559
3.26M
  ctxt->input->col += in - ctxt->input->cur;
3560
3.26M
  ctxt->input->cur = in;
3561
3.26M
  return (const xmlChar*) 1;
3562
3.26M
    }
3563
    /* failure (or end of input buffer), check with full function */
3564
4.91k
    ret = xmlParseName (ctxt);
3565
    /* strings coming from the dictionary direct compare possible */
3566
4.91k
    if (ret == other) {
3567
70
  return (const xmlChar*) 1;
3568
70
    }
3569
4.84k
    return ret;
3570
4.91k
}
3571
3572
/**
3573
 * xmlParseStringName:
3574
 * @ctxt:  an XML parser context
3575
 * @str:  a pointer to the string pointer (IN/OUT)
3576
 *
3577
 * parse an XML name.
3578
 *
3579
 * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' |
3580
 *                  CombiningChar | Extender
3581
 *
3582
 * [5] Name ::= (Letter | '_' | ':') (NameChar)*
3583
 *
3584
 * [6] Names ::= Name (#x20 Name)*
3585
 *
3586
 * Returns the Name parsed or NULL. The @str pointer
3587
 * is updated to the current location in the string.
3588
 */
3589
3590
static xmlChar *
3591
29.9k
xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) {
3592
29.9k
    xmlChar buf[XML_MAX_NAMELEN + 5];
3593
29.9k
    xmlChar *ret;
3594
29.9k
    const xmlChar *cur = *str;
3595
29.9k
    int len = 0, l;
3596
29.9k
    int c;
3597
29.9k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3598
29.9k
                    XML_MAX_TEXT_LENGTH :
3599
29.9k
                    XML_MAX_NAME_LENGTH;
3600
3601
29.9k
    c = CUR_SCHAR(cur, l);
3602
29.9k
    if (!xmlIsNameStartChar(ctxt, c)) {
3603
4.79k
  return(NULL);
3604
4.79k
    }
3605
3606
25.1k
    COPY_BUF(buf, len, c);
3607
25.1k
    cur += l;
3608
25.1k
    c = CUR_SCHAR(cur, l);
3609
87.5k
    while (xmlIsNameChar(ctxt, c)) {
3610
63.1k
  COPY_BUF(buf, len, c);
3611
63.1k
  cur += l;
3612
63.1k
  c = CUR_SCHAR(cur, l);
3613
63.1k
  if (len >= XML_MAX_NAMELEN) { /* test bigentname.xml */
3614
      /*
3615
       * Okay someone managed to make a huge name, so he's ready to pay
3616
       * for the processing speed.
3617
       */
3618
770
      xmlChar *buffer;
3619
770
      int max = len * 2;
3620
3621
770
      buffer = xmlMalloc(max);
3622
770
      if (buffer == NULL) {
3623
0
          xmlErrMemory(ctxt);
3624
0
    return(NULL);
3625
0
      }
3626
770
      memcpy(buffer, buf, len);
3627
42.8k
      while (xmlIsNameChar(ctxt, c)) {
3628
42.1k
    if (len + 10 > max) {
3629
358
        xmlChar *tmp;
3630
358
                    int newSize;
3631
3632
358
                    newSize = xmlGrowCapacity(max, 1, 1, maxLength);
3633
358
                    if (newSize < 0) {
3634
0
                        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3635
0
                        xmlFree(buffer);
3636
0
                        return(NULL);
3637
0
                    }
3638
358
        tmp = xmlRealloc(buffer, newSize);
3639
358
        if (tmp == NULL) {
3640
0
      xmlErrMemory(ctxt);
3641
0
      xmlFree(buffer);
3642
0
      return(NULL);
3643
0
        }
3644
358
        buffer = tmp;
3645
358
                    max = newSize;
3646
358
    }
3647
42.1k
    COPY_BUF(buffer, len, c);
3648
42.1k
    cur += l;
3649
42.1k
    c = CUR_SCHAR(cur, l);
3650
42.1k
      }
3651
770
      buffer[len] = 0;
3652
770
      *str = cur;
3653
770
      return(buffer);
3654
770
  }
3655
63.1k
    }
3656
24.3k
    if (len > maxLength) {
3657
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3658
0
        return(NULL);
3659
0
    }
3660
24.3k
    *str = cur;
3661
24.3k
    ret = xmlStrndup(buf, len);
3662
24.3k
    if (ret == NULL)
3663
0
        xmlErrMemory(ctxt);
3664
24.3k
    return(ret);
3665
24.3k
}
3666
3667
/**
3668
 * xmlParseNmtoken:
3669
 * @ctxt:  an XML parser context
3670
 *
3671
 * DEPRECATED: Internal function, don't use.
3672
 *
3673
 * parse an XML Nmtoken.
3674
 *
3675
 * [7] Nmtoken ::= (NameChar)+
3676
 *
3677
 * [8] Nmtokens ::= Nmtoken (#x20 Nmtoken)*
3678
 *
3679
 * Returns the Nmtoken parsed or NULL
3680
 */
3681
3682
xmlChar *
3683
526k
xmlParseNmtoken(xmlParserCtxtPtr ctxt) {
3684
526k
    xmlChar buf[XML_MAX_NAMELEN + 5];
3685
526k
    xmlChar *ret;
3686
526k
    int len = 0, l;
3687
526k
    int c;
3688
526k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3689
520k
                    XML_MAX_TEXT_LENGTH :
3690
526k
                    XML_MAX_NAME_LENGTH;
3691
3692
526k
    c = xmlCurrentChar(ctxt, &l);
3693
3694
4.65M
    while (xmlIsNameChar(ctxt, c)) {
3695
4.13M
  COPY_BUF(buf, len, c);
3696
4.13M
  NEXTL(l);
3697
4.13M
  c = xmlCurrentChar(ctxt, &l);
3698
4.13M
  if (len >= XML_MAX_NAMELEN) {
3699
      /*
3700
       * Okay someone managed to make a huge token, so he's ready to pay
3701
       * for the processing speed.
3702
       */
3703
4.03k
      xmlChar *buffer;
3704
4.03k
      int max = len * 2;
3705
3706
4.03k
      buffer = xmlMalloc(max);
3707
4.03k
      if (buffer == NULL) {
3708
0
          xmlErrMemory(ctxt);
3709
0
    return(NULL);
3710
0
      }
3711
4.03k
      memcpy(buffer, buf, len);
3712
232k
      while (xmlIsNameChar(ctxt, c)) {
3713
228k
    if (len + 10 > max) {
3714
1.00k
        xmlChar *tmp;
3715
1.00k
                    int newSize;
3716
3717
1.00k
                    newSize = xmlGrowCapacity(max, 1, 1, maxLength);
3718
1.00k
                    if (newSize < 0) {
3719
0
                        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3720
0
                        xmlFree(buffer);
3721
0
                        return(NULL);
3722
0
                    }
3723
1.00k
        tmp = xmlRealloc(buffer, newSize);
3724
1.00k
        if (tmp == NULL) {
3725
0
      xmlErrMemory(ctxt);
3726
0
      xmlFree(buffer);
3727
0
      return(NULL);
3728
0
        }
3729
1.00k
        buffer = tmp;
3730
1.00k
                    max = newSize;
3731
1.00k
    }
3732
228k
    COPY_BUF(buffer, len, c);
3733
228k
    NEXTL(l);
3734
228k
    c = xmlCurrentChar(ctxt, &l);
3735
228k
      }
3736
4.03k
      buffer[len] = 0;
3737
4.03k
      return(buffer);
3738
4.03k
  }
3739
4.13M
    }
3740
521k
    if (len == 0)
3741
80.5k
        return(NULL);
3742
441k
    if (len > maxLength) {
3743
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3744
0
        return(NULL);
3745
0
    }
3746
441k
    ret = xmlStrndup(buf, len);
3747
441k
    if (ret == NULL)
3748
0
        xmlErrMemory(ctxt);
3749
441k
    return(ret);
3750
441k
}
3751
3752
/**
3753
 * xmlExpandPEsInEntityValue:
3754
 * @ctxt:  parser context
3755
 * @buf:  string buffer
3756
 * @str:  entity value
3757
 * @length:  size of entity value
3758
 * @depth:  nesting depth
3759
 *
3760
 * Validate an entity value and expand parameter entities.
3761
 */
3762
static void
3763
xmlExpandPEsInEntityValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf,
3764
24.4k
                          const xmlChar *str, int length, int depth) {
3765
24.4k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3766
24.4k
    const xmlChar *end, *chunk;
3767
24.4k
    int c, l;
3768
3769
24.4k
    if (str == NULL)
3770
0
        return;
3771
3772
24.4k
    depth += 1;
3773
24.4k
    if (depth > maxDepth) {
3774
0
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3775
0
                       "Maximum entity nesting depth exceeded");
3776
0
  return;
3777
0
    }
3778
3779
24.4k
    end = str + length;
3780
24.4k
    chunk = str;
3781
3782
2.86M
    while ((str < end) && (!PARSER_STOPPED(ctxt))) {
3783
2.85M
        c = *str;
3784
3785
2.85M
        if (c >= 0x80) {
3786
505k
            l = xmlUTF8MultibyteLen(ctxt, str,
3787
505k
                    "invalid character in entity value\n");
3788
505k
            if (l == 0) {
3789
249k
                if (chunk < str)
3790
25.0k
                    xmlSBufAddString(buf, chunk, str - chunk);
3791
249k
                xmlSBufAddReplChar(buf);
3792
249k
                str += 1;
3793
249k
                chunk = str;
3794
255k
            } else {
3795
255k
                str += l;
3796
255k
            }
3797
2.34M
        } else if (c == '&') {
3798
62.6k
            if (str[1] == '#') {
3799
35.6k
                if (chunk < str)
3800
21.0k
                    xmlSBufAddString(buf, chunk, str - chunk);
3801
3802
35.6k
                c = xmlParseStringCharRef(ctxt, &str);
3803
35.6k
                if (c == 0)
3804
4.46k
                    return;
3805
3806
31.1k
                xmlSBufAddChar(buf, c);
3807
3808
31.1k
                chunk = str;
3809
31.1k
            } else {
3810
27.0k
                xmlChar *name;
3811
3812
                /*
3813
                 * General entity references are checked for
3814
                 * syntactic validity.
3815
                 */
3816
27.0k
                str++;
3817
27.0k
                name = xmlParseStringName(ctxt, &str);
3818
3819
27.0k
                if ((name == NULL) || (*str++ != ';')) {
3820
6.01k
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR,
3821
6.01k
                            "EntityValue: '&' forbidden except for entities "
3822
6.01k
                            "references\n");
3823
6.01k
                    xmlFree(name);
3824
6.01k
                    return;
3825
6.01k
                }
3826
3827
21.0k
                xmlFree(name);
3828
21.0k
            }
3829
2.28M
        } else if (c == '%') {
3830
2.91k
            xmlEntityPtr ent;
3831
3832
2.91k
            if (chunk < str)
3833
2.88k
                xmlSBufAddString(buf, chunk, str - chunk);
3834
3835
2.91k
            ent = xmlParseStringPEReference(ctxt, &str);
3836
2.91k
            if (ent == NULL)
3837
2.91k
                return;
3838
3839
0
            if (!PARSER_EXTERNAL(ctxt)) {
3840
0
                xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL);
3841
0
                return;
3842
0
            }
3843
3844
0
            if (ent->content == NULL) {
3845
                /*
3846
                 * Note: external parsed entities will not be loaded,
3847
                 * it is not required for a non-validating parser to
3848
                 * complete external PEReferences coming from the
3849
                 * internal subset
3850
                 */
3851
0
                if (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
3852
0
                    ((ctxt->replaceEntities) ||
3853
0
                     (ctxt->validate))) {
3854
0
                    xmlLoadEntityContent(ctxt, ent);
3855
0
                } else {
3856
0
                    xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING,
3857
0
                                  "not validating will not read content for "
3858
0
                                  "PE entity %s\n", ent->name, NULL);
3859
0
                }
3860
0
            }
3861
3862
            /*
3863
             * TODO: Skip if ent->content is still NULL.
3864
             */
3865
3866
0
            if (xmlParserEntityCheck(ctxt, ent->length))
3867
0
                return;
3868
3869
0
            if (ent->flags & XML_ENT_EXPANDING) {
3870
0
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3871
0
                xmlHaltParser(ctxt);
3872
0
                return;
3873
0
            }
3874
3875
0
            ent->flags |= XML_ENT_EXPANDING;
3876
0
            xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length,
3877
0
                                      depth);
3878
0
            ent->flags &= ~XML_ENT_EXPANDING;
3879
3880
0
            chunk = str;
3881
2.28M
        } else {
3882
            /* Normal ASCII char */
3883
2.28M
            if (!IS_BYTE_CHAR(c)) {
3884
120k
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3885
120k
                        "invalid character in entity value\n");
3886
120k
                if (chunk < str)
3887
5.18k
                    xmlSBufAddString(buf, chunk, str - chunk);
3888
120k
                xmlSBufAddReplChar(buf);
3889
120k
                str += 1;
3890
120k
                chunk = str;
3891
2.16M
            } else {
3892
2.16M
                str += 1;
3893
2.16M
            }
3894
2.28M
        }
3895
2.85M
    }
3896
3897
11.1k
    if (chunk < str)
3898
7.74k
        xmlSBufAddString(buf, chunk, str - chunk);
3899
11.1k
}
3900
3901
/**
3902
 * xmlParseEntityValue:
3903
 * @ctxt:  an XML parser context
3904
 * @orig:  if non-NULL store a copy of the original entity value
3905
 *
3906
 * DEPRECATED: Internal function, don't use.
3907
 *
3908
 * parse a value for ENTITY declarations
3909
 *
3910
 * [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' |
3911
 *                 "'" ([^%&'] | PEReference | Reference)* "'"
3912
 *
3913
 * Returns the EntityValue parsed with reference substituted or NULL
3914
 */
3915
xmlChar *
3916
24.8k
xmlParseEntityValue(xmlParserCtxtPtr ctxt, xmlChar **orig) {
3917
24.8k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3918
24.8k
                         XML_MAX_HUGE_LENGTH :
3919
24.8k
                         XML_MAX_TEXT_LENGTH;
3920
24.8k
    xmlSBuf buf;
3921
24.8k
    const xmlChar *start;
3922
24.8k
    int quote, length;
3923
3924
24.8k
    xmlSBufInit(&buf, maxLength);
3925
3926
24.8k
    GROW;
3927
3928
24.8k
    quote = CUR;
3929
24.8k
    if ((quote != '"') && (quote != '\'')) {
3930
0
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
3931
0
  return(NULL);
3932
0
    }
3933
24.8k
    CUR_PTR++;
3934
3935
24.8k
    length = 0;
3936
3937
    /*
3938
     * Copy raw content of the entity into a buffer
3939
     */
3940
5.91M
    while (1) {
3941
5.91M
        int c;
3942
3943
5.91M
        if (PARSER_STOPPED(ctxt))
3944
0
            goto error;
3945
3946
5.91M
        if (CUR_PTR >= ctxt->input->end) {
3947
212
            xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL);
3948
212
            goto error;
3949
212
        }
3950
3951
5.91M
        c = CUR;
3952
3953
5.91M
        if (c == 0) {
3954
107
            xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3955
107
                    "invalid character in entity value\n");
3956
107
            goto error;
3957
107
        }
3958
5.91M
        if (c == quote)
3959
24.4k
            break;
3960
5.89M
        NEXTL(1);
3961
5.89M
        length += 1;
3962
3963
        /*
3964
         * TODO: Check growth threshold
3965
         */
3966
5.89M
        if (ctxt->input->end - CUR_PTR < 10)
3967
3.03k
            GROW;
3968
5.89M
    }
3969
3970
24.4k
    start = CUR_PTR - length;
3971
3972
24.4k
    if (orig != NULL) {
3973
24.4k
        *orig = xmlStrndup(start, length);
3974
24.4k
        if (*orig == NULL)
3975
0
            xmlErrMemory(ctxt);
3976
24.4k
    }
3977
3978
24.4k
    xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr);
3979
3980
24.4k
    NEXTL(1);
3981
3982
24.4k
    return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long"));
3983
3984
319
error:
3985
319
    xmlSBufCleanup(&buf, ctxt, "entity length too long");
3986
319
    return(NULL);
3987
24.8k
}
3988
3989
/**
3990
 * xmlCheckEntityInAttValue:
3991
 * @ctxt:  parser context
3992
 * @pent:  entity
3993
 * @depth:  nesting depth
3994
 *
3995
 * Check an entity reference in an attribute value for validity
3996
 * without expanding it.
3997
 */
3998
static void
3999
0
xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) {
4000
0
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
4001
0
    const xmlChar *str;
4002
0
    unsigned long expandedSize = pent->length;
4003
0
    int c, flags;
4004
4005
0
    depth += 1;
4006
0
    if (depth > maxDepth) {
4007
0
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
4008
0
                       "Maximum entity nesting depth exceeded");
4009
0
  return;
4010
0
    }
4011
4012
0
    if (pent->flags & XML_ENT_EXPANDING) {
4013
0
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
4014
0
        xmlHaltParser(ctxt);
4015
0
        return;
4016
0
    }
4017
4018
    /*
4019
     * If we're parsing a default attribute value in DTD content,
4020
     * the entity might reference other entities which weren't
4021
     * defined yet, so the check isn't reliable.
4022
     */
4023
0
    if (ctxt->inSubset == 0)
4024
0
        flags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
4025
0
    else
4026
0
        flags = XML_ENT_VALIDATED;
4027
4028
0
    str = pent->content;
4029
0
    if (str == NULL)
4030
0
        goto done;
4031
4032
    /*
4033
     * Note that entity values are already validated. We only check
4034
     * for illegal less-than signs and compute the expanded size
4035
     * of the entity. No special handling for multi-byte characters
4036
     * is needed.
4037
     */
4038
0
    while (!PARSER_STOPPED(ctxt)) {
4039
0
        c = *str;
4040
4041
0
  if (c != '&') {
4042
0
            if (c == 0)
4043
0
                break;
4044
4045
0
            if (c == '<')
4046
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
4047
0
                        "'<' in entity '%s' is not allowed in attributes "
4048
0
                        "values\n", pent->name);
4049
4050
0
            str += 1;
4051
0
        } else if (str[1] == '#') {
4052
0
            int val;
4053
4054
0
      val = xmlParseStringCharRef(ctxt, &str);
4055
0
      if (val == 0) {
4056
0
                pent->content[0] = 0;
4057
0
                break;
4058
0
            }
4059
0
  } else {
4060
0
            xmlChar *name;
4061
0
            xmlEntityPtr ent;
4062
4063
0
      name = xmlParseStringEntityRef(ctxt, &str);
4064
0
      if (name == NULL) {
4065
0
                pent->content[0] = 0;
4066
0
                break;
4067
0
            }
4068
4069
0
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
4070
0
            xmlFree(name);
4071
4072
0
            if ((ent != NULL) &&
4073
0
                (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) {
4074
0
                if ((ent->flags & flags) != flags) {
4075
0
                    pent->flags |= XML_ENT_EXPANDING;
4076
0
                    xmlCheckEntityInAttValue(ctxt, ent, depth);
4077
0
                    pent->flags &= ~XML_ENT_EXPANDING;
4078
0
                }
4079
4080
0
                xmlSaturatedAdd(&expandedSize, ent->expandedSize);
4081
0
                xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST);
4082
0
            }
4083
0
        }
4084
0
    }
4085
4086
0
done:
4087
0
    if (ctxt->inSubset == 0)
4088
0
        pent->expandedSize = expandedSize;
4089
4090
0
    pent->flags |= flags;
4091
0
}
4092
4093
/**
4094
 * xmlExpandEntityInAttValue:
4095
 * @ctxt:  parser context
4096
 * @buf:  string buffer
4097
 * @str:  entity or attribute value
4098
 * @pent:  entity for entity value, NULL for attribute values
4099
 * @normalize:  whether to collapse whitespace
4100
 * @inSpace:  whitespace state
4101
 * @depth:  nesting depth
4102
 * @check:  whether to check for amplification
4103
 *
4104
 * Expand general entity references in an entity or attribute value.
4105
 * Perform attribute value normalization.
4106
 */
4107
static void
4108
xmlExpandEntityInAttValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf,
4109
                          const xmlChar *str, xmlEntityPtr pent, int normalize,
4110
0
                          int *inSpace, int depth, int check) {
4111
0
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
4112
0
    int c, chunkSize;
4113
4114
0
    if (str == NULL)
4115
0
        return;
4116
4117
0
    depth += 1;
4118
0
    if (depth > maxDepth) {
4119
0
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
4120
0
                       "Maximum entity nesting depth exceeded");
4121
0
  return;
4122
0
    }
4123
4124
0
    if (pent != NULL) {
4125
0
        if (pent->flags & XML_ENT_EXPANDING) {
4126
0
            xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
4127
0
            xmlHaltParser(ctxt);
4128
0
            return;
4129
0
        }
4130
4131
0
        if (check) {
4132
0
            if (xmlParserEntityCheck(ctxt, pent->length))
4133
0
                return;
4134
0
        }
4135
0
    }
4136
4137
0
    chunkSize = 0;
4138
4139
    /*
4140
     * Note that entity values are already validated. No special
4141
     * handling for multi-byte characters is needed.
4142
     */
4143
0
    while (!PARSER_STOPPED(ctxt)) {
4144
0
        c = *str;
4145
4146
0
  if (c != '&') {
4147
0
            if (c == 0)
4148
0
                break;
4149
4150
            /*
4151
             * If this function is called without an entity, it is used to
4152
             * expand entities in an attribute content where less-than was
4153
             * already unscaped and is allowed.
4154
             */
4155
0
            if ((pent != NULL) && (c == '<')) {
4156
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
4157
0
                        "'<' in entity '%s' is not allowed in attributes "
4158
0
                        "values\n", pent->name);
4159
0
                break;
4160
0
            }
4161
4162
0
            if (c <= 0x20) {
4163
0
                if ((normalize) && (*inSpace)) {
4164
                    /* Skip char */
4165
0
                    if (chunkSize > 0) {
4166
0
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
4167
0
                        chunkSize = 0;
4168
0
                    }
4169
0
                } else if (c < 0x20) {
4170
0
                    if (chunkSize > 0) {
4171
0
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
4172
0
                        chunkSize = 0;
4173
0
                    }
4174
4175
0
                    xmlSBufAddCString(buf, " ", 1);
4176
0
                } else {
4177
0
                    chunkSize += 1;
4178
0
                }
4179
4180
0
                *inSpace = 1;
4181
0
            } else {
4182
0
                chunkSize += 1;
4183
0
                *inSpace = 0;
4184
0
            }
4185
4186
0
            str += 1;
4187
0
        } else if (str[1] == '#') {
4188
0
            int val;
4189
4190
0
            if (chunkSize > 0) {
4191
0
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
4192
0
                chunkSize = 0;
4193
0
            }
4194
4195
0
      val = xmlParseStringCharRef(ctxt, &str);
4196
0
      if (val == 0) {
4197
0
                if (pent != NULL)
4198
0
                    pent->content[0] = 0;
4199
0
                break;
4200
0
            }
4201
4202
0
            if (val == ' ') {
4203
0
                if ((!normalize) || (!*inSpace))
4204
0
                    xmlSBufAddCString(buf, " ", 1);
4205
0
                *inSpace = 1;
4206
0
            } else {
4207
0
                xmlSBufAddChar(buf, val);
4208
0
                *inSpace = 0;
4209
0
            }
4210
0
  } else {
4211
0
            xmlChar *name;
4212
0
            xmlEntityPtr ent;
4213
4214
0
            if (chunkSize > 0) {
4215
0
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
4216
0
                chunkSize = 0;
4217
0
            }
4218
4219
0
      name = xmlParseStringEntityRef(ctxt, &str);
4220
0
            if (name == NULL) {
4221
0
                if (pent != NULL)
4222
0
                    pent->content[0] = 0;
4223
0
                break;
4224
0
            }
4225
4226
0
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
4227
0
            xmlFree(name);
4228
4229
0
      if ((ent != NULL) &&
4230
0
    (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
4231
0
    if (ent->content == NULL) {
4232
0
        xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
4233
0
          "predefined entity has no content\n");
4234
0
                    break;
4235
0
                }
4236
4237
0
                xmlSBufAddString(buf, ent->content, ent->length);
4238
4239
0
                *inSpace = 0;
4240
0
      } else if ((ent != NULL) && (ent->content != NULL)) {
4241
0
                if (pent != NULL)
4242
0
                    pent->flags |= XML_ENT_EXPANDING;
4243
0
    xmlExpandEntityInAttValue(ctxt, buf, ent->content, ent,
4244
0
                                          normalize, inSpace, depth, check);
4245
0
                if (pent != NULL)
4246
0
                    pent->flags &= ~XML_ENT_EXPANDING;
4247
0
      }
4248
0
        }
4249
0
    }
4250
4251
0
    if (chunkSize > 0)
4252
0
        xmlSBufAddString(buf, str - chunkSize, chunkSize);
4253
0
}
4254
4255
/**
4256
 * xmlExpandEntitiesInAttValue:
4257
 * @ctxt:  parser context
4258
 * @str:  entity or attribute value
4259
 * @normalize:  whether to collapse whitespace
4260
 *
4261
 * Expand general entity references in an entity or attribute value.
4262
 * Perform attribute value normalization.
4263
 *
4264
 * Returns the expanded attribtue value.
4265
 */
4266
xmlChar *
4267
xmlExpandEntitiesInAttValue(xmlParserCtxtPtr ctxt, const xmlChar *str,
4268
0
                            int normalize) {
4269
0
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4270
0
                         XML_MAX_HUGE_LENGTH :
4271
0
                         XML_MAX_TEXT_LENGTH;
4272
0
    xmlSBuf buf;
4273
0
    int inSpace = 1;
4274
4275
0
    xmlSBufInit(&buf, maxLength);
4276
4277
0
    xmlExpandEntityInAttValue(ctxt, &buf, str, NULL, normalize, &inSpace,
4278
0
                              ctxt->inputNr, /* check */ 0);
4279
4280
0
    if ((normalize) && (inSpace) && (buf.size > 0))
4281
0
        buf.size--;
4282
4283
0
    return(xmlSBufFinish(&buf, NULL, ctxt, "AttValue length too long"));
4284
0
}
4285
4286
/**
4287
 * xmlParseAttValueInternal:
4288
 * @ctxt:  an XML parser context
4289
 * @len:  attribute len result
4290
 * @alloc:  whether the attribute was reallocated as a new string
4291
 * @normalize:  if 1 then further non-CDATA normalization must be done
4292
 *
4293
 * parse a value for an attribute.
4294
 * NOTE: if no normalization is needed, the routine will return pointers
4295
 *       directly from the data buffer.
4296
 *
4297
 * 3.3.3 Attribute-Value Normalization:
4298
 * Before the value of an attribute is passed to the application or
4299
 * checked for validity, the XML processor must normalize it as follows:
4300
 * - a character reference is processed by appending the referenced
4301
 *   character to the attribute value
4302
 * - an entity reference is processed by recursively processing the
4303
 *   replacement text of the entity
4304
 * - a whitespace character (#x20, #xD, #xA, #x9) is processed by
4305
 *   appending #x20 to the normalized value, except that only a single
4306
 *   #x20 is appended for a "#xD#xA" sequence that is part of an external
4307
 *   parsed entity or the literal entity value of an internal parsed entity
4308
 * - other characters are processed by appending them to the normalized value
4309
 * If the declared value is not CDATA, then the XML processor must further
4310
 * process the normalized attribute value by discarding any leading and
4311
 * trailing space (#x20) characters, and by replacing sequences of space
4312
 * (#x20) characters by a single space (#x20) character.
4313
 * All attributes for which no declaration has been read should be treated
4314
 * by a non-validating parser as if declared CDATA.
4315
 *
4316
 * Returns the AttValue parsed or NULL. The value has to be freed by the
4317
 *     caller if it was copied, this can be detected by val[*len] == 0.
4318
 */
4319
static xmlChar *
4320
xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *attlen, int *alloc,
4321
28.1M
                         int normalize, int isNamespace) {
4322
28.1M
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4323
26.7M
                         XML_MAX_HUGE_LENGTH :
4324
28.1M
                         XML_MAX_TEXT_LENGTH;
4325
28.1M
    xmlSBuf buf;
4326
28.1M
    xmlChar *ret;
4327
28.1M
    int c, l, quote, flags, chunkSize;
4328
28.1M
    int inSpace = 1;
4329
28.1M
    int replaceEntities;
4330
4331
    /* Always expand namespace URIs */
4332
28.1M
    replaceEntities = (ctxt->replaceEntities) || (isNamespace);
4333
4334
28.1M
    xmlSBufInit(&buf, maxLength);
4335
4336
28.1M
    GROW;
4337
4338
28.1M
    quote = CUR;
4339
28.1M
    if ((quote != '"') && (quote != '\'')) {
4340
3.75k
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
4341
3.75k
  return(NULL);
4342
3.75k
    }
4343
28.1M
    NEXTL(1);
4344
4345
28.1M
    if (ctxt->inSubset == 0)
4346
28.1M
        flags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
4347
32.8k
    else
4348
32.8k
        flags = XML_ENT_VALIDATED;
4349
4350
28.1M
    inSpace = 1;
4351
28.1M
    chunkSize = 0;
4352
4353
307M
    while (1) {
4354
307M
        if (PARSER_STOPPED(ctxt))
4355
0
            goto error;
4356
4357
307M
        if (CUR_PTR >= ctxt->input->end) {
4358
5.06k
            xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED,
4359
5.06k
                           "AttValue: ' expected\n");
4360
5.06k
            goto error;
4361
5.06k
        }
4362
4363
        /*
4364
         * TODO: Check growth threshold
4365
         */
4366
307M
        if (ctxt->input->end - CUR_PTR < 10)
4367
57.8k
            GROW;
4368
4369
307M
        c = CUR;
4370
4371
307M
        if (c >= 0x80) {
4372
5.96M
            l = xmlUTF8MultibyteLen(ctxt, CUR_PTR,
4373
5.96M
                    "invalid character in attribute value\n");
4374
5.96M
            if (l == 0) {
4375
2.60M
                if (chunkSize > 0) {
4376
212k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4377
212k
                    chunkSize = 0;
4378
212k
                }
4379
2.60M
                xmlSBufAddReplChar(&buf);
4380
2.60M
                NEXTL(1);
4381
3.35M
            } else {
4382
3.35M
                chunkSize += l;
4383
3.35M
                NEXTL(l);
4384
3.35M
            }
4385
4386
5.96M
            inSpace = 0;
4387
301M
        } else if (c != '&') {
4388
300M
            if (c > 0x20) {
4389
295M
                if (c == quote)
4390
28.1M
                    break;
4391
4392
267M
                if (c == '<')
4393
199k
                    xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL);
4394
4395
267M
                chunkSize += 1;
4396
267M
                inSpace = 0;
4397
267M
            } else if (!IS_BYTE_CHAR(c)) {
4398
1.39M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
4399
1.39M
                        "invalid character in attribute value\n");
4400
1.39M
                if (chunkSize > 0) {
4401
64.3k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4402
64.3k
                    chunkSize = 0;
4403
64.3k
                }
4404
1.39M
                xmlSBufAddReplChar(&buf);
4405
1.39M
                inSpace = 0;
4406
3.39M
            } else {
4407
                /* Whitespace */
4408
3.39M
                if ((normalize) && (inSpace)) {
4409
                    /* Skip char */
4410
34.7k
                    if (chunkSize > 0) {
4411
4.21k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4412
4.21k
                        chunkSize = 0;
4413
4.21k
                    }
4414
3.36M
                } else if (c < 0x20) {
4415
                    /* Convert to space */
4416
1.03M
                    if (chunkSize > 0) {
4417
175k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4418
175k
                        chunkSize = 0;
4419
175k
                    }
4420
4421
1.03M
                    xmlSBufAddCString(&buf, " ", 1);
4422
2.33M
                } else {
4423
2.33M
                    chunkSize += 1;
4424
2.33M
                }
4425
4426
3.39M
                inSpace = 1;
4427
4428
3.39M
                if ((c == 0xD) && (NXT(1) == 0xA))
4429
36.1k
                    CUR_PTR++;
4430
3.39M
            }
4431
4432
272M
            NEXTL(1);
4433
272M
        } else if (NXT(1) == '#') {
4434
61.1k
            int val;
4435
4436
61.1k
            if (chunkSize > 0) {
4437
28.8k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4438
28.8k
                chunkSize = 0;
4439
28.8k
            }
4440
4441
61.1k
            val = xmlParseCharRef(ctxt);
4442
61.1k
            if (val == 0)
4443
912
                goto error;
4444
4445
60.2k
            if ((val == '&') && (!replaceEntities)) {
4446
                /*
4447
                 * The reparsing will be done in xmlNodeParseContent()
4448
                 * called from SAX2.c
4449
                 */
4450
0
                xmlSBufAddCString(&buf, "&#38;", 5);
4451
0
                inSpace = 0;
4452
60.2k
            } else if (val == ' ') {
4453
31.5k
                if ((!normalize) || (!inSpace))
4454
25.1k
                    xmlSBufAddCString(&buf, " ", 1);
4455
31.5k
                inSpace = 1;
4456
31.5k
            } else {
4457
28.7k
                xmlSBufAddChar(&buf, val);
4458
28.7k
                inSpace = 0;
4459
28.7k
            }
4460
702k
        } else {
4461
702k
            const xmlChar *name;
4462
702k
            xmlEntityPtr ent;
4463
4464
702k
            if (chunkSize > 0) {
4465
299k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4466
299k
                chunkSize = 0;
4467
299k
            }
4468
4469
702k
            name = xmlParseEntityRefInternal(ctxt);
4470
702k
            if (name == NULL) {
4471
                /*
4472
                 * Probably a literal '&' which wasn't escaped.
4473
                 * TODO: Handle gracefully in recovery mode.
4474
                 */
4475
471k
                continue;
4476
471k
            }
4477
4478
230k
            ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1);
4479
230k
            if (ent == NULL)
4480
18.2k
                continue;
4481
4482
212k
            if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) {
4483
212k
                if ((ent->content[0] == '&') && (!replaceEntities))
4484
0
                    xmlSBufAddCString(&buf, "&#38;", 5);
4485
212k
                else
4486
212k
                    xmlSBufAddString(&buf, ent->content, ent->length);
4487
212k
                inSpace = 0;
4488
18.4E
            } else if (replaceEntities) {
4489
0
                xmlExpandEntityInAttValue(ctxt, &buf, ent->content, ent,
4490
0
                                          normalize, &inSpace, ctxt->inputNr,
4491
0
                                          /* check */ 1);
4492
18.4E
            } else {
4493
18.4E
                if ((ent->flags & flags) != flags)
4494
0
                    xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr);
4495
4496
18.4E
                if (xmlParserEntityCheck(ctxt, ent->expandedSize)) {
4497
0
                    ent->content[0] = 0;
4498
0
                    goto error;
4499
0
                }
4500
4501
                /*
4502
                 * Just output the reference
4503
                 */
4504
18.4E
                xmlSBufAddCString(&buf, "&", 1);
4505
18.4E
                xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name));
4506
18.4E
                xmlSBufAddCString(&buf, ";", 1);
4507
4508
18.4E
                inSpace = 0;
4509
18.4E
            }
4510
212k
  }
4511
307M
    }
4512
4513
28.1M
    if ((buf.mem == NULL) && (alloc != NULL)) {
4514
27.8M
        ret = (xmlChar *) CUR_PTR - chunkSize;
4515
4516
27.8M
        if (attlen != NULL)
4517
27.8M
            *attlen = chunkSize;
4518
27.8M
        if ((normalize) && (inSpace) && (chunkSize > 0))
4519
602
            *attlen -= 1;
4520
27.8M
        *alloc = 0;
4521
4522
        /* Report potential error */
4523
27.8M
        xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4524
27.8M
    } else {
4525
254k
        if (chunkSize > 0)
4526
164k
            xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4527
4528
254k
        if ((normalize) && (inSpace) && (buf.size > 0))
4529
4.33k
            buf.size--;
4530
4531
254k
        ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long");
4532
4533
254k
        if (ret != NULL) {
4534
254k
            if (attlen != NULL)
4535
222k
                *attlen = buf.size;
4536
254k
            if (alloc != NULL)
4537
222k
                *alloc = 1;
4538
254k
        }
4539
254k
    }
4540
4541
28.1M
    NEXTL(1);
4542
4543
28.1M
    return(ret);
4544
4545
5.98k
error:
4546
5.98k
    xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4547
5.98k
    return(NULL);
4548
28.1M
}
4549
4550
/**
4551
 * xmlParseAttValue:
4552
 * @ctxt:  an XML parser context
4553
 *
4554
 * DEPRECATED: Internal function, don't use.
4555
 *
4556
 * parse a value for an attribute
4557
 * Note: the parser won't do substitution of entities here, this
4558
 * will be handled later in xmlStringGetNodeList
4559
 *
4560
 * [10] AttValue ::= '"' ([^<&"] | Reference)* '"' |
4561
 *                   "'" ([^<&'] | Reference)* "'"
4562
 *
4563
 * 3.3.3 Attribute-Value Normalization:
4564
 * Before the value of an attribute is passed to the application or
4565
 * checked for validity, the XML processor must normalize it as follows:
4566
 * - a character reference is processed by appending the referenced
4567
 *   character to the attribute value
4568
 * - an entity reference is processed by recursively processing the
4569
 *   replacement text of the entity
4570
 * - a whitespace character (#x20, #xD, #xA, #x9) is processed by
4571
 *   appending #x20 to the normalized value, except that only a single
4572
 *   #x20 is appended for a "#xD#xA" sequence that is part of an external
4573
 *   parsed entity or the literal entity value of an internal parsed entity
4574
 * - other characters are processed by appending them to the normalized value
4575
 * If the declared value is not CDATA, then the XML processor must further
4576
 * process the normalized attribute value by discarding any leading and
4577
 * trailing space (#x20) characters, and by replacing sequences of space
4578
 * (#x20) characters by a single space (#x20) character.
4579
 * All attributes for which no declaration has been read should be treated
4580
 * by a non-validating parser as if declared CDATA.
4581
 *
4582
 * Returns the AttValue parsed or NULL. The value has to be freed by the caller.
4583
 */
4584
4585
4586
xmlChar *
4587
33.4k
xmlParseAttValue(xmlParserCtxtPtr ctxt) {
4588
33.4k
    if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL);
4589
33.4k
    return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0));
4590
33.4k
}
4591
4592
/**
4593
 * xmlParseSystemLiteral:
4594
 * @ctxt:  an XML parser context
4595
 *
4596
 * DEPRECATED: Internal function, don't use.
4597
 *
4598
 * parse an XML Literal
4599
 *
4600
 * [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'")
4601
 *
4602
 * Returns the SystemLiteral parsed or NULL
4603
 */
4604
4605
xmlChar *
4606
7.73k
xmlParseSystemLiteral(xmlParserCtxtPtr ctxt) {
4607
7.73k
    xmlChar *buf = NULL;
4608
7.73k
    int len = 0;
4609
7.73k
    int size = XML_PARSER_BUFFER_SIZE;
4610
7.73k
    int cur, l;
4611
7.73k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4612
7.73k
                    XML_MAX_TEXT_LENGTH :
4613
7.73k
                    XML_MAX_NAME_LENGTH;
4614
7.73k
    xmlChar stop;
4615
4616
7.73k
    if (RAW == '"') {
4617
5.20k
        NEXT;
4618
5.20k
  stop = '"';
4619
5.20k
    } else if (RAW == '\'') {
4620
2.29k
        NEXT;
4621
2.29k
  stop = '\'';
4622
2.29k
    } else {
4623
237
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4624
237
  return(NULL);
4625
237
    }
4626
4627
7.49k
    buf = xmlMalloc(size);
4628
7.49k
    if (buf == NULL) {
4629
0
        xmlErrMemory(ctxt);
4630
0
  return(NULL);
4631
0
    }
4632
7.49k
    cur = xmlCurrentCharRecover(ctxt, &l);
4633
419k
    while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */
4634
412k
  if (len + 5 >= size) {
4635
3.08k
      xmlChar *tmp;
4636
3.08k
            int newSize;
4637
4638
3.08k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4639
3.08k
            if (newSize < 0) {
4640
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral");
4641
0
                xmlFree(buf);
4642
0
                return(NULL);
4643
0
            }
4644
3.08k
      tmp = xmlRealloc(buf, newSize);
4645
3.08k
      if (tmp == NULL) {
4646
0
          xmlFree(buf);
4647
0
    xmlErrMemory(ctxt);
4648
0
    return(NULL);
4649
0
      }
4650
3.08k
      buf = tmp;
4651
3.08k
            size = newSize;
4652
3.08k
  }
4653
412k
  COPY_BUF(buf, len, cur);
4654
412k
  NEXTL(l);
4655
412k
  cur = xmlCurrentCharRecover(ctxt, &l);
4656
412k
    }
4657
7.49k
    buf[len] = 0;
4658
7.49k
    if (!IS_CHAR(cur)) {
4659
190
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4660
7.30k
    } else {
4661
7.30k
  NEXT;
4662
7.30k
    }
4663
7.49k
    return(buf);
4664
7.49k
}
4665
4666
/**
4667
 * xmlParsePubidLiteral:
4668
 * @ctxt:  an XML parser context
4669
 *
4670
 * DEPRECATED: Internal function, don't use.
4671
 *
4672
 * parse an XML public literal
4673
 *
4674
 * [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'"
4675
 *
4676
 * Returns the PubidLiteral parsed or NULL.
4677
 */
4678
4679
xmlChar *
4680
3.72k
xmlParsePubidLiteral(xmlParserCtxtPtr ctxt) {
4681
3.72k
    xmlChar *buf = NULL;
4682
3.72k
    int len = 0;
4683
3.72k
    int size = XML_PARSER_BUFFER_SIZE;
4684
3.72k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4685
3.72k
                    XML_MAX_TEXT_LENGTH :
4686
3.72k
                    XML_MAX_NAME_LENGTH;
4687
3.72k
    xmlChar cur;
4688
3.72k
    xmlChar stop;
4689
4690
3.72k
    if (RAW == '"') {
4691
1.40k
        NEXT;
4692
1.40k
  stop = '"';
4693
2.32k
    } else if (RAW == '\'') {
4694
2.12k
        NEXT;
4695
2.12k
  stop = '\'';
4696
2.12k
    } else {
4697
208
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4698
208
  return(NULL);
4699
208
    }
4700
3.52k
    buf = xmlMalloc(size);
4701
3.52k
    if (buf == NULL) {
4702
0
  xmlErrMemory(ctxt);
4703
0
  return(NULL);
4704
0
    }
4705
3.52k
    cur = CUR;
4706
109k
    while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) &&
4707
109k
           (PARSER_STOPPED(ctxt) == 0)) { /* checked */
4708
105k
  if (len + 1 >= size) {
4709
83
      xmlChar *tmp;
4710
83
            int newSize;
4711
4712
83
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4713
83
            if (newSize < 0) {
4714
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID");
4715
0
                xmlFree(buf);
4716
0
                return(NULL);
4717
0
            }
4718
83
      tmp = xmlRealloc(buf, newSize);
4719
83
      if (tmp == NULL) {
4720
0
    xmlErrMemory(ctxt);
4721
0
    xmlFree(buf);
4722
0
    return(NULL);
4723
0
      }
4724
83
      buf = tmp;
4725
83
            size = newSize;
4726
83
  }
4727
105k
  buf[len++] = cur;
4728
105k
  NEXT;
4729
105k
  cur = CUR;
4730
105k
    }
4731
3.52k
    buf[len] = 0;
4732
3.52k
    if (cur != stop) {
4733
78
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4734
3.44k
    } else {
4735
3.44k
  NEXTL(1);
4736
3.44k
    }
4737
3.52k
    return(buf);
4738
3.52k
}
4739
4740
static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial);
4741
4742
/*
4743
 * used for the test in the inner loop of the char data testing
4744
 */
4745
static const unsigned char test_char_data[256] = {
4746
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4747
    0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */
4748
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4749
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4750
    0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */
4751
    0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F,
4752
    0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37,
4753
    0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */
4754
    0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47,
4755
    0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F,
4756
    0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57,
4757
    0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */
4758
    0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67,
4759
    0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F,
4760
    0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77,
4761
    0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F,
4762
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */
4763
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4764
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4765
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4766
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4767
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4768
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4769
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4770
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4771
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4772
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4773
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4774
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4775
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4776
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4777
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
4778
};
4779
4780
static void
4781
xmlCharacters(xmlParserCtxtPtr ctxt, const xmlChar *buf, int size,
4782
14.5M
              int isBlank) {
4783
14.5M
    int checkBlanks;
4784
4785
14.5M
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
4786
58.3k
        return;
4787
4788
14.4M
    checkBlanks = (!ctxt->keepBlanks) ||
4789
14.4M
                  (ctxt->sax->ignorableWhitespace != ctxt->sax->characters);
4790
4791
    /*
4792
     * Calling areBlanks with only parts of a text node
4793
     * is fundamentally broken, making the NOBLANKS option
4794
     * essentially unusable.
4795
     */
4796
14.4M
    if ((checkBlanks) &&
4797
14.4M
        (areBlanks(ctxt, buf, size, isBlank))) {
4798
0
        if ((ctxt->sax->ignorableWhitespace != NULL) &&
4799
0
            (ctxt->keepBlanks))
4800
0
            ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size);
4801
14.4M
    } else {
4802
14.4M
        if (ctxt->sax->characters != NULL)
4803
14.4M
            ctxt->sax->characters(ctxt->userData, buf, size);
4804
4805
        /*
4806
         * The old code used to update this value for "complex" data
4807
         * even if checkBlanks was false. This was probably a bug.
4808
         */
4809
14.4M
        if ((checkBlanks) && (*ctxt->space == -1))
4810
8.02M
            *ctxt->space = -2;
4811
14.4M
    }
4812
14.4M
}
4813
4814
/**
4815
 * xmlParseCharDataInternal:
4816
 * @ctxt:  an XML parser context
4817
 * @partial:  buffer may contain partial UTF-8 sequences
4818
 *
4819
 * Parse character data. Always makes progress if the first char isn't
4820
 * '<' or '&'.
4821
 *
4822
 * The right angle bracket (>) may be represented using the string "&gt;",
4823
 * and must, for compatibility, be escaped using "&gt;" or a character
4824
 * reference when it appears in the string "]]>" in content, when that
4825
 * string is not marking the end of a CDATA section.
4826
 *
4827
 * [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*)
4828
 */
4829
static void
4830
13.7M
xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) {
4831
13.7M
    const xmlChar *in;
4832
13.7M
    int nbchar = 0;
4833
13.7M
    int line = ctxt->input->line;
4834
13.7M
    int col = ctxt->input->col;
4835
13.7M
    int ccol;
4836
4837
13.7M
    GROW;
4838
    /*
4839
     * Accelerated common case where input don't need to be
4840
     * modified before passing it to the handler.
4841
     */
4842
13.7M
    in = ctxt->input->cur;
4843
14.2M
    do {
4844
19.3M
get_more_space:
4845
47.6M
        while (*in == 0x20) { in++; ctxt->input->col++; }
4846
19.3M
        if (*in == 0xA) {
4847
5.29M
            do {
4848
5.29M
                ctxt->input->line++; ctxt->input->col = 1;
4849
5.29M
                in++;
4850
5.29M
            } while (*in == 0xA);
4851
5.10M
            goto get_more_space;
4852
5.10M
        }
4853
14.2M
        if (*in == '<') {
4854
5.20M
            nbchar = in - ctxt->input->cur;
4855
5.20M
            if (nbchar > 0) {
4856
5.20M
                const xmlChar *tmp = ctxt->input->cur;
4857
5.20M
                ctxt->input->cur = in;
4858
4859
5.20M
                xmlCharacters(ctxt, tmp, nbchar, 1);
4860
5.20M
            }
4861
5.20M
            return;
4862
5.20M
        }
4863
4864
9.96M
get_more:
4865
9.96M
        ccol = ctxt->input->col;
4866
130M
        while (test_char_data[*in]) {
4867
120M
            in++;
4868
120M
            ccol++;
4869
120M
        }
4870
9.96M
        ctxt->input->col = ccol;
4871
9.96M
        if (*in == 0xA) {
4872
955k
            do {
4873
955k
                ctxt->input->line++; ctxt->input->col = 1;
4874
955k
                in++;
4875
955k
            } while (*in == 0xA);
4876
753k
            goto get_more;
4877
753k
        }
4878
9.21M
        if (*in == ']') {
4879
183k
            if ((in[1] == ']') && (in[2] == '>')) {
4880
10
                xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4881
10
                ctxt->input->cur = in + 1;
4882
10
                return;
4883
10
            }
4884
183k
            if ((!partial) || (ctxt->input->end - in >= 2)) {
4885
183k
                in++;
4886
183k
                ctxt->input->col++;
4887
183k
                goto get_more;
4888
183k
            }
4889
183k
        }
4890
9.02M
        nbchar = in - ctxt->input->cur;
4891
9.02M
        if (nbchar > 0) {
4892
8.25M
            const xmlChar *tmp = ctxt->input->cur;
4893
8.25M
            ctxt->input->cur = in;
4894
4895
8.25M
            xmlCharacters(ctxt, tmp, nbchar, 0);
4896
4897
8.25M
            line = ctxt->input->line;
4898
8.25M
            col = ctxt->input->col;
4899
8.25M
        }
4900
9.02M
        ctxt->input->cur = in;
4901
9.02M
        if (*in == 0xD) {
4902
564k
            in++;
4903
564k
            if (*in == 0xA) {
4904
531k
                ctxt->input->cur = in;
4905
531k
                in++;
4906
531k
                ctxt->input->line++; ctxt->input->col = 1;
4907
531k
                continue; /* while */
4908
531k
            }
4909
32.6k
            in--;
4910
32.6k
        }
4911
8.49M
        if (*in == '<') {
4912
7.39M
            return;
4913
7.39M
        }
4914
1.10M
        if (*in == '&') {
4915
236k
            return;
4916
236k
        }
4917
864k
        if ((partial) && (*in == ']') && (ctxt->input->end - in < 2)) {
4918
0
            return;
4919
0
        }
4920
864k
        SHRINK;
4921
864k
        GROW;
4922
864k
        in = ctxt->input->cur;
4923
1.39M
    } while (((*in >= 0x20) && (*in <= 0x7F)) ||
4924
1.39M
             (*in == 0x09) || (*in == 0x0a));
4925
888k
    ctxt->input->line = line;
4926
888k
    ctxt->input->col = col;
4927
888k
    xmlParseCharDataComplex(ctxt, partial);
4928
888k
}
4929
4930
/**
4931
 * xmlParseCharDataComplex:
4932
 * @ctxt:  an XML parser context
4933
 * @cdata:  int indicating whether we are within a CDATA section
4934
 *
4935
 * Always makes progress if the first char isn't '<' or '&'.
4936
 *
4937
 * parse a CharData section.this is the fallback function
4938
 * of xmlParseCharData() when the parsing requires handling
4939
 * of non-ASCII characters.
4940
 */
4941
static void
4942
888k
xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) {
4943
888k
    xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5];
4944
888k
    int nbchar = 0;
4945
888k
    int cur, l;
4946
4947
888k
    cur = xmlCurrentCharRecover(ctxt, &l);
4948
64.3M
    while ((cur != '<') && /* checked */
4949
64.3M
           (cur != '&') &&
4950
64.3M
           ((!partial) || (cur != ']') ||
4951
63.4M
            (ctxt->input->end - ctxt->input->cur >= 2)) &&
4952
64.3M
     (IS_CHAR(cur))) {
4953
63.4M
  if ((cur == ']') && (NXT(1) == ']') && (NXT(2) == '>')) {
4954
349
      xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4955
349
  }
4956
63.4M
  COPY_BUF(buf, nbchar, cur);
4957
  /* move current position before possible calling of ctxt->sax->characters */
4958
63.4M
  NEXTL(l);
4959
63.4M
  if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) {
4960
185k
      buf[nbchar] = 0;
4961
4962
185k
            xmlCharacters(ctxt, buf, nbchar, 0);
4963
185k
      nbchar = 0;
4964
185k
            SHRINK;
4965
185k
  }
4966
63.4M
  cur = xmlCurrentCharRecover(ctxt, &l);
4967
63.4M
    }
4968
888k
    if (nbchar != 0) {
4969
879k
        buf[nbchar] = 0;
4970
4971
879k
        xmlCharacters(ctxt, buf, nbchar, 0);
4972
879k
    }
4973
    /*
4974
     * cur == 0 can mean
4975
     *
4976
     * - End of buffer.
4977
     * - An actual 0 character.
4978
     * - An incomplete UTF-8 sequence. This is allowed if partial is set.
4979
     */
4980
888k
    if (ctxt->input->cur < ctxt->input->end) {
4981
880k
        if ((cur == 0) && (CUR != 0)) {
4982
294
            if (partial == 0) {
4983
288
                xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4984
288
                        "Incomplete UTF-8 sequence starting with %02X\n", CUR);
4985
288
                NEXTL(1);
4986
288
            }
4987
880k
        } else if ((cur != '<') && (cur != '&') && (cur != ']')) {
4988
            /* Generate the error and skip the offending character */
4989
3.43k
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4990
3.43k
                              "PCDATA invalid Char value %d\n", cur);
4991
3.43k
            NEXTL(l);
4992
3.43k
        }
4993
880k
    }
4994
888k
}
4995
4996
/**
4997
 * xmlParseCharData:
4998
 * @ctxt:  an XML parser context
4999
 * @cdata:  unused
5000
 *
5001
 * DEPRECATED: Internal function, don't use.
5002
 */
5003
void
5004
0
xmlParseCharData(xmlParserCtxtPtr ctxt, ATTRIBUTE_UNUSED int cdata) {
5005
0
    xmlParseCharDataInternal(ctxt, 0);
5006
0
}
5007
5008
/**
5009
 * xmlParseExternalID:
5010
 * @ctxt:  an XML parser context
5011
 * @publicID:  a xmlChar** receiving PubidLiteral
5012
 * @strict: indicate whether we should restrict parsing to only
5013
 *          production [75], see NOTE below
5014
 *
5015
 * DEPRECATED: Internal function, don't use.
5016
 *
5017
 * Parse an External ID or a Public ID
5018
 *
5019
 * NOTE: Productions [75] and [83] interact badly since [75] can generate
5020
 *       'PUBLIC' S PubidLiteral S SystemLiteral
5021
 *
5022
 * [75] ExternalID ::= 'SYSTEM' S SystemLiteral
5023
 *                   | 'PUBLIC' S PubidLiteral S SystemLiteral
5024
 *
5025
 * [83] PublicID ::= 'PUBLIC' S PubidLiteral
5026
 *
5027
 * Returns the function returns SystemLiteral and in the second
5028
 *                case publicID receives PubidLiteral, is strict is off
5029
 *                it is possible to return NULL and have publicID set.
5030
 */
5031
5032
xmlChar *
5033
17.2k
xmlParseExternalID(xmlParserCtxtPtr ctxt, xmlChar **publicID, int strict) {
5034
17.2k
    xmlChar *URI = NULL;
5035
5036
17.2k
    *publicID = NULL;
5037
17.2k
    if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) {
5038
5.85k
        SKIP(6);
5039
5.85k
  if (SKIP_BLANKS == 0) {
5040
946
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5041
946
                     "Space required after 'SYSTEM'\n");
5042
946
  }
5043
5.85k
  URI = xmlParseSystemLiteral(ctxt);
5044
5.85k
  if (URI == NULL) {
5045
146
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
5046
146
        }
5047
11.4k
    } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) {
5048
3.72k
        SKIP(6);
5049
3.72k
  if (SKIP_BLANKS == 0) {
5050
23
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5051
23
        "Space required after 'PUBLIC'\n");
5052
23
  }
5053
3.72k
  *publicID = xmlParsePubidLiteral(ctxt);
5054
3.72k
  if (*publicID == NULL) {
5055
208
      xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL);
5056
208
  }
5057
3.72k
  if (strict) {
5058
      /*
5059
       * We don't handle [83] so "S SystemLiteral" is required.
5060
       */
5061
1.34k
      if (SKIP_BLANKS == 0) {
5062
93
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5063
93
      "Space required after the Public Identifier\n");
5064
93
      }
5065
2.38k
  } else {
5066
      /*
5067
       * We handle [83] so we return immediately, if
5068
       * "S SystemLiteral" is not detected. We skip blanks if no
5069
             * system literal was found, but this is harmless since we must
5070
             * be at the end of a NotationDecl.
5071
       */
5072
2.38k
      if (SKIP_BLANKS == 0) return(NULL);
5073
1.31k
      if ((CUR != '\'') && (CUR != '"')) return(NULL);
5074
1.31k
  }
5075
1.88k
  URI = xmlParseSystemLiteral(ctxt);
5076
1.88k
  if (URI == NULL) {
5077
91
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
5078
91
        }
5079
1.88k
    }
5080
15.4k
    return(URI);
5081
17.2k
}
5082
5083
/**
5084
 * xmlParseCommentComplex:
5085
 * @ctxt:  an XML parser context
5086
 * @buf:  the already parsed part of the buffer
5087
 * @len:  number of bytes in the buffer
5088
 * @size:  allocated size of the buffer
5089
 *
5090
 * Skip an XML (SGML) comment <!-- .... -->
5091
 *  The spec says that "For compatibility, the string "--" (double-hyphen)
5092
 *  must not occur within comments. "
5093
 * This is the slow routine in case the accelerator for ascii didn't work
5094
 *
5095
 * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->'
5096
 */
5097
static void
5098
xmlParseCommentComplex(xmlParserCtxtPtr ctxt, xmlChar *buf,
5099
34.8k
                       size_t len, size_t size) {
5100
34.8k
    int q, ql;
5101
34.8k
    int r, rl;
5102
34.8k
    int cur, l;
5103
34.8k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
5104
34.8k
                    XML_MAX_HUGE_LENGTH :
5105
34.8k
                    XML_MAX_TEXT_LENGTH;
5106
5107
34.8k
    if (buf == NULL) {
5108
6.56k
        len = 0;
5109
6.56k
  size = XML_PARSER_BUFFER_SIZE;
5110
6.56k
  buf = xmlMalloc(size);
5111
6.56k
  if (buf == NULL) {
5112
0
      xmlErrMemory(ctxt);
5113
0
      return;
5114
0
  }
5115
6.56k
    }
5116
34.8k
    q = xmlCurrentCharRecover(ctxt, &ql);
5117
34.8k
    if (q == 0)
5118
292
        goto not_terminated;
5119
34.6k
    if (!IS_CHAR(q)) {
5120
60
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
5121
60
                          "xmlParseComment: invalid xmlChar value %d\n",
5122
60
                    q);
5123
60
  xmlFree (buf);
5124
60
  return;
5125
60
    }
5126
34.5k
    NEXTL(ql);
5127
34.5k
    r = xmlCurrentCharRecover(ctxt, &rl);
5128
34.5k
    if (r == 0)
5129
33
        goto not_terminated;
5130
34.5k
    if (!IS_CHAR(r)) {
5131
33
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
5132
33
                          "xmlParseComment: invalid xmlChar value %d\n",
5133
33
                    r);
5134
33
  xmlFree (buf);
5135
33
  return;
5136
33
    }
5137
34.4k
    NEXTL(rl);
5138
34.4k
    cur = xmlCurrentCharRecover(ctxt, &l);
5139
34.4k
    if (cur == 0)
5140
33
        goto not_terminated;
5141
3.26M
    while (IS_CHAR(cur) && /* checked */
5142
3.26M
           ((cur != '>') ||
5143
3.26M
      (r != '-') || (q != '-'))) {
5144
3.23M
  if ((r == '-') && (q == '-')) {
5145
22.1k
      xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL);
5146
22.1k
  }
5147
3.23M
  if (len + 5 >= size) {
5148
16.1k
      xmlChar *tmp;
5149
16.1k
            int newSize;
5150
5151
16.1k
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
5152
16.1k
            if (newSize < 0) {
5153
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
5154
0
                             "Comment too big found", NULL);
5155
0
                xmlFree (buf);
5156
0
                return;
5157
0
            }
5158
16.1k
      tmp = xmlRealloc(buf, newSize);
5159
16.1k
      if (tmp == NULL) {
5160
0
    xmlErrMemory(ctxt);
5161
0
    xmlFree(buf);
5162
0
    return;
5163
0
      }
5164
16.1k
      buf = tmp;
5165
16.1k
            size = newSize;
5166
16.1k
  }
5167
3.23M
  COPY_BUF(buf, len, q);
5168
5169
3.23M
  q = r;
5170
3.23M
  ql = rl;
5171
3.23M
  r = cur;
5172
3.23M
  rl = l;
5173
5174
3.23M
  NEXTL(l);
5175
3.23M
  cur = xmlCurrentCharRecover(ctxt, &l);
5176
5177
3.23M
    }
5178
34.4k
    buf[len] = 0;
5179
34.4k
    if (cur == 0) {
5180
599
  xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
5181
599
                       "Comment not terminated \n<!--%.50s\n", buf);
5182
33.8k
    } else if (!IS_CHAR(cur)) {
5183
145
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
5184
145
                          "xmlParseComment: invalid xmlChar value %d\n",
5185
145
                    cur);
5186
33.7k
    } else {
5187
33.7k
        NEXT;
5188
33.7k
  if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
5189
33.7k
      (!ctxt->disableSAX))
5190
0
      ctxt->sax->comment(ctxt->userData, buf);
5191
33.7k
    }
5192
34.4k
    xmlFree(buf);
5193
34.4k
    return;
5194
358
not_terminated:
5195
358
    xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
5196
358
       "Comment not terminated\n", NULL);
5197
358
    xmlFree(buf);
5198
358
}
5199
5200
/**
5201
 * xmlParseComment:
5202
 * @ctxt:  an XML parser context
5203
 *
5204
 * DEPRECATED: Internal function, don't use.
5205
 *
5206
 * Parse an XML (SGML) comment. Always consumes '<!'.
5207
 *
5208
 *  The spec says that "For compatibility, the string "--" (double-hyphen)
5209
 *  must not occur within comments. "
5210
 *
5211
 * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->'
5212
 */
5213
void
5214
83.8k
xmlParseComment(xmlParserCtxtPtr ctxt) {
5215
83.8k
    xmlChar *buf = NULL;
5216
83.8k
    size_t size = XML_PARSER_BUFFER_SIZE;
5217
83.8k
    size_t len = 0;
5218
83.8k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
5219
83.8k
                       XML_MAX_HUGE_LENGTH :
5220
83.8k
                       XML_MAX_TEXT_LENGTH;
5221
83.8k
    const xmlChar *in;
5222
83.8k
    size_t nbchar = 0;
5223
83.8k
    int ccol;
5224
5225
    /*
5226
     * Check that there is a comment right here.
5227
     */
5228
83.8k
    if ((RAW != '<') || (NXT(1) != '!'))
5229
0
        return;
5230
83.8k
    SKIP(2);
5231
83.8k
    if ((RAW != '-') || (NXT(1) != '-'))
5232
24
        return;
5233
83.8k
    SKIP(2);
5234
83.8k
    GROW;
5235
5236
    /*
5237
     * Accelerated common case where input don't need to be
5238
     * modified before passing it to the handler.
5239
     */
5240
83.8k
    in = ctxt->input->cur;
5241
83.8k
    do {
5242
83.8k
  if (*in == 0xA) {
5243
12.0k
      do {
5244
12.0k
    ctxt->input->line++; ctxt->input->col = 1;
5245
12.0k
    in++;
5246
12.0k
      } while (*in == 0xA);
5247
4.86k
  }
5248
186k
get_more:
5249
186k
        ccol = ctxt->input->col;
5250
2.06M
  while (((*in > '-') && (*in <= 0x7F)) ||
5251
2.06M
         ((*in >= 0x20) && (*in < '-')) ||
5252
2.06M
         (*in == 0x09)) {
5253
1.88M
        in++;
5254
1.88M
        ccol++;
5255
1.88M
  }
5256
186k
  ctxt->input->col = ccol;
5257
186k
  if (*in == 0xA) {
5258
28.2k
      do {
5259
28.2k
    ctxt->input->line++; ctxt->input->col = 1;
5260
28.2k
    in++;
5261
28.2k
      } while (*in == 0xA);
5262
18.0k
      goto get_more;
5263
18.0k
  }
5264
168k
  nbchar = in - ctxt->input->cur;
5265
  /*
5266
   * save current set of data
5267
   */
5268
168k
  if (nbchar > 0) {
5269
142k
            if (nbchar > maxLength - len) {
5270
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
5271
0
                                  "Comment too big found", NULL);
5272
0
                xmlFree(buf);
5273
0
                return;
5274
0
            }
5275
142k
            if (buf == NULL) {
5276
72.0k
                if ((*in == '-') && (in[1] == '-'))
5277
36.3k
                    size = nbchar + 1;
5278
35.7k
                else
5279
35.7k
                    size = XML_PARSER_BUFFER_SIZE + nbchar;
5280
72.0k
                buf = xmlMalloc(size);
5281
72.0k
                if (buf == NULL) {
5282
0
                    xmlErrMemory(ctxt);
5283
0
                    return;
5284
0
                }
5285
72.0k
                len = 0;
5286
72.0k
            } else if (len + nbchar + 1 >= size) {
5287
3.62k
                xmlChar *new_buf;
5288
3.62k
                size += len + nbchar + XML_PARSER_BUFFER_SIZE;
5289
3.62k
                new_buf = xmlRealloc(buf, size);
5290
3.62k
                if (new_buf == NULL) {
5291
0
                    xmlErrMemory(ctxt);
5292
0
                    xmlFree(buf);
5293
0
                    return;
5294
0
                }
5295
3.62k
                buf = new_buf;
5296
3.62k
            }
5297
142k
            memcpy(&buf[len], ctxt->input->cur, nbchar);
5298
142k
            len += nbchar;
5299
142k
            buf[len] = 0;
5300
142k
  }
5301
168k
  ctxt->input->cur = in;
5302
168k
  if (*in == 0xA) {
5303
0
      in++;
5304
0
      ctxt->input->line++; ctxt->input->col = 1;
5305
0
  }
5306
168k
  if (*in == 0xD) {
5307
13.8k
      in++;
5308
13.8k
      if (*in == 0xA) {
5309
6.60k
    ctxt->input->cur = in;
5310
6.60k
    in++;
5311
6.60k
    ctxt->input->line++; ctxt->input->col = 1;
5312
6.60k
    goto get_more;
5313
6.60k
      }
5314
7.19k
      in--;
5315
7.19k
  }
5316
161k
  SHRINK;
5317
161k
  GROW;
5318
161k
  in = ctxt->input->cur;
5319
161k
  if (*in == '-') {
5320
126k
      if (in[1] == '-') {
5321
92.6k
          if (in[2] == '>') {
5322
48.9k
        SKIP(3);
5323
48.9k
        if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
5324
48.9k
            (!ctxt->disableSAX)) {
5325
0
      if (buf != NULL)
5326
0
          ctxt->sax->comment(ctxt->userData, buf);
5327
0
      else
5328
0
          ctxt->sax->comment(ctxt->userData, BAD_CAST "");
5329
0
        }
5330
48.9k
        if (buf != NULL)
5331
43.7k
            xmlFree(buf);
5332
48.9k
        return;
5333
48.9k
    }
5334
43.6k
    if (buf != NULL) {
5335
38.4k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5336
38.4k
                          "Double hyphen within comment: "
5337
38.4k
                                      "<!--%.50s\n",
5338
38.4k
              buf);
5339
38.4k
    } else
5340
5.22k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5341
5.22k
                          "Double hyphen within comment\n", NULL);
5342
43.6k
    in++;
5343
43.6k
    ctxt->input->col++;
5344
43.6k
      }
5345
77.7k
      in++;
5346
77.7k
      ctxt->input->col++;
5347
77.7k
      goto get_more;
5348
126k
  }
5349
161k
    } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a));
5350
34.8k
    xmlParseCommentComplex(ctxt, buf, len, size);
5351
34.8k
}
5352
5353
5354
/**
5355
 * xmlParsePITarget:
5356
 * @ctxt:  an XML parser context
5357
 *
5358
 * DEPRECATED: Internal function, don't use.
5359
 *
5360
 * parse the name of a PI
5361
 *
5362
 * [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l'))
5363
 *
5364
 * Returns the PITarget name or NULL
5365
 */
5366
5367
const xmlChar *
5368
79.0k
xmlParsePITarget(xmlParserCtxtPtr ctxt) {
5369
79.0k
    const xmlChar *name;
5370
5371
79.0k
    name = xmlParseName(ctxt);
5372
79.0k
    if ((name != NULL) &&
5373
79.0k
        ((name[0] == 'x') || (name[0] == 'X')) &&
5374
79.0k
        ((name[1] == 'm') || (name[1] == 'M')) &&
5375
79.0k
        ((name[2] == 'l') || (name[2] == 'L'))) {
5376
25.9k
  int i;
5377
25.9k
  if ((name[0] == 'x') && (name[1] == 'm') &&
5378
25.9k
      (name[2] == 'l') && (name[3] == 0)) {
5379
1.83k
      xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5380
1.83k
     "XML declaration allowed only at the start of the document\n");
5381
1.83k
      return(name);
5382
24.0k
  } else if (name[3] == 0) {
5383
434
      xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL);
5384
434
      return(name);
5385
434
  }
5386
67.9k
  for (i = 0;;i++) {
5387
67.9k
      if (xmlW3CPIs[i] == NULL) break;
5388
45.9k
      if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i]))
5389
1.66k
          return(name);
5390
45.9k
  }
5391
22.0k
  xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5392
22.0k
          "xmlParsePITarget: invalid name prefix 'xml'\n",
5393
22.0k
          NULL, NULL);
5394
22.0k
    }
5395
75.0k
    if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) {
5396
5.19k
  xmlNsErr(ctxt, XML_NS_ERR_COLON,
5397
5.19k
     "colons are forbidden from PI names '%s'\n", name, NULL, NULL);
5398
5.19k
    }
5399
75.0k
    return(name);
5400
79.0k
}
5401
5402
#ifdef LIBXML_CATALOG_ENABLED
5403
/**
5404
 * xmlParseCatalogPI:
5405
 * @ctxt:  an XML parser context
5406
 * @catalog:  the PI value string
5407
 *
5408
 * parse an XML Catalog Processing Instruction.
5409
 *
5410
 * <?oasis-xml-catalog catalog="http://example.com/catalog.xml"?>
5411
 *
5412
 * Occurs only if allowed by the user and if happening in the Misc
5413
 * part of the document before any doctype information
5414
 * This will add the given catalog to the parsing context in order
5415
 * to be used if there is a resolution need further down in the document
5416
 */
5417
5418
static void
5419
0
xmlParseCatalogPI(xmlParserCtxtPtr ctxt, const xmlChar *catalog) {
5420
0
    xmlChar *URL = NULL;
5421
0
    const xmlChar *tmp, *base;
5422
0
    xmlChar marker;
5423
5424
0
    tmp = catalog;
5425
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5426
0
    if (xmlStrncmp(tmp, BAD_CAST"catalog", 7))
5427
0
  goto error;
5428
0
    tmp += 7;
5429
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5430
0
    if (*tmp != '=') {
5431
0
  return;
5432
0
    }
5433
0
    tmp++;
5434
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5435
0
    marker = *tmp;
5436
0
    if ((marker != '\'') && (marker != '"'))
5437
0
  goto error;
5438
0
    tmp++;
5439
0
    base = tmp;
5440
0
    while ((*tmp != 0) && (*tmp != marker)) tmp++;
5441
0
    if (*tmp == 0)
5442
0
  goto error;
5443
0
    URL = xmlStrndup(base, tmp - base);
5444
0
    tmp++;
5445
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5446
0
    if (*tmp != 0)
5447
0
  goto error;
5448
5449
0
    if (URL != NULL) {
5450
        /*
5451
         * Unfortunately, the catalog API doesn't report OOM errors.
5452
         * xmlGetLastError isn't very helpful since we don't know
5453
         * where the last error came from. We'd have to reset it
5454
         * before this call and restore it afterwards.
5455
         */
5456
0
  ctxt->catalogs = xmlCatalogAddLocal(ctxt->catalogs, URL);
5457
0
  xmlFree(URL);
5458
0
    }
5459
0
    return;
5460
5461
0
error:
5462
0
    xmlWarningMsg(ctxt, XML_WAR_CATALOG_PI,
5463
0
            "Catalog PI syntax error: %s\n",
5464
0
      catalog, NULL);
5465
0
    if (URL != NULL)
5466
0
  xmlFree(URL);
5467
0
}
5468
#endif
5469
5470
/**
5471
 * xmlParsePI:
5472
 * @ctxt:  an XML parser context
5473
 *
5474
 * DEPRECATED: Internal function, don't use.
5475
 *
5476
 * parse an XML Processing Instruction.
5477
 *
5478
 * [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>'
5479
 *
5480
 * The processing is transferred to SAX once parsed.
5481
 */
5482
5483
void
5484
79.0k
xmlParsePI(xmlParserCtxtPtr ctxt) {
5485
79.0k
    xmlChar *buf = NULL;
5486
79.0k
    size_t len = 0;
5487
79.0k
    size_t size = XML_PARSER_BUFFER_SIZE;
5488
79.0k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
5489
78.9k
                       XML_MAX_HUGE_LENGTH :
5490
79.0k
                       XML_MAX_TEXT_LENGTH;
5491
79.0k
    int cur, l;
5492
79.0k
    const xmlChar *target;
5493
5494
79.0k
    if ((RAW == '<') && (NXT(1) == '?')) {
5495
  /*
5496
   * this is a Processing Instruction.
5497
   */
5498
79.0k
  SKIP(2);
5499
5500
  /*
5501
   * Parse the target name and check for special support like
5502
   * namespace.
5503
   */
5504
79.0k
        target = xmlParsePITarget(ctxt);
5505
79.0k
  if (target != NULL) {
5506
78.2k
      if ((RAW == '?') && (NXT(1) == '>')) {
5507
18.1k
    SKIP(2);
5508
5509
    /*
5510
     * SAX: PI detected.
5511
     */
5512
18.1k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5513
18.1k
        (ctxt->sax->processingInstruction != NULL))
5514
12.0k
        ctxt->sax->processingInstruction(ctxt->userData,
5515
12.0k
                                         target, NULL);
5516
18.1k
    return;
5517
18.1k
      }
5518
60.1k
      buf = xmlMalloc(size);
5519
60.1k
      if (buf == NULL) {
5520
0
    xmlErrMemory(ctxt);
5521
0
    return;
5522
0
      }
5523
60.1k
      if (SKIP_BLANKS == 0) {
5524
9.12k
    xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED,
5525
9.12k
        "ParsePI: PI %s space expected\n", target);
5526
9.12k
      }
5527
60.1k
      cur = xmlCurrentCharRecover(ctxt, &l);
5528
7.08M
      while (IS_CHAR(cur) && /* checked */
5529
7.08M
       ((cur != '?') || (NXT(1) != '>'))) {
5530
7.02M
    if (len + 5 >= size) {
5531
43.3k
        xmlChar *tmp;
5532
43.3k
                    int newSize;
5533
5534
43.3k
                    newSize = xmlGrowCapacity(size, 1, 1, maxLength);
5535
43.3k
                    if (newSize < 0) {
5536
0
                        xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5537
0
                                          "PI %s too big found", target);
5538
0
                        xmlFree(buf);
5539
0
                        return;
5540
0
                    }
5541
43.3k
        tmp = xmlRealloc(buf, newSize);
5542
43.3k
        if (tmp == NULL) {
5543
0
      xmlErrMemory(ctxt);
5544
0
      xmlFree(buf);
5545
0
      return;
5546
0
        }
5547
43.3k
        buf = tmp;
5548
43.3k
                    size = newSize;
5549
43.3k
    }
5550
7.02M
    COPY_BUF(buf, len, cur);
5551
7.02M
    NEXTL(l);
5552
7.02M
    cur = xmlCurrentCharRecover(ctxt, &l);
5553
7.02M
      }
5554
60.1k
      buf[len] = 0;
5555
60.1k
      if (cur != '?') {
5556
2.69k
    xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5557
2.69k
          "ParsePI: PI %s never end ...\n", target);
5558
57.4k
      } else {
5559
57.4k
    SKIP(2);
5560
5561
57.4k
#ifdef LIBXML_CATALOG_ENABLED
5562
57.4k
    if ((ctxt->inSubset == 0) &&
5563
57.4k
        (xmlStrEqual(target, XML_CATALOG_PI))) {
5564
10.9k
        xmlCatalogAllow allow = xmlCatalogGetDefaults();
5565
5566
10.9k
        if ((ctxt->options & XML_PARSE_CATALOG_PI) &&
5567
10.9k
                        ((allow == XML_CATA_ALLOW_DOCUMENT) ||
5568
0
       (allow == XML_CATA_ALLOW_ALL)))
5569
0
      xmlParseCatalogPI(ctxt, buf);
5570
10.9k
    }
5571
57.4k
#endif
5572
5573
    /*
5574
     * SAX: PI detected.
5575
     */
5576
57.4k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5577
57.4k
        (ctxt->sax->processingInstruction != NULL))
5578
47.7k
        ctxt->sax->processingInstruction(ctxt->userData,
5579
47.7k
                                         target, buf);
5580
57.4k
      }
5581
60.1k
      xmlFree(buf);
5582
60.1k
  } else {
5583
733
      xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL);
5584
733
  }
5585
79.0k
    }
5586
79.0k
}
5587
5588
/**
5589
 * xmlParseNotationDecl:
5590
 * @ctxt:  an XML parser context
5591
 *
5592
 * DEPRECATED: Internal function, don't use.
5593
 *
5594
 * Parse a notation declaration. Always consumes '<!'.
5595
 *
5596
 * [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID |  PublicID) S? '>'
5597
 *
5598
 * Hence there is actually 3 choices:
5599
 *     'PUBLIC' S PubidLiteral
5600
 *     'PUBLIC' S PubidLiteral S SystemLiteral
5601
 * and 'SYSTEM' S SystemLiteral
5602
 *
5603
 * See the NOTE on xmlParseExternalID().
5604
 */
5605
5606
void
5607
3.47k
xmlParseNotationDecl(xmlParserCtxtPtr ctxt) {
5608
3.47k
    const xmlChar *name;
5609
3.47k
    xmlChar *Pubid;
5610
3.47k
    xmlChar *Systemid;
5611
5612
3.47k
    if ((CUR != '<') || (NXT(1) != '!'))
5613
0
        return;
5614
3.47k
    SKIP(2);
5615
5616
3.47k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5617
3.40k
  int inputid = ctxt->input->id;
5618
3.40k
  SKIP(8);
5619
3.40k
  if (SKIP_BLANKS_PE == 0) {
5620
11
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5621
11
         "Space required after '<!NOTATION'\n");
5622
11
      return;
5623
11
  }
5624
5625
3.39k
        name = xmlParseName(ctxt);
5626
3.39k
  if (name == NULL) {
5627
87
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5628
87
      return;
5629
87
  }
5630
3.30k
  if (xmlStrchr(name, ':') != NULL) {
5631
9
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5632
9
         "colons are forbidden from notation names '%s'\n",
5633
9
         name, NULL, NULL);
5634
9
  }
5635
3.30k
  if (SKIP_BLANKS_PE == 0) {
5636
16
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5637
16
         "Space required after the NOTATION name'\n");
5638
16
      return;
5639
16
  }
5640
5641
  /*
5642
   * Parse the IDs.
5643
   */
5644
3.29k
  Systemid = xmlParseExternalID(ctxt, &Pubid, 0);
5645
3.29k
  SKIP_BLANKS_PE;
5646
5647
3.29k
  if (RAW == '>') {
5648
2.92k
      if (inputid != ctxt->input->id) {
5649
0
    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
5650
0
                         "Notation declaration doesn't start and stop"
5651
0
                               " in the same entity\n");
5652
0
      }
5653
2.92k
      NEXT;
5654
2.92k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5655
2.92k
    (ctxt->sax->notationDecl != NULL))
5656
0
    ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid);
5657
2.92k
  } else {
5658
364
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5659
364
  }
5660
3.29k
  if (Systemid != NULL) xmlFree(Systemid);
5661
3.29k
  if (Pubid != NULL) xmlFree(Pubid);
5662
3.29k
    }
5663
3.47k
}
5664
5665
/**
5666
 * xmlParseEntityDecl:
5667
 * @ctxt:  an XML parser context
5668
 *
5669
 * DEPRECATED: Internal function, don't use.
5670
 *
5671
 * Parse an entity declaration. Always consumes '<!'.
5672
 *
5673
 * [70] EntityDecl ::= GEDecl | PEDecl
5674
 *
5675
 * [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>'
5676
 *
5677
 * [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
5678
 *
5679
 * [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?)
5680
 *
5681
 * [74] PEDef ::= EntityValue | ExternalID
5682
 *
5683
 * [76] NDataDecl ::= S 'NDATA' S Name
5684
 *
5685
 * [ VC: Notation Declared ]
5686
 * The Name must match the declared name of a notation.
5687
 */
5688
5689
void
5690
31.1k
xmlParseEntityDecl(xmlParserCtxtPtr ctxt) {
5691
31.1k
    const xmlChar *name = NULL;
5692
31.1k
    xmlChar *value = NULL;
5693
31.1k
    xmlChar *URI = NULL, *literal = NULL;
5694
31.1k
    const xmlChar *ndata = NULL;
5695
31.1k
    int isParameter = 0;
5696
31.1k
    xmlChar *orig = NULL;
5697
5698
31.1k
    if ((CUR != '<') || (NXT(1) != '!'))
5699
0
        return;
5700
31.1k
    SKIP(2);
5701
5702
    /* GROW; done in the caller */
5703
31.1k
    if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5704
31.0k
  int inputid = ctxt->input->id;
5705
31.0k
  SKIP(6);
5706
31.0k
  if (SKIP_BLANKS_PE == 0) {
5707
2.97k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5708
2.97k
         "Space required after '<!ENTITY'\n");
5709
2.97k
  }
5710
5711
31.0k
  if (RAW == '%') {
5712
7.17k
      NEXT;
5713
7.17k
      if (SKIP_BLANKS_PE == 0) {
5714
1.95k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5715
1.95k
             "Space required after '%%'\n");
5716
1.95k
      }
5717
7.17k
      isParameter = 1;
5718
7.17k
  }
5719
5720
31.0k
        name = xmlParseName(ctxt);
5721
31.0k
  if (name == NULL) {
5722
195
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5723
195
                     "xmlParseEntityDecl: no name\n");
5724
195
            return;
5725
195
  }
5726
30.8k
  if (xmlStrchr(name, ':') != NULL) {
5727
1.11k
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5728
1.11k
         "colons are forbidden from entities names '%s'\n",
5729
1.11k
         name, NULL, NULL);
5730
1.11k
  }
5731
30.8k
  if (SKIP_BLANKS_PE == 0) {
5732
9.13k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5733
9.13k
         "Space required after the entity name\n");
5734
9.13k
  }
5735
5736
  /*
5737
   * handle the various case of definitions...
5738
   */
5739
30.8k
  if (isParameter) {
5740
7.14k
      if ((RAW == '"') || (RAW == '\'')) {
5741
5.70k
          value = xmlParseEntityValue(ctxt, &orig);
5742
5.70k
    if (value) {
5743
5.66k
        if ((ctxt->sax != NULL) &&
5744
5.66k
      (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5745
0
      ctxt->sax->entityDecl(ctxt->userData, name,
5746
0
                        XML_INTERNAL_PARAMETER_ENTITY,
5747
0
            NULL, NULL, value);
5748
5.66k
    }
5749
5.70k
      } else {
5750
1.44k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5751
1.44k
    if ((URI == NULL) && (literal == NULL)) {
5752
48
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5753
48
    }
5754
1.44k
    if (URI) {
5755
1.38k
                    if (xmlStrchr(URI, '#')) {
5756
6
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5757
1.38k
                    } else {
5758
1.38k
                        if ((ctxt->sax != NULL) &&
5759
1.38k
                            (!ctxt->disableSAX) &&
5760
1.38k
                            (ctxt->sax->entityDecl != NULL))
5761
0
                            ctxt->sax->entityDecl(ctxt->userData, name,
5762
0
                                        XML_EXTERNAL_PARAMETER_ENTITY,
5763
0
                                        literal, URI, NULL);
5764
1.38k
                    }
5765
1.38k
    }
5766
1.44k
      }
5767
23.7k
  } else {
5768
23.7k
      if ((RAW == '"') || (RAW == '\'')) {
5769
19.1k
          value = xmlParseEntityValue(ctxt, &orig);
5770
19.1k
    if ((ctxt->sax != NULL) &&
5771
19.1k
        (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5772
0
        ctxt->sax->entityDecl(ctxt->userData, name,
5773
0
        XML_INTERNAL_GENERAL_ENTITY,
5774
0
        NULL, NULL, value);
5775
    /*
5776
     * For expat compatibility in SAX mode.
5777
     */
5778
19.1k
    if ((ctxt->myDoc == NULL) ||
5779
19.1k
        (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) {
5780
19.1k
        if (ctxt->myDoc == NULL) {
5781
1.15k
      ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5782
1.15k
      if (ctxt->myDoc == NULL) {
5783
0
          xmlErrMemory(ctxt);
5784
0
          goto done;
5785
0
      }
5786
1.15k
      ctxt->myDoc->properties = XML_DOC_INTERNAL;
5787
1.15k
        }
5788
19.1k
        if (ctxt->myDoc->intSubset == NULL) {
5789
1.15k
      ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5790
1.15k
              BAD_CAST "fake", NULL, NULL);
5791
1.15k
                        if (ctxt->myDoc->intSubset == NULL) {
5792
0
                            xmlErrMemory(ctxt);
5793
0
                            goto done;
5794
0
                        }
5795
1.15k
                    }
5796
5797
19.1k
        xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY,
5798
19.1k
                    NULL, NULL, value);
5799
19.1k
    }
5800
19.1k
      } else {
5801
4.59k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5802
4.59k
    if ((URI == NULL) && (literal == NULL)) {
5803
393
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5804
393
    }
5805
4.59k
    if (URI) {
5806
4.17k
                    if (xmlStrchr(URI, '#')) {
5807
560
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5808
560
                    }
5809
4.17k
    }
5810
4.59k
    if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) {
5811
322
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5812
322
           "Space required before 'NDATA'\n");
5813
322
    }
5814
4.59k
    if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) {
5815
1.15k
        SKIP(5);
5816
1.15k
        if (SKIP_BLANKS_PE == 0) {
5817
10
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5818
10
               "Space required after 'NDATA'\n");
5819
10
        }
5820
1.15k
        ndata = xmlParseName(ctxt);
5821
1.15k
        if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5822
1.15k
            (ctxt->sax->unparsedEntityDecl != NULL))
5823
0
      ctxt->sax->unparsedEntityDecl(ctxt->userData, name,
5824
0
            literal, URI, ndata);
5825
3.43k
    } else {
5826
3.43k
        if ((ctxt->sax != NULL) &&
5827
3.43k
            (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5828
0
      ctxt->sax->entityDecl(ctxt->userData, name,
5829
0
            XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5830
0
            literal, URI, NULL);
5831
        /*
5832
         * For expat compatibility in SAX mode.
5833
         * assuming the entity replacement was asked for
5834
         */
5835
3.43k
        if ((ctxt->replaceEntities != 0) &&
5836
3.43k
      ((ctxt->myDoc == NULL) ||
5837
3.43k
      (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) {
5838
3.43k
      if (ctxt->myDoc == NULL) {
5839
316
          ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5840
316
          if (ctxt->myDoc == NULL) {
5841
0
              xmlErrMemory(ctxt);
5842
0
        goto done;
5843
0
          }
5844
316
          ctxt->myDoc->properties = XML_DOC_INTERNAL;
5845
316
      }
5846
5847
3.43k
      if (ctxt->myDoc->intSubset == NULL) {
5848
316
          ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5849
316
            BAD_CAST "fake", NULL, NULL);
5850
316
                            if (ctxt->myDoc->intSubset == NULL) {
5851
0
                                xmlErrMemory(ctxt);
5852
0
                                goto done;
5853
0
                            }
5854
316
                        }
5855
3.43k
      xmlSAX2EntityDecl(ctxt, name,
5856
3.43k
                  XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5857
3.43k
                  literal, URI, NULL);
5858
3.43k
        }
5859
3.43k
    }
5860
4.59k
      }
5861
23.7k
  }
5862
30.8k
  SKIP_BLANKS_PE;
5863
30.8k
  if (RAW != '>') {
5864
1.10k
      xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED,
5865
1.10k
              "xmlParseEntityDecl: entity %s not terminated\n", name);
5866
1.10k
      xmlHaltParser(ctxt);
5867
29.7k
  } else {
5868
29.7k
      if (inputid != ctxt->input->id) {
5869
0
    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
5870
0
                         "Entity declaration doesn't start and stop in"
5871
0
                               " the same entity\n");
5872
0
      }
5873
29.7k
      NEXT;
5874
29.7k
  }
5875
30.8k
  if (orig != NULL) {
5876
      /*
5877
       * Ugly mechanism to save the raw entity value.
5878
       */
5879
24.4k
      xmlEntityPtr cur = NULL;
5880
5881
24.4k
      if (isParameter) {
5882
5.66k
          if ((ctxt->sax != NULL) &&
5883
5.66k
        (ctxt->sax->getParameterEntity != NULL))
5884
0
        cur = ctxt->sax->getParameterEntity(ctxt->userData, name);
5885
18.8k
      } else {
5886
18.8k
          if ((ctxt->sax != NULL) &&
5887
18.8k
        (ctxt->sax->getEntity != NULL))
5888
18.8k
        cur = ctxt->sax->getEntity(ctxt->userData, name);
5889
18.8k
    if ((cur == NULL) && (ctxt->userData==ctxt)) {
5890
0
        cur = xmlSAX2GetEntity(ctxt, name);
5891
0
    }
5892
18.8k
      }
5893
24.4k
            if ((cur != NULL) && (cur->orig == NULL)) {
5894
0
    cur->orig = orig;
5895
0
                orig = NULL;
5896
0
      }
5897
24.4k
  }
5898
5899
30.8k
done:
5900
30.8k
  if (value != NULL) xmlFree(value);
5901
30.8k
  if (URI != NULL) xmlFree(URI);
5902
30.8k
  if (literal != NULL) xmlFree(literal);
5903
30.8k
        if (orig != NULL) xmlFree(orig);
5904
30.8k
    }
5905
31.1k
}
5906
5907
/**
5908
 * xmlParseDefaultDecl:
5909
 * @ctxt:  an XML parser context
5910
 * @value:  Receive a possible fixed default value for the attribute
5911
 *
5912
 * DEPRECATED: Internal function, don't use.
5913
 *
5914
 * Parse an attribute default declaration
5915
 *
5916
 * [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue)
5917
 *
5918
 * [ VC: Required Attribute ]
5919
 * if the default declaration is the keyword #REQUIRED, then the
5920
 * attribute must be specified for all elements of the type in the
5921
 * attribute-list declaration.
5922
 *
5923
 * [ VC: Attribute Default Legal ]
5924
 * The declared default value must meet the lexical constraints of
5925
 * the declared attribute type c.f. xmlValidateAttributeDecl()
5926
 *
5927
 * [ VC: Fixed Attribute Default ]
5928
 * if an attribute has a default value declared with the #FIXED
5929
 * keyword, instances of that attribute must match the default value.
5930
 *
5931
 * [ WFC: No < in Attribute Values ]
5932
 * handled in xmlParseAttValue()
5933
 *
5934
 * returns: XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED
5935
 *          or XML_ATTRIBUTE_FIXED.
5936
 */
5937
5938
int
5939
43.9k
xmlParseDefaultDecl(xmlParserCtxtPtr ctxt, xmlChar **value) {
5940
43.9k
    int val;
5941
43.9k
    xmlChar *ret;
5942
5943
43.9k
    *value = NULL;
5944
43.9k
    if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) {
5945
2.85k
  SKIP(9);
5946
2.85k
  return(XML_ATTRIBUTE_REQUIRED);
5947
2.85k
    }
5948
41.0k
    if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) {
5949
7.60k
  SKIP(8);
5950
7.60k
  return(XML_ATTRIBUTE_IMPLIED);
5951
7.60k
    }
5952
33.4k
    val = XML_ATTRIBUTE_NONE;
5953
33.4k
    if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) {
5954
1.22k
  SKIP(6);
5955
1.22k
  val = XML_ATTRIBUTE_FIXED;
5956
1.22k
  if (SKIP_BLANKS_PE == 0) {
5957
225
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5958
225
         "Space required after '#FIXED'\n");
5959
225
  }
5960
1.22k
    }
5961
33.4k
    ret = xmlParseAttValue(ctxt);
5962
33.4k
    if (ret == NULL) {
5963
639
  xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo,
5964
639
           "Attribute default value declaration error\n");
5965
639
    } else
5966
32.8k
        *value = ret;
5967
33.4k
    return(val);
5968
41.0k
}
5969
5970
/**
5971
 * xmlParseNotationType:
5972
 * @ctxt:  an XML parser context
5973
 *
5974
 * DEPRECATED: Internal function, don't use.
5975
 *
5976
 * parse an Notation attribute type.
5977
 *
5978
 * Note: the leading 'NOTATION' S part has already being parsed...
5979
 *
5980
 * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
5981
 *
5982
 * [ VC: Notation Attributes ]
5983
 * Values of this type must match one of the notation names included
5984
 * in the declaration; all notation names in the declaration must be declared.
5985
 *
5986
 * Returns: the notation attribute tree built while parsing
5987
 */
5988
5989
xmlEnumerationPtr
5990
854
xmlParseNotationType(xmlParserCtxtPtr ctxt) {
5991
854
    const xmlChar *name;
5992
854
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5993
5994
854
    if (RAW != '(') {
5995
8
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5996
8
  return(NULL);
5997
8
    }
5998
2.40k
    do {
5999
2.40k
        NEXT;
6000
2.40k
  SKIP_BLANKS_PE;
6001
2.40k
        name = xmlParseName(ctxt);
6002
2.40k
  if (name == NULL) {
6003
41
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6004
41
         "Name expected in NOTATION declaration\n");
6005
41
            xmlFreeEnumeration(ret);
6006
41
      return(NULL);
6007
41
  }
6008
2.36k
        tmp = NULL;
6009
2.36k
#ifdef LIBXML_VALID_ENABLED
6010
2.36k
        if (ctxt->validate) {
6011
0
            tmp = ret;
6012
0
            while (tmp != NULL) {
6013
0
                if (xmlStrEqual(name, tmp->name)) {
6014
0
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
6015
0
              "standalone: attribute notation value token %s duplicated\n",
6016
0
                                     name, NULL);
6017
0
                    if (!xmlDictOwns(ctxt->dict, name))
6018
0
                        xmlFree((xmlChar *) name);
6019
0
                    break;
6020
0
                }
6021
0
                tmp = tmp->next;
6022
0
            }
6023
0
        }
6024
2.36k
#endif /* LIBXML_VALID_ENABLED */
6025
2.36k
  if (tmp == NULL) {
6026
2.36k
      cur = xmlCreateEnumeration(name);
6027
2.36k
      if (cur == NULL) {
6028
0
                xmlErrMemory(ctxt);
6029
0
                xmlFreeEnumeration(ret);
6030
0
                return(NULL);
6031
0
            }
6032
2.36k
      if (last == NULL) ret = last = cur;
6033
1.53k
      else {
6034
1.53k
    last->next = cur;
6035
1.53k
    last = cur;
6036
1.53k
      }
6037
2.36k
  }
6038
2.36k
  SKIP_BLANKS_PE;
6039
2.36k
    } while (RAW == '|');
6040
805
    if (RAW != ')') {
6041
46
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
6042
46
        xmlFreeEnumeration(ret);
6043
46
  return(NULL);
6044
46
    }
6045
759
    NEXT;
6046
759
    return(ret);
6047
805
}
6048
6049
/**
6050
 * xmlParseEnumerationType:
6051
 * @ctxt:  an XML parser context
6052
 *
6053
 * DEPRECATED: Internal function, don't use.
6054
 *
6055
 * parse an Enumeration attribute type.
6056
 *
6057
 * [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')'
6058
 *
6059
 * [ VC: Enumeration ]
6060
 * Values of this type must match one of the Nmtoken tokens in
6061
 * the declaration
6062
 *
6063
 * Returns: the enumeration attribute tree built while parsing
6064
 */
6065
6066
xmlEnumerationPtr
6067
11.8k
xmlParseEnumerationType(xmlParserCtxtPtr ctxt) {
6068
11.8k
    xmlChar *name;
6069
11.8k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
6070
6071
11.8k
    if (RAW != '(') {
6072
383
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL);
6073
383
  return(NULL);
6074
383
    }
6075
15.3k
    do {
6076
15.3k
        NEXT;
6077
15.3k
  SKIP_BLANKS_PE;
6078
15.3k
        name = xmlParseNmtoken(ctxt);
6079
15.3k
  if (name == NULL) {
6080
71
      xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL);
6081
71
      return(ret);
6082
71
  }
6083
15.3k
        tmp = NULL;
6084
15.3k
#ifdef LIBXML_VALID_ENABLED
6085
15.3k
        if (ctxt->validate) {
6086
0
            tmp = ret;
6087
0
            while (tmp != NULL) {
6088
0
                if (xmlStrEqual(name, tmp->name)) {
6089
0
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
6090
0
              "standalone: attribute enumeration value token %s duplicated\n",
6091
0
                                     name, NULL);
6092
0
                    if (!xmlDictOwns(ctxt->dict, name))
6093
0
                        xmlFree(name);
6094
0
                    break;
6095
0
                }
6096
0
                tmp = tmp->next;
6097
0
            }
6098
0
        }
6099
15.3k
#endif /* LIBXML_VALID_ENABLED */
6100
15.3k
  if (tmp == NULL) {
6101
15.3k
      cur = xmlCreateEnumeration(name);
6102
15.3k
      if (!xmlDictOwns(ctxt->dict, name))
6103
15.3k
    xmlFree(name);
6104
15.3k
      if (cur == NULL) {
6105
0
                xmlErrMemory(ctxt);
6106
0
                xmlFreeEnumeration(ret);
6107
0
                return(NULL);
6108
0
            }
6109
15.3k
      if (last == NULL) ret = last = cur;
6110
3.82k
      else {
6111
3.82k
    last->next = cur;
6112
3.82k
    last = cur;
6113
3.82k
      }
6114
15.3k
  }
6115
15.3k
  SKIP_BLANKS_PE;
6116
15.3k
    } while (RAW == '|');
6117
11.4k
    if (RAW != ')') {
6118
104
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL);
6119
104
  return(ret);
6120
104
    }
6121
11.3k
    NEXT;
6122
11.3k
    return(ret);
6123
11.4k
}
6124
6125
/**
6126
 * xmlParseEnumeratedType:
6127
 * @ctxt:  an XML parser context
6128
 * @tree:  the enumeration tree built while parsing
6129
 *
6130
 * DEPRECATED: Internal function, don't use.
6131
 *
6132
 * parse an Enumerated attribute type.
6133
 *
6134
 * [57] EnumeratedType ::= NotationType | Enumeration
6135
 *
6136
 * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
6137
 *
6138
 *
6139
 * Returns: XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION
6140
 */
6141
6142
int
6143
12.7k
xmlParseEnumeratedType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) {
6144
12.7k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
6145
860
  SKIP(8);
6146
860
  if (SKIP_BLANKS_PE == 0) {
6147
6
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6148
6
         "Space required after 'NOTATION'\n");
6149
6
      return(0);
6150
6
  }
6151
854
  *tree = xmlParseNotationType(ctxt);
6152
854
  if (*tree == NULL) return(0);
6153
759
  return(XML_ATTRIBUTE_NOTATION);
6154
854
    }
6155
11.8k
    *tree = xmlParseEnumerationType(ctxt);
6156
11.8k
    if (*tree == NULL) return(0);
6157
11.5k
    return(XML_ATTRIBUTE_ENUMERATION);
6158
11.8k
}
6159
6160
/**
6161
 * xmlParseAttributeType:
6162
 * @ctxt:  an XML parser context
6163
 * @tree:  the enumeration tree built while parsing
6164
 *
6165
 * DEPRECATED: Internal function, don't use.
6166
 *
6167
 * parse the Attribute list def for an element
6168
 *
6169
 * [54] AttType ::= StringType | TokenizedType | EnumeratedType
6170
 *
6171
 * [55] StringType ::= 'CDATA'
6172
 *
6173
 * [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' |
6174
 *                        'ENTITIES' | 'NMTOKEN' | 'NMTOKENS'
6175
 *
6176
 * Validity constraints for attribute values syntax are checked in
6177
 * xmlValidateAttributeValue()
6178
 *
6179
 * [ VC: ID ]
6180
 * Values of type ID must match the Name production. A name must not
6181
 * appear more than once in an XML document as a value of this type;
6182
 * i.e., ID values must uniquely identify the elements which bear them.
6183
 *
6184
 * [ VC: One ID per Element Type ]
6185
 * No element type may have more than one ID attribute specified.
6186
 *
6187
 * [ VC: ID Attribute Default ]
6188
 * An ID attribute must have a declared default of #IMPLIED or #REQUIRED.
6189
 *
6190
 * [ VC: IDREF ]
6191
 * Values of type IDREF must match the Name production, and values
6192
 * of type IDREFS must match Names; each IDREF Name must match the value
6193
 * of an ID attribute on some element in the XML document; i.e. IDREF
6194
 * values must match the value of some ID attribute.
6195
 *
6196
 * [ VC: Entity Name ]
6197
 * Values of type ENTITY must match the Name production, values
6198
 * of type ENTITIES must match Names; each Entity Name must match the
6199
 * name of an unparsed entity declared in the DTD.
6200
 *
6201
 * [ VC: Name Token ]
6202
 * Values of type NMTOKEN must match the Nmtoken production; values
6203
 * of type NMTOKENS must match Nmtokens.
6204
 *
6205
 * Returns the attribute type
6206
 */
6207
int
6208
44.6k
xmlParseAttributeType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) {
6209
44.6k
    if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) {
6210
3.28k
  SKIP(5);
6211
3.28k
  return(XML_ATTRIBUTE_CDATA);
6212
41.4k
     } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) {
6213
1.06k
  SKIP(6);
6214
1.06k
  return(XML_ATTRIBUTE_IDREFS);
6215
40.3k
     } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) {
6216
813
  SKIP(5);
6217
813
  return(XML_ATTRIBUTE_IDREF);
6218
39.5k
     } else if ((RAW == 'I') && (NXT(1) == 'D')) {
6219
19.7k
        SKIP(2);
6220
19.7k
  return(XML_ATTRIBUTE_ID);
6221
19.8k
     } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
6222
373
  SKIP(6);
6223
373
  return(XML_ATTRIBUTE_ENTITY);
6224
19.4k
     } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) {
6225
5.37k
  SKIP(8);
6226
5.37k
  return(XML_ATTRIBUTE_ENTITIES);
6227
14.0k
     } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) {
6228
365
  SKIP(8);
6229
365
  return(XML_ATTRIBUTE_NMTOKENS);
6230
13.7k
     } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) {
6231
950
  SKIP(7);
6232
950
  return(XML_ATTRIBUTE_NMTOKEN);
6233
950
     }
6234
12.7k
     return(xmlParseEnumeratedType(ctxt, tree));
6235
44.6k
}
6236
6237
/**
6238
 * xmlParseAttributeListDecl:
6239
 * @ctxt:  an XML parser context
6240
 *
6241
 * DEPRECATED: Internal function, don't use.
6242
 *
6243
 * Parse an attribute list declaration for an element. Always consumes '<!'.
6244
 *
6245
 * [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>'
6246
 *
6247
 * [53] AttDef ::= S Name S AttType S DefaultDecl
6248
 *
6249
 */
6250
void
6251
14.7k
xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt) {
6252
14.7k
    const xmlChar *elemName;
6253
14.7k
    const xmlChar *attrName;
6254
14.7k
    xmlEnumerationPtr tree;
6255
6256
14.7k
    if ((CUR != '<') || (NXT(1) != '!'))
6257
0
        return;
6258
14.7k
    SKIP(2);
6259
6260
14.7k
    if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) {
6261
14.6k
  int inputid = ctxt->input->id;
6262
6263
14.6k
  SKIP(7);
6264
14.6k
  if (SKIP_BLANKS_PE == 0) {
6265
188
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6266
188
                     "Space required after '<!ATTLIST'\n");
6267
188
  }
6268
14.6k
        elemName = xmlParseName(ctxt);
6269
14.6k
  if (elemName == NULL) {
6270
134
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6271
134
         "ATTLIST: no name for Element\n");
6272
134
      return;
6273
134
  }
6274
14.5k
  SKIP_BLANKS_PE;
6275
14.5k
  GROW;
6276
57.7k
  while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) {
6277
45.2k
      int type;
6278
45.2k
      int def;
6279
45.2k
      xmlChar *defaultValue = NULL;
6280
6281
45.2k
      GROW;
6282
45.2k
            tree = NULL;
6283
45.2k
      attrName = xmlParseName(ctxt);
6284
45.2k
      if (attrName == NULL) {
6285
309
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6286
309
             "ATTLIST: no name for Attribute\n");
6287
309
    break;
6288
309
      }
6289
44.9k
      GROW;
6290
44.9k
      if (SKIP_BLANKS_PE == 0) {
6291
288
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6292
288
            "Space required after the attribute name\n");
6293
288
    break;
6294
288
      }
6295
6296
44.6k
      type = xmlParseAttributeType(ctxt, &tree);
6297
44.6k
      if (type <= 0) {
6298
494
          break;
6299
494
      }
6300
6301
44.1k
      GROW;
6302
44.1k
      if (SKIP_BLANKS_PE == 0) {
6303
262
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6304
262
             "Space required after the attribute type\n");
6305
262
          if (tree != NULL)
6306
186
        xmlFreeEnumeration(tree);
6307
262
    break;
6308
262
      }
6309
6310
43.9k
      def = xmlParseDefaultDecl(ctxt, &defaultValue);
6311
43.9k
      if (def <= 0) {
6312
0
                if (defaultValue != NULL)
6313
0
        xmlFree(defaultValue);
6314
0
          if (tree != NULL)
6315
0
        xmlFreeEnumeration(tree);
6316
0
          break;
6317
0
      }
6318
43.9k
      if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL))
6319
29.9k
          xmlAttrNormalizeSpace(defaultValue, defaultValue);
6320
6321
43.9k
      GROW;
6322
43.9k
            if (RAW != '>') {
6323
35.0k
    if (SKIP_BLANKS_PE == 0) {
6324
730
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6325
730
      "Space required after the attribute default value\n");
6326
730
        if (defaultValue != NULL)
6327
85
      xmlFree(defaultValue);
6328
730
        if (tree != NULL)
6329
77
      xmlFreeEnumeration(tree);
6330
730
        break;
6331
730
    }
6332
35.0k
      }
6333
43.1k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6334
43.1k
    (ctxt->sax->attributeDecl != NULL))
6335
0
    ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName,
6336
0
                          type, def, defaultValue, tree);
6337
43.1k
      else if (tree != NULL)
6338
11.9k
    xmlFreeEnumeration(tree);
6339
6340
43.1k
      if ((ctxt->sax2) && (defaultValue != NULL) &&
6341
43.1k
          (def != XML_ATTRIBUTE_IMPLIED) &&
6342
43.1k
    (def != XML_ATTRIBUTE_REQUIRED)) {
6343
32.7k
    xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue);
6344
32.7k
      }
6345
43.1k
      if (ctxt->sax2) {
6346
43.1k
    xmlAddSpecialAttr(ctxt, elemName, attrName, type);
6347
43.1k
      }
6348
43.1k
      if (defaultValue != NULL)
6349
32.7k
          xmlFree(defaultValue);
6350
43.1k
      GROW;
6351
43.1k
  }
6352
14.5k
  if (RAW == '>') {
6353
12.5k
      if (inputid != ctxt->input->id) {
6354
0
    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
6355
0
                               "Attribute list declaration doesn't start and"
6356
0
                               " stop in the same entity\n");
6357
0
      }
6358
12.5k
      NEXT;
6359
12.5k
  }
6360
14.5k
    }
6361
14.7k
}
6362
6363
/**
6364
 * xmlParseElementMixedContentDecl:
6365
 * @ctxt:  an XML parser context
6366
 * @inputchk:  the input used for the current entity, needed for boundary checks
6367
 *
6368
 * DEPRECATED: Internal function, don't use.
6369
 *
6370
 * parse the declaration for a Mixed Element content
6371
 * The leading '(' and spaces have been skipped in xmlParseElementContentDecl
6372
 *
6373
 * [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' |
6374
 *                '(' S? '#PCDATA' S? ')'
6375
 *
6376
 * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49])
6377
 *
6378
 * [ VC: No Duplicate Types ]
6379
 * The same name must not appear more than once in a single
6380
 * mixed-content declaration.
6381
 *
6382
 * returns: the list of the xmlElementContentPtr describing the element choices
6383
 */
6384
xmlElementContentPtr
6385
3.83k
xmlParseElementMixedContentDecl(xmlParserCtxtPtr ctxt, int inputchk) {
6386
3.83k
    xmlElementContentPtr ret = NULL, cur = NULL, n;
6387
3.83k
    const xmlChar *elem = NULL;
6388
6389
3.83k
    GROW;
6390
3.83k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6391
3.83k
  SKIP(7);
6392
3.83k
  SKIP_BLANKS_PE;
6393
3.83k
  if (RAW == ')') {
6394
1.61k
      if (ctxt->input->id != inputchk) {
6395
0
    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
6396
0
                               "Element content declaration doesn't start and"
6397
0
                               " stop in the same entity\n");
6398
0
      }
6399
1.61k
      NEXT;
6400
1.61k
      ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6401
1.61k
      if (ret == NULL)
6402
0
                goto mem_error;
6403
1.61k
      if (RAW == '*') {
6404
210
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6405
210
    NEXT;
6406
210
      }
6407
1.61k
      return(ret);
6408
1.61k
  }
6409
2.21k
  if ((RAW == '(') || (RAW == '|')) {
6410
2.13k
      ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6411
2.13k
      if (ret == NULL)
6412
0
                goto mem_error;
6413
2.13k
  }
6414
6.96k
  while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) {
6415
4.89k
      NEXT;
6416
4.89k
            n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6417
4.89k
            if (n == NULL)
6418
0
                goto mem_error;
6419
4.89k
      if (elem == NULL) {
6420
2.13k
    n->c1 = cur;
6421
2.13k
    if (cur != NULL)
6422
2.13k
        cur->parent = n;
6423
2.13k
    ret = cur = n;
6424
2.75k
      } else {
6425
2.75k
          cur->c2 = n;
6426
2.75k
    n->parent = cur;
6427
2.75k
    n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6428
2.75k
                if (n->c1 == NULL)
6429
0
                    goto mem_error;
6430
2.75k
    n->c1->parent = n;
6431
2.75k
    cur = n;
6432
2.75k
      }
6433
4.89k
      SKIP_BLANKS_PE;
6434
4.89k
      elem = xmlParseName(ctxt);
6435
4.89k
      if (elem == NULL) {
6436
145
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6437
145
      "xmlParseElementMixedContentDecl : Name expected\n");
6438
145
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6439
145
    return(NULL);
6440
145
      }
6441
4.74k
      SKIP_BLANKS_PE;
6442
4.74k
      GROW;
6443
4.74k
  }
6444
2.07k
  if ((RAW == ')') && (NXT(1) == '*')) {
6445
1.33k
      if (elem != NULL) {
6446
1.33k
    cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem,
6447
1.33k
                                   XML_ELEMENT_CONTENT_ELEMENT);
6448
1.33k
    if (cur->c2 == NULL)
6449
0
                    goto mem_error;
6450
1.33k
    cur->c2->parent = cur;
6451
1.33k
            }
6452
1.33k
            if (ret != NULL)
6453
1.33k
                ret->ocur = XML_ELEMENT_CONTENT_MULT;
6454
1.33k
      if (ctxt->input->id != inputchk) {
6455
0
    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
6456
0
                               "Element content declaration doesn't start and"
6457
0
                               " stop in the same entity\n");
6458
0
      }
6459
1.33k
      SKIP(2);
6460
1.33k
  } else {
6461
739
      xmlFreeDocElementContent(ctxt->myDoc, ret);
6462
739
      xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL);
6463
739
      return(NULL);
6464
739
  }
6465
6466
2.07k
    } else {
6467
0
  xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL);
6468
0
    }
6469
1.33k
    return(ret);
6470
6471
0
mem_error:
6472
0
    xmlErrMemory(ctxt);
6473
0
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6474
0
    return(NULL);
6475
3.83k
}
6476
6477
/**
6478
 * xmlParseElementChildrenContentDeclPriv:
6479
 * @ctxt:  an XML parser context
6480
 * @inputchk:  the input used for the current entity, needed for boundary checks
6481
 * @depth: the level of recursion
6482
 *
6483
 * parse the declaration for a Mixed Element content
6484
 * The leading '(' and spaces have been skipped in xmlParseElementContentDecl
6485
 *
6486
 *
6487
 * [47] children ::= (choice | seq) ('?' | '*' | '+')?
6488
 *
6489
 * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')?
6490
 *
6491
 * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')'
6492
 *
6493
 * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
6494
 *
6495
 * [ VC: Proper Group/PE Nesting ] applies to [49] and [50]
6496
 * TODO Parameter-entity replacement text must be properly nested
6497
 *  with parenthesized groups. That is to say, if either of the
6498
 *  opening or closing parentheses in a choice, seq, or Mixed
6499
 *  construct is contained in the replacement text for a parameter
6500
 *  entity, both must be contained in the same replacement text. For
6501
 *  interoperability, if a parameter-entity reference appears in a
6502
 *  choice, seq, or Mixed construct, its replacement text should not
6503
 *  be empty, and neither the first nor last non-blank character of
6504
 *  the replacement text should be a connector (| or ,).
6505
 *
6506
 * Returns the tree of xmlElementContentPtr describing the element
6507
 *          hierarchy.
6508
 */
6509
static xmlElementContentPtr
6510
xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int inputchk,
6511
117k
                                       int depth) {
6512
117k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
6513
117k
    xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL;
6514
117k
    const xmlChar *elem;
6515
117k
    xmlChar type = 0;
6516
6517
117k
    if (depth > maxDepth) {
6518
7
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
6519
7
                "xmlParseElementChildrenContentDecl : depth %d too deep, "
6520
7
                "use XML_PARSE_HUGE\n", depth);
6521
7
  return(NULL);
6522
7
    }
6523
117k
    SKIP_BLANKS_PE;
6524
117k
    GROW;
6525
117k
    if (RAW == '(') {
6526
105k
  int inputid = ctxt->input->id;
6527
6528
        /* Recurse on first child */
6529
105k
  NEXT;
6530
105k
  SKIP_BLANKS_PE;
6531
105k
        cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, inputid,
6532
105k
                                                           depth + 1);
6533
105k
        if (cur == NULL)
6534
90.7k
            return(NULL);
6535
14.9k
  SKIP_BLANKS_PE;
6536
14.9k
  GROW;
6537
14.9k
    } else {
6538
12.1k
  elem = xmlParseName(ctxt);
6539
12.1k
  if (elem == NULL) {
6540
199
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6541
199
      return(NULL);
6542
199
  }
6543
11.9k
        cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6544
11.9k
  if (cur == NULL) {
6545
0
      xmlErrMemory(ctxt);
6546
0
      return(NULL);
6547
0
  }
6548
11.9k
  GROW;
6549
11.9k
  if (RAW == '?') {
6550
643
      cur->ocur = XML_ELEMENT_CONTENT_OPT;
6551
643
      NEXT;
6552
11.3k
  } else if (RAW == '*') {
6553
653
      cur->ocur = XML_ELEMENT_CONTENT_MULT;
6554
653
      NEXT;
6555
10.6k
  } else if (RAW == '+') {
6556
1.56k
      cur->ocur = XML_ELEMENT_CONTENT_PLUS;
6557
1.56k
      NEXT;
6558
9.10k
  } else {
6559
9.10k
      cur->ocur = XML_ELEMENT_CONTENT_ONCE;
6560
9.10k
  }
6561
11.9k
  GROW;
6562
11.9k
    }
6563
26.9k
    SKIP_BLANKS_PE;
6564
78.2k
    while ((RAW != ')') && (PARSER_STOPPED(ctxt) == 0)) {
6565
        /*
6566
   * Each loop we parse one separator and one element.
6567
   */
6568
58.0k
        if (RAW == ',') {
6569
43.3k
      if (type == 0) type = CUR;
6570
6571
      /*
6572
       * Detect "Name | Name , Name" error
6573
       */
6574
38.9k
      else if (type != CUR) {
6575
7
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6576
7
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6577
7
                      type);
6578
7
    if ((last != NULL) && (last != ret))
6579
7
        xmlFreeDocElementContent(ctxt->myDoc, last);
6580
7
    if (ret != NULL)
6581
7
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6582
7
    return(NULL);
6583
7
      }
6584
43.3k
      NEXT;
6585
6586
43.3k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ);
6587
43.3k
      if (op == NULL) {
6588
0
                xmlErrMemory(ctxt);
6589
0
    if ((last != NULL) && (last != ret))
6590
0
        xmlFreeDocElementContent(ctxt->myDoc, last);
6591
0
          xmlFreeDocElementContent(ctxt->myDoc, ret);
6592
0
    return(NULL);
6593
0
      }
6594
43.3k
      if (last == NULL) {
6595
4.41k
    op->c1 = ret;
6596
4.41k
    if (ret != NULL)
6597
4.41k
        ret->parent = op;
6598
4.41k
    ret = cur = op;
6599
38.9k
      } else {
6600
38.9k
          cur->c2 = op;
6601
38.9k
    if (op != NULL)
6602
38.9k
        op->parent = cur;
6603
38.9k
    op->c1 = last;
6604
38.9k
    if (last != NULL)
6605
38.9k
        last->parent = op;
6606
38.9k
    cur =op;
6607
38.9k
    last = NULL;
6608
38.9k
      }
6609
43.3k
  } else if (RAW == '|') {
6610
13.8k
      if (type == 0) type = CUR;
6611
6612
      /*
6613
       * Detect "Name , Name | Name" error
6614
       */
6615
6.51k
      else if (type != CUR) {
6616
6
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6617
6
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6618
6
          type);
6619
6
    if ((last != NULL) && (last != ret))
6620
6
        xmlFreeDocElementContent(ctxt->myDoc, last);
6621
6
    if (ret != NULL)
6622
6
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6623
6
    return(NULL);
6624
6
      }
6625
13.8k
      NEXT;
6626
6627
13.8k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6628
13.8k
      if (op == NULL) {
6629
0
                xmlErrMemory(ctxt);
6630
0
    if ((last != NULL) && (last != ret))
6631
0
        xmlFreeDocElementContent(ctxt->myDoc, last);
6632
0
    if (ret != NULL)
6633
0
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6634
0
    return(NULL);
6635
0
      }
6636
13.8k
      if (last == NULL) {
6637
7.35k
    op->c1 = ret;
6638
7.35k
    if (ret != NULL)
6639
7.35k
        ret->parent = op;
6640
7.35k
    ret = cur = op;
6641
7.35k
      } else {
6642
6.50k
          cur->c2 = op;
6643
6.50k
    if (op != NULL)
6644
6.50k
        op->parent = cur;
6645
6.50k
    op->c1 = last;
6646
6.50k
    if (last != NULL)
6647
6.50k
        last->parent = op;
6648
6.50k
    cur =op;
6649
6.50k
    last = NULL;
6650
6.50k
      }
6651
13.8k
  } else {
6652
759
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL);
6653
759
      if ((last != NULL) && (last != ret))
6654
514
          xmlFreeDocElementContent(ctxt->myDoc, last);
6655
759
      if (ret != NULL)
6656
759
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6657
759
      return(NULL);
6658
759
  }
6659
57.2k
  GROW;
6660
57.2k
  SKIP_BLANKS_PE;
6661
57.2k
  GROW;
6662
57.2k
  if (RAW == '(') {
6663
8.71k
      int inputid = ctxt->input->id;
6664
      /* Recurse on second child */
6665
8.71k
      NEXT;
6666
8.71k
      SKIP_BLANKS_PE;
6667
8.71k
      last = xmlParseElementChildrenContentDeclPriv(ctxt, inputid,
6668
8.71k
                                                          depth + 1);
6669
8.71k
            if (last == NULL) {
6670
5.82k
    if (ret != NULL)
6671
5.82k
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6672
5.82k
    return(NULL);
6673
5.82k
            }
6674
2.89k
      SKIP_BLANKS_PE;
6675
48.5k
  } else {
6676
48.5k
      elem = xmlParseName(ctxt);
6677
48.5k
      if (elem == NULL) {
6678
90
    xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6679
90
    if (ret != NULL)
6680
90
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6681
90
    return(NULL);
6682
90
      }
6683
48.4k
      last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6684
48.4k
      if (last == NULL) {
6685
0
                xmlErrMemory(ctxt);
6686
0
    if (ret != NULL)
6687
0
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6688
0
    return(NULL);
6689
0
      }
6690
48.4k
      if (RAW == '?') {
6691
464
    last->ocur = XML_ELEMENT_CONTENT_OPT;
6692
464
    NEXT;
6693
47.9k
      } else if (RAW == '*') {
6694
724
    last->ocur = XML_ELEMENT_CONTENT_MULT;
6695
724
    NEXT;
6696
47.2k
      } else if (RAW == '+') {
6697
1.17k
    last->ocur = XML_ELEMENT_CONTENT_PLUS;
6698
1.17k
    NEXT;
6699
46.0k
      } else {
6700
46.0k
    last->ocur = XML_ELEMENT_CONTENT_ONCE;
6701
46.0k
      }
6702
48.4k
  }
6703
51.3k
  SKIP_BLANKS_PE;
6704
51.3k
  GROW;
6705
51.3k
    }
6706
20.2k
    if ((cur != NULL) && (last != NULL)) {
6707
5.33k
        cur->c2 = last;
6708
5.33k
  if (last != NULL)
6709
5.33k
      last->parent = cur;
6710
5.33k
    }
6711
20.2k
    if (ctxt->input->id != inputchk) {
6712
0
  xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
6713
0
                       "Element content declaration doesn't start and stop in"
6714
0
                       " the same entity\n");
6715
0
    }
6716
20.2k
    NEXT;
6717
20.2k
    if (RAW == '?') {
6718
2.33k
  if (ret != NULL) {
6719
2.33k
      if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) ||
6720
2.33k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6721
368
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6722
1.96k
      else
6723
1.96k
          ret->ocur = XML_ELEMENT_CONTENT_OPT;
6724
2.33k
  }
6725
2.33k
  NEXT;
6726
17.9k
    } else if (RAW == '*') {
6727
2.78k
  if (ret != NULL) {
6728
2.78k
      ret->ocur = XML_ELEMENT_CONTENT_MULT;
6729
2.78k
      cur = ret;
6730
      /*
6731
       * Some normalization:
6732
       * (a | b* | c?)* == (a | b | c)*
6733
       */
6734
6.99k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6735
4.20k
    if ((cur->c1 != NULL) &&
6736
4.20k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6737
4.20k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT)))
6738
1.39k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6739
4.20k
    if ((cur->c2 != NULL) &&
6740
4.20k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6741
4.20k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT)))
6742
327
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6743
4.20k
    cur = cur->c2;
6744
4.20k
      }
6745
2.78k
  }
6746
2.78k
  NEXT;
6747
15.1k
    } else if (RAW == '+') {
6748
4.33k
  if (ret != NULL) {
6749
4.33k
      int found = 0;
6750
6751
4.33k
      if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) ||
6752
4.33k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6753
1.27k
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6754
3.06k
      else
6755
3.06k
          ret->ocur = XML_ELEMENT_CONTENT_PLUS;
6756
      /*
6757
       * Some normalization:
6758
       * (a | b*)+ == (a | b)*
6759
       * (a | b?)+ == (a | b)*
6760
       */
6761
6.82k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6762
2.48k
    if ((cur->c1 != NULL) &&
6763
2.48k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6764
2.48k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) {
6765
580
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6766
580
        found = 1;
6767
580
    }
6768
2.48k
    if ((cur->c2 != NULL) &&
6769
2.48k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6770
2.48k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) {
6771
403
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6772
403
        found = 1;
6773
403
    }
6774
2.48k
    cur = cur->c2;
6775
2.48k
      }
6776
4.33k
      if (found)
6777
911
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6778
4.33k
  }
6779
4.33k
  NEXT;
6780
4.33k
    }
6781
20.2k
    return(ret);
6782
26.9k
}
6783
6784
/**
6785
 * xmlParseElementChildrenContentDecl:
6786
 * @ctxt:  an XML parser context
6787
 * @inputchk:  the input used for the current entity, needed for boundary checks
6788
 *
6789
 * DEPRECATED: Internal function, don't use.
6790
 *
6791
 * parse the declaration for a Mixed Element content
6792
 * The leading '(' and spaces have been skipped in xmlParseElementContentDecl
6793
 *
6794
 * [47] children ::= (choice | seq) ('?' | '*' | '+')?
6795
 *
6796
 * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')?
6797
 *
6798
 * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')'
6799
 *
6800
 * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
6801
 *
6802
 * [ VC: Proper Group/PE Nesting ] applies to [49] and [50]
6803
 * TODO Parameter-entity replacement text must be properly nested
6804
 *  with parenthesized groups. That is to say, if either of the
6805
 *  opening or closing parentheses in a choice, seq, or Mixed
6806
 *  construct is contained in the replacement text for a parameter
6807
 *  entity, both must be contained in the same replacement text. For
6808
 *  interoperability, if a parameter-entity reference appears in a
6809
 *  choice, seq, or Mixed construct, its replacement text should not
6810
 *  be empty, and neither the first nor last non-blank character of
6811
 *  the replacement text should be a connector (| or ,).
6812
 *
6813
 * Returns the tree of xmlElementContentPtr describing the element
6814
 *          hierarchy.
6815
 */
6816
xmlElementContentPtr
6817
0
xmlParseElementChildrenContentDecl(xmlParserCtxtPtr ctxt, int inputchk) {
6818
    /* stub left for API/ABI compat */
6819
0
    return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1));
6820
0
}
6821
6822
/**
6823
 * xmlParseElementContentDecl:
6824
 * @ctxt:  an XML parser context
6825
 * @name:  the name of the element being defined.
6826
 * @result:  the Element Content pointer will be stored here if any
6827
 *
6828
 * DEPRECATED: Internal function, don't use.
6829
 *
6830
 * parse the declaration for an Element content either Mixed or Children,
6831
 * the cases EMPTY and ANY are handled directly in xmlParseElementDecl
6832
 *
6833
 * [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children
6834
 *
6835
 * returns: the type of element content XML_ELEMENT_TYPE_xxx
6836
 */
6837
6838
int
6839
xmlParseElementContentDecl(xmlParserCtxtPtr ctxt, const xmlChar *name,
6840
7.29k
                           xmlElementContentPtr *result) {
6841
6842
7.29k
    xmlElementContentPtr tree = NULL;
6843
7.29k
    int inputid = ctxt->input->id;
6844
7.29k
    int res;
6845
6846
7.29k
    *result = NULL;
6847
6848
7.29k
    if (RAW != '(') {
6849
0
  xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6850
0
    "xmlParseElementContentDecl : %s '(' expected\n", name);
6851
0
  return(-1);
6852
0
    }
6853
7.29k
    NEXT;
6854
7.29k
    GROW;
6855
7.29k
    SKIP_BLANKS_PE;
6856
7.29k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6857
3.83k
        tree = xmlParseElementMixedContentDecl(ctxt, inputid);
6858
3.83k
  res = XML_ELEMENT_TYPE_MIXED;
6859
3.83k
    } else {
6860
3.46k
        tree = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, 1);
6861
3.46k
  res = XML_ELEMENT_TYPE_ELEMENT;
6862
3.46k
    }
6863
7.29k
    SKIP_BLANKS_PE;
6864
7.29k
    *result = tree;
6865
7.29k
    return(res);
6866
7.29k
}
6867
6868
/**
6869
 * xmlParseElementDecl:
6870
 * @ctxt:  an XML parser context
6871
 *
6872
 * DEPRECATED: Internal function, don't use.
6873
 *
6874
 * Parse an element declaration. Always consumes '<!'.
6875
 *
6876
 * [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>'
6877
 *
6878
 * [ VC: Unique Element Type Declaration ]
6879
 * No element type may be declared more than once
6880
 *
6881
 * Returns the type of the element, or -1 in case of error
6882
 */
6883
int
6884
9.44k
xmlParseElementDecl(xmlParserCtxtPtr ctxt) {
6885
9.44k
    const xmlChar *name;
6886
9.44k
    int ret = -1;
6887
9.44k
    xmlElementContentPtr content  = NULL;
6888
6889
9.44k
    if ((CUR != '<') || (NXT(1) != '!'))
6890
0
        return(ret);
6891
9.44k
    SKIP(2);
6892
6893
    /* GROW; done in the caller */
6894
9.44k
    if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) {
6895
9.38k
  int inputid = ctxt->input->id;
6896
6897
9.38k
  SKIP(7);
6898
9.38k
  if (SKIP_BLANKS_PE == 0) {
6899
71
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6900
71
               "Space required after 'ELEMENT'\n");
6901
71
      return(-1);
6902
71
  }
6903
9.31k
        name = xmlParseName(ctxt);
6904
9.31k
  if (name == NULL) {
6905
238
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6906
238
         "xmlParseElementDecl: no name for Element\n");
6907
238
      return(-1);
6908
238
  }
6909
9.07k
  if (SKIP_BLANKS_PE == 0) {
6910
229
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6911
229
         "Space required after the element name\n");
6912
229
  }
6913
9.07k
  if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) {
6914
834
      SKIP(5);
6915
      /*
6916
       * Element must always be empty.
6917
       */
6918
834
      ret = XML_ELEMENT_TYPE_EMPTY;
6919
8.24k
  } else if ((RAW == 'A') && (NXT(1) == 'N') &&
6920
8.24k
             (NXT(2) == 'Y')) {
6921
697
      SKIP(3);
6922
      /*
6923
       * Element is a generic container.
6924
       */
6925
697
      ret = XML_ELEMENT_TYPE_ANY;
6926
7.54k
  } else if (RAW == '(') {
6927
7.29k
      ret = xmlParseElementContentDecl(ctxt, name, &content);
6928
7.29k
  } else {
6929
      /*
6930
       * [ WFC: PEs in Internal Subset ] error handling.
6931
       */
6932
252
            xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6933
252
                  "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n");
6934
252
      return(-1);
6935
252
  }
6936
6937
8.82k
  SKIP_BLANKS_PE;
6938
6939
8.82k
  if (RAW != '>') {
6940
1.03k
      xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
6941
1.03k
      if (content != NULL) {
6942
58
    xmlFreeDocElementContent(ctxt->myDoc, content);
6943
58
      }
6944
7.78k
  } else {
6945
7.78k
      if (inputid != ctxt->input->id) {
6946
0
    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
6947
0
                               "Element declaration doesn't start and stop in"
6948
0
                               " the same entity\n");
6949
0
      }
6950
6951
7.78k
      NEXT;
6952
7.78k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6953
7.78k
    (ctxt->sax->elementDecl != NULL)) {
6954
0
    if (content != NULL)
6955
0
        content->parent = NULL;
6956
0
          ctxt->sax->elementDecl(ctxt->userData, name, ret,
6957
0
                           content);
6958
0
    if ((content != NULL) && (content->parent == NULL)) {
6959
        /*
6960
         * this is a trick: if xmlAddElementDecl is called,
6961
         * instead of copying the full tree it is plugged directly
6962
         * if called from the parser. Avoid duplicating the
6963
         * interfaces or change the API/ABI
6964
         */
6965
0
        xmlFreeDocElementContent(ctxt->myDoc, content);
6966
0
    }
6967
7.78k
      } else if (content != NULL) {
6968
5.28k
    xmlFreeDocElementContent(ctxt->myDoc, content);
6969
5.28k
      }
6970
7.78k
  }
6971
8.82k
    }
6972
8.88k
    return(ret);
6973
9.44k
}
6974
6975
/**
6976
 * xmlParseConditionalSections
6977
 * @ctxt:  an XML parser context
6978
 *
6979
 * Parse a conditional section. Always consumes '<!['.
6980
 *
6981
 * [61] conditionalSect ::= includeSect | ignoreSect
6982
 * [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>'
6983
 * [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>'
6984
 * [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' Ignore)*
6985
 * [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*)
6986
 */
6987
6988
static void
6989
0
xmlParseConditionalSections(xmlParserCtxtPtr ctxt) {
6990
0
    int *inputIds = NULL;
6991
0
    size_t inputIdsSize = 0;
6992
0
    size_t depth = 0;
6993
6994
0
    while (PARSER_STOPPED(ctxt) == 0) {
6995
0
        if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6996
0
            int id = ctxt->input->id;
6997
6998
0
            SKIP(3);
6999
0
            SKIP_BLANKS_PE;
7000
7001
0
            if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) {
7002
0
                SKIP(7);
7003
0
                SKIP_BLANKS_PE;
7004
0
                if (RAW != '[') {
7005
0
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
7006
0
                    xmlHaltParser(ctxt);
7007
0
                    goto error;
7008
0
                }
7009
0
                if (ctxt->input->id != id) {
7010
0
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
7011
0
                                   "All markup of the conditional section is"
7012
0
                                   " not in the same entity\n");
7013
0
                }
7014
0
                NEXT;
7015
7016
0
                if (inputIdsSize <= depth) {
7017
0
                    int *tmp;
7018
0
                    int newSize;
7019
7020
0
                    newSize = xmlGrowCapacity(inputIdsSize, sizeof(tmp[0]),
7021
0
                                              4, 1000);
7022
0
                    if (newSize < 0) {
7023
0
                        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
7024
0
                                       "Maximum conditional section nesting"
7025
0
                                       " depth exceeded\n");
7026
0
                        goto error;
7027
0
                    }
7028
0
                    tmp = xmlRealloc(inputIds, newSize * sizeof(tmp[0]));
7029
0
                    if (tmp == NULL) {
7030
0
                        xmlErrMemory(ctxt);
7031
0
                        goto error;
7032
0
                    }
7033
0
                    inputIds = tmp;
7034
0
                    inputIdsSize = newSize;
7035
0
                }
7036
0
                inputIds[depth] = id;
7037
0
                depth++;
7038
0
            } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) {
7039
0
                size_t ignoreDepth = 0;
7040
7041
0
                SKIP(6);
7042
0
                SKIP_BLANKS_PE;
7043
0
                if (RAW != '[') {
7044
0
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
7045
0
                    xmlHaltParser(ctxt);
7046
0
                    goto error;
7047
0
                }
7048
0
                if (ctxt->input->id != id) {
7049
0
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
7050
0
                                   "All markup of the conditional section is"
7051
0
                                   " not in the same entity\n");
7052
0
                }
7053
0
                NEXT;
7054
7055
0
                while (PARSER_STOPPED(ctxt) == 0) {
7056
0
                    if (RAW == 0) {
7057
0
                        xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL);
7058
0
                        goto error;
7059
0
                    }
7060
0
                    if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
7061
0
                        SKIP(3);
7062
0
                        ignoreDepth++;
7063
                        /* Check for integer overflow */
7064
0
                        if (ignoreDepth == 0) {
7065
0
                            xmlErrMemory(ctxt);
7066
0
                            goto error;
7067
0
                        }
7068
0
                    } else if ((RAW == ']') && (NXT(1) == ']') &&
7069
0
                               (NXT(2) == '>')) {
7070
0
                        SKIP(3);
7071
0
                        if (ignoreDepth == 0)
7072
0
                            break;
7073
0
                        ignoreDepth--;
7074
0
                    } else {
7075
0
                        NEXT;
7076
0
                    }
7077
0
                }
7078
7079
0
                if (ctxt->input->id != id) {
7080
0
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
7081
0
                                   "All markup of the conditional section is"
7082
0
                                   " not in the same entity\n");
7083
0
                }
7084
0
            } else {
7085
0
                xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL);
7086
0
                xmlHaltParser(ctxt);
7087
0
                goto error;
7088
0
            }
7089
0
        } else if ((depth > 0) &&
7090
0
                   (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) {
7091
0
            depth--;
7092
0
            if (ctxt->input->id != inputIds[depth]) {
7093
0
                xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY,
7094
0
                               "All markup of the conditional section is not"
7095
0
                               " in the same entity\n");
7096
0
            }
7097
0
            SKIP(3);
7098
0
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
7099
0
            xmlParseMarkupDecl(ctxt);
7100
0
        } else {
7101
0
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
7102
0
            xmlHaltParser(ctxt);
7103
0
            goto error;
7104
0
        }
7105
7106
0
        if (depth == 0)
7107
0
            break;
7108
7109
0
        SKIP_BLANKS_PE;
7110
0
        SHRINK;
7111
0
        GROW;
7112
0
    }
7113
7114
0
error:
7115
0
    xmlFree(inputIds);
7116
0
}
7117
7118
/**
7119
 * xmlParseMarkupDecl:
7120
 * @ctxt:  an XML parser context
7121
 *
7122
 * DEPRECATED: Internal function, don't use.
7123
 *
7124
 * Parse markup declarations. Always consumes '<!' or '<?'.
7125
 *
7126
 * [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl |
7127
 *                     NotationDecl | PI | Comment
7128
 *
7129
 * [ VC: Proper Declaration/PE Nesting ]
7130
 * Parameter-entity replacement text must be properly nested with
7131
 * markup declarations. That is to say, if either the first character
7132
 * or the last character of a markup declaration (markupdecl above) is
7133
 * contained in the replacement text for a parameter-entity reference,
7134
 * both must be contained in the same replacement text.
7135
 *
7136
 * [ WFC: PEs in Internal Subset ]
7137
 * In the internal DTD subset, parameter-entity references can occur
7138
 * only where markup declarations can occur, not within markup declarations.
7139
 * (This does not apply to references that occur in external parameter
7140
 * entities or to the external subset.)
7141
 */
7142
void
7143
96.2k
xmlParseMarkupDecl(xmlParserCtxtPtr ctxt) {
7144
96.2k
    GROW;
7145
96.2k
    if (CUR == '<') {
7146
96.2k
        if (NXT(1) == '!') {
7147
77.4k
      switch (NXT(2)) {
7148
40.6k
          case 'E':
7149
40.6k
        if (NXT(3) == 'L')
7150
9.44k
      xmlParseElementDecl(ctxt);
7151
31.2k
        else if (NXT(3) == 'N')
7152
31.1k
      xmlParseEntityDecl(ctxt);
7153
47
                    else
7154
47
                        SKIP(2);
7155
40.6k
        break;
7156
14.7k
          case 'A':
7157
14.7k
        xmlParseAttributeListDecl(ctxt);
7158
14.7k
        break;
7159
3.47k
          case 'N':
7160
3.47k
        xmlParseNotationDecl(ctxt);
7161
3.47k
        break;
7162
18.3k
          case '-':
7163
18.3k
        xmlParseComment(ctxt);
7164
18.3k
        break;
7165
164
    default:
7166
164
                    xmlFatalErr(ctxt,
7167
164
                                ctxt->inSubset == 2 ?
7168
0
                                    XML_ERR_EXT_SUBSET_NOT_FINISHED :
7169
164
                                    XML_ERR_INT_SUBSET_NOT_FINISHED,
7170
164
                                NULL);
7171
164
                    SKIP(2);
7172
164
        break;
7173
77.4k
      }
7174
77.4k
  } else if (NXT(1) == '?') {
7175
18.8k
      xmlParsePI(ctxt);
7176
18.8k
  }
7177
96.2k
    }
7178
96.2k
}
7179
7180
/**
7181
 * xmlParseTextDecl:
7182
 * @ctxt:  an XML parser context
7183
 *
7184
 * DEPRECATED: Internal function, don't use.
7185
 *
7186
 * parse an XML declaration header for external entities
7187
 *
7188
 * [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>'
7189
 */
7190
7191
void
7192
0
xmlParseTextDecl(xmlParserCtxtPtr ctxt) {
7193
0
    xmlChar *version;
7194
7195
    /*
7196
     * We know that '<?xml' is here.
7197
     */
7198
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
7199
0
  SKIP(5);
7200
0
    } else {
7201
0
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL);
7202
0
  return;
7203
0
    }
7204
7205
0
    if (SKIP_BLANKS == 0) {
7206
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7207
0
           "Space needed after '<?xml'\n");
7208
0
    }
7209
7210
    /*
7211
     * We may have the VersionInfo here.
7212
     */
7213
0
    version = xmlParseVersionInfo(ctxt);
7214
0
    if (version == NULL) {
7215
0
  version = xmlCharStrdup(XML_DEFAULT_VERSION);
7216
0
        if (version == NULL) {
7217
0
            xmlErrMemory(ctxt);
7218
0
            return;
7219
0
        }
7220
0
    } else {
7221
0
  if (SKIP_BLANKS == 0) {
7222
0
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7223
0
               "Space needed here\n");
7224
0
  }
7225
0
    }
7226
0
    ctxt->input->version = version;
7227
7228
    /*
7229
     * We must have the encoding declaration
7230
     */
7231
0
    xmlParseEncodingDecl(ctxt);
7232
7233
0
    SKIP_BLANKS;
7234
0
    if ((RAW == '?') && (NXT(1) == '>')) {
7235
0
        SKIP(2);
7236
0
    } else if (RAW == '>') {
7237
        /* Deprecated old WD ... */
7238
0
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7239
0
  NEXT;
7240
0
    } else {
7241
0
        int c;
7242
7243
0
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7244
0
        while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) {
7245
0
            NEXT;
7246
0
            if (c == '>')
7247
0
                break;
7248
0
        }
7249
0
    }
7250
0
}
7251
7252
/**
7253
 * xmlParseExternalSubset:
7254
 * @ctxt:  an XML parser context
7255
 * @ExternalID: the external identifier
7256
 * @SystemID: the system identifier (or URL)
7257
 *
7258
 * DEPRECATED: Internal function, don't use.
7259
 *
7260
 * parse Markup declarations from an external subset
7261
 *
7262
 * [30] extSubset ::= textDecl? extSubsetDecl
7263
 *
7264
 * [31] extSubsetDecl ::= (markupdecl | conditionalSect | PEReference | S) *
7265
 */
7266
void
7267
xmlParseExternalSubset(xmlParserCtxtPtr ctxt, const xmlChar *ExternalID,
7268
0
                       const xmlChar *SystemID) {
7269
0
    int oldInputNr;
7270
7271
0
    xmlCtxtInitializeLate(ctxt);
7272
7273
0
    xmlDetectEncoding(ctxt);
7274
7275
0
    if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) {
7276
0
  xmlParseTextDecl(ctxt);
7277
0
    }
7278
0
    if (ctxt->myDoc == NULL) {
7279
0
        ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0");
7280
0
  if (ctxt->myDoc == NULL) {
7281
0
      xmlErrMemory(ctxt);
7282
0
      return;
7283
0
  }
7284
0
  ctxt->myDoc->properties = XML_DOC_INTERNAL;
7285
0
    }
7286
0
    if ((ctxt->myDoc != NULL) && (ctxt->myDoc->intSubset == NULL) &&
7287
0
        (xmlCreateIntSubset(ctxt->myDoc, NULL, ExternalID, SystemID) == NULL)) {
7288
0
        xmlErrMemory(ctxt);
7289
0
    }
7290
7291
0
    ctxt->inSubset = 2;
7292
0
    oldInputNr = ctxt->inputNr;
7293
7294
0
    SKIP_BLANKS_PE;
7295
0
    while (((RAW != 0) || (ctxt->inputNr > oldInputNr)) &&
7296
0
           (!PARSER_STOPPED(ctxt))) {
7297
0
  GROW;
7298
0
        if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
7299
0
            xmlParseConditionalSections(ctxt);
7300
0
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
7301
0
            xmlParseMarkupDecl(ctxt);
7302
0
        } else {
7303
0
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
7304
0
            xmlHaltParser(ctxt);
7305
0
            return;
7306
0
        }
7307
0
        SKIP_BLANKS_PE;
7308
0
        SHRINK;
7309
0
    }
7310
7311
0
    while (ctxt->inputNr > oldInputNr)
7312
0
        xmlPopPE(ctxt);
7313
7314
0
    xmlParserCheckEOF(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED);
7315
0
}
7316
7317
/**
7318
 * xmlParseReference:
7319
 * @ctxt:  an XML parser context
7320
 *
7321
 * DEPRECATED: Internal function, don't use.
7322
 *
7323
 * parse and handle entity references in content, depending on the SAX
7324
 * interface, this may end-up in a call to character() if this is a
7325
 * CharRef, a predefined entity, if there is no reference() callback.
7326
 * or if the parser was asked to switch to that mode.
7327
 *
7328
 * Always consumes '&'.
7329
 *
7330
 * [67] Reference ::= EntityRef | CharRef
7331
 */
7332
void
7333
363k
xmlParseReference(xmlParserCtxtPtr ctxt) {
7334
363k
    xmlEntityPtr ent = NULL;
7335
363k
    const xmlChar *name;
7336
363k
    xmlChar *val;
7337
7338
363k
    if (RAW != '&')
7339
0
        return;
7340
7341
    /*
7342
     * Simple case of a CharRef
7343
     */
7344
363k
    if (NXT(1) == '#') {
7345
65.0k
  int i = 0;
7346
65.0k
  xmlChar out[16];
7347
65.0k
  int value = xmlParseCharRef(ctxt);
7348
7349
65.0k
  if (value == 0)
7350
286
      return;
7351
7352
        /*
7353
         * Just encode the value in UTF-8
7354
         */
7355
64.7k
        COPY_BUF(out, i, value);
7356
64.7k
        out[i] = 0;
7357
64.7k
        if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7358
64.7k
            (!ctxt->disableSAX))
7359
64.7k
            ctxt->sax->characters(ctxt->userData, out, i);
7360
64.7k
  return;
7361
65.0k
    }
7362
7363
    /*
7364
     * We are seeing an entity reference
7365
     */
7366
298k
    name = xmlParseEntityRefInternal(ctxt);
7367
298k
    if (name == NULL)
7368
293
        return;
7369
298k
    ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0);
7370
298k
    if (ent == NULL) {
7371
        /*
7372
         * Create a reference for undeclared entities.
7373
         */
7374
1.34k
        if ((ctxt->replaceEntities == 0) &&
7375
1.34k
            (ctxt->sax != NULL) &&
7376
1.34k
            (ctxt->disableSAX == 0) &&
7377
1.34k
            (ctxt->sax->reference != NULL)) {
7378
0
            ctxt->sax->reference(ctxt->userData, name);
7379
0
        }
7380
1.34k
        return;
7381
1.34k
    }
7382
296k
    if (!ctxt->wellFormed)
7383
0
  return;
7384
7385
    /* special case of predefined entities */
7386
296k
    if ((ent->name == NULL) ||
7387
296k
        (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
7388
296k
  val = ent->content;
7389
296k
  if (val == NULL) return;
7390
  /*
7391
   * inline the entity.
7392
   */
7393
296k
  if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7394
296k
      (!ctxt->disableSAX))
7395
296k
      ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val));
7396
296k
  return;
7397
296k
    }
7398
7399
    /*
7400
     * Some users try to parse entities on their own and used to set
7401
     * the renamed "checked" member. Fix the flags to cover this
7402
     * case.
7403
     */
7404
0
    if (((ent->flags & XML_ENT_PARSED) == 0) && (ent->children != NULL))
7405
0
        ent->flags |= XML_ENT_PARSED;
7406
7407
    /*
7408
     * The first reference to the entity trigger a parsing phase
7409
     * where the ent->children is filled with the result from
7410
     * the parsing.
7411
     * Note: external parsed entities will not be loaded, it is not
7412
     * required for a non-validating parser, unless the parsing option
7413
     * of validating, or substituting entities were given. Doing so is
7414
     * far more secure as the parser will only process data coming from
7415
     * the document entity by default.
7416
     *
7417
     * FIXME: This doesn't work correctly since entities can be
7418
     * expanded with different namespace declarations in scope.
7419
     * For example:
7420
     *
7421
     * <!DOCTYPE doc [
7422
     *   <!ENTITY ent "<ns:elem/>">
7423
     * ]>
7424
     * <doc>
7425
     *   <decl1 xmlns:ns="urn:ns1">
7426
     *     &ent;
7427
     *   </decl1>
7428
     *   <decl2 xmlns:ns="urn:ns2">
7429
     *     &ent;
7430
     *   </decl2>
7431
     * </doc>
7432
     *
7433
     * Proposed fix:
7434
     *
7435
     * - Ignore current namespace declarations when parsing the
7436
     *   entity. If a prefix can't be resolved, don't report an error
7437
     *   but mark it as unresolved.
7438
     * - Try to resolve these prefixes when expanding the entity.
7439
     *   This will require a specialized version of xmlStaticCopyNode
7440
     *   which can also make use of the namespace hash table to avoid
7441
     *   quadratic behavior.
7442
     *
7443
     * Alternatively, we could simply reparse the entity on each
7444
     * expansion like we already do with custom SAX callbacks.
7445
     * External entity content should be cached in this case.
7446
     */
7447
0
    if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) ||
7448
0
        (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
7449
0
         ((ctxt->replaceEntities) ||
7450
0
          (ctxt->validate)))) {
7451
0
        if ((ent->flags & XML_ENT_PARSED) == 0) {
7452
0
            xmlCtxtParseEntity(ctxt, ent);
7453
0
        } else if (ent->children == NULL) {
7454
            /*
7455
             * Probably running in SAX mode and the callbacks don't
7456
             * build the entity content. Parse the entity again.
7457
             *
7458
             * This will also be triggered in normal tree builder mode
7459
             * if an entity happens to be empty, causing unnecessary
7460
             * reloads. It's hard to come up with a reliable check in
7461
             * which mode we're running.
7462
             */
7463
0
            xmlCtxtParseEntity(ctxt, ent);
7464
0
        }
7465
0
    }
7466
7467
    /*
7468
     * We also check for amplification if entities aren't substituted.
7469
     * They might be expanded later.
7470
     */
7471
0
    if (xmlParserEntityCheck(ctxt, ent->expandedSize))
7472
0
        return;
7473
7474
0
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
7475
0
        return;
7476
7477
0
    if (ctxt->replaceEntities == 0) {
7478
  /*
7479
   * Create a reference
7480
   */
7481
0
        if (ctxt->sax->reference != NULL)
7482
0
      ctxt->sax->reference(ctxt->userData, ent->name);
7483
0
    } else if ((ent->children != NULL) && (ctxt->node != NULL)) {
7484
0
        xmlNodePtr copy, cur;
7485
7486
        /*
7487
         * Seems we are generating the DOM content, copy the tree
7488
   */
7489
0
        cur = ent->children;
7490
7491
        /*
7492
         * Handle first text node with SAX to coalesce text efficiently
7493
         */
7494
0
        if ((cur->type == XML_TEXT_NODE) ||
7495
0
            (cur->type == XML_CDATA_SECTION_NODE)) {
7496
0
            int len = xmlStrlen(cur->content);
7497
7498
0
            if ((cur->type == XML_TEXT_NODE) ||
7499
0
                (ctxt->options & XML_PARSE_NOCDATA)) {
7500
0
                if (ctxt->sax->characters != NULL)
7501
0
                    ctxt->sax->characters(ctxt, cur->content, len);
7502
0
            } else {
7503
0
                if (ctxt->sax->cdataBlock != NULL)
7504
0
                    ctxt->sax->cdataBlock(ctxt, cur->content, len);
7505
0
            }
7506
7507
0
            cur = cur->next;
7508
0
        }
7509
7510
0
        while (cur != NULL) {
7511
0
            xmlNodePtr last;
7512
7513
            /*
7514
             * Handle last text node with SAX to coalesce text efficiently
7515
             */
7516
0
            if ((cur->next == NULL) &&
7517
0
                ((cur->type == XML_TEXT_NODE) ||
7518
0
                 (cur->type == XML_CDATA_SECTION_NODE))) {
7519
0
                int len = xmlStrlen(cur->content);
7520
7521
0
                if ((cur->type == XML_TEXT_NODE) ||
7522
0
                    (ctxt->options & XML_PARSE_NOCDATA)) {
7523
0
                    if (ctxt->sax->characters != NULL)
7524
0
                        ctxt->sax->characters(ctxt, cur->content, len);
7525
0
                } else {
7526
0
                    if (ctxt->sax->cdataBlock != NULL)
7527
0
                        ctxt->sax->cdataBlock(ctxt, cur->content, len);
7528
0
                }
7529
7530
0
                break;
7531
0
            }
7532
7533
            /*
7534
             * Reset coalesce buffer stats only for non-text nodes.
7535
             */
7536
0
            ctxt->nodemem = 0;
7537
0
            ctxt->nodelen = 0;
7538
7539
0
            copy = xmlDocCopyNode(cur, ctxt->myDoc, 1);
7540
7541
0
            if (copy == NULL) {
7542
0
                xmlErrMemory(ctxt);
7543
0
                break;
7544
0
            }
7545
7546
0
            if (ctxt->parseMode == XML_PARSE_READER) {
7547
                /* Needed for reader */
7548
0
                copy->extra = cur->extra;
7549
                /* Maybe needed for reader */
7550
0
                copy->_private = cur->_private;
7551
0
            }
7552
7553
0
            copy->parent = ctxt->node;
7554
0
            last = ctxt->node->last;
7555
0
            if (last == NULL) {
7556
0
                ctxt->node->children = copy;
7557
0
            } else {
7558
0
                last->next = copy;
7559
0
                copy->prev = last;
7560
0
            }
7561
0
            ctxt->node->last = copy;
7562
7563
0
            cur = cur->next;
7564
0
        }
7565
0
    }
7566
0
}
7567
7568
static void
7569
29.8k
xmlHandleUndeclaredEntity(xmlParserCtxtPtr ctxt, const xmlChar *name) {
7570
    /*
7571
     * [ WFC: Entity Declared ]
7572
     * In a document without any DTD, a document with only an
7573
     * internal DTD subset which contains no parameter entity
7574
     * references, or a document with "standalone='yes'", the
7575
     * Name given in the entity reference must match that in an
7576
     * entity declaration, except that well-formed documents
7577
     * need not declare any of the following entities: amp, lt,
7578
     * gt, apos, quot.
7579
     * The declaration of a parameter entity must precede any
7580
     * reference to it.
7581
     * Similarly, the declaration of a general entity must
7582
     * precede any reference to it which appears in a default
7583
     * value in an attribute-list declaration. Note that if
7584
     * entities are declared in the external subset or in
7585
     * external parameter entities, a non-validating processor
7586
     * is not obligated to read and process their declarations;
7587
     * for such documents, the rule that an entity must be
7588
     * declared is a well-formedness constraint only if
7589
     * standalone='yes'.
7590
     */
7591
29.8k
    if ((ctxt->standalone == 1) ||
7592
29.8k
        ((ctxt->hasExternalSubset == 0) &&
7593
29.5k
         (ctxt->hasPErefs == 0))) {
7594
15.9k
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY,
7595
15.9k
                          "Entity '%s' not defined\n", name);
7596
15.9k
    } else if (ctxt->validate) {
7597
        /*
7598
         * [ VC: Entity Declared ]
7599
         * In a document with an external subset or external
7600
         * parameter entities with "standalone='no'", ...
7601
         * ... The declaration of a parameter entity must
7602
         * precede any reference to it...
7603
         */
7604
0
        xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY,
7605
0
                         "Entity '%s' not defined\n", name, NULL);
7606
13.8k
    } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) ||
7607
13.8k
               ((ctxt->replaceEntities) &&
7608
13.8k
                ((ctxt->options & XML_PARSE_NO_XXE) == 0))) {
7609
        /*
7610
         * Also raise a non-fatal error
7611
         *
7612
         * - if the external subset is loaded and all entity declarations
7613
         *   should be available, or
7614
         * - entity substition was requested without restricting
7615
         *   external entity access.
7616
         */
7617
13.8k
        xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY,
7618
13.8k
                     "Entity '%s' not defined\n", name);
7619
13.8k
    } else {
7620
0
        xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7621
0
                      "Entity '%s' not defined\n", name, NULL);
7622
0
    }
7623
7624
29.8k
    ctxt->valid = 0;
7625
29.8k
}
7626
7627
static xmlEntityPtr
7628
529k
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) {
7629
529k
    xmlEntityPtr ent = NULL;
7630
7631
    /*
7632
     * Predefined entities override any extra definition
7633
     */
7634
529k
    if ((ctxt->options & XML_PARSE_OLDSAX) == 0) {
7635
529k
        ent = xmlGetPredefinedEntity(name);
7636
529k
        if (ent != NULL)
7637
509k
            return(ent);
7638
529k
    }
7639
7640
    /*
7641
     * Ask first SAX for entity resolution, otherwise try the
7642
     * entities which may have stored in the parser context.
7643
     */
7644
19.5k
    if (ctxt->sax != NULL) {
7645
19.5k
  if (ctxt->sax->getEntity != NULL)
7646
19.5k
      ent = ctxt->sax->getEntity(ctxt->userData, name);
7647
19.5k
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7648
19.5k
      (ctxt->options & XML_PARSE_OLDSAX))
7649
0
      ent = xmlGetPredefinedEntity(name);
7650
19.5k
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7651
19.5k
      (ctxt->userData==ctxt)) {
7652
0
      ent = xmlSAX2GetEntity(ctxt, name);
7653
0
  }
7654
19.5k
    }
7655
7656
19.5k
    if (ent == NULL) {
7657
19.5k
        xmlHandleUndeclaredEntity(ctxt, name);
7658
19.5k
    }
7659
7660
    /*
7661
     * [ WFC: Parsed Entity ]
7662
     * An entity reference must not contain the name of an
7663
     * unparsed entity
7664
     */
7665
0
    else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) {
7666
0
  xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY,
7667
0
     "Entity reference to unparsed entity %s\n", name);
7668
0
        ent = NULL;
7669
0
    }
7670
7671
    /*
7672
     * [ WFC: No External Entity References ]
7673
     * Attribute values cannot contain direct or indirect
7674
     * entity references to external entities.
7675
     */
7676
0
    else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) {
7677
0
        if (inAttr) {
7678
0
            xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL,
7679
0
                 "Attribute references external entity '%s'\n", name);
7680
0
            ent = NULL;
7681
0
        }
7682
0
    }
7683
7684
19.5k
    return(ent);
7685
529k
}
7686
7687
/**
7688
 * xmlParseEntityRefInternal:
7689
 * @ctxt:  an XML parser context
7690
 * @inAttr:  whether we are in an attribute value
7691
 *
7692
 * Parse an entity reference. Always consumes '&'.
7693
 *
7694
 * [68] EntityRef ::= '&' Name ';'
7695
 *
7696
 * Returns the name, or NULL in case of error.
7697
 */
7698
static const xmlChar *
7699
1.00M
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) {
7700
1.00M
    const xmlChar *name;
7701
7702
1.00M
    GROW;
7703
7704
1.00M
    if (RAW != '&')
7705
0
        return(NULL);
7706
1.00M
    NEXT;
7707
1.00M
    name = xmlParseName(ctxt);
7708
1.00M
    if (name == NULL) {
7709
287k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7710
287k
           "xmlParseEntityRef: no name\n");
7711
287k
        return(NULL);
7712
287k
    }
7713
713k
    if (RAW != ';') {
7714
184k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7715
184k
  return(NULL);
7716
184k
    }
7717
529k
    NEXT;
7718
7719
529k
    return(name);
7720
713k
}
7721
7722
/**
7723
 * xmlParseEntityRef:
7724
 * @ctxt:  an XML parser context
7725
 *
7726
 * DEPRECATED: Internal function, don't use.
7727
 *
7728
 * Returns the xmlEntityPtr if found, or NULL otherwise.
7729
 */
7730
xmlEntityPtr
7731
0
xmlParseEntityRef(xmlParserCtxtPtr ctxt) {
7732
0
    const xmlChar *name;
7733
7734
0
    if (ctxt == NULL)
7735
0
        return(NULL);
7736
7737
0
    name = xmlParseEntityRefInternal(ctxt);
7738
0
    if (name == NULL)
7739
0
        return(NULL);
7740
7741
0
    return(xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 0));
7742
0
}
7743
7744
/**
7745
 * xmlParseStringEntityRef:
7746
 * @ctxt:  an XML parser context
7747
 * @str:  a pointer to an index in the string
7748
 *
7749
 * parse ENTITY references declarations, but this version parses it from
7750
 * a string value.
7751
 *
7752
 * [68] EntityRef ::= '&' Name ';'
7753
 *
7754
 * [ WFC: Entity Declared ]
7755
 * In a document without any DTD, a document with only an internal DTD
7756
 * subset which contains no parameter entity references, or a document
7757
 * with "standalone='yes'", the Name given in the entity reference
7758
 * must match that in an entity declaration, except that well-formed
7759
 * documents need not declare any of the following entities: amp, lt,
7760
 * gt, apos, quot.  The declaration of a parameter entity must precede
7761
 * any reference to it.  Similarly, the declaration of a general entity
7762
 * must precede any reference to it which appears in a default value in an
7763
 * attribute-list declaration. Note that if entities are declared in the
7764
 * external subset or in external parameter entities, a non-validating
7765
 * processor is not obligated to read and process their declarations;
7766
 * for such documents, the rule that an entity must be declared is a
7767
 * well-formedness constraint only if standalone='yes'.
7768
 *
7769
 * [ WFC: Parsed Entity ]
7770
 * An entity reference must not contain the name of an unparsed entity
7771
 *
7772
 * Returns the xmlEntityPtr if found, or NULL otherwise. The str pointer
7773
 * is updated to the current location in the string.
7774
 */
7775
static xmlChar *
7776
0
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) {
7777
0
    xmlChar *name;
7778
0
    const xmlChar *ptr;
7779
0
    xmlChar cur;
7780
7781
0
    if ((str == NULL) || (*str == NULL))
7782
0
        return(NULL);
7783
0
    ptr = *str;
7784
0
    cur = *ptr;
7785
0
    if (cur != '&')
7786
0
  return(NULL);
7787
7788
0
    ptr++;
7789
0
    name = xmlParseStringName(ctxt, &ptr);
7790
0
    if (name == NULL) {
7791
0
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7792
0
           "xmlParseStringEntityRef: no name\n");
7793
0
  *str = ptr;
7794
0
  return(NULL);
7795
0
    }
7796
0
    if (*ptr != ';') {
7797
0
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7798
0
        xmlFree(name);
7799
0
  *str = ptr;
7800
0
  return(NULL);
7801
0
    }
7802
0
    ptr++;
7803
7804
0
    *str = ptr;
7805
0
    return(name);
7806
0
}
7807
7808
/**
7809
 * xmlParsePEReference:
7810
 * @ctxt:  an XML parser context
7811
 *
7812
 * DEPRECATED: Internal function, don't use.
7813
 *
7814
 * Parse a parameter entity reference. Always consumes '%'.
7815
 *
7816
 * The entity content is handled directly by pushing it's content as
7817
 * a new input stream.
7818
 *
7819
 * [69] PEReference ::= '%' Name ';'
7820
 *
7821
 * [ WFC: No Recursion ]
7822
 * A parsed entity must not contain a recursive
7823
 * reference to itself, either directly or indirectly.
7824
 *
7825
 * [ WFC: Entity Declared ]
7826
 * In a document without any DTD, a document with only an internal DTD
7827
 * subset which contains no parameter entity references, or a document
7828
 * with "standalone='yes'", ...  ... The declaration of a parameter
7829
 * entity must precede any reference to it...
7830
 *
7831
 * [ VC: Entity Declared ]
7832
 * In a document with an external subset or external parameter entities
7833
 * with "standalone='no'", ...  ... The declaration of a parameter entity
7834
 * must precede any reference to it...
7835
 *
7836
 * [ WFC: In DTD ]
7837
 * Parameter-entity references may only appear in the DTD.
7838
 * NOTE: misleading but this is handled.
7839
 */
7840
void
7841
xmlParsePEReference(xmlParserCtxtPtr ctxt)
7842
11.0k
{
7843
11.0k
    const xmlChar *name;
7844
11.0k
    xmlEntityPtr entity = NULL;
7845
11.0k
    xmlParserInputPtr input;
7846
7847
11.0k
    if (RAW != '%')
7848
0
        return;
7849
11.0k
    NEXT;
7850
11.0k
    name = xmlParseName(ctxt);
7851
11.0k
    if (name == NULL) {
7852
1.70k
  xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n");
7853
1.70k
  return;
7854
1.70k
    }
7855
9.38k
    if (RAW != ';') {
7856
150
  xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL);
7857
150
        return;
7858
150
    }
7859
7860
9.23k
    NEXT;
7861
7862
    /* Must be set before xmlHandleUndeclaredEntity */
7863
9.23k
    ctxt->hasPErefs = 1;
7864
7865
    /*
7866
     * Request the entity from SAX
7867
     */
7868
9.23k
    if ((ctxt->sax != NULL) &&
7869
9.23k
  (ctxt->sax->getParameterEntity != NULL))
7870
0
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7871
7872
9.23k
    if (entity == NULL) {
7873
9.23k
        xmlHandleUndeclaredEntity(ctxt, name);
7874
9.23k
    } else {
7875
  /*
7876
   * Internal checking in case the entity quest barfed
7877
   */
7878
0
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7879
0
      (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) {
7880
0
      xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7881
0
      "Internal: %%%s; is not a parameter entity\n",
7882
0
        name, NULL);
7883
0
  } else {
7884
0
      if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) &&
7885
0
                ((ctxt->options & XML_PARSE_NO_XXE) ||
7886
0
     ((ctxt->loadsubset == 0) &&
7887
0
      (ctxt->replaceEntities == 0) &&
7888
0
      (ctxt->validate == 0))))
7889
0
    return;
7890
7891
0
            if (entity->flags & XML_ENT_EXPANDING) {
7892
0
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
7893
0
                xmlHaltParser(ctxt);
7894
0
                return;
7895
0
            }
7896
7897
0
      input = xmlNewEntityInputStream(ctxt, entity);
7898
0
      if (xmlCtxtPushInput(ctxt, input) < 0) {
7899
0
                xmlFreeInputStream(input);
7900
0
    return;
7901
0
            }
7902
7903
0
            entity->flags |= XML_ENT_EXPANDING;
7904
7905
0
            GROW;
7906
7907
0
      if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
7908
0
                xmlDetectEncoding(ctxt);
7909
7910
0
                if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
7911
0
                    (IS_BLANK_CH(NXT(5)))) {
7912
0
                    xmlParseTextDecl(ctxt);
7913
0
                }
7914
0
            }
7915
0
  }
7916
0
    }
7917
9.23k
}
7918
7919
/**
7920
 * xmlLoadEntityContent:
7921
 * @ctxt:  an XML parser context
7922
 * @entity: an unloaded system entity
7923
 *
7924
 * Load the content of an entity.
7925
 *
7926
 * Returns 0 in case of success and -1 in case of failure
7927
 */
7928
static int
7929
0
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) {
7930
0
    xmlParserInputPtr oldinput, input = NULL;
7931
0
    xmlParserInputPtr *oldinputTab;
7932
0
    const xmlChar *oldencoding;
7933
0
    xmlChar *content = NULL;
7934
0
    xmlResourceType rtype;
7935
0
    size_t length, i;
7936
0
    int oldinputNr, oldinputMax;
7937
0
    int ret = -1;
7938
0
    int res;
7939
7940
0
    if ((ctxt == NULL) || (entity == NULL) ||
7941
0
        ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) &&
7942
0
   (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) ||
7943
0
  (entity->content != NULL)) {
7944
0
  xmlFatalErr(ctxt, XML_ERR_ARGUMENT,
7945
0
              "xmlLoadEntityContent parameter error");
7946
0
        return(-1);
7947
0
    }
7948
7949
0
    if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY)
7950
0
        rtype = XML_RESOURCE_PARAMETER_ENTITY;
7951
0
    else
7952
0
        rtype = XML_RESOURCE_GENERAL_ENTITY;
7953
7954
0
    input = xmlLoadResource(ctxt, (char *) entity->URI,
7955
0
                            (char *) entity->ExternalID, rtype);
7956
0
    if (input == NULL)
7957
0
        return(-1);
7958
7959
0
    oldinput = ctxt->input;
7960
0
    oldinputNr = ctxt->inputNr;
7961
0
    oldinputMax = ctxt->inputMax;
7962
0
    oldinputTab = ctxt->inputTab;
7963
0
    oldencoding = ctxt->encoding;
7964
7965
0
    ctxt->input = NULL;
7966
0
    ctxt->inputNr = 0;
7967
0
    ctxt->inputMax = 1;
7968
0
    ctxt->encoding = NULL;
7969
0
    ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr));
7970
0
    if (ctxt->inputTab == NULL) {
7971
0
        xmlErrMemory(ctxt);
7972
0
        xmlFreeInputStream(input);
7973
0
        goto error;
7974
0
    }
7975
7976
0
    xmlBufResetInput(input->buf->buffer, input);
7977
7978
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
7979
0
        xmlFreeInputStream(input);
7980
0
        goto error;
7981
0
    }
7982
7983
0
    xmlDetectEncoding(ctxt);
7984
7985
    /*
7986
     * Parse a possible text declaration first
7987
     */
7988
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
7989
0
  xmlParseTextDecl(ctxt);
7990
        /*
7991
         * An XML-1.0 document can't reference an entity not XML-1.0
7992
         */
7993
0
        if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
7994
0
            (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
7995
0
            xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
7996
0
                           "Version mismatch between document and entity\n");
7997
0
        }
7998
0
    }
7999
8000
0
    length = input->cur - input->base;
8001
0
    xmlBufShrink(input->buf->buffer, length);
8002
0
    xmlSaturatedAdd(&ctxt->sizeentities, length);
8003
8004
0
    while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0)
8005
0
        ;
8006
8007
0
    xmlBufResetInput(input->buf->buffer, input);
8008
8009
0
    if (res < 0) {
8010
0
        xmlCtxtErrIO(ctxt, input->buf->error, NULL);
8011
0
        goto error;
8012
0
    }
8013
8014
0
    length = xmlBufUse(input->buf->buffer);
8015
0
    if (length > INT_MAX) {
8016
0
        xmlErrMemory(ctxt);
8017
0
        goto error;
8018
0
    }
8019
8020
0
    content = xmlStrndup(xmlBufContent(input->buf->buffer), length);
8021
0
    if (content == NULL) {
8022
0
        xmlErrMemory(ctxt);
8023
0
        goto error;
8024
0
    }
8025
8026
0
    for (i = 0; i < length; ) {
8027
0
        int clen = length - i;
8028
0
        int c = xmlGetUTF8Char(content + i, &clen);
8029
8030
0
        if ((c < 0) || (!IS_CHAR(c))) {
8031
0
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
8032
0
                              "xmlLoadEntityContent: invalid char value %d\n",
8033
0
                              content[i]);
8034
0
            goto error;
8035
0
        }
8036
0
        i += clen;
8037
0
    }
8038
8039
0
    xmlSaturatedAdd(&ctxt->sizeentities, length);
8040
0
    entity->content = content;
8041
0
    entity->length = length;
8042
0
    content = NULL;
8043
0
    ret = 0;
8044
8045
0
error:
8046
0
    while (ctxt->inputNr > 0)
8047
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
8048
0
    xmlFree(ctxt->inputTab);
8049
0
    xmlFree((xmlChar *) ctxt->encoding);
8050
8051
0
    ctxt->input = oldinput;
8052
0
    ctxt->inputNr = oldinputNr;
8053
0
    ctxt->inputMax = oldinputMax;
8054
0
    ctxt->inputTab = oldinputTab;
8055
0
    ctxt->encoding = oldencoding;
8056
8057
0
    xmlFree(content);
8058
8059
0
    return(ret);
8060
0
}
8061
8062
/**
8063
 * xmlParseStringPEReference:
8064
 * @ctxt:  an XML parser context
8065
 * @str:  a pointer to an index in the string
8066
 *
8067
 * parse PEReference declarations
8068
 *
8069
 * [69] PEReference ::= '%' Name ';'
8070
 *
8071
 * [ WFC: No Recursion ]
8072
 * A parsed entity must not contain a recursive
8073
 * reference to itself, either directly or indirectly.
8074
 *
8075
 * [ WFC: Entity Declared ]
8076
 * In a document without any DTD, a document with only an internal DTD
8077
 * subset which contains no parameter entity references, or a document
8078
 * with "standalone='yes'", ...  ... The declaration of a parameter
8079
 * entity must precede any reference to it...
8080
 *
8081
 * [ VC: Entity Declared ]
8082
 * In a document with an external subset or external parameter entities
8083
 * with "standalone='no'", ...  ... The declaration of a parameter entity
8084
 * must precede any reference to it...
8085
 *
8086
 * [ WFC: In DTD ]
8087
 * Parameter-entity references may only appear in the DTD.
8088
 * NOTE: misleading but this is handled.
8089
 *
8090
 * Returns the string of the entity content.
8091
 *         str is updated to the current value of the index
8092
 */
8093
static xmlEntityPtr
8094
2.91k
xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) {
8095
2.91k
    const xmlChar *ptr;
8096
2.91k
    xmlChar cur;
8097
2.91k
    xmlChar *name;
8098
2.91k
    xmlEntityPtr entity = NULL;
8099
8100
2.91k
    if ((str == NULL) || (*str == NULL)) return(NULL);
8101
2.91k
    ptr = *str;
8102
2.91k
    cur = *ptr;
8103
2.91k
    if (cur != '%')
8104
0
        return(NULL);
8105
2.91k
    ptr++;
8106
2.91k
    name = xmlParseStringName(ctxt, &ptr);
8107
2.91k
    if (name == NULL) {
8108
1.12k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8109
1.12k
           "xmlParseStringPEReference: no name\n");
8110
1.12k
  *str = ptr;
8111
1.12k
  return(NULL);
8112
1.12k
    }
8113
1.78k
    cur = *ptr;
8114
1.78k
    if (cur != ';') {
8115
788
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
8116
788
  xmlFree(name);
8117
788
  *str = ptr;
8118
788
  return(NULL);
8119
788
    }
8120
1.00k
    ptr++;
8121
8122
    /* Must be set before xmlHandleUndeclaredEntity */
8123
1.00k
    ctxt->hasPErefs = 1;
8124
8125
    /*
8126
     * Request the entity from SAX
8127
     */
8128
1.00k
    if ((ctxt->sax != NULL) &&
8129
1.00k
  (ctxt->sax->getParameterEntity != NULL))
8130
0
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
8131
8132
1.00k
    if (entity == NULL) {
8133
1.00k
        xmlHandleUndeclaredEntity(ctxt, name);
8134
1.00k
    } else {
8135
  /*
8136
   * Internal checking in case the entity quest barfed
8137
   */
8138
0
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
8139
0
      (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) {
8140
0
      xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
8141
0
        "%%%s; is not a parameter entity\n",
8142
0
        name, NULL);
8143
0
  }
8144
0
    }
8145
8146
1.00k
    xmlFree(name);
8147
1.00k
    *str = ptr;
8148
1.00k
    return(entity);
8149
1.78k
}
8150
8151
/**
8152
 * xmlParseDocTypeDecl:
8153
 * @ctxt:  an XML parser context
8154
 *
8155
 * DEPRECATED: Internal function, don't use.
8156
 *
8157
 * parse a DOCTYPE declaration
8158
 *
8159
 * [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S?
8160
 *                      ('[' (markupdecl | PEReference | S)* ']' S?)? '>'
8161
 *
8162
 * [ VC: Root Element Type ]
8163
 * The Name in the document type declaration must match the element
8164
 * type of the root element.
8165
 */
8166
8167
void
8168
7.93k
xmlParseDocTypeDecl(xmlParserCtxtPtr ctxt) {
8169
7.93k
    const xmlChar *name = NULL;
8170
7.93k
    xmlChar *ExternalID = NULL;
8171
7.93k
    xmlChar *URI = NULL;
8172
8173
    /*
8174
     * We know that '<!DOCTYPE' has been detected.
8175
     */
8176
7.93k
    SKIP(9);
8177
8178
7.93k
    if (SKIP_BLANKS == 0) {
8179
256
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
8180
256
                       "Space required after 'DOCTYPE'\n");
8181
256
    }
8182
8183
    /*
8184
     * Parse the DOCTYPE name.
8185
     */
8186
7.93k
    name = xmlParseName(ctxt);
8187
7.93k
    if (name == NULL) {
8188
13
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8189
13
           "xmlParseDocTypeDecl : no DOCTYPE name !\n");
8190
13
    }
8191
7.93k
    ctxt->intSubName = name;
8192
8193
7.93k
    SKIP_BLANKS;
8194
8195
    /*
8196
     * Check for SystemID and ExternalID
8197
     */
8198
7.93k
    URI = xmlParseExternalID(ctxt, &ExternalID, 1);
8199
8200
7.93k
    if ((URI != NULL) || (ExternalID != NULL)) {
8201
664
        ctxt->hasExternalSubset = 1;
8202
664
    }
8203
7.93k
    ctxt->extSubURI = URI;
8204
7.93k
    ctxt->extSubSystem = ExternalID;
8205
8206
7.93k
    SKIP_BLANKS;
8207
8208
    /*
8209
     * Create and update the internal subset.
8210
     */
8211
7.93k
    if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) &&
8212
7.93k
  (!ctxt->disableSAX))
8213
0
  ctxt->sax->internalSubset(ctxt->userData, name, ExternalID, URI);
8214
8215
7.93k
    if ((RAW != '[') && (RAW != '>')) {
8216
292
  xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
8217
292
    }
8218
7.93k
}
8219
8220
/**
8221
 * xmlParseInternalSubset:
8222
 * @ctxt:  an XML parser context
8223
 *
8224
 * parse the internal subset declaration
8225
 *
8226
 * [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>'
8227
 */
8228
8229
static void
8230
7.02k
xmlParseInternalSubset(xmlParserCtxtPtr ctxt) {
8231
    /*
8232
     * Is there any DTD definition ?
8233
     */
8234
7.02k
    if (RAW == '[') {
8235
7.02k
        int oldInputNr = ctxt->inputNr;
8236
8237
7.02k
        NEXT;
8238
  /*
8239
   * Parse the succession of Markup declarations and
8240
   * PEReferences.
8241
   * Subsequence (markupdecl | PEReference | S)*
8242
   */
8243
7.02k
  SKIP_BLANKS;
8244
114k
  while (((RAW != ']') || (ctxt->inputNr > oldInputNr)) &&
8245
114k
               (PARSER_STOPPED(ctxt) == 0)) {
8246
8247
            /*
8248
             * Conditional sections are allowed from external entities included
8249
             * by PE References in the internal subset.
8250
             */
8251
110k
            if ((PARSER_EXTERNAL(ctxt)) &&
8252
110k
                (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
8253
0
                xmlParseConditionalSections(ctxt);
8254
110k
            } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
8255
96.2k
          xmlParseMarkupDecl(ctxt);
8256
96.2k
            } else if (RAW == '%') {
8257
11.0k
          xmlParsePEReference(ctxt);
8258
11.0k
            } else {
8259
3.54k
    xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8260
3.54k
                break;
8261
3.54k
            }
8262
107k
      SKIP_BLANKS_PE;
8263
107k
            SHRINK;
8264
107k
            GROW;
8265
107k
  }
8266
8267
7.02k
        while (ctxt->inputNr > oldInputNr)
8268
0
            xmlPopPE(ctxt);
8269
8270
7.02k
  if (RAW == ']') {
8271
2.38k
      NEXT;
8272
2.38k
      SKIP_BLANKS;
8273
2.38k
  }
8274
7.02k
    }
8275
8276
    /*
8277
     * We should be at the end of the DOCTYPE declaration.
8278
     */
8279
7.02k
    if ((ctxt->wellFormed) && (RAW != '>')) {
8280
26
  xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
8281
26
  return;
8282
26
    }
8283
7.00k
    NEXT;
8284
7.00k
}
8285
8286
#ifdef LIBXML_SAX1_ENABLED
8287
/**
8288
 * xmlParseAttribute:
8289
 * @ctxt:  an XML parser context
8290
 * @value:  a xmlChar ** used to store the value of the attribute
8291
 *
8292
 * DEPRECATED: Internal function, don't use.
8293
 *
8294
 * parse an attribute
8295
 *
8296
 * [41] Attribute ::= Name Eq AttValue
8297
 *
8298
 * [ WFC: No External Entity References ]
8299
 * Attribute values cannot contain direct or indirect entity references
8300
 * to external entities.
8301
 *
8302
 * [ WFC: No < in Attribute Values ]
8303
 * The replacement text of any entity referred to directly or indirectly in
8304
 * an attribute value (other than "&lt;") must not contain a <.
8305
 *
8306
 * [ VC: Attribute Value Type ]
8307
 * The attribute must have been declared; the value must be of the type
8308
 * declared for it.
8309
 *
8310
 * [25] Eq ::= S? '=' S?
8311
 *
8312
 * With namespace:
8313
 *
8314
 * [NS 11] Attribute ::= QName Eq AttValue
8315
 *
8316
 * Also the case QName == xmlns:??? is handled independently as a namespace
8317
 * definition.
8318
 *
8319
 * Returns the attribute name, and the value in *value.
8320
 */
8321
8322
const xmlChar *
8323
0
xmlParseAttribute(xmlParserCtxtPtr ctxt, xmlChar **value) {
8324
0
    const xmlChar *name;
8325
0
    xmlChar *val;
8326
8327
0
    *value = NULL;
8328
0
    GROW;
8329
0
    name = xmlParseName(ctxt);
8330
0
    if (name == NULL) {
8331
0
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8332
0
                 "error parsing attribute name\n");
8333
0
        return(NULL);
8334
0
    }
8335
8336
    /*
8337
     * read the value
8338
     */
8339
0
    SKIP_BLANKS;
8340
0
    if (RAW == '=') {
8341
0
        NEXT;
8342
0
  SKIP_BLANKS;
8343
0
  val = xmlParseAttValue(ctxt);
8344
0
    } else {
8345
0
  xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8346
0
         "Specification mandates value for attribute %s\n", name);
8347
0
  return(name);
8348
0
    }
8349
8350
    /*
8351
     * Check that xml:lang conforms to the specification
8352
     * No more registered as an error, just generate a warning now
8353
     * since this was deprecated in XML second edition
8354
     */
8355
0
    if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) {
8356
0
  if (!xmlCheckLanguageID(val)) {
8357
0
      xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8358
0
              "Malformed value for xml:lang : %s\n",
8359
0
        val, NULL);
8360
0
  }
8361
0
    }
8362
8363
    /*
8364
     * Check that xml:space conforms to the specification
8365
     */
8366
0
    if (xmlStrEqual(name, BAD_CAST "xml:space")) {
8367
0
  if (xmlStrEqual(val, BAD_CAST "default"))
8368
0
      *(ctxt->space) = 0;
8369
0
  else if (xmlStrEqual(val, BAD_CAST "preserve"))
8370
0
      *(ctxt->space) = 1;
8371
0
  else {
8372
0
    xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8373
0
"Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8374
0
                                 val, NULL);
8375
0
  }
8376
0
    }
8377
8378
0
    *value = val;
8379
0
    return(name);
8380
0
}
8381
8382
/**
8383
 * xmlParseStartTag:
8384
 * @ctxt:  an XML parser context
8385
 *
8386
 * DEPRECATED: Internal function, don't use.
8387
 *
8388
 * Parse a start tag. Always consumes '<'.
8389
 *
8390
 * [40] STag ::= '<' Name (S Attribute)* S? '>'
8391
 *
8392
 * [ WFC: Unique Att Spec ]
8393
 * No attribute name may appear more than once in the same start-tag or
8394
 * empty-element tag.
8395
 *
8396
 * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
8397
 *
8398
 * [ WFC: Unique Att Spec ]
8399
 * No attribute name may appear more than once in the same start-tag or
8400
 * empty-element tag.
8401
 *
8402
 * With namespace:
8403
 *
8404
 * [NS 8] STag ::= '<' QName (S Attribute)* S? '>'
8405
 *
8406
 * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>'
8407
 *
8408
 * Returns the element name parsed
8409
 */
8410
8411
const xmlChar *
8412
0
xmlParseStartTag(xmlParserCtxtPtr ctxt) {
8413
0
    const xmlChar *name;
8414
0
    const xmlChar *attname;
8415
0
    xmlChar *attvalue;
8416
0
    const xmlChar **atts = ctxt->atts;
8417
0
    int nbatts = 0;
8418
0
    int maxatts = ctxt->maxatts;
8419
0
    int i;
8420
8421
0
    if (RAW != '<') return(NULL);
8422
0
    NEXT1;
8423
8424
0
    name = xmlParseName(ctxt);
8425
0
    if (name == NULL) {
8426
0
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8427
0
       "xmlParseStartTag: invalid element name\n");
8428
0
        return(NULL);
8429
0
    }
8430
8431
    /*
8432
     * Now parse the attributes, it ends up with the ending
8433
     *
8434
     * (S Attribute)* S?
8435
     */
8436
0
    SKIP_BLANKS;
8437
0
    GROW;
8438
8439
0
    while (((RAW != '>') &&
8440
0
     ((RAW != '/') || (NXT(1) != '>')) &&
8441
0
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8442
0
  attname = xmlParseAttribute(ctxt, &attvalue);
8443
0
        if (attname == NULL)
8444
0
      break;
8445
0
        if (attvalue != NULL) {
8446
      /*
8447
       * [ WFC: Unique Att Spec ]
8448
       * No attribute name may appear more than once in the same
8449
       * start-tag or empty-element tag.
8450
       */
8451
0
      for (i = 0; i < nbatts;i += 2) {
8452
0
          if (xmlStrEqual(atts[i], attname)) {
8453
0
        xmlErrAttributeDup(ctxt, NULL, attname);
8454
0
        goto failed;
8455
0
    }
8456
0
      }
8457
      /*
8458
       * Add the pair to atts
8459
       */
8460
0
      if (nbatts + 4 > maxatts) {
8461
0
          const xmlChar **n;
8462
0
                int newSize;
8463
8464
0
                newSize = xmlGrowCapacity(maxatts, sizeof(n[0]) * 2,
8465
0
                                          11, XML_MAX_ATTRS);
8466
0
                if (newSize < 0) {
8467
0
        xmlErrMemory(ctxt);
8468
0
        goto failed;
8469
0
    }
8470
0
#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
8471
0
                if (newSize < 2)
8472
0
                    newSize = 2;
8473
0
#endif
8474
0
          n = xmlRealloc(atts, newSize * sizeof(n[0]) * 2);
8475
0
    if (n == NULL) {
8476
0
        xmlErrMemory(ctxt);
8477
0
        goto failed;
8478
0
    }
8479
0
    atts = n;
8480
0
                maxatts = newSize * 2;
8481
0
    ctxt->atts = atts;
8482
0
    ctxt->maxatts = maxatts;
8483
0
      }
8484
8485
0
      atts[nbatts++] = attname;
8486
0
      atts[nbatts++] = attvalue;
8487
0
      atts[nbatts] = NULL;
8488
0
      atts[nbatts + 1] = NULL;
8489
8490
0
            attvalue = NULL;
8491
0
  }
8492
8493
0
failed:
8494
8495
0
        if (attvalue != NULL)
8496
0
            xmlFree(attvalue);
8497
8498
0
  GROW
8499
0
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
8500
0
      break;
8501
0
  if (SKIP_BLANKS == 0) {
8502
0
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
8503
0
         "attributes construct error\n");
8504
0
  }
8505
0
  SHRINK;
8506
0
        GROW;
8507
0
    }
8508
8509
    /*
8510
     * SAX: Start of Element !
8511
     */
8512
0
    if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) &&
8513
0
  (!ctxt->disableSAX)) {
8514
0
  if (nbatts > 0)
8515
0
      ctxt->sax->startElement(ctxt->userData, name, atts);
8516
0
  else
8517
0
      ctxt->sax->startElement(ctxt->userData, name, NULL);
8518
0
    }
8519
8520
0
    if (atts != NULL) {
8521
        /* Free only the content strings */
8522
0
        for (i = 1;i < nbatts;i+=2)
8523
0
      if (atts[i] != NULL)
8524
0
         xmlFree((xmlChar *) atts[i]);
8525
0
    }
8526
0
    return(name);
8527
0
}
8528
8529
/**
8530
 * xmlParseEndTag1:
8531
 * @ctxt:  an XML parser context
8532
 * @line:  line of the start tag
8533
 * @nsNr:  number of namespaces on the start tag
8534
 *
8535
 * Parse an end tag. Always consumes '</'.
8536
 *
8537
 * [42] ETag ::= '</' Name S? '>'
8538
 *
8539
 * With namespace
8540
 *
8541
 * [NS 9] ETag ::= '</' QName S? '>'
8542
 */
8543
8544
static void
8545
0
xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) {
8546
0
    const xmlChar *name;
8547
8548
0
    GROW;
8549
0
    if ((RAW != '<') || (NXT(1) != '/')) {
8550
0
  xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED,
8551
0
           "xmlParseEndTag: '</' not found\n");
8552
0
  return;
8553
0
    }
8554
0
    SKIP(2);
8555
8556
0
    name = xmlParseNameAndCompare(ctxt,ctxt->name);
8557
8558
    /*
8559
     * We should definitely be at the ending "S? '>'" part
8560
     */
8561
0
    GROW;
8562
0
    SKIP_BLANKS;
8563
0
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
8564
0
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
8565
0
    } else
8566
0
  NEXT1;
8567
8568
    /*
8569
     * [ WFC: Element Type Match ]
8570
     * The Name in an element's end-tag must match the element type in the
8571
     * start-tag.
8572
     *
8573
     */
8574
0
    if (name != (xmlChar*)1) {
8575
0
        if (name == NULL) name = BAD_CAST "unparsable";
8576
0
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
8577
0
         "Opening and ending tag mismatch: %s line %d and %s\n",
8578
0
                    ctxt->name, line, name);
8579
0
    }
8580
8581
    /*
8582
     * SAX: End of Tag
8583
     */
8584
0
    if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
8585
0
  (!ctxt->disableSAX))
8586
0
        ctxt->sax->endElement(ctxt->userData, ctxt->name);
8587
8588
0
    namePop(ctxt);
8589
0
    spacePop(ctxt);
8590
0
}
8591
8592
/**
8593
 * xmlParseEndTag:
8594
 * @ctxt:  an XML parser context
8595
 *
8596
 * DEPRECATED: Internal function, don't use.
8597
 *
8598
 * parse an end of tag
8599
 *
8600
 * [42] ETag ::= '</' Name S? '>'
8601
 *
8602
 * With namespace
8603
 *
8604
 * [NS 9] ETag ::= '</' QName S? '>'
8605
 */
8606
8607
void
8608
0
xmlParseEndTag(xmlParserCtxtPtr ctxt) {
8609
0
    xmlParseEndTag1(ctxt, 0);
8610
0
}
8611
#endif /* LIBXML_SAX1_ENABLED */
8612
8613
/************************************************************************
8614
 *                  *
8615
 *          SAX 2 specific operations       *
8616
 *                  *
8617
 ************************************************************************/
8618
8619
/**
8620
 * xmlParseQNameHashed:
8621
 * @ctxt:  an XML parser context
8622
 * @prefix:  pointer to store the prefix part
8623
 *
8624
 * parse an XML Namespace QName
8625
 *
8626
 * [6]  QName  ::= (Prefix ':')? LocalPart
8627
 * [7]  Prefix  ::= NCName
8628
 * [8]  LocalPart  ::= NCName
8629
 *
8630
 * Returns the Name parsed or NULL
8631
 */
8632
8633
static xmlHashedString
8634
64.1M
xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) {
8635
64.1M
    xmlHashedString l, p;
8636
64.1M
    int start, isNCName = 0;
8637
8638
64.1M
    l.name = NULL;
8639
64.1M
    p.name = NULL;
8640
8641
64.1M
    GROW;
8642
64.1M
    start = CUR_PTR - BASE_PTR;
8643
8644
64.1M
    l = xmlParseNCName(ctxt);
8645
64.1M
    if (l.name != NULL) {
8646
63.8M
        isNCName = 1;
8647
63.8M
        if (CUR == ':') {
8648
26.7M
            NEXT;
8649
26.7M
            p = l;
8650
26.7M
            l = xmlParseNCName(ctxt);
8651
26.7M
        }
8652
63.8M
    }
8653
64.1M
    if ((l.name == NULL) || (CUR == ':')) {
8654
567k
        xmlChar *tmp;
8655
8656
567k
        l.name = NULL;
8657
567k
        p.name = NULL;
8658
567k
        if ((isNCName == 0) && (CUR != ':'))
8659
57.0k
            return(l);
8660
510k
        tmp = xmlParseNmtoken(ctxt);
8661
510k
        if (tmp != NULL)
8662
430k
            xmlFree(tmp);
8663
510k
        l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start,
8664
510k
                                CUR_PTR - (BASE_PTR + start));
8665
510k
        if (l.name == NULL) {
8666
0
            xmlErrMemory(ctxt);
8667
0
            return(l);
8668
0
        }
8669
510k
        xmlNsErr(ctxt, XML_NS_ERR_QNAME,
8670
510k
                 "Failed to parse QName '%s'\n", l.name, NULL, NULL);
8671
510k
    }
8672
8673
64.0M
    *prefix = p;
8674
64.0M
    return(l);
8675
64.1M
}
8676
8677
/**
8678
 * xmlParseQName:
8679
 * @ctxt:  an XML parser context
8680
 * @prefix:  pointer to store the prefix part
8681
 *
8682
 * parse an XML Namespace QName
8683
 *
8684
 * [6]  QName  ::= (Prefix ':')? LocalPart
8685
 * [7]  Prefix  ::= NCName
8686
 * [8]  LocalPart  ::= NCName
8687
 *
8688
 * Returns the Name parsed or NULL
8689
 */
8690
8691
static const xmlChar *
8692
13.8k
xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) {
8693
13.8k
    xmlHashedString n, p;
8694
8695
13.8k
    n = xmlParseQNameHashed(ctxt, &p);
8696
13.8k
    if (n.name == NULL)
8697
425
        return(NULL);
8698
13.4k
    *prefix = p.name;
8699
13.4k
    return(n.name);
8700
13.8k
}
8701
8702
/**
8703
 * xmlParseQNameAndCompare:
8704
 * @ctxt:  an XML parser context
8705
 * @name:  the localname
8706
 * @prefix:  the prefix, if any.
8707
 *
8708
 * parse an XML name and compares for match
8709
 * (specialized for endtag parsing)
8710
 *
8711
 * Returns NULL for an illegal name, (xmlChar*) 1 for success
8712
 * and the name for mismatch
8713
 */
8714
8715
static const xmlChar *
8716
xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name,
8717
7.09M
                        xmlChar const *prefix) {
8718
7.09M
    const xmlChar *cmp;
8719
7.09M
    const xmlChar *in;
8720
7.09M
    const xmlChar *ret;
8721
7.09M
    const xmlChar *prefix2;
8722
8723
7.09M
    if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name));
8724
8725
7.09M
    GROW;
8726
7.09M
    in = ctxt->input->cur;
8727
8728
7.09M
    cmp = prefix;
8729
24.0M
    while (*in != 0 && *in == *cmp) {
8730
16.9M
  ++in;
8731
16.9M
  ++cmp;
8732
16.9M
    }
8733
7.09M
    if ((*cmp == 0) && (*in == ':')) {
8734
7.09M
        in++;
8735
7.09M
  cmp = name;
8736
51.8M
  while (*in != 0 && *in == *cmp) {
8737
44.7M
      ++in;
8738
44.7M
      ++cmp;
8739
44.7M
  }
8740
7.09M
  if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
8741
      /* success */
8742
7.08M
            ctxt->input->col += in - ctxt->input->cur;
8743
7.08M
      ctxt->input->cur = in;
8744
7.08M
      return((const xmlChar*) 1);
8745
7.08M
  }
8746
7.09M
    }
8747
    /*
8748
     * all strings coms from the dictionary, equality can be done directly
8749
     */
8750
13.8k
    ret = xmlParseQName (ctxt, &prefix2);
8751
13.8k
    if (ret == NULL)
8752
425
        return(NULL);
8753
13.4k
    if ((ret == name) && (prefix == prefix2))
8754
282
  return((const xmlChar*) 1);
8755
13.1k
    return ret;
8756
13.4k
}
8757
8758
/**
8759
 * xmlParseAttribute2:
8760
 * @ctxt:  an XML parser context
8761
 * @pref:  the element prefix
8762
 * @elem:  the element name
8763
 * @prefix:  a xmlChar ** used to store the value of the attribute prefix
8764
 * @value:  a xmlChar ** used to store the value of the attribute
8765
 * @len:  an int * to save the length of the attribute
8766
 * @alloc:  an int * to indicate if the attribute was allocated
8767
 *
8768
 * parse an attribute in the new SAX2 framework.
8769
 *
8770
 * Returns the attribute name, and the value in *value, .
8771
 */
8772
8773
static xmlHashedString
8774
xmlParseAttribute2(xmlParserCtxtPtr ctxt,
8775
                   const xmlChar * pref, const xmlChar * elem,
8776
                   xmlHashedString * hprefix, xmlChar ** value,
8777
                   int *len, int *alloc)
8778
28.1M
{
8779
28.1M
    xmlHashedString hname;
8780
28.1M
    const xmlChar *prefix, *name;
8781
28.1M
    xmlChar *val = NULL, *internal_val = NULL;
8782
28.1M
    int normalize = 0;
8783
28.1M
    int isNamespace;
8784
8785
28.1M
    *value = NULL;
8786
28.1M
    GROW;
8787
28.1M
    hname = xmlParseQNameHashed(ctxt, hprefix);
8788
28.1M
    if (hname.name == NULL) {
8789
47.3k
        xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8790
47.3k
                       "error parsing attribute name\n");
8791
47.3k
        return(hname);
8792
47.3k
    }
8793
28.1M
    name = hname.name;
8794
28.1M
    prefix = hprefix->name;
8795
8796
    /*
8797
     * get the type if needed
8798
     */
8799
28.1M
    if (ctxt->attsSpecial != NULL) {
8800
21.1k
        int type;
8801
8802
21.1k
        type = XML_PTR_TO_INT(xmlHashQLookup2(ctxt->attsSpecial, pref, elem,
8803
21.1k
                                              prefix, name));
8804
21.1k
        if (type != 0)
8805
9.32k
            normalize = 1;
8806
21.1k
    }
8807
8808
    /*
8809
     * read the value
8810
     */
8811
28.1M
    SKIP_BLANKS;
8812
28.1M
    if (RAW == '=') {
8813
28.1M
        NEXT;
8814
28.1M
        SKIP_BLANKS;
8815
28.1M
        isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) ||
8816
28.1M
                       (prefix == ctxt->str_xmlns));
8817
28.1M
        val = xmlParseAttValueInternal(ctxt, len, alloc, normalize,
8818
28.1M
                                       isNamespace);
8819
28.1M
        if (val == NULL)
8820
9.09k
            goto error;
8821
28.1M
    } else {
8822
18.0k
        xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8823
18.0k
                          "Specification mandates value for attribute %s\n",
8824
18.0k
                          name);
8825
18.0k
        goto error;
8826
18.0k
    }
8827
8828
28.0M
    if (prefix == ctxt->str_xml) {
8829
        /*
8830
         * Check that xml:lang conforms to the specification
8831
         * No more registered as an error, just generate a warning now
8832
         * since this was deprecated in XML second edition
8833
         */
8834
111k
        if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) {
8835
0
            internal_val = xmlStrndup(val, *len);
8836
0
            if (internal_val == NULL)
8837
0
                goto mem_error;
8838
0
            if (!xmlCheckLanguageID(internal_val)) {
8839
0
                xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8840
0
                              "Malformed value for xml:lang : %s\n",
8841
0
                              internal_val, NULL);
8842
0
            }
8843
0
        }
8844
8845
        /*
8846
         * Check that xml:space conforms to the specification
8847
         */
8848
111k
        if (xmlStrEqual(name, BAD_CAST "space")) {
8849
89.8k
            internal_val = xmlStrndup(val, *len);
8850
89.8k
            if (internal_val == NULL)
8851
0
                goto mem_error;
8852
89.8k
            if (xmlStrEqual(internal_val, BAD_CAST "default"))
8853
37
                *(ctxt->space) = 0;
8854
89.8k
            else if (xmlStrEqual(internal_val, BAD_CAST "preserve"))
8855
88.7k
                *(ctxt->space) = 1;
8856
1.12k
            else {
8857
1.12k
                xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8858
1.12k
                              "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8859
1.12k
                              internal_val, NULL);
8860
1.12k
            }
8861
89.8k
        }
8862
111k
        if (internal_val) {
8863
89.8k
            xmlFree(internal_val);
8864
89.8k
        }
8865
111k
    }
8866
8867
28.0M
    *value = val;
8868
28.0M
    return (hname);
8869
8870
0
mem_error:
8871
0
    xmlErrMemory(ctxt);
8872
26.3k
error:
8873
26.3k
    if ((val != NULL) && (*alloc != 0))
8874
0
        xmlFree(val);
8875
26.3k
    return(hname);
8876
0
}
8877
8878
/**
8879
 * xmlAttrHashInsert:
8880
 * @ctxt: parser context
8881
 * @size: size of the hash table
8882
 * @name: attribute name
8883
 * @uri: namespace uri
8884
 * @hashValue: combined hash value of name and uri
8885
 * @aindex: attribute index (this is a multiple of 5)
8886
 *
8887
 * Inserts a new attribute into the hash table.
8888
 *
8889
 * Returns INT_MAX if no existing attribute was found, the attribute
8890
 * index if an attribute was found, -1 if a memory allocation failed.
8891
 */
8892
static int
8893
xmlAttrHashInsert(xmlParserCtxtPtr ctxt, unsigned size, const xmlChar *name,
8894
20.8M
                  const xmlChar *uri, unsigned hashValue, int aindex) {
8895
20.8M
    xmlAttrHashBucket *table = ctxt->attrHash;
8896
20.8M
    xmlAttrHashBucket *bucket;
8897
20.8M
    unsigned hindex;
8898
8899
20.8M
    hindex = hashValue & (size - 1);
8900
20.8M
    bucket = &table[hindex];
8901
8902
24.5M
    while (bucket->index >= 0) {
8903
3.76M
        const xmlChar **atts = &ctxt->atts[bucket->index];
8904
8905
3.76M
        if (name == atts[0]) {
8906
152k
            int nsIndex = XML_PTR_TO_INT(atts[2]);
8907
8908
152k
            if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) :
8909
152k
                (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) :
8910
2.81k
                (uri == ctxt->nsTab[nsIndex * 2 + 1]))
8911
29.5k
                return(bucket->index);
8912
152k
        }
8913
8914
3.73M
        hindex++;
8915
3.73M
        bucket++;
8916
3.73M
        if (hindex >= size) {
8917
1.16M
            hindex = 0;
8918
1.16M
            bucket = table;
8919
1.16M
        }
8920
3.73M
    }
8921
8922
20.8M
    bucket->index = aindex;
8923
8924
20.8M
    return(INT_MAX);
8925
20.8M
}
8926
8927
static int
8928
xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size,
8929
                       const xmlChar *name, const xmlChar *prefix,
8930
110
                       unsigned hashValue, int aindex) {
8931
110
    xmlAttrHashBucket *table = ctxt->attrHash;
8932
110
    xmlAttrHashBucket *bucket;
8933
110
    unsigned hindex;
8934
8935
110
    hindex = hashValue & (size - 1);
8936
110
    bucket = &table[hindex];
8937
8938
148
    while (bucket->index >= 0) {
8939
94
        const xmlChar **atts = &ctxt->atts[bucket->index];
8940
8941
94
        if ((name == atts[0]) && (prefix == atts[1]))
8942
56
            return(bucket->index);
8943
8944
38
        hindex++;
8945
38
        bucket++;
8946
38
        if (hindex >= size) {
8947
0
            hindex = 0;
8948
0
            bucket = table;
8949
0
        }
8950
38
    }
8951
8952
54
    bucket->index = aindex;
8953
8954
54
    return(INT_MAX);
8955
110
}
8956
/**
8957
 * xmlParseStartTag2:
8958
 * @ctxt:  an XML parser context
8959
 *
8960
 * Parse a start tag. Always consumes '<'.
8961
 *
8962
 * This routine is called when running SAX2 parsing
8963
 *
8964
 * [40] STag ::= '<' Name (S Attribute)* S? '>'
8965
 *
8966
 * [ WFC: Unique Att Spec ]
8967
 * No attribute name may appear more than once in the same start-tag or
8968
 * empty-element tag.
8969
 *
8970
 * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
8971
 *
8972
 * [ WFC: Unique Att Spec ]
8973
 * No attribute name may appear more than once in the same start-tag or
8974
 * empty-element tag.
8975
 *
8976
 * With namespace:
8977
 *
8978
 * [NS 8] STag ::= '<' QName (S Attribute)* S? '>'
8979
 *
8980
 * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>'
8981
 *
8982
 * Returns the element name parsed
8983
 */
8984
8985
static const xmlChar *
8986
xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref,
8987
35.9M
                  const xmlChar **URI, int *nbNsPtr) {
8988
35.9M
    xmlHashedString hlocalname;
8989
35.9M
    xmlHashedString hprefix;
8990
35.9M
    xmlHashedString hattname;
8991
35.9M
    xmlHashedString haprefix;
8992
35.9M
    const xmlChar *localname;
8993
35.9M
    const xmlChar *prefix;
8994
35.9M
    const xmlChar *attname;
8995
35.9M
    const xmlChar *aprefix;
8996
35.9M
    const xmlChar *uri;
8997
35.9M
    xmlChar *attvalue = NULL;
8998
35.9M
    const xmlChar **atts = ctxt->atts;
8999
35.9M
    unsigned attrHashSize = 0;
9000
35.9M
    int maxatts = ctxt->maxatts;
9001
35.9M
    int nratts, nbatts, nbdef;
9002
35.9M
    int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts;
9003
35.9M
    int alloc = 0;
9004
35.9M
    int numNsErr = 0;
9005
35.9M
    int numDupErr = 0;
9006
9007
35.9M
    if (RAW != '<') return(NULL);
9008
35.9M
    NEXT1;
9009
9010
35.9M
    nbatts = 0;
9011
35.9M
    nratts = 0;
9012
35.9M
    nbdef = 0;
9013
35.9M
    nbNs = 0;
9014
35.9M
    nbTotalDef = 0;
9015
35.9M
    attval = 0;
9016
9017
35.9M
    if (xmlParserNsStartElement(ctxt->nsdb) < 0) {
9018
0
        xmlErrMemory(ctxt);
9019
0
        return(NULL);
9020
0
    }
9021
9022
35.9M
    hlocalname = xmlParseQNameHashed(ctxt, &hprefix);
9023
35.9M
    if (hlocalname.name == NULL) {
9024
9.21k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
9025
9.21k
           "StartTag: invalid element name\n");
9026
9.21k
        return(NULL);
9027
9.21k
    }
9028
35.9M
    localname = hlocalname.name;
9029
35.9M
    prefix = hprefix.name;
9030
9031
    /*
9032
     * Now parse the attributes, it ends up with the ending
9033
     *
9034
     * (S Attribute)* S?
9035
     */
9036
35.9M
    SKIP_BLANKS;
9037
35.9M
    GROW;
9038
9039
    /*
9040
     * The ctxt->atts array will be ultimately passed to the SAX callback
9041
     * containing five xmlChar pointers for each attribute:
9042
     *
9043
     * [0] attribute name
9044
     * [1] attribute prefix
9045
     * [2] namespace URI
9046
     * [3] attribute value
9047
     * [4] end of attribute value
9048
     *
9049
     * To save memory, we reuse this array temporarily and store integers
9050
     * in these pointer variables.
9051
     *
9052
     * [0] attribute name
9053
     * [1] attribute prefix
9054
     * [2] hash value of attribute prefix, and later namespace index
9055
     * [3] for non-allocated values: ptrdiff_t offset into input buffer
9056
     * [4] for non-allocated values: ptrdiff_t offset into input buffer
9057
     *
9058
     * The ctxt->attallocs array contains an additional unsigned int for
9059
     * each attribute, containing the hash value of the attribute name
9060
     * and the alloc flag in bit 31.
9061
     */
9062
9063
51.4M
    while (((RAW != '>') &&
9064
51.4M
     ((RAW != '/') || (NXT(1) != '>')) &&
9065
51.4M
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
9066
28.1M
  int len = -1;
9067
9068
28.1M
  hattname = xmlParseAttribute2(ctxt, prefix, localname,
9069
28.1M
                                          &haprefix, &attvalue, &len,
9070
28.1M
                                          &alloc);
9071
28.1M
        if (hattname.name == NULL)
9072
47.3k
      break;
9073
28.1M
        if (attvalue == NULL)
9074
26.3k
            goto next_attr;
9075
28.0M
        attname = hattname.name;
9076
28.0M
        aprefix = haprefix.name;
9077
28.0M
  if (len < 0) len = xmlStrlen(attvalue);
9078
9079
28.0M
        if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
9080
321k
            xmlHashedString huri;
9081
321k
            xmlURIPtr parsedUri;
9082
9083
321k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
9084
321k
            uri = huri.name;
9085
321k
            if (uri == NULL) {
9086
0
                xmlErrMemory(ctxt);
9087
0
                goto next_attr;
9088
0
            }
9089
321k
            if (*uri != 0) {
9090
317k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
9091
0
                    xmlErrMemory(ctxt);
9092
0
                    goto next_attr;
9093
0
                }
9094
317k
                if (parsedUri == NULL) {
9095
89.3k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
9096
89.3k
                             "xmlns: '%s' is not a valid URI\n",
9097
89.3k
                                       uri, NULL, NULL);
9098
228k
                } else {
9099
228k
                    if (parsedUri->scheme == NULL) {
9100
71.3k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
9101
71.3k
                                  "xmlns: URI %s is not absolute\n",
9102
71.3k
                                  uri, NULL, NULL);
9103
71.3k
                    }
9104
228k
                    xmlFreeURI(parsedUri);
9105
228k
                }
9106
317k
                if (uri == ctxt->str_xml_ns) {
9107
123
                    if (attname != ctxt->str_xml) {
9108
123
                        xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
9109
123
                     "xml namespace URI cannot be the default namespace\n",
9110
123
                                 NULL, NULL, NULL);
9111
123
                    }
9112
123
                    goto next_attr;
9113
123
                }
9114
317k
                if ((len == 29) &&
9115
317k
                    (xmlStrEqual(uri,
9116
8.05k
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
9117
214
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
9118
214
                         "reuse of the xmlns namespace name is forbidden\n",
9119
214
                             NULL, NULL, NULL);
9120
214
                    goto next_attr;
9121
214
                }
9122
317k
            }
9123
9124
320k
            if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0)
9125
237k
                nbNs++;
9126
27.7M
        } else if (aprefix == ctxt->str_xmlns) {
9127
1.32M
            xmlHashedString huri;
9128
1.32M
            xmlURIPtr parsedUri;
9129
9130
1.32M
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
9131
1.32M
            uri = huri.name;
9132
1.32M
            if (uri == NULL) {
9133
0
                xmlErrMemory(ctxt);
9134
0
                goto next_attr;
9135
0
            }
9136
9137
1.32M
            if (attname == ctxt->str_xml) {
9138
509
                if (uri != ctxt->str_xml_ns) {
9139
453
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
9140
453
                             "xml namespace prefix mapped to wrong URI\n",
9141
453
                             NULL, NULL, NULL);
9142
453
                }
9143
                /*
9144
                 * Do not keep a namespace definition node
9145
                 */
9146
509
                goto next_attr;
9147
509
            }
9148
1.32M
            if (uri == ctxt->str_xml_ns) {
9149
11
                if (attname != ctxt->str_xml) {
9150
11
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
9151
11
                             "xml namespace URI mapped to wrong prefix\n",
9152
11
                             NULL, NULL, NULL);
9153
11
                }
9154
11
                goto next_attr;
9155
11
            }
9156
1.32M
            if (attname == ctxt->str_xmlns) {
9157
429
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
9158
429
                         "redefinition of the xmlns prefix is forbidden\n",
9159
429
                         NULL, NULL, NULL);
9160
429
                goto next_attr;
9161
429
            }
9162
1.32M
            if ((len == 29) &&
9163
1.32M
                (xmlStrEqual(uri,
9164
29.1k
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
9165
117
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
9166
117
                         "reuse of the xmlns namespace name is forbidden\n",
9167
117
                         NULL, NULL, NULL);
9168
117
                goto next_attr;
9169
117
            }
9170
1.32M
            if ((uri == NULL) || (uri[0] == 0)) {
9171
271
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
9172
271
                         "xmlns:%s: Empty XML namespace is not allowed\n",
9173
271
                              attname, NULL, NULL);
9174
271
                goto next_attr;
9175
1.32M
            } else {
9176
1.32M
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
9177
0
                    xmlErrMemory(ctxt);
9178
0
                    goto next_attr;
9179
0
                }
9180
1.32M
                if (parsedUri == NULL) {
9181
112k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
9182
112k
                         "xmlns:%s: '%s' is not a valid URI\n",
9183
112k
                                       attname, uri, NULL);
9184
1.21M
                } else {
9185
1.21M
                    if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) {
9186
0
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
9187
0
                                  "xmlns:%s: URI %s is not absolute\n",
9188
0
                                  attname, uri, NULL);
9189
0
                    }
9190
1.21M
                    xmlFreeURI(parsedUri);
9191
1.21M
                }
9192
1.32M
            }
9193
9194
1.32M
            if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0)
9195
1.24M
                nbNs++;
9196
26.4M
        } else {
9197
            /*
9198
             * Populate attributes array, see above for repurposing
9199
             * of xmlChar pointers.
9200
             */
9201
26.4M
            if ((atts == NULL) || (nbatts + 5 > maxatts)) {
9202
732k
                int res = xmlCtxtGrowAttrs(ctxt);
9203
9204
732k
                maxatts = ctxt->maxatts;
9205
732k
                atts = ctxt->atts;
9206
9207
732k
                if (res < 0)
9208
0
                    goto next_attr;
9209
732k
            }
9210
26.4M
            ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) |
9211
26.4M
                                        ((unsigned) alloc << 31);
9212
26.4M
            atts[nbatts++] = attname;
9213
26.4M
            atts[nbatts++] = aprefix;
9214
26.4M
            atts[nbatts++] = (const xmlChar *) (size_t) haprefix.hashValue;
9215
26.4M
            if (alloc) {
9216
126k
                atts[nbatts++] = attvalue;
9217
126k
                attvalue += len;
9218
126k
                atts[nbatts++] = attvalue;
9219
26.3M
            } else {
9220
                /*
9221
                 * attvalue points into the input buffer which can be
9222
                 * reallocated. Store differences to input->base instead.
9223
                 * The pointers will be reconstructed later.
9224
                 */
9225
26.3M
                atts[nbatts++] = (void *) (attvalue - BASE_PTR);
9226
26.3M
                attvalue += len;
9227
26.3M
                atts[nbatts++] = (void *) (attvalue - BASE_PTR);
9228
26.3M
            }
9229
            /*
9230
             * tag if some deallocation is needed
9231
             */
9232
26.4M
            if (alloc != 0) attval = 1;
9233
26.4M
            attvalue = NULL; /* moved into atts */
9234
26.4M
        }
9235
9236
28.1M
next_attr:
9237
28.1M
        if ((attvalue != NULL) && (alloc != 0)) {
9238
95.4k
            xmlFree(attvalue);
9239
95.4k
            attvalue = NULL;
9240
95.4k
        }
9241
9242
28.1M
  GROW
9243
28.1M
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
9244
12.5M
      break;
9245
15.5M
  if (SKIP_BLANKS == 0) {
9246
35.5k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
9247
35.5k
         "attributes construct error\n");
9248
35.5k
      break;
9249
35.5k
  }
9250
15.5M
        GROW;
9251
15.5M
    }
9252
9253
    /*
9254
     * Namespaces from default attributes
9255
     */
9256
35.9M
    if (ctxt->attsDefault != NULL) {
9257
117k
        xmlDefAttrsPtr defaults;
9258
9259
117k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9260
117k
  if (defaults != NULL) {
9261
722k
      for (i = 0; i < defaults->nbAttrs; i++) {
9262
639k
                xmlDefAttr *attr = &defaults->attrs[i];
9263
9264
639k
          attname = attr->name.name;
9265
639k
    aprefix = attr->prefix.name;
9266
9267
639k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
9268
13.6k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9269
9270
13.6k
                    if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0)
9271
12.4k
                        nbNs++;
9272
625k
    } else if (aprefix == ctxt->str_xmlns) {
9273
252k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9274
9275
252k
                    if (xmlParserNsPush(ctxt, &attr->name, &attr->value,
9276
252k
                                      NULL, 1) > 0)
9277
252k
                        nbNs++;
9278
372k
    } else {
9279
372k
                    if (nratts + nbTotalDef >= XML_MAX_ATTRS) {
9280
0
                        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
9281
0
                                    "Maximum number of attributes exceeded");
9282
0
                        break;
9283
0
                    }
9284
372k
                    nbTotalDef += 1;
9285
372k
                }
9286
639k
      }
9287
83.4k
  }
9288
117k
    }
9289
9290
    /*
9291
     * Resolve attribute namespaces
9292
     */
9293
62.3M
    for (i = 0; i < nbatts; i += 5) {
9294
26.4M
        attname = atts[i];
9295
26.4M
        aprefix = atts[i+1];
9296
9297
        /*
9298
  * The default namespace does not apply to attribute names.
9299
  */
9300
26.4M
  if (aprefix == NULL) {
9301
16.4M
            nsIndex = NS_INDEX_EMPTY;
9302
16.4M
        } else if (aprefix == ctxt->str_xml) {
9303
111k
            nsIndex = NS_INDEX_XML;
9304
9.92M
        } else {
9305
9.92M
            haprefix.name = aprefix;
9306
9.92M
            haprefix.hashValue = (size_t) atts[i+2];
9307
9.92M
            nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL);
9308
9309
9.92M
      if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) {
9310
440k
                xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9311
440k
        "Namespace prefix %s for %s on %s is not defined\n",
9312
440k
        aprefix, attname, localname);
9313
440k
                nsIndex = NS_INDEX_EMPTY;
9314
440k
            }
9315
9.92M
        }
9316
9317
26.4M
        atts[i+2] = XML_INT_TO_PTR(nsIndex);
9318
26.4M
    }
9319
9320
    /*
9321
     * Maximum number of attributes including default attributes.
9322
     */
9323
35.9M
    maxAtts = nratts + nbTotalDef;
9324
9325
    /*
9326
     * Verify that attribute names are unique.
9327
     */
9328
35.9M
    if (maxAtts > 1) {
9329
7.08M
        attrHashSize = 4;
9330
11.0M
        while (attrHashSize / 2 < (unsigned) maxAtts)
9331
3.99M
            attrHashSize *= 2;
9332
9333
7.08M
        if (attrHashSize > ctxt->attrHashMax) {
9334
301k
            xmlAttrHashBucket *tmp;
9335
9336
301k
            tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0]));
9337
301k
            if (tmp == NULL) {
9338
0
                xmlErrMemory(ctxt);
9339
0
                goto done;
9340
0
            }
9341
9342
301k
            ctxt->attrHash = tmp;
9343
301k
            ctxt->attrHashMax = attrHashSize;
9344
301k
        }
9345
9346
7.08M
        memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0]));
9347
9348
27.9M
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9349
20.8M
            const xmlChar *nsuri;
9350
20.8M
            unsigned hashValue, nameHashValue, uriHashValue;
9351
20.8M
            int res;
9352
9353
20.8M
            attname = atts[i];
9354
20.8M
            aprefix = atts[i+1];
9355
20.8M
            nsIndex = XML_PTR_TO_INT(atts[i+2]);
9356
            /* Hash values always have bit 31 set, see dict.c */
9357
20.8M
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9358
9359
20.8M
            if (nsIndex == NS_INDEX_EMPTY) {
9360
                /*
9361
                 * Prefix with empty namespace means an undeclared
9362
                 * prefix which was already reported above.
9363
                 */
9364
13.3M
                if (aprefix != NULL)
9365
377k
                    continue;
9366
12.9M
                nsuri = NULL;
9367
12.9M
                uriHashValue = URI_HASH_EMPTY;
9368
12.9M
            } else if (nsIndex == NS_INDEX_XML) {
9369
2.32k
                nsuri = ctxt->str_xml_ns;
9370
2.32k
                uriHashValue = URI_HASH_XML;
9371
7.49M
            } else {
9372
7.49M
                nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9373
7.49M
                uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9374
7.49M
            }
9375
9376
20.4M
            hashValue = xmlDictCombineHash(nameHashValue, uriHashValue);
9377
20.4M
            res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9378
20.4M
                                    hashValue, i);
9379
20.4M
            if (res < 0)
9380
0
                continue;
9381
9382
            /*
9383
             * [ WFC: Unique Att Spec ]
9384
             * No attribute name may appear more than once in the same
9385
             * start-tag or empty-element tag.
9386
             * As extended by the Namespace in XML REC.
9387
             */
9388
20.4M
            if (res < INT_MAX) {
9389
27.4k
                if (aprefix == atts[res+1]) {
9390
27.1k
                    xmlErrAttributeDup(ctxt, aprefix, attname);
9391
27.1k
                    numDupErr += 1;
9392
27.1k
                } else {
9393
311
                    xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9394
311
                             "Namespaced Attribute %s in '%s' redefined\n",
9395
311
                             attname, nsuri, NULL);
9396
311
                    numNsErr += 1;
9397
311
                }
9398
27.4k
            }
9399
20.4M
        }
9400
7.08M
    }
9401
9402
    /*
9403
     * Default attributes
9404
     */
9405
35.9M
    if (ctxt->attsDefault != NULL) {
9406
117k
        xmlDefAttrsPtr defaults;
9407
9408
117k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9409
117k
  if (defaults != NULL) {
9410
722k
      for (i = 0; i < defaults->nbAttrs; i++) {
9411
639k
                xmlDefAttr *attr = &defaults->attrs[i];
9412
639k
                const xmlChar *nsuri = NULL;
9413
639k
                unsigned hashValue, uriHashValue = 0;
9414
639k
                int res;
9415
9416
639k
          attname = attr->name.name;
9417
639k
    aprefix = attr->prefix.name;
9418
9419
639k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL))
9420
13.6k
                    continue;
9421
625k
    if (aprefix == ctxt->str_xmlns)
9422
252k
                    continue;
9423
9424
372k
                if (aprefix == NULL) {
9425
200k
                    nsIndex = NS_INDEX_EMPTY;
9426
200k
                    nsuri = NULL;
9427
200k
                    uriHashValue = URI_HASH_EMPTY;
9428
200k
                } else if (aprefix == ctxt->str_xml) {
9429
62.8k
                    nsIndex = NS_INDEX_XML;
9430
62.8k
                    nsuri = ctxt->str_xml_ns;
9431
62.8k
                    uriHashValue = URI_HASH_XML;
9432
109k
                } else {
9433
109k
                    nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL);
9434
109k
                    if ((nsIndex == INT_MAX) ||
9435
109k
                        (nsIndex < ctxt->nsdb->minNsIndex)) {
9436
108k
                        xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9437
108k
                                 "Namespace prefix %s for %s on %s is not "
9438
108k
                                 "defined\n",
9439
108k
                                 aprefix, attname, localname);
9440
108k
                        nsIndex = NS_INDEX_EMPTY;
9441
108k
                        nsuri = NULL;
9442
108k
                        uriHashValue = URI_HASH_EMPTY;
9443
108k
                    } else {
9444
780
                        nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9445
780
                        uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9446
780
                    }
9447
109k
                }
9448
9449
                /*
9450
                 * Check whether the attribute exists
9451
                 */
9452
372k
                if (maxAtts > 1) {
9453
357k
                    hashValue = xmlDictCombineHash(attr->name.hashValue,
9454
357k
                                                   uriHashValue);
9455
357k
                    res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9456
357k
                                            hashValue, nbatts);
9457
357k
                    if (res < 0)
9458
0
                        continue;
9459
357k
                    if (res < INT_MAX) {
9460
2.13k
                        if (aprefix == atts[res+1])
9461
838
                            continue;
9462
1.30k
                        xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9463
1.30k
                                 "Namespaced Attribute %s in '%s' redefined\n",
9464
1.30k
                                 attname, nsuri, NULL);
9465
1.30k
                    }
9466
357k
                }
9467
9468
372k
                xmlParserEntityCheck(ctxt, attr->expandedSize);
9469
9470
372k
                if ((atts == NULL) || (nbatts + 5 > maxatts)) {
9471
5.32k
                    res = xmlCtxtGrowAttrs(ctxt);
9472
9473
5.32k
                    maxatts = ctxt->maxatts;
9474
5.32k
                    atts = ctxt->atts;
9475
9476
5.32k
                    if (res < 0) {
9477
0
                        localname = NULL;
9478
0
                        goto done;
9479
0
                    }
9480
5.32k
                }
9481
9482
372k
                atts[nbatts++] = attname;
9483
372k
                atts[nbatts++] = aprefix;
9484
372k
                atts[nbatts++] = XML_INT_TO_PTR(nsIndex);
9485
372k
                atts[nbatts++] = attr->value.name;
9486
372k
                atts[nbatts++] = attr->valueEnd;
9487
372k
                if ((ctxt->standalone == 1) && (attr->external != 0)) {
9488
0
                    xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED,
9489
0
                            "standalone: attribute %s on %s defaulted "
9490
0
                            "from external subset\n",
9491
0
                            attname, localname);
9492
0
                }
9493
372k
                nbdef++;
9494
372k
      }
9495
83.4k
  }
9496
117k
    }
9497
9498
    /*
9499
     * Using a single hash table for nsUri/localName pairs cannot
9500
     * detect duplicate QNames reliably. The following example will
9501
     * only result in two namespace errors.
9502
     *
9503
     * <doc xmlns:a="a" xmlns:b="a">
9504
     *   <elem a:a="" b:a="" b:a=""/>
9505
     * </doc>
9506
     *
9507
     * If we saw more than one namespace error but no duplicate QNames
9508
     * were found, we have to scan for duplicate QNames.
9509
     */
9510
35.9M
    if ((numDupErr == 0) && (numNsErr > 1)) {
9511
18
        memset(ctxt->attrHash, -1,
9512
18
               attrHashSize * sizeof(ctxt->attrHash[0]));
9513
9514
150
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9515
132
            unsigned hashValue, nameHashValue, prefixHashValue;
9516
132
            int res;
9517
9518
132
            aprefix = atts[i+1];
9519
132
            if (aprefix == NULL)
9520
22
                continue;
9521
9522
110
            attname = atts[i];
9523
            /* Hash values always have bit 31 set, see dict.c */
9524
110
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9525
110
            prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix);
9526
9527
110
            hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue);
9528
110
            res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname,
9529
110
                                         aprefix, hashValue, i);
9530
110
            if (res < INT_MAX)
9531
56
                xmlErrAttributeDup(ctxt, aprefix, attname);
9532
110
        }
9533
18
    }
9534
9535
    /*
9536
     * Reconstruct attribute pointers
9537
     */
9538
62.7M
    for (i = 0, j = 0; i < nbatts; i += 5, j++) {
9539
        /* namespace URI */
9540
26.8M
        nsIndex = XML_PTR_TO_INT(atts[i+2]);
9541
26.8M
        if (nsIndex == INT_MAX)
9542
17.1M
            atts[i+2] = NULL;
9543
9.66M
        else if (nsIndex == INT_MAX - 1)
9544
174k
            atts[i+2] = ctxt->str_xml_ns;
9545
9.48M
        else
9546
9.48M
            atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1];
9547
9548
26.8M
        if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) {
9549
26.3M
            atts[i+3] = BASE_PTR + XML_PTR_TO_INT(atts[i+3]);  /* value */
9550
26.3M
            atts[i+4] = BASE_PTR + XML_PTR_TO_INT(atts[i+4]);  /* valuend */
9551
26.3M
        }
9552
26.8M
    }
9553
9554
35.9M
    uri = xmlParserNsLookupUri(ctxt, &hprefix);
9555
35.9M
    if ((prefix != NULL) && (uri == NULL)) {
9556
520k
  xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9557
520k
           "Namespace prefix %s on %s is not defined\n",
9558
520k
     prefix, localname, NULL);
9559
520k
    }
9560
35.9M
    *pref = prefix;
9561
35.9M
    *URI = uri;
9562
9563
    /*
9564
     * SAX callback
9565
     */
9566
35.9M
    if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) &&
9567
35.9M
  (!ctxt->disableSAX)) {
9568
35.7M
  if (nbNs > 0)
9569
527k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9570
527k
                          nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs),
9571
527k
        nbatts / 5, nbdef, atts);
9572
35.2M
  else
9573
35.2M
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9574
35.2M
                          0, NULL, nbatts / 5, nbdef, atts);
9575
35.7M
    }
9576
9577
35.9M
done:
9578
    /*
9579
     * Free allocated attribute values
9580
     */
9581
35.9M
    if (attval != 0) {
9582
510k
  for (i = 0, j = 0; j < nratts; i += 5, j++)
9583
396k
      if (ctxt->attallocs[j] & 0x80000000)
9584
126k
          xmlFree((xmlChar *) atts[i+3]);
9585
113k
    }
9586
9587
35.9M
    *nbNsPtr = nbNs;
9588
35.9M
    return(localname);
9589
35.9M
}
9590
9591
/**
9592
 * xmlParseEndTag2:
9593
 * @ctxt:  an XML parser context
9594
 * @line:  line of the start tag
9595
 * @nsNr:  number of namespaces on the start tag
9596
 *
9597
 * Parse an end tag. Always consumes '</'.
9598
 *
9599
 * [42] ETag ::= '</' Name S? '>'
9600
 *
9601
 * With namespace
9602
 *
9603
 * [NS 9] ETag ::= '</' QName S? '>'
9604
 */
9605
9606
static void
9607
10.3M
xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) {
9608
10.3M
    const xmlChar *name;
9609
9610
10.3M
    GROW;
9611
10.3M
    if ((RAW != '<') || (NXT(1) != '/')) {
9612
4
  xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL);
9613
4
  return;
9614
4
    }
9615
10.3M
    SKIP(2);
9616
9617
10.3M
    if (tag->prefix == NULL)
9618
3.27M
        name = xmlParseNameAndCompare(ctxt, ctxt->name);
9619
7.09M
    else
9620
7.09M
        name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix);
9621
9622
    /*
9623
     * We should definitely be at the ending "S? '>'" part
9624
     */
9625
10.3M
    GROW;
9626
10.3M
    SKIP_BLANKS;
9627
10.3M
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
9628
7.38k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
9629
7.38k
    } else
9630
10.3M
  NEXT1;
9631
9632
    /*
9633
     * [ WFC: Element Type Match ]
9634
     * The Name in an element's end-tag must match the element type in the
9635
     * start-tag.
9636
     *
9637
     */
9638
10.3M
    if (name != (xmlChar*)1) {
9639
17.7k
        if (name == NULL) name = BAD_CAST "unparsable";
9640
17.7k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
9641
17.7k
         "Opening and ending tag mismatch: %s line %d and %s\n",
9642
17.7k
                    ctxt->name, tag->line, name);
9643
17.7k
    }
9644
9645
    /*
9646
     * SAX: End of Tag
9647
     */
9648
10.3M
    if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9649
10.3M
  (!ctxt->disableSAX))
9650
10.3M
  ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix,
9651
10.3M
                                tag->URI);
9652
9653
10.3M
    spacePop(ctxt);
9654
10.3M
    if (tag->nsNr != 0)
9655
211k
  xmlParserNsPop(ctxt, tag->nsNr);
9656
10.3M
}
9657
9658
/**
9659
 * xmlParseCDSect:
9660
 * @ctxt:  an XML parser context
9661
 *
9662
 * DEPRECATED: Internal function, don't use.
9663
 *
9664
 * Parse escaped pure raw content. Always consumes '<!['.
9665
 *
9666
 * [18] CDSect ::= CDStart CData CDEnd
9667
 *
9668
 * [19] CDStart ::= '<![CDATA['
9669
 *
9670
 * [20] Data ::= (Char* - (Char* ']]>' Char*))
9671
 *
9672
 * [21] CDEnd ::= ']]>'
9673
 */
9674
void
9675
14.0k
xmlParseCDSect(xmlParserCtxtPtr ctxt) {
9676
14.0k
    xmlChar *buf = NULL;
9677
14.0k
    int len = 0;
9678
14.0k
    int size = XML_PARSER_BUFFER_SIZE;
9679
14.0k
    int r, rl;
9680
14.0k
    int s, sl;
9681
14.0k
    int cur, l;
9682
14.0k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9683
14.0k
                    XML_MAX_HUGE_LENGTH :
9684
14.0k
                    XML_MAX_TEXT_LENGTH;
9685
9686
14.0k
    if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '['))
9687
0
        return;
9688
14.0k
    SKIP(3);
9689
9690
14.0k
    if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '['))
9691
0
        return;
9692
14.0k
    SKIP(6);
9693
9694
14.0k
    r = xmlCurrentCharRecover(ctxt, &rl);
9695
14.0k
    if (!IS_CHAR(r)) {
9696
15
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9697
15
        goto out;
9698
15
    }
9699
14.0k
    NEXTL(rl);
9700
14.0k
    s = xmlCurrentCharRecover(ctxt, &sl);
9701
14.0k
    if (!IS_CHAR(s)) {
9702
26
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9703
26
        goto out;
9704
26
    }
9705
14.0k
    NEXTL(sl);
9706
14.0k
    cur = xmlCurrentCharRecover(ctxt, &l);
9707
14.0k
    buf = xmlMalloc(size);
9708
14.0k
    if (buf == NULL) {
9709
0
  xmlErrMemory(ctxt);
9710
0
        goto out;
9711
0
    }
9712
62.6M
    while (IS_CHAR(cur) &&
9713
62.6M
           ((r != ']') || (s != ']') || (cur != '>'))) {
9714
62.6M
  if (len + 5 >= size) {
9715
75.0k
      xmlChar *tmp;
9716
75.0k
            int newSize;
9717
9718
75.0k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9719
75.0k
            if (newSize < 0) {
9720
0
                xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9721
0
                               "CData section too big found\n");
9722
0
                goto out;
9723
0
            }
9724
75.0k
      tmp = xmlRealloc(buf, newSize);
9725
75.0k
      if (tmp == NULL) {
9726
0
    xmlErrMemory(ctxt);
9727
0
                goto out;
9728
0
      }
9729
75.0k
      buf = tmp;
9730
75.0k
      size = newSize;
9731
75.0k
  }
9732
62.6M
  COPY_BUF(buf, len, r);
9733
62.6M
  r = s;
9734
62.6M
  rl = sl;
9735
62.6M
  s = cur;
9736
62.6M
  sl = l;
9737
62.6M
  NEXTL(l);
9738
62.6M
  cur = xmlCurrentCharRecover(ctxt, &l);
9739
62.6M
    }
9740
14.0k
    buf[len] = 0;
9741
14.0k
    if (cur != '>') {
9742
362
  xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9743
362
                       "CData section not finished\n%.50s\n", buf);
9744
362
        goto out;
9745
362
    }
9746
13.6k
    NEXTL(l);
9747
9748
    /*
9749
     * OK the buffer is to be consumed as cdata.
9750
     */
9751
13.6k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) {
9752
13.6k
        if ((ctxt->sax->cdataBlock != NULL) &&
9753
13.6k
            ((ctxt->options & XML_PARSE_NOCDATA) == 0)) {
9754
0
            ctxt->sax->cdataBlock(ctxt->userData, buf, len);
9755
13.6k
        } else if (ctxt->sax->characters != NULL) {
9756
13.6k
            ctxt->sax->characters(ctxt->userData, buf, len);
9757
13.6k
        }
9758
13.6k
    }
9759
9760
14.0k
out:
9761
14.0k
    xmlFree(buf);
9762
14.0k
}
9763
9764
/**
9765
 * xmlParseContentInternal:
9766
 * @ctxt:  an XML parser context
9767
 *
9768
 * Parse a content sequence. Stops at EOF or '</'. Leaves checking of
9769
 * unexpected EOF to the caller.
9770
 */
9771
9772
static void
9773
6.52k
xmlParseContentInternal(xmlParserCtxtPtr ctxt) {
9774
6.52k
    int oldNameNr = ctxt->nameNr;
9775
6.52k
    int oldSpaceNr = ctxt->spaceNr;
9776
6.52k
    int oldNodeNr = ctxt->nodeNr;
9777
9778
6.52k
    GROW;
9779
2.06M
    while ((ctxt->input->cur < ctxt->input->end) &&
9780
2.06M
     (PARSER_STOPPED(ctxt) == 0)) {
9781
2.06M
  const xmlChar *cur = ctxt->input->cur;
9782
9783
  /*
9784
   * First case : a Processing Instruction.
9785
   */
9786
2.06M
  if ((*cur == '<') && (cur[1] == '?')) {
9787
51
      xmlParsePI(ctxt);
9788
51
  }
9789
9790
  /*
9791
   * Second case : a CDSection
9792
   */
9793
  /* 2.6.0 test was *cur not RAW */
9794
2.06M
  else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) {
9795
0
      xmlParseCDSect(ctxt);
9796
0
  }
9797
9798
  /*
9799
   * Third case :  a comment
9800
   */
9801
2.06M
  else if ((*cur == '<') && (NXT(1) == '!') &&
9802
2.06M
     (NXT(2) == '-') && (NXT(3) == '-')) {
9803
0
      xmlParseComment(ctxt);
9804
0
  }
9805
9806
  /*
9807
   * Fourth case :  a sub-element.
9808
   */
9809
2.06M
  else if (*cur == '<') {
9810
1.65M
            if (NXT(1) == '/') {
9811
482k
                if (ctxt->nameNr <= oldNameNr)
9812
6.34k
                    break;
9813
475k
          xmlParseElementEnd(ctxt);
9814
1.16M
            } else {
9815
1.16M
          xmlParseElementStart(ctxt);
9816
1.16M
            }
9817
1.65M
  }
9818
9819
  /*
9820
   * Fifth case : a reference. If if has not been resolved,
9821
   *    parsing returns it's Name, create the node
9822
   */
9823
9824
413k
  else if (*cur == '&') {
9825
8
      xmlParseReference(ctxt);
9826
8
  }
9827
9828
  /*
9829
   * Last case, text. Note that References are handled directly.
9830
   */
9831
413k
  else {
9832
413k
      xmlParseCharDataInternal(ctxt, 0);
9833
413k
  }
9834
9835
2.05M
  SHRINK;
9836
2.05M
  GROW;
9837
2.05M
    }
9838
9839
6.52k
    if ((ctxt->nameNr > oldNameNr) &&
9840
6.52k
        (ctxt->input->cur >= ctxt->input->end) &&
9841
6.52k
        (ctxt->wellFormed)) {
9842
1
        const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9843
1
        int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9844
1
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9845
1
                "Premature end of data in tag %s line %d\n",
9846
1
                name, line, NULL);
9847
1
    }
9848
9849
    /*
9850
     * Clean up in error case
9851
     */
9852
9853
6.60k
    while (ctxt->nodeNr > oldNodeNr)
9854
80
        nodePop(ctxt);
9855
9856
16.8k
    while (ctxt->nameNr > oldNameNr) {
9857
10.3k
        xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1];
9858
9859
10.3k
        if (tag->nsNr != 0)
9860
14
            xmlParserNsPop(ctxt, tag->nsNr);
9861
9862
10.3k
        namePop(ctxt);
9863
10.3k
    }
9864
9865
16.8k
    while (ctxt->spaceNr > oldSpaceNr)
9866
10.3k
        spacePop(ctxt);
9867
6.52k
}
9868
9869
/**
9870
 * xmlParseContent:
9871
 * @ctxt:  an XML parser context
9872
 *
9873
 * Parse XML element content. This is useful if you're only interested
9874
 * in custom SAX callbacks. If you want a node list, use
9875
 * xmlCtxtParseContent.
9876
 */
9877
void
9878
0
xmlParseContent(xmlParserCtxtPtr ctxt) {
9879
0
    if ((ctxt == NULL) || (ctxt->input == NULL))
9880
0
        return;
9881
9882
0
    xmlCtxtInitializeLate(ctxt);
9883
9884
0
    xmlParseContentInternal(ctxt);
9885
9886
0
    xmlParserCheckEOF(ctxt, XML_ERR_NOT_WELL_BALANCED);
9887
0
}
9888
9889
/**
9890
 * xmlParseElement:
9891
 * @ctxt:  an XML parser context
9892
 *
9893
 * DEPRECATED: Internal function, don't use.
9894
 *
9895
 * parse an XML element
9896
 *
9897
 * [39] element ::= EmptyElemTag | STag content ETag
9898
 *
9899
 * [ WFC: Element Type Match ]
9900
 * The Name in an element's end-tag must match the element type in the
9901
 * start-tag.
9902
 *
9903
 */
9904
9905
void
9906
6.72k
xmlParseElement(xmlParserCtxtPtr ctxt) {
9907
6.72k
    if (xmlParseElementStart(ctxt) != 0)
9908
199
        return;
9909
9910
6.52k
    xmlParseContentInternal(ctxt);
9911
9912
6.52k
    if (ctxt->input->cur >= ctxt->input->end) {
9913
182
        if (ctxt->wellFormed) {
9914
0
            const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9915
0
            int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9916
0
            xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9917
0
                    "Premature end of data in tag %s line %d\n",
9918
0
                    name, line, NULL);
9919
0
        }
9920
182
        return;
9921
182
    }
9922
9923
6.34k
    xmlParseElementEnd(ctxt);
9924
6.34k
}
9925
9926
/**
9927
 * xmlParseElementStart:
9928
 * @ctxt:  an XML parser context
9929
 *
9930
 * Parse the start of an XML element. Returns -1 in case of error, 0 if an
9931
 * opening tag was parsed, 1 if an empty element was parsed.
9932
 *
9933
 * Always consumes '<'.
9934
 */
9935
static int
9936
1.17M
xmlParseElementStart(xmlParserCtxtPtr ctxt) {
9937
1.17M
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
9938
1.17M
    const xmlChar *name;
9939
1.17M
    const xmlChar *prefix = NULL;
9940
1.17M
    const xmlChar *URI = NULL;
9941
1.17M
    xmlParserNodeInfo node_info;
9942
1.17M
    int line;
9943
1.17M
    xmlNodePtr cur;
9944
1.17M
    int nbNs = 0;
9945
9946
1.17M
    if (ctxt->nameNr > maxDepth) {
9947
4
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
9948
4
                "Excessive depth in document: %d use XML_PARSE_HUGE option\n",
9949
4
                ctxt->nameNr);
9950
4
  xmlHaltParser(ctxt);
9951
4
  return(-1);
9952
4
    }
9953
9954
    /* Capture start position */
9955
1.17M
    if (ctxt->record_info) {
9956
0
        node_info.begin_pos = ctxt->input->consumed +
9957
0
                          (CUR_PTR - ctxt->input->base);
9958
0
  node_info.begin_line = ctxt->input->line;
9959
0
    }
9960
9961
1.17M
    if (ctxt->spaceNr == 0)
9962
6.72k
  spacePush(ctxt, -1);
9963
1.16M
    else if (*ctxt->space == -2)
9964
0
  spacePush(ctxt, -1);
9965
1.16M
    else
9966
1.16M
  spacePush(ctxt, *ctxt->space);
9967
9968
1.17M
    line = ctxt->input->line;
9969
1.17M
#ifdef LIBXML_SAX1_ENABLED
9970
1.17M
    if (ctxt->sax2)
9971
1.17M
#endif /* LIBXML_SAX1_ENABLED */
9972
1.17M
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
9973
0
#ifdef LIBXML_SAX1_ENABLED
9974
0
    else
9975
0
  name = xmlParseStartTag(ctxt);
9976
1.17M
#endif /* LIBXML_SAX1_ENABLED */
9977
1.17M
    if (name == NULL) {
9978
4.56k
  spacePop(ctxt);
9979
4.56k
        return(-1);
9980
4.56k
    }
9981
1.17M
    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
9982
1.17M
    cur = ctxt->node;
9983
9984
1.17M
#ifdef LIBXML_VALID_ENABLED
9985
    /*
9986
     * [ VC: Root Element Type ]
9987
     * The Name in the document type declaration must match the element
9988
     * type of the root element.
9989
     */
9990
1.17M
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
9991
1.17M
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
9992
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
9993
1.17M
#endif /* LIBXML_VALID_ENABLED */
9994
9995
    /*
9996
     * Check for an Empty Element.
9997
     */
9998
1.17M
    if ((RAW == '/') && (NXT(1) == '>')) {
9999
633k
        SKIP(2);
10000
633k
  if (ctxt->sax2) {
10001
633k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
10002
633k
    (!ctxt->disableSAX))
10003
618k
    ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI);
10004
633k
#ifdef LIBXML_SAX1_ENABLED
10005
633k
  } else {
10006
0
      if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
10007
0
    (!ctxt->disableSAX))
10008
0
    ctxt->sax->endElement(ctxt->userData, name);
10009
0
#endif /* LIBXML_SAX1_ENABLED */
10010
0
  }
10011
633k
  namePop(ctxt);
10012
633k
  spacePop(ctxt);
10013
633k
  if (nbNs > 0)
10014
1.27k
      xmlParserNsPop(ctxt, nbNs);
10015
633k
  if (cur != NULL && ctxt->record_info) {
10016
0
            node_info.node = cur;
10017
0
            node_info.end_pos = ctxt->input->consumed +
10018
0
                                (CUR_PTR - ctxt->input->base);
10019
0
            node_info.end_line = ctxt->input->line;
10020
0
            xmlParserAddNodeInfo(ctxt, &node_info);
10021
0
  }
10022
633k
  return(1);
10023
633k
    }
10024
537k
    if (RAW == '>') {
10025
492k
        NEXT1;
10026
492k
        if (cur != NULL && ctxt->record_info) {
10027
0
            node_info.node = cur;
10028
0
            node_info.end_pos = 0;
10029
0
            node_info.end_line = 0;
10030
0
            xmlParserAddNodeInfo(ctxt, &node_info);
10031
0
        }
10032
492k
    } else {
10033
45.0k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED,
10034
45.0k
         "Couldn't find end of Start Tag %s line %d\n",
10035
45.0k
                    name, line, NULL);
10036
10037
  /*
10038
   * end of parsing of this node.
10039
   */
10040
45.0k
  nodePop(ctxt);
10041
45.0k
  namePop(ctxt);
10042
45.0k
  spacePop(ctxt);
10043
45.0k
  if (nbNs > 0)
10044
107
      xmlParserNsPop(ctxt, nbNs);
10045
45.0k
  return(-1);
10046
45.0k
    }
10047
10048
492k
    return(0);
10049
537k
}
10050
10051
/**
10052
 * xmlParseElementEnd:
10053
 * @ctxt:  an XML parser context
10054
 *
10055
 * Parse the end of an XML element. Always consumes '</'.
10056
 */
10057
static void
10058
482k
xmlParseElementEnd(xmlParserCtxtPtr ctxt) {
10059
482k
    xmlNodePtr cur = ctxt->node;
10060
10061
482k
    if (ctxt->nameNr <= 0) {
10062
0
        if ((RAW == '<') && (NXT(1) == '/'))
10063
0
            SKIP(2);
10064
0
        return;
10065
0
    }
10066
10067
    /*
10068
     * parse the end of tag: '</' should be here.
10069
     */
10070
482k
    if (ctxt->sax2) {
10071
482k
  xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
10072
482k
  namePop(ctxt);
10073
482k
    }
10074
0
#ifdef LIBXML_SAX1_ENABLED
10075
0
    else
10076
0
  xmlParseEndTag1(ctxt, 0);
10077
482k
#endif /* LIBXML_SAX1_ENABLED */
10078
10079
    /*
10080
     * Capture end position
10081
     */
10082
482k
    if (cur != NULL && ctxt->record_info) {
10083
0
        xmlParserNodeInfoPtr node_info;
10084
10085
0
        node_info = (xmlParserNodeInfoPtr) xmlParserFindNodeInfo(ctxt, cur);
10086
0
        if (node_info != NULL) {
10087
0
            node_info->end_pos = ctxt->input->consumed +
10088
0
                                 (CUR_PTR - ctxt->input->base);
10089
0
            node_info->end_line = ctxt->input->line;
10090
0
        }
10091
0
    }
10092
482k
}
10093
10094
/**
10095
 * xmlParseVersionNum:
10096
 * @ctxt:  an XML parser context
10097
 *
10098
 * DEPRECATED: Internal function, don't use.
10099
 *
10100
 * parse the XML version value.
10101
 *
10102
 * [26] VersionNum ::= '1.' [0-9]+
10103
 *
10104
 * In practice allow [0-9].[0-9]+ at that level
10105
 *
10106
 * Returns the string giving the XML version number, or NULL
10107
 */
10108
xmlChar *
10109
205k
xmlParseVersionNum(xmlParserCtxtPtr ctxt) {
10110
205k
    xmlChar *buf = NULL;
10111
205k
    int len = 0;
10112
205k
    int size = 10;
10113
205k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
10114
199k
                    XML_MAX_TEXT_LENGTH :
10115
205k
                    XML_MAX_NAME_LENGTH;
10116
205k
    xmlChar cur;
10117
10118
205k
    buf = xmlMalloc(size);
10119
205k
    if (buf == NULL) {
10120
0
  xmlErrMemory(ctxt);
10121
0
  return(NULL);
10122
0
    }
10123
205k
    cur = CUR;
10124
205k
    if (!((cur >= '0') && (cur <= '9'))) {
10125
131
  xmlFree(buf);
10126
131
  return(NULL);
10127
131
    }
10128
205k
    buf[len++] = cur;
10129
205k
    NEXT;
10130
205k
    cur=CUR;
10131
205k
    if (cur != '.') {
10132
51
  xmlFree(buf);
10133
51
  return(NULL);
10134
51
    }
10135
205k
    buf[len++] = cur;
10136
205k
    NEXT;
10137
205k
    cur=CUR;
10138
416k
    while ((cur >= '0') && (cur <= '9')) {
10139
211k
  if (len + 1 >= size) {
10140
309
      xmlChar *tmp;
10141
309
            int newSize;
10142
10143
309
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
10144
309
            if (newSize < 0) {
10145
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "VersionNum");
10146
0
                xmlFree(buf);
10147
0
                return(NULL);
10148
0
            }
10149
309
      tmp = xmlRealloc(buf, newSize);
10150
309
      if (tmp == NULL) {
10151
0
    xmlErrMemory(ctxt);
10152
0
          xmlFree(buf);
10153
0
    return(NULL);
10154
0
      }
10155
309
      buf = tmp;
10156
309
            size = newSize;
10157
309
  }
10158
211k
  buf[len++] = cur;
10159
211k
  NEXT;
10160
211k
  cur=CUR;
10161
211k
    }
10162
205k
    buf[len] = 0;
10163
205k
    return(buf);
10164
205k
}
10165
10166
/**
10167
 * xmlParseVersionInfo:
10168
 * @ctxt:  an XML parser context
10169
 *
10170
 * DEPRECATED: Internal function, don't use.
10171
 *
10172
 * parse the XML version.
10173
 *
10174
 * [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ")
10175
 *
10176
 * [25] Eq ::= S? '=' S?
10177
 *
10178
 * Returns the version string, e.g. "1.0"
10179
 */
10180
10181
xmlChar *
10182
206k
xmlParseVersionInfo(xmlParserCtxtPtr ctxt) {
10183
206k
    xmlChar *version = NULL;
10184
10185
206k
    if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) {
10186
205k
  SKIP(7);
10187
205k
  SKIP_BLANKS;
10188
205k
  if (RAW != '=') {
10189
58
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10190
58
      return(NULL);
10191
58
        }
10192
205k
  NEXT;
10193
205k
  SKIP_BLANKS;
10194
205k
  if (RAW == '"') {
10195
204k
      NEXT;
10196
204k
      version = xmlParseVersionNum(ctxt);
10197
204k
      if (RAW != '"') {
10198
211
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10199
211
      } else
10200
204k
          NEXT;
10201
204k
  } else if (RAW == '\''){
10202
462
      NEXT;
10203
462
      version = xmlParseVersionNum(ctxt);
10204
462
      if (RAW != '\'') {
10205
17
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10206
17
      } else
10207
445
          NEXT;
10208
462
  } else {
10209
77
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10210
77
  }
10211
205k
    }
10212
206k
    return(version);
10213
206k
}
10214
10215
/**
10216
 * xmlParseEncName:
10217
 * @ctxt:  an XML parser context
10218
 *
10219
 * DEPRECATED: Internal function, don't use.
10220
 *
10221
 * parse the XML encoding name
10222
 *
10223
 * [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')*
10224
 *
10225
 * Returns the encoding name value or NULL
10226
 */
10227
xmlChar *
10228
180k
xmlParseEncName(xmlParserCtxtPtr ctxt) {
10229
180k
    xmlChar *buf = NULL;
10230
180k
    int len = 0;
10231
180k
    int size = 10;
10232
180k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
10233
174k
                    XML_MAX_TEXT_LENGTH :
10234
180k
                    XML_MAX_NAME_LENGTH;
10235
180k
    xmlChar cur;
10236
10237
180k
    cur = CUR;
10238
180k
    if (((cur >= 'a') && (cur <= 'z')) ||
10239
180k
        ((cur >= 'A') && (cur <= 'Z'))) {
10240
180k
  buf = xmlMalloc(size);
10241
180k
  if (buf == NULL) {
10242
0
      xmlErrMemory(ctxt);
10243
0
      return(NULL);
10244
0
  }
10245
10246
180k
  buf[len++] = cur;
10247
180k
  NEXT;
10248
180k
  cur = CUR;
10249
925k
  while (((cur >= 'a') && (cur <= 'z')) ||
10250
925k
         ((cur >= 'A') && (cur <= 'Z')) ||
10251
925k
         ((cur >= '0') && (cur <= '9')) ||
10252
925k
         (cur == '.') || (cur == '_') ||
10253
925k
         (cur == '-')) {
10254
744k
      if (len + 1 >= size) {
10255
1.40k
          xmlChar *tmp;
10256
1.40k
                int newSize;
10257
10258
1.40k
                newSize = xmlGrowCapacity(size, 1, 1, maxLength);
10259
1.40k
                if (newSize < 0) {
10260
0
                    xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName");
10261
0
                    xmlFree(buf);
10262
0
                    return(NULL);
10263
0
                }
10264
1.40k
    tmp = xmlRealloc(buf, newSize);
10265
1.40k
    if (tmp == NULL) {
10266
0
        xmlErrMemory(ctxt);
10267
0
        xmlFree(buf);
10268
0
        return(NULL);
10269
0
    }
10270
1.40k
    buf = tmp;
10271
1.40k
                size = newSize;
10272
1.40k
      }
10273
744k
      buf[len++] = cur;
10274
744k
      NEXT;
10275
744k
      cur = CUR;
10276
744k
        }
10277
180k
  buf[len] = 0;
10278
180k
    } else {
10279
38
  xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL);
10280
38
    }
10281
180k
    return(buf);
10282
180k
}
10283
10284
/**
10285
 * xmlParseEncodingDecl:
10286
 * @ctxt:  an XML parser context
10287
 *
10288
 * DEPRECATED: Internal function, don't use.
10289
 *
10290
 * parse the XML encoding declaration
10291
 *
10292
 * [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' |  "'" EncName "'")
10293
 *
10294
 * this setups the conversion filters.
10295
 *
10296
 * Returns the encoding value or NULL
10297
 */
10298
10299
const xmlChar *
10300
183k
xmlParseEncodingDecl(xmlParserCtxtPtr ctxt) {
10301
183k
    xmlChar *encoding = NULL;
10302
10303
183k
    SKIP_BLANKS;
10304
183k
    if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0)
10305
2.77k
        return(NULL);
10306
10307
180k
    SKIP(8);
10308
180k
    SKIP_BLANKS;
10309
180k
    if (RAW != '=') {
10310
36
        xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10311
36
        return(NULL);
10312
36
    }
10313
180k
    NEXT;
10314
180k
    SKIP_BLANKS;
10315
180k
    if (RAW == '"') {
10316
180k
        NEXT;
10317
180k
        encoding = xmlParseEncName(ctxt);
10318
180k
        if (RAW != '"') {
10319
165
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10320
165
            xmlFree((xmlChar *) encoding);
10321
165
            return(NULL);
10322
165
        } else
10323
180k
            NEXT;
10324
180k
    } else if (RAW == '\''){
10325
39
        NEXT;
10326
39
        encoding = xmlParseEncName(ctxt);
10327
39
        if (RAW != '\'') {
10328
18
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10329
18
            xmlFree((xmlChar *) encoding);
10330
18
            return(NULL);
10331
18
        } else
10332
21
            NEXT;
10333
39
    } else {
10334
22
        xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10335
22
    }
10336
10337
180k
    if (encoding == NULL)
10338
31
        return(NULL);
10339
10340
180k
    xmlSetDeclaredEncoding(ctxt, encoding);
10341
10342
180k
    return(ctxt->encoding);
10343
180k
}
10344
10345
/**
10346
 * xmlParseSDDecl:
10347
 * @ctxt:  an XML parser context
10348
 *
10349
 * DEPRECATED: Internal function, don't use.
10350
 *
10351
 * parse the XML standalone declaration
10352
 *
10353
 * [32] SDDecl ::= S 'standalone' Eq
10354
 *                 (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"'))
10355
 *
10356
 * [ VC: Standalone Document Declaration ]
10357
 * TODO The standalone document declaration must have the value "no"
10358
 * if any external markup declarations contain declarations of:
10359
 *  - attributes with default values, if elements to which these
10360
 *    attributes apply appear in the document without specifications
10361
 *    of values for these attributes, or
10362
 *  - entities (other than amp, lt, gt, apos, quot), if references
10363
 *    to those entities appear in the document, or
10364
 *  - attributes with values subject to normalization, where the
10365
 *    attribute appears in the document with a value which will change
10366
 *    as a result of normalization, or
10367
 *  - element types with element content, if white space occurs directly
10368
 *    within any instance of those types.
10369
 *
10370
 * Returns:
10371
 *   1 if standalone="yes"
10372
 *   0 if standalone="no"
10373
 *  -2 if standalone attribute is missing or invalid
10374
 *    (A standalone value of -2 means that the XML declaration was found,
10375
 *     but no value was specified for the standalone attribute).
10376
 */
10377
10378
int
10379
119k
xmlParseSDDecl(xmlParserCtxtPtr ctxt) {
10380
119k
    int standalone = -2;
10381
10382
119k
    SKIP_BLANKS;
10383
119k
    if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) {
10384
117k
  SKIP(10);
10385
117k
        SKIP_BLANKS;
10386
117k
  if (RAW != '=') {
10387
12
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10388
12
      return(standalone);
10389
12
        }
10390
117k
  NEXT;
10391
117k
  SKIP_BLANKS;
10392
117k
        if (RAW == '\''){
10393
80
      NEXT;
10394
80
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10395
65
          standalone = 0;
10396
65
                SKIP(2);
10397
65
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10398
15
                 (NXT(2) == 's')) {
10399
4
          standalone = 1;
10400
4
    SKIP(3);
10401
11
            } else {
10402
11
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10403
11
      }
10404
80
      if (RAW != '\'') {
10405
14
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10406
14
      } else
10407
66
          NEXT;
10408
116k
  } else if (RAW == '"'){
10409
116k
      NEXT;
10410
116k
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10411
520
          standalone = 0;
10412
520
    SKIP(2);
10413
116k
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10414
116k
                 (NXT(2) == 's')) {
10415
116k
          standalone = 1;
10416
116k
                SKIP(3);
10417
116k
            } else {
10418
40
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10419
40
      }
10420
116k
      if (RAW != '"') {
10421
52
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10422
52
      } else
10423
116k
          NEXT;
10424
116k
  } else {
10425
10
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10426
10
        }
10427
117k
    }
10428
119k
    return(standalone);
10429
119k
}
10430
10431
/**
10432
 * xmlParseXMLDecl:
10433
 * @ctxt:  an XML parser context
10434
 *
10435
 * DEPRECATED: Internal function, don't use.
10436
 *
10437
 * parse an XML declaration header
10438
 *
10439
 * [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>'
10440
 */
10441
10442
void
10443
206k
xmlParseXMLDecl(xmlParserCtxtPtr ctxt) {
10444
206k
    xmlChar *version;
10445
10446
    /*
10447
     * This value for standalone indicates that the document has an
10448
     * XML declaration but it does not have a standalone attribute.
10449
     * It will be overwritten later if a standalone attribute is found.
10450
     */
10451
10452
206k
    ctxt->standalone = -2;
10453
10454
    /*
10455
     * We know that '<?xml' is here.
10456
     */
10457
206k
    SKIP(5);
10458
10459
206k
    if (!IS_BLANK_CH(RAW)) {
10460
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
10461
0
                 "Blank needed after '<?xml'\n");
10462
0
    }
10463
206k
    SKIP_BLANKS;
10464
10465
    /*
10466
     * We must have the VersionInfo here.
10467
     */
10468
206k
    version = xmlParseVersionInfo(ctxt);
10469
206k
    if (version == NULL) {
10470
1.33k
  xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL);
10471
205k
    } else {
10472
205k
  if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) {
10473
      /*
10474
       * Changed here for XML-1.0 5th edition
10475
       */
10476
5.70k
      if (ctxt->options & XML_PARSE_OLD10) {
10477
0
    xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10478
0
                "Unsupported version '%s'\n",
10479
0
                version);
10480
5.70k
      } else {
10481
5.70k
          if ((version[0] == '1') && ((version[1] == '.'))) {
10482
5.52k
        xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION,
10483
5.52k
                      "Unsupported version '%s'\n",
10484
5.52k
          version, NULL);
10485
5.52k
    } else {
10486
176
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10487
176
              "Unsupported version '%s'\n",
10488
176
              version);
10489
176
    }
10490
5.70k
      }
10491
5.70k
  }
10492
205k
  if (ctxt->version != NULL)
10493
0
      xmlFree((void *) ctxt->version);
10494
205k
  ctxt->version = version;
10495
205k
    }
10496
10497
    /*
10498
     * We may have the encoding declaration
10499
     */
10500
206k
    if (!IS_BLANK_CH(RAW)) {
10501
24.7k
        if ((RAW == '?') && (NXT(1) == '>')) {
10502
23.3k
      SKIP(2);
10503
23.3k
      return;
10504
23.3k
  }
10505
1.42k
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10506
1.42k
    }
10507
183k
    xmlParseEncodingDecl(ctxt);
10508
10509
    /*
10510
     * We may have the standalone status.
10511
     */
10512
183k
    if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) {
10513
63.6k
        if ((RAW == '?') && (NXT(1) == '>')) {
10514
63.5k
      SKIP(2);
10515
63.5k
      return;
10516
63.5k
  }
10517
56
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10518
56
    }
10519
10520
    /*
10521
     * We can grow the input buffer freely at that point
10522
     */
10523
119k
    GROW;
10524
10525
119k
    SKIP_BLANKS;
10526
119k
    ctxt->standalone = xmlParseSDDecl(ctxt);
10527
10528
119k
    SKIP_BLANKS;
10529
119k
    if ((RAW == '?') && (NXT(1) == '>')) {
10530
117k
        SKIP(2);
10531
117k
    } else if (RAW == '>') {
10532
        /* Deprecated old WD ... */
10533
65
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10534
65
  NEXT;
10535
2.01k
    } else {
10536
2.01k
        int c;
10537
10538
2.01k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10539
921k
        while ((PARSER_STOPPED(ctxt) == 0) &&
10540
921k
               ((c = CUR) != 0)) {
10541
920k
            NEXT;
10542
920k
            if (c == '>')
10543
1.40k
                break;
10544
920k
        }
10545
2.01k
    }
10546
119k
}
10547
10548
/**
10549
 * xmlCtxtGetVersion:
10550
 * @ctxt:  parser context
10551
 *
10552
 * Available since 2.14.0.
10553
 *
10554
 * Returns the version from the XML declaration.
10555
 */
10556
const xmlChar *
10557
0
xmlCtxtGetVersion(xmlParserCtxtPtr ctxt) {
10558
0
    if (ctxt == NULL)
10559
0
        return(NULL);
10560
10561
0
    return(ctxt->version);
10562
0
}
10563
10564
/**
10565
 * xmlCtxtGetStandalone:
10566
 * @ctxt:  parser context
10567
 *
10568
 * Available since 2.14.0.
10569
 *
10570
 * Returns the value from the standalone document declaration.
10571
 */
10572
int
10573
0
xmlCtxtGetStandalone(xmlParserCtxtPtr ctxt) {
10574
0
    if (ctxt == NULL)
10575
0
        return(0);
10576
10577
0
    return(ctxt->standalone);
10578
0
}
10579
10580
/**
10581
 * xmlParseMisc:
10582
 * @ctxt:  an XML parser context
10583
 *
10584
 * DEPRECATED: Internal function, don't use.
10585
 *
10586
 * parse an XML Misc* optional field.
10587
 *
10588
 * [27] Misc ::= Comment | PI |  S
10589
 */
10590
10591
void
10592
13.4k
xmlParseMisc(xmlParserCtxtPtr ctxt) {
10593
13.5k
    while (PARSER_STOPPED(ctxt) == 0) {
10594
13.5k
        SKIP_BLANKS;
10595
13.5k
        GROW;
10596
13.5k
        if ((RAW == '<') && (NXT(1) == '?')) {
10597
61
      xmlParsePI(ctxt);
10598
13.4k
        } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) {
10599
0
      xmlParseComment(ctxt);
10600
13.4k
        } else {
10601
13.4k
            break;
10602
13.4k
        }
10603
13.5k
    }
10604
13.4k
}
10605
10606
static void
10607
178k
xmlFinishDocument(xmlParserCtxtPtr ctxt) {
10608
178k
    xmlDocPtr doc;
10609
10610
    /*
10611
     * SAX: end of the document processing.
10612
     */
10613
178k
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10614
6.72k
        ctxt->sax->endDocument(ctxt->userData);
10615
10616
178k
    doc = ctxt->myDoc;
10617
178k
    if (doc != NULL) {
10618
6.73k
        if (ctxt->wellFormed) {
10619
6.42k
            doc->properties |= XML_DOC_WELLFORMED;
10620
6.42k
            if (ctxt->valid)
10621
6.42k
                doc->properties |= XML_DOC_DTDVALID;
10622
6.42k
            if (ctxt->nsWellFormed)
10623
6.41k
                doc->properties |= XML_DOC_NSVALID;
10624
6.42k
        }
10625
10626
6.73k
        if (ctxt->options & XML_PARSE_OLD10)
10627
0
            doc->properties |= XML_DOC_OLD10;
10628
10629
        /*
10630
         * Remove locally kept entity definitions if the tree was not built
10631
         */
10632
6.73k
  if (xmlStrEqual(doc->version, SAX_COMPAT_MODE)) {
10633
15
            xmlFreeDoc(doc);
10634
15
            ctxt->myDoc = NULL;
10635
15
        }
10636
6.73k
    }
10637
178k
}
10638
10639
/**
10640
 * xmlParseDocument:
10641
 * @ctxt:  an XML parser context
10642
 *
10643
 * Parse an XML document and invoke the SAX handlers. This is useful
10644
 * if you're only interested in custom SAX callbacks. If you want a
10645
 * document tree, use xmlCtxtParseDocument.
10646
 *
10647
 * Returns 0, -1 in case of error.
10648
 */
10649
10650
int
10651
6.77k
xmlParseDocument(xmlParserCtxtPtr ctxt) {
10652
6.77k
    if ((ctxt == NULL) || (ctxt->input == NULL))
10653
0
        return(-1);
10654
10655
6.77k
    GROW;
10656
10657
    /*
10658
     * SAX: detecting the level.
10659
     */
10660
6.77k
    xmlCtxtInitializeLate(ctxt);
10661
10662
6.77k
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10663
6.77k
        ctxt->sax->setDocumentLocator(ctxt->userData,
10664
6.77k
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10665
6.77k
    }
10666
10667
6.77k
    xmlDetectEncoding(ctxt);
10668
10669
6.77k
    if (CUR == 0) {
10670
47
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10671
47
  return(-1);
10672
47
    }
10673
10674
6.72k
    GROW;
10675
6.72k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
10676
10677
  /*
10678
   * Note that we will switch encoding on the fly.
10679
   */
10680
6.39k
  xmlParseXMLDecl(ctxt);
10681
6.39k
  SKIP_BLANKS;
10682
6.39k
    } else {
10683
330
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10684
330
        if (ctxt->version == NULL) {
10685
0
            xmlErrMemory(ctxt);
10686
0
            return(-1);
10687
0
        }
10688
330
    }
10689
6.72k
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10690
6.71k
        ctxt->sax->startDocument(ctxt->userData);
10691
6.72k
    if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) &&
10692
6.72k
        (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) {
10693
0
  ctxt->myDoc->compression = ctxt->input->buf->compressed;
10694
0
    }
10695
10696
    /*
10697
     * The Misc part of the Prolog
10698
     */
10699
6.72k
    xmlParseMisc(ctxt);
10700
10701
    /*
10702
     * Then possibly doc type declaration(s) and more Misc
10703
     * (doctypedecl Misc*)?
10704
     */
10705
6.72k
    GROW;
10706
6.72k
    if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) {
10707
10708
0
  ctxt->inSubset = 1;
10709
0
  xmlParseDocTypeDecl(ctxt);
10710
0
  if (RAW == '[') {
10711
0
      xmlParseInternalSubset(ctxt);
10712
0
  } else if (RAW == '>') {
10713
0
            NEXT;
10714
0
        }
10715
10716
  /*
10717
   * Create and update the external subset.
10718
   */
10719
0
  ctxt->inSubset = 2;
10720
0
  if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) &&
10721
0
      (!ctxt->disableSAX))
10722
0
      ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
10723
0
                                ctxt->extSubSystem, ctxt->extSubURI);
10724
0
  ctxt->inSubset = 0;
10725
10726
0
        xmlCleanSpecialAttr(ctxt);
10727
10728
0
  xmlParseMisc(ctxt);
10729
0
    }
10730
10731
    /*
10732
     * Time to start parsing the tree itself
10733
     */
10734
6.72k
    GROW;
10735
6.72k
    if (RAW != '<') {
10736
1
        if (ctxt->wellFormed)
10737
0
            xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10738
0
                           "Start tag expected, '<' not found\n");
10739
6.72k
    } else {
10740
6.72k
  xmlParseElement(ctxt);
10741
10742
  /*
10743
   * The Misc part at the end
10744
   */
10745
6.72k
  xmlParseMisc(ctxt);
10746
10747
6.72k
        xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
10748
6.72k
    }
10749
10750
6.72k
    ctxt->instate = XML_PARSER_EOF;
10751
6.72k
    xmlFinishDocument(ctxt);
10752
10753
6.72k
    if (! ctxt->wellFormed) {
10754
310
  ctxt->valid = 0;
10755
310
  return(-1);
10756
310
    }
10757
10758
6.41k
    return(0);
10759
6.72k
}
10760
10761
/**
10762
 * xmlParseExtParsedEnt:
10763
 * @ctxt:  an XML parser context
10764
 *
10765
 * DEPRECATED: Internal function, don't use.
10766
 *
10767
 * parse a general parsed entity
10768
 * An external general parsed entity is well-formed if it matches the
10769
 * production labeled extParsedEnt.
10770
 *
10771
 * [78] extParsedEnt ::= TextDecl? content
10772
 *
10773
 * Returns 0, -1 in case of error. the parser context is augmented
10774
 *                as a result of the parsing.
10775
 */
10776
10777
int
10778
0
xmlParseExtParsedEnt(xmlParserCtxtPtr ctxt) {
10779
0
    if ((ctxt == NULL) || (ctxt->input == NULL))
10780
0
        return(-1);
10781
10782
0
    xmlCtxtInitializeLate(ctxt);
10783
10784
0
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10785
0
        ctxt->sax->setDocumentLocator(ctxt->userData,
10786
0
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10787
0
    }
10788
10789
0
    xmlDetectEncoding(ctxt);
10790
10791
0
    if (CUR == 0) {
10792
0
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10793
0
    }
10794
10795
    /*
10796
     * Check for the XMLDecl in the Prolog.
10797
     */
10798
0
    GROW;
10799
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
10800
10801
  /*
10802
   * Note that we will switch encoding on the fly.
10803
   */
10804
0
  xmlParseXMLDecl(ctxt);
10805
0
  SKIP_BLANKS;
10806
0
    } else {
10807
0
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10808
0
    }
10809
0
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10810
0
        ctxt->sax->startDocument(ctxt->userData);
10811
10812
    /*
10813
     * Doing validity checking on chunk doesn't make sense
10814
     */
10815
0
    ctxt->options &= ~XML_PARSE_DTDVALID;
10816
0
    ctxt->validate = 0;
10817
0
    ctxt->depth = 0;
10818
10819
0
    xmlParseContentInternal(ctxt);
10820
10821
0
    if (ctxt->input->cur < ctxt->input->end)
10822
0
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
10823
10824
    /*
10825
     * SAX: end of the document processing.
10826
     */
10827
0
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10828
0
        ctxt->sax->endDocument(ctxt->userData);
10829
10830
0
    if (! ctxt->wellFormed) return(-1);
10831
0
    return(0);
10832
0
}
10833
10834
#ifdef LIBXML_PUSH_ENABLED
10835
/************************************************************************
10836
 *                  *
10837
 *    Progressive parsing interfaces        *
10838
 *                  *
10839
 ************************************************************************/
10840
10841
/**
10842
 * xmlParseLookupChar:
10843
 * @ctxt:  an XML parser context
10844
 * @c:  character
10845
 *
10846
 * Check whether the input buffer contains a character.
10847
 */
10848
static int
10849
5.09M
xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) {
10850
5.09M
    const xmlChar *cur;
10851
10852
5.09M
    if (ctxt->checkIndex == 0) {
10853
5.09M
        cur = ctxt->input->cur + 1;
10854
5.09M
    } else {
10855
723
        cur = ctxt->input->cur + ctxt->checkIndex;
10856
723
    }
10857
10858
5.09M
    if (memchr(cur, c, ctxt->input->end - cur) == NULL) {
10859
892
        size_t index = ctxt->input->end - ctxt->input->cur;
10860
10861
892
        if (index > LONG_MAX) {
10862
0
            ctxt->checkIndex = 0;
10863
0
            return(1);
10864
0
        }
10865
892
        ctxt->checkIndex = index;
10866
892
        return(0);
10867
5.09M
    } else {
10868
5.09M
        ctxt->checkIndex = 0;
10869
5.09M
        return(1);
10870
5.09M
    }
10871
5.09M
}
10872
10873
/**
10874
 * xmlParseLookupString:
10875
 * @ctxt:  an XML parser context
10876
 * @startDelta: delta to apply at the start
10877
 * @str:  string
10878
 * @strLen:  length of string
10879
 *
10880
 * Check whether the input buffer contains a string.
10881
 */
10882
static const xmlChar *
10883
xmlParseLookupString(xmlParserCtxtPtr ctxt, size_t startDelta,
10884
106k
                     const char *str, size_t strLen) {
10885
106k
    const xmlChar *cur, *term;
10886
10887
106k
    if (ctxt->checkIndex == 0) {
10888
105k
        cur = ctxt->input->cur + startDelta;
10889
105k
    } else {
10890
1.11k
        cur = ctxt->input->cur + ctxt->checkIndex;
10891
1.11k
    }
10892
10893
106k
    term = BAD_CAST strstr((const char *) cur, str);
10894
106k
    if (term == NULL) {
10895
1.53k
        const xmlChar *end = ctxt->input->end;
10896
1.53k
        size_t index;
10897
10898
        /* Rescan (strLen - 1) characters. */
10899
1.53k
        if ((size_t) (end - cur) < strLen)
10900
27
            end = cur;
10901
1.50k
        else
10902
1.50k
            end -= strLen - 1;
10903
1.53k
        index = end - ctxt->input->cur;
10904
1.53k
        if (index > LONG_MAX) {
10905
0
            ctxt->checkIndex = 0;
10906
0
            return(ctxt->input->end - strLen);
10907
0
        }
10908
1.53k
        ctxt->checkIndex = index;
10909
104k
    } else {
10910
104k
        ctxt->checkIndex = 0;
10911
104k
    }
10912
10913
106k
    return(term);
10914
106k
}
10915
10916
/**
10917
 * xmlParseLookupCharData:
10918
 * @ctxt:  an XML parser context
10919
 *
10920
 * Check whether the input buffer contains terminated char data.
10921
 */
10922
static int
10923
93.7k
xmlParseLookupCharData(xmlParserCtxtPtr ctxt) {
10924
93.7k
    const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex;
10925
93.7k
    const xmlChar *end = ctxt->input->end;
10926
93.7k
    size_t index;
10927
10928
1.12M
    while (cur < end) {
10929
1.12M
        if ((*cur == '<') || (*cur == '&')) {
10930
91.5k
            ctxt->checkIndex = 0;
10931
91.5k
            return(1);
10932
91.5k
        }
10933
1.03M
        cur++;
10934
1.03M
    }
10935
10936
2.25k
    index = cur - ctxt->input->cur;
10937
2.25k
    if (index > LONG_MAX) {
10938
0
        ctxt->checkIndex = 0;
10939
0
        return(1);
10940
0
    }
10941
2.25k
    ctxt->checkIndex = index;
10942
2.25k
    return(0);
10943
2.25k
}
10944
10945
/**
10946
 * xmlParseLookupGt:
10947
 * @ctxt:  an XML parser context
10948
 *
10949
 * Check whether there's enough data in the input buffer to finish parsing
10950
 * a start tag. This has to take quotes into account.
10951
 */
10952
static int
10953
19.8M
xmlParseLookupGt(xmlParserCtxtPtr ctxt) {
10954
19.8M
    const xmlChar *cur;
10955
19.8M
    const xmlChar *end = ctxt->input->end;
10956
19.8M
    int state = ctxt->endCheckState;
10957
19.8M
    size_t index;
10958
10959
19.8M
    if (ctxt->checkIndex == 0)
10960
19.8M
        cur = ctxt->input->cur + 1;
10961
7.41k
    else
10962
7.41k
        cur = ctxt->input->cur + ctxt->checkIndex;
10963
10964
567M
    while (cur < end) {
10965
567M
        if (state) {
10966
161M
            if (*cur == state)
10967
14.3M
                state = 0;
10968
405M
        } else if (*cur == '\'' || *cur == '"') {
10969
14.3M
            state = *cur;
10970
391M
        } else if (*cur == '>') {
10971
19.8M
            ctxt->checkIndex = 0;
10972
19.8M
            ctxt->endCheckState = 0;
10973
19.8M
            return(1);
10974
19.8M
        }
10975
547M
        cur++;
10976
547M
    }
10977
10978
12.3k
    index = cur - ctxt->input->cur;
10979
12.3k
    if (index > LONG_MAX) {
10980
0
        ctxt->checkIndex = 0;
10981
0
        ctxt->endCheckState = 0;
10982
0
        return(1);
10983
0
    }
10984
12.3k
    ctxt->checkIndex = index;
10985
12.3k
    ctxt->endCheckState = state;
10986
12.3k
    return(0);
10987
12.3k
}
10988
10989
/**
10990
 * xmlParseLookupInternalSubset:
10991
 * @ctxt:  an XML parser context
10992
 *
10993
 * Check whether there's enough data in the input buffer to finish parsing
10994
 * the internal subset.
10995
 */
10996
static int
10997
836
xmlParseLookupInternalSubset(xmlParserCtxtPtr ctxt) {
10998
    /*
10999
     * Sorry, but progressive parsing of the internal subset is not
11000
     * supported. We first check that the full content of the internal
11001
     * subset is available and parsing is launched only at that point.
11002
     * Internal subset ends with "']' S? '>'" in an unescaped section and
11003
     * not in a ']]>' sequence which are conditional sections.
11004
     */
11005
836
    const xmlChar *cur, *start;
11006
836
    const xmlChar *end = ctxt->input->end;
11007
836
    int state = ctxt->endCheckState;
11008
836
    size_t index;
11009
11010
836
    if (ctxt->checkIndex == 0) {
11011
652
        cur = ctxt->input->cur + 1;
11012
652
    } else {
11013
184
        cur = ctxt->input->cur + ctxt->checkIndex;
11014
184
    }
11015
836
    start = cur;
11016
11017
15.8M
    while (cur < end) {
11018
15.8M
        if (state == '-') {
11019
2.54M
            if ((*cur == '-') &&
11020
2.54M
                (cur[1] == '-') &&
11021
2.54M
                (cur[2] == '>')) {
11022
28.9k
                state = 0;
11023
28.9k
                cur += 3;
11024
28.9k
                start = cur;
11025
28.9k
                continue;
11026
28.9k
            }
11027
2.54M
        }
11028
13.2M
        else if (state == ']') {
11029
15.0k
            if (*cur == '>') {
11030
128
                ctxt->checkIndex = 0;
11031
128
                ctxt->endCheckState = 0;
11032
128
                return(1);
11033
128
            }
11034
14.9k
            if (IS_BLANK_CH(*cur)) {
11035
4.53k
                state = ' ';
11036
10.4k
            } else if (*cur != ']') {
11037
4.47k
                state = 0;
11038
4.47k
                start = cur;
11039
4.47k
                continue;
11040
4.47k
            }
11041
14.9k
        }
11042
13.2M
        else if (state == ' ') {
11043
19.6k
            if (*cur == '>') {
11044
3
                ctxt->checkIndex = 0;
11045
3
                ctxt->endCheckState = 0;
11046
3
                return(1);
11047
3
            }
11048
19.6k
            if (!IS_BLANK_CH(*cur)) {
11049
4.53k
                state = 0;
11050
4.53k
                start = cur;
11051
4.53k
                continue;
11052
4.53k
            }
11053
19.6k
        }
11054
13.2M
        else if (state != 0) {
11055
8.87M
            if (*cur == state) {
11056
79.4k
                state = 0;
11057
79.4k
                start = cur + 1;
11058
79.4k
            }
11059
8.87M
        }
11060
4.35M
        else if (*cur == '<') {
11061
130k
            if ((cur[1] == '!') &&
11062
130k
                (cur[2] == '-') &&
11063
130k
                (cur[3] == '-')) {
11064
29.0k
                state = '-';
11065
29.0k
                cur += 4;
11066
                /* Don't treat <!--> as comment */
11067
29.0k
                start = cur;
11068
29.0k
                continue;
11069
29.0k
            }
11070
130k
        }
11071
4.22M
        else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) {
11072
88.8k
            state = *cur;
11073
88.8k
        }
11074
11075
15.7M
        cur++;
11076
15.7M
    }
11077
11078
    /*
11079
     * Rescan the three last characters to detect "<!--" and "-->"
11080
     * split across chunks.
11081
     */
11082
705
    if ((state == 0) || (state == '-')) {
11083
334
        if (cur - start < 3)
11084
36
            cur = start;
11085
298
        else
11086
298
            cur -= 3;
11087
334
    }
11088
705
    index = cur - ctxt->input->cur;
11089
705
    if (index > LONG_MAX) {
11090
0
        ctxt->checkIndex = 0;
11091
0
        ctxt->endCheckState = 0;
11092
0
        return(1);
11093
0
    }
11094
705
    ctxt->checkIndex = index;
11095
705
    ctxt->endCheckState = state;
11096
705
    return(0);
11097
705
}
11098
11099
/**
11100
 * xmlParseTryOrFinish:
11101
 * @ctxt:  an XML parser context
11102
 * @terminate:  last chunk indicator
11103
 *
11104
 * Try to progress on parsing
11105
 *
11106
 * Returns zero if no parsing was possible
11107
 */
11108
static int
11109
291k
xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) {
11110
291k
    int ret = 0;
11111
291k
    size_t avail;
11112
291k
    xmlChar cur, next;
11113
11114
291k
    if (ctxt->input == NULL)
11115
0
        return(0);
11116
11117
291k
    if ((ctxt->input != NULL) &&
11118
291k
        (ctxt->input->cur - ctxt->input->base > 4096)) {
11119
22.0k
        xmlParserShrink(ctxt);
11120
22.0k
    }
11121
11122
103M
    while (ctxt->disableSAX == 0) {
11123
103M
        avail = ctxt->input->end - ctxt->input->cur;
11124
103M
        if (avail < 1)
11125
162k
      goto done;
11126
103M
        switch (ctxt->instate) {
11127
1.39k
            case XML_PARSER_EOF:
11128
          /*
11129
     * Document parsing is done !
11130
     */
11131
1.39k
          goto done;
11132
265k
            case XML_PARSER_START:
11133
                /*
11134
                 * Very first chars read from the document flow.
11135
                 */
11136
265k
                if ((!terminate) && (avail < 4))
11137
0
                    goto done;
11138
11139
                /*
11140
                 * We need more bytes to detect EBCDIC code pages.
11141
                 * See xmlDetectEBCDIC.
11142
                 */
11143
265k
                if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) &&
11144
265k
                    (!terminate) && (avail < 200))
11145
0
                    goto done;
11146
11147
265k
                xmlDetectEncoding(ctxt);
11148
265k
                ctxt->instate = XML_PARSER_XML_DECL;
11149
265k
    break;
11150
11151
265k
            case XML_PARSER_XML_DECL:
11152
265k
    if ((!terminate) && (avail < 2))
11153
0
        goto done;
11154
265k
    cur = ctxt->input->cur[0];
11155
265k
    next = ctxt->input->cur[1];
11156
265k
          if ((cur == '<') && (next == '?')) {
11157
        /* PI or XML decl */
11158
219k
        if ((!terminate) &&
11159
219k
                        (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11160
215
      goto done;
11161
218k
        if ((ctxt->input->cur[2] == 'x') &&
11162
218k
      (ctxt->input->cur[3] == 'm') &&
11163
218k
      (ctxt->input->cur[4] == 'l') &&
11164
218k
      (IS_BLANK_CH(ctxt->input->cur[5]))) {
11165
200k
      ret += 5;
11166
200k
      xmlParseXMLDecl(ctxt);
11167
200k
        } else {
11168
18.6k
      ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
11169
18.6k
                        if (ctxt->version == NULL) {
11170
0
                            xmlErrMemory(ctxt);
11171
0
                            break;
11172
0
                        }
11173
18.6k
        }
11174
218k
    } else {
11175
46.7k
        ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
11176
46.7k
        if (ctxt->version == NULL) {
11177
0
            xmlErrMemory(ctxt);
11178
0
      break;
11179
0
        }
11180
46.7k
    }
11181
265k
                if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
11182
0
                    ctxt->sax->setDocumentLocator(ctxt->userData,
11183
0
                            (xmlSAXLocator *) &xmlDefaultSAXLocator);
11184
0
                }
11185
265k
                if ((ctxt->sax) && (ctxt->sax->startDocument) &&
11186
265k
                    (!ctxt->disableSAX))
11187
0
                    ctxt->sax->startDocument(ctxt->userData);
11188
265k
                ctxt->instate = XML_PARSER_MISC;
11189
265k
    break;
11190
34.7M
            case XML_PARSER_START_TAG: {
11191
34.7M
          const xmlChar *name;
11192
34.7M
    const xmlChar *prefix = NULL;
11193
34.7M
    const xmlChar *URI = NULL;
11194
34.7M
                int line = ctxt->input->line;
11195
34.7M
    int nbNs = 0;
11196
11197
34.7M
    if ((!terminate) && (avail < 2))
11198
2
        goto done;
11199
34.7M
    cur = ctxt->input->cur[0];
11200
34.7M
          if (cur != '<') {
11201
501
        xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
11202
501
                                   "Start tag expected, '<' not found");
11203
501
                    ctxt->instate = XML_PARSER_EOF;
11204
501
                    xmlFinishDocument(ctxt);
11205
501
        goto done;
11206
501
    }
11207
34.7M
    if ((!terminate) && (!xmlParseLookupGt(ctxt)))
11208
11.1k
                    goto done;
11209
34.7M
    if (ctxt->spaceNr == 0)
11210
0
        spacePush(ctxt, -1);
11211
34.7M
    else if (*ctxt->space == -2)
11212
9.44M
        spacePush(ctxt, -1);
11213
25.2M
    else
11214
25.2M
        spacePush(ctxt, *ctxt->space);
11215
34.7M
#ifdef LIBXML_SAX1_ENABLED
11216
34.7M
    if (ctxt->sax2)
11217
34.7M
#endif /* LIBXML_SAX1_ENABLED */
11218
34.7M
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
11219
956
#ifdef LIBXML_SAX1_ENABLED
11220
956
    else
11221
956
        name = xmlParseStartTag(ctxt);
11222
34.7M
#endif /* LIBXML_SAX1_ENABLED */
11223
34.7M
    if (name == NULL) {
11224
4.65k
        spacePop(ctxt);
11225
4.65k
                    ctxt->instate = XML_PARSER_EOF;
11226
4.65k
                    xmlFinishDocument(ctxt);
11227
4.65k
        goto done;
11228
4.65k
    }
11229
34.7M
#ifdef LIBXML_VALID_ENABLED
11230
    /*
11231
     * [ VC: Root Element Type ]
11232
     * The Name in the document type declaration must match
11233
     * the element type of the root element.
11234
     */
11235
34.7M
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
11236
34.7M
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
11237
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
11238
34.7M
#endif /* LIBXML_VALID_ENABLED */
11239
11240
    /*
11241
     * Check for an Empty Element.
11242
     */
11243
34.7M
    if ((RAW == '/') && (NXT(1) == '>')) {
11244
9.67M
        SKIP(2);
11245
11246
9.67M
        if (ctxt->sax2) {
11247
9.67M
      if ((ctxt->sax != NULL) &&
11248
9.67M
          (ctxt->sax->endElementNs != NULL) &&
11249
9.67M
          (!ctxt->disableSAX))
11250
9.67M
          ctxt->sax->endElementNs(ctxt->userData, name,
11251
9.67M
                                  prefix, URI);
11252
9.67M
      if (nbNs > 0)
11253
160k
          xmlParserNsPop(ctxt, nbNs);
11254
9.67M
#ifdef LIBXML_SAX1_ENABLED
11255
9.67M
        } else {
11256
1
      if ((ctxt->sax != NULL) &&
11257
1
          (ctxt->sax->endElement != NULL) &&
11258
1
          (!ctxt->disableSAX))
11259
0
          ctxt->sax->endElement(ctxt->userData, name);
11260
1
#endif /* LIBXML_SAX1_ENABLED */
11261
1
        }
11262
9.67M
        spacePop(ctxt);
11263
25.0M
    } else if (RAW == '>') {
11264
25.0M
        NEXT;
11265
25.0M
                    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
11266
25.0M
    } else {
11267
59.8k
        xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED,
11268
59.8k
           "Couldn't find end of Start Tag %s\n",
11269
59.8k
           name);
11270
59.8k
        nodePop(ctxt);
11271
59.8k
        spacePop(ctxt);
11272
59.8k
                    if (nbNs > 0)
11273
6.59k
                        xmlParserNsPop(ctxt, nbNs);
11274
59.8k
    }
11275
11276
34.7M
                if (ctxt->nameNr == 0)
11277
15.5k
                    ctxt->instate = XML_PARSER_EPILOG;
11278
34.7M
                else
11279
34.7M
                    ctxt->instate = XML_PARSER_CONTENT;
11280
34.7M
                break;
11281
34.7M
      }
11282
58.1M
            case XML_PARSER_CONTENT: {
11283
58.1M
    cur = ctxt->input->cur[0];
11284
11285
58.1M
    if (cur == '<') {
11286
44.4M
                    if ((!terminate) && (avail < 2))
11287
608
                        goto done;
11288
44.4M
        next = ctxt->input->cur[1];
11289
11290
44.4M
                    if (next == '/') {
11291
9.88M
                        ctxt->instate = XML_PARSER_END_TAG;
11292
9.88M
                        break;
11293
34.5M
                    } else if (next == '?') {
11294
15.0k
                        if ((!terminate) &&
11295
15.0k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11296
103
                            goto done;
11297
14.9k
                        xmlParsePI(ctxt);
11298
14.9k
                        ctxt->instate = XML_PARSER_CONTENT;
11299
14.9k
                        break;
11300
34.5M
                    } else if (next == '!') {
11301
75.6k
                        if ((!terminate) && (avail < 3))
11302
8
                            goto done;
11303
75.6k
                        next = ctxt->input->cur[2];
11304
11305
75.6k
                        if (next == '-') {
11306
60.4k
                            if ((!terminate) && (avail < 4))
11307
10
                                goto done;
11308
60.4k
                            if (ctxt->input->cur[3] == '-') {
11309
60.3k
                                if ((!terminate) &&
11310
60.3k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11311
136
                                    goto done;
11312
60.2k
                                xmlParseComment(ctxt);
11313
60.2k
                                ctxt->instate = XML_PARSER_CONTENT;
11314
60.2k
                                break;
11315
60.3k
                            }
11316
60.4k
                        } else if (next == '[') {
11317
15.0k
                            if ((!terminate) && (avail < 9))
11318
4
                                goto done;
11319
15.0k
                            if ((ctxt->input->cur[2] == '[') &&
11320
15.0k
                                (ctxt->input->cur[3] == 'C') &&
11321
15.0k
                                (ctxt->input->cur[4] == 'D') &&
11322
15.0k
                                (ctxt->input->cur[5] == 'A') &&
11323
15.0k
                                (ctxt->input->cur[6] == 'T') &&
11324
15.0k
                                (ctxt->input->cur[7] == 'A') &&
11325
15.0k
                                (ctxt->input->cur[8] == '[')) {
11326
14.9k
                                if ((!terminate) &&
11327
14.9k
                                    (!xmlParseLookupString(ctxt, 9, "]]>", 3)))
11328
936
                                    goto done;
11329
14.0k
                                ctxt->instate = XML_PARSER_CDATA_SECTION;
11330
14.0k
                                xmlParseCDSect(ctxt);
11331
14.0k
                                ctxt->instate = XML_PARSER_CONTENT;
11332
14.0k
                                break;
11333
14.9k
                            }
11334
15.0k
                        }
11335
75.6k
                    }
11336
44.4M
    } else if (cur == '&') {
11337
363k
        if ((!terminate) && (!xmlParseLookupChar(ctxt, ';')))
11338
80
      goto done;
11339
363k
        xmlParseReference(ctxt);
11340
363k
                    break;
11341
13.3M
    } else {
11342
        /* TODO Avoid the extra copy, handle directly !!! */
11343
        /*
11344
         * Goal of the following test is:
11345
         *  - minimize calls to the SAX 'character' callback
11346
         *    when they are mergeable
11347
         *  - handle an problem for isBlank when we only parse
11348
         *    a sequence of blank chars and the next one is
11349
         *    not available to check against '<' presence.
11350
         *  - tries to homogenize the differences in SAX
11351
         *    callbacks between the push and pull versions
11352
         *    of the parser.
11353
         */
11354
13.3M
        if (avail < XML_PARSER_BIG_BUFFER_SIZE) {
11355
516k
      if ((!terminate) && (!xmlParseLookupCharData(ctxt)))
11356
2.25k
          goto done;
11357
516k
                    }
11358
13.3M
                    ctxt->checkIndex = 0;
11359
13.3M
        xmlParseCharDataInternal(ctxt, !terminate);
11360
13.3M
                    break;
11361
13.3M
    }
11362
11363
34.4M
                ctxt->instate = XML_PARSER_START_TAG;
11364
34.4M
    break;
11365
58.1M
      }
11366
9.88M
            case XML_PARSER_END_TAG:
11367
9.88M
    if ((!terminate) && (!xmlParseLookupChar(ctxt, '>')))
11368
812
        goto done;
11369
9.88M
    if (ctxt->sax2) {
11370
9.88M
              xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
11371
9.88M
        nameNsPop(ctxt);
11372
9.88M
    }
11373
5
#ifdef LIBXML_SAX1_ENABLED
11374
5
      else
11375
5
        xmlParseEndTag1(ctxt, 0);
11376
9.88M
#endif /* LIBXML_SAX1_ENABLED */
11377
9.88M
    if (ctxt->nameNr == 0) {
11378
156k
        ctxt->instate = XML_PARSER_EPILOG;
11379
9.73M
    } else {
11380
9.73M
        ctxt->instate = XML_PARSER_CONTENT;
11381
9.73M
    }
11382
9.88M
    break;
11383
307k
            case XML_PARSER_MISC:
11384
310k
            case XML_PARSER_PROLOG:
11385
327k
            case XML_PARSER_EPILOG:
11386
327k
    SKIP_BLANKS;
11387
327k
                avail = ctxt->input->end - ctxt->input->cur;
11388
327k
    if (avail < 1)
11389
12.7k
        goto done;
11390
314k
    if (ctxt->input->cur[0] == '<') {
11391
313k
                    if ((!terminate) && (avail < 2))
11392
8
                        goto done;
11393
313k
                    next = ctxt->input->cur[1];
11394
313k
                    if (next == '?') {
11395
45.1k
                        if ((!terminate) &&
11396
45.1k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11397
74
                            goto done;
11398
45.0k
                        xmlParsePI(ctxt);
11399
45.0k
                        break;
11400
268k
                    } else if (next == '!') {
11401
13.4k
                        if ((!terminate) && (avail < 3))
11402
4
                            goto done;
11403
11404
13.4k
                        if (ctxt->input->cur[2] == '-') {
11405
5.36k
                            if ((!terminate) && (avail < 4))
11406
0
                                goto done;
11407
5.36k
                            if (ctxt->input->cur[3] == '-') {
11408
5.34k
                                if ((!terminate) &&
11409
5.34k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11410
71
                                    goto done;
11411
5.27k
                                xmlParseComment(ctxt);
11412
5.27k
                                break;
11413
5.34k
                            }
11414
8.09k
                        } else if (ctxt->instate == XML_PARSER_MISC) {
11415
8.07k
                            if ((!terminate) && (avail < 9))
11416
0
                                goto done;
11417
8.07k
                            if ((ctxt->input->cur[2] == 'D') &&
11418
8.07k
                                (ctxt->input->cur[3] == 'O') &&
11419
8.07k
                                (ctxt->input->cur[4] == 'C') &&
11420
8.07k
                                (ctxt->input->cur[5] == 'T') &&
11421
8.07k
                                (ctxt->input->cur[6] == 'Y') &&
11422
8.07k
                                (ctxt->input->cur[7] == 'P') &&
11423
8.07k
                                (ctxt->input->cur[8] == 'E')) {
11424
8.00k
                                if ((!terminate) && (!xmlParseLookupGt(ctxt)))
11425
70
                                    goto done;
11426
7.93k
                                ctxt->inSubset = 1;
11427
7.93k
                                xmlParseDocTypeDecl(ctxt);
11428
7.93k
                                if (RAW == '[') {
11429
7.18k
                                    ctxt->instate = XML_PARSER_DTD;
11430
7.18k
                                } else {
11431
745
                                    if (RAW == '>')
11432
453
                                        NEXT;
11433
                                    /*
11434
                                     * Create and update the external subset.
11435
                                     */
11436
745
                                    ctxt->inSubset = 2;
11437
745
                                    if ((ctxt->sax != NULL) &&
11438
745
                                        (!ctxt->disableSAX) &&
11439
745
                                        (ctxt->sax->externalSubset != NULL))
11440
0
                                        ctxt->sax->externalSubset(
11441
0
                                                ctxt->userData,
11442
0
                                                ctxt->intSubName,
11443
0
                                                ctxt->extSubSystem,
11444
0
                                                ctxt->extSubURI);
11445
745
                                    ctxt->inSubset = 0;
11446
745
                                    xmlCleanSpecialAttr(ctxt);
11447
745
                                    ctxt->instate = XML_PARSER_PROLOG;
11448
745
                                }
11449
7.93k
                                break;
11450
8.00k
                            }
11451
8.07k
                        }
11452
13.4k
                    }
11453
313k
                }
11454
11455
256k
                if (ctxt->instate == XML_PARSER_EPILOG) {
11456
1.54k
                    if (ctxt->errNo == XML_ERR_OK)
11457
329
                        xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL);
11458
1.54k
        ctxt->instate = XML_PARSER_EOF;
11459
1.54k
                    xmlFinishDocument(ctxt);
11460
254k
                } else {
11461
254k
        ctxt->instate = XML_PARSER_START_TAG;
11462
254k
    }
11463
256k
    break;
11464
7.73k
            case XML_PARSER_DTD: {
11465
7.73k
                if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt)))
11466
705
                    goto done;
11467
7.02k
    xmlParseInternalSubset(ctxt);
11468
7.02k
    ctxt->inSubset = 2;
11469
7.02k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
11470
7.02k
        (ctxt->sax->externalSubset != NULL))
11471
0
        ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
11472
0
          ctxt->extSubSystem, ctxt->extSubURI);
11473
7.02k
    ctxt->inSubset = 0;
11474
7.02k
    xmlCleanSpecialAttr(ctxt);
11475
7.02k
    ctxt->instate = XML_PARSER_PROLOG;
11476
7.02k
                break;
11477
7.73k
      }
11478
0
            default:
11479
0
                xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
11480
0
      "PP: internal error\n");
11481
0
    ctxt->instate = XML_PARSER_EOF;
11482
0
    break;
11483
103M
  }
11484
103M
    }
11485
291k
done:
11486
291k
    return(ret);
11487
291k
}
11488
11489
/**
11490
 * xmlParseChunk:
11491
 * @ctxt:  an XML parser context
11492
 * @chunk:  chunk of memory
11493
 * @size:  size of chunk in bytes
11494
 * @terminate:  last chunk indicator
11495
 *
11496
 * Parse a chunk of memory in push parser mode.
11497
 *
11498
 * Assumes that the parser context was initialized with
11499
 * xmlCreatePushParserCtxt.
11500
 *
11501
 * The last chunk, which will often be empty, must be marked with
11502
 * the @terminate flag. With the default SAX callbacks, the resulting
11503
 * document will be available in ctxt->myDoc. This pointer will not
11504
 * be freed when calling xmlFreeParserCtxt and must be freed by the
11505
 * caller. If the document isn't well-formed, it will still be returned
11506
 * in ctxt->myDoc.
11507
 *
11508
 * As an exception, xmlCtxtResetPush will free the document in
11509
 * ctxt->myDoc. So ctxt->myDoc should be set to NULL after extracting
11510
 * the document.
11511
 *
11512
 * Returns an xmlParserErrors code (0 on success).
11513
 */
11514
int
11515
xmlParseChunk(xmlParserCtxtPtr ctxt, const char *chunk, int size,
11516
291k
              int terminate) {
11517
291k
    size_t curBase;
11518
291k
    size_t maxLength;
11519
291k
    size_t pos;
11520
291k
    int end_in_lf = 0;
11521
291k
    int res;
11522
11523
291k
    if ((ctxt == NULL) || (size < 0))
11524
0
        return(XML_ERR_ARGUMENT);
11525
291k
    if ((chunk == NULL) && (size > 0))
11526
0
        return(XML_ERR_ARGUMENT);
11527
291k
    if ((ctxt->input == NULL) || (ctxt->input->buf == NULL))
11528
0
        return(XML_ERR_ARGUMENT);
11529
291k
    if (ctxt->disableSAX != 0)
11530
0
        return(ctxt->errNo);
11531
11532
291k
    ctxt->input->flags |= XML_INPUT_PROGRESSIVE;
11533
291k
    if (ctxt->instate == XML_PARSER_START)
11534
265k
        xmlCtxtInitializeLate(ctxt);
11535
291k
    if ((size > 0) && (chunk != NULL) && (!terminate) &&
11536
291k
        (chunk[size - 1] == '\r')) {
11537
195
  end_in_lf = 1;
11538
195
  size--;
11539
195
    }
11540
11541
    /*
11542
     * Also push an empty chunk to make sure that the raw buffer
11543
     * will be flushed if there is an encoder.
11544
     */
11545
291k
    pos = ctxt->input->cur - ctxt->input->base;
11546
291k
    res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk);
11547
291k
    xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11548
291k
    if (res < 0) {
11549
4
        xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11550
4
        xmlHaltParser(ctxt);
11551
4
        return(ctxt->errNo);
11552
4
    }
11553
11554
291k
    xmlParseTryOrFinish(ctxt, terminate);
11555
11556
291k
    curBase = ctxt->input->cur - ctxt->input->base;
11557
291k
    maxLength = (ctxt->options & XML_PARSE_HUGE) ?
11558
291k
                XML_MAX_HUGE_LENGTH :
11559
291k
                XML_MAX_LOOKUP_LIMIT;
11560
291k
    if (curBase > maxLength) {
11561
0
        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
11562
0
                    "Buffer size limit exceeded, try XML_PARSE_HUGE\n");
11563
0
        xmlHaltParser(ctxt);
11564
0
    }
11565
11566
291k
    if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0))
11567
97.4k
        return(ctxt->errNo);
11568
11569
194k
    if (end_in_lf == 1) {
11570
162
  pos = ctxt->input->cur - ctxt->input->base;
11571
162
  res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r");
11572
162
  xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11573
162
        if (res < 0) {
11574
0
            xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11575
0
            xmlHaltParser(ctxt);
11576
0
            return(ctxt->errNo);
11577
0
        }
11578
162
    }
11579
194k
    if (terminate) {
11580
  /*
11581
   * Check for termination
11582
   */
11583
166k
        if ((ctxt->instate != XML_PARSER_EOF) &&
11584
166k
            (ctxt->instate != XML_PARSER_EPILOG)) {
11585
9.62k
            if (ctxt->nameNr > 0) {
11586
9.45k
                const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
11587
9.45k
                int line = ctxt->pushTab[ctxt->nameNr - 1].line;
11588
9.45k
                xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
11589
9.45k
                        "Premature end of data in tag %s line %d\n",
11590
9.45k
                        name, line, NULL);
11591
9.45k
            } else if (ctxt->instate == XML_PARSER_START) {
11592
0
                xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
11593
168
            } else {
11594
168
                xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
11595
168
                               "Start tag expected, '<' not found\n");
11596
168
            }
11597
156k
        } else {
11598
156k
            xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
11599
156k
        }
11600
166k
  if (ctxt->instate != XML_PARSER_EOF) {
11601
165k
            ctxt->instate = XML_PARSER_EOF;
11602
165k
            xmlFinishDocument(ctxt);
11603
165k
  }
11604
166k
    }
11605
194k
    if (ctxt->wellFormed == 0)
11606
9.62k
  return((xmlParserErrors) ctxt->errNo);
11607
184k
    else
11608
184k
        return(0);
11609
194k
}
11610
11611
/************************************************************************
11612
 *                  *
11613
 *    I/O front end functions to the parser     *
11614
 *                  *
11615
 ************************************************************************/
11616
11617
/**
11618
 * xmlCreatePushParserCtxt:
11619
 * @sax:  a SAX handler (optional)
11620
 * @user_data:  user data for SAX callbacks (optional)
11621
 * @chunk:  initial chunk (optional, deprecated)
11622
 * @size:  size of initial chunk in bytes
11623
 * @filename:  file name or URI (optional)
11624
 *
11625
 * Create a parser context for using the XML parser in push mode.
11626
 * See xmlParseChunk.
11627
 *
11628
 * Passing an initial chunk is useless and deprecated.
11629
 *
11630
 * The push parser doesn't support recovery mode or the
11631
 * XML_PARSE_NOBLANKS option.
11632
 *
11633
 * @filename is used as base URI to fetch external entities and for
11634
 * error reports.
11635
 *
11636
 * Returns the new parser context or NULL if a memory allocation
11637
 * failed.
11638
 */
11639
11640
xmlParserCtxtPtr
11641
xmlCreatePushParserCtxt(xmlSAXHandlerPtr sax, void *user_data,
11642
265k
                        const char *chunk, int size, const char *filename) {
11643
265k
    xmlParserCtxtPtr ctxt;
11644
265k
    xmlParserInputPtr input;
11645
11646
265k
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11647
265k
    if (ctxt == NULL)
11648
0
  return(NULL);
11649
11650
265k
    ctxt->options &= ~XML_PARSE_NODICT;
11651
265k
    ctxt->dictNames = 1;
11652
11653
265k
    input = xmlNewPushInput(filename, chunk, size);
11654
265k
    if (input == NULL) {
11655
0
  xmlFreeParserCtxt(ctxt);
11656
0
  return(NULL);
11657
0
    }
11658
265k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11659
0
        xmlFreeInputStream(input);
11660
0
        xmlFreeParserCtxt(ctxt);
11661
0
        return(NULL);
11662
0
    }
11663
11664
265k
    return(ctxt);
11665
265k
}
11666
#endif /* LIBXML_PUSH_ENABLED */
11667
11668
/**
11669
 * xmlStopParser:
11670
 * @ctxt:  an XML parser context
11671
 *
11672
 * Blocks further parser processing
11673
 */
11674
void
11675
0
xmlStopParser(xmlParserCtxtPtr ctxt) {
11676
0
    if (ctxt == NULL)
11677
0
        return;
11678
0
    xmlHaltParser(ctxt);
11679
0
    if (ctxt->errNo != XML_ERR_NO_MEMORY)
11680
0
        ctxt->errNo = XML_ERR_USER_STOP;
11681
0
}
11682
11683
/**
11684
 * xmlCreateIOParserCtxt:
11685
 * @sax:  a SAX handler (optional)
11686
 * @user_data:  user data for SAX callbacks (optional)
11687
 * @ioread:  an I/O read function
11688
 * @ioclose:  an I/O close function (optional)
11689
 * @ioctx:  an I/O handler
11690
 * @enc:  the charset encoding if known (deprecated)
11691
 *
11692
 * Create a parser context for using the XML parser with an existing
11693
 * I/O stream
11694
 *
11695
 * Returns the new parser context or NULL
11696
 */
11697
xmlParserCtxtPtr
11698
xmlCreateIOParserCtxt(xmlSAXHandlerPtr sax, void *user_data,
11699
                      xmlInputReadCallback ioread,
11700
                      xmlInputCloseCallback ioclose,
11701
0
                      void *ioctx, xmlCharEncoding enc) {
11702
0
    xmlParserCtxtPtr ctxt;
11703
0
    xmlParserInputPtr input;
11704
0
    const char *encoding;
11705
11706
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11707
0
    if (ctxt == NULL)
11708
0
  return(NULL);
11709
11710
0
    encoding = xmlGetCharEncodingName(enc);
11711
0
    input = xmlCtxtNewInputFromIO(ctxt, NULL, ioread, ioclose, ioctx,
11712
0
                                  encoding, 0);
11713
0
    if (input == NULL) {
11714
0
  xmlFreeParserCtxt(ctxt);
11715
0
        return (NULL);
11716
0
    }
11717
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11718
0
        xmlFreeInputStream(input);
11719
0
        xmlFreeParserCtxt(ctxt);
11720
0
        return(NULL);
11721
0
    }
11722
11723
0
    return(ctxt);
11724
0
}
11725
11726
#ifdef LIBXML_VALID_ENABLED
11727
/************************************************************************
11728
 *                  *
11729
 *    Front ends when parsing a DTD       *
11730
 *                  *
11731
 ************************************************************************/
11732
11733
/**
11734
 * xmlCtxtParseDtd:
11735
 * @ctxt:  a parser context
11736
 * @input:  a parser input
11737
 * @publicId:  public ID of the DTD (optional)
11738
 * @systemId:  system ID of the DTD (optional)
11739
 *
11740
 * Parse a DTD.
11741
 *
11742
 * Option XML_PARSE_DTDLOAD should be enabled in the parser context
11743
 * to make external entities work.
11744
 *
11745
 * Availabe since 2.14.0.
11746
 *
11747
 * Returns the resulting xmlDtdPtr or NULL in case of error.
11748
 * @input will be freed by the function in any case.
11749
 */
11750
xmlDtdPtr
11751
xmlCtxtParseDtd(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
11752
0
                const xmlChar *publicId, const xmlChar *systemId) {
11753
0
    xmlDtdPtr ret = NULL;
11754
11755
0
    if ((ctxt == NULL) || (input == NULL)) {
11756
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
11757
0
        xmlFreeInputStream(input);
11758
0
        return(NULL);
11759
0
    }
11760
11761
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11762
0
        xmlFreeInputStream(input);
11763
0
        return(NULL);
11764
0
    }
11765
11766
0
    if (publicId == NULL)
11767
0
        publicId = BAD_CAST "none";
11768
0
    if (systemId == NULL)
11769
0
        systemId = BAD_CAST "none";
11770
11771
0
    ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0");
11772
0
    if (ctxt->myDoc == NULL) {
11773
0
        xmlErrMemory(ctxt);
11774
0
        goto error;
11775
0
    }
11776
0
    ctxt->myDoc->properties = XML_DOC_INTERNAL;
11777
0
    ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none",
11778
0
                                       publicId, systemId);
11779
0
    if (ctxt->myDoc->extSubset == NULL) {
11780
0
        xmlErrMemory(ctxt);
11781
0
        xmlFreeDoc(ctxt->myDoc);
11782
0
        goto error;
11783
0
    }
11784
11785
0
    xmlParseExternalSubset(ctxt, publicId, systemId);
11786
11787
0
    if (ctxt->wellFormed) {
11788
0
        ret = ctxt->myDoc->extSubset;
11789
0
        ctxt->myDoc->extSubset = NULL;
11790
0
        if (ret != NULL) {
11791
0
            xmlNodePtr tmp;
11792
11793
0
            ret->doc = NULL;
11794
0
            tmp = ret->children;
11795
0
            while (tmp != NULL) {
11796
0
                tmp->doc = NULL;
11797
0
                tmp = tmp->next;
11798
0
            }
11799
0
        }
11800
0
    } else {
11801
0
        ret = NULL;
11802
0
    }
11803
0
    xmlFreeDoc(ctxt->myDoc);
11804
0
    ctxt->myDoc = NULL;
11805
11806
0
error:
11807
0
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
11808
11809
0
    return(ret);
11810
0
}
11811
11812
/**
11813
 * xmlIOParseDTD:
11814
 * @sax:  the SAX handler block or NULL
11815
 * @input:  an Input Buffer
11816
 * @enc:  the charset encoding if known
11817
 *
11818
 * DEPRECATED: Use xmlCtxtParseDtd.
11819
 *
11820
 * Load and parse a DTD
11821
 *
11822
 * Returns the resulting xmlDtdPtr or NULL in case of error.
11823
 * @input will be freed by the function in any case.
11824
 */
11825
11826
xmlDtdPtr
11827
xmlIOParseDTD(xmlSAXHandlerPtr sax, xmlParserInputBufferPtr input,
11828
0
        xmlCharEncoding enc) {
11829
0
    xmlDtdPtr ret = NULL;
11830
0
    xmlParserCtxtPtr ctxt;
11831
0
    xmlParserInputPtr pinput = NULL;
11832
11833
0
    if (input == NULL)
11834
0
  return(NULL);
11835
11836
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
11837
0
    if (ctxt == NULL) {
11838
0
        xmlFreeParserInputBuffer(input);
11839
0
  return(NULL);
11840
0
    }
11841
0
    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
11842
11843
    /*
11844
     * generate a parser input from the I/O handler
11845
     */
11846
11847
0
    pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE);
11848
0
    if (pinput == NULL) {
11849
0
        xmlFreeParserInputBuffer(input);
11850
0
  xmlFreeParserCtxt(ctxt);
11851
0
  return(NULL);
11852
0
    }
11853
11854
0
    if (enc != XML_CHAR_ENCODING_NONE) {
11855
0
        xmlSwitchEncoding(ctxt, enc);
11856
0
    }
11857
11858
0
    ret = xmlCtxtParseDtd(ctxt, pinput, NULL, NULL);
11859
11860
0
    xmlFreeParserCtxt(ctxt);
11861
0
    return(ret);
11862
0
}
11863
11864
/**
11865
 * xmlSAXParseDTD:
11866
 * @sax:  the SAX handler block
11867
 * @ExternalID:  a NAME* containing the External ID of the DTD
11868
 * @SystemID:  a NAME* containing the URL to the DTD
11869
 *
11870
 * DEPRECATED: Use xmlCtxtParseDtd.
11871
 *
11872
 * Load and parse an external subset.
11873
 *
11874
 * Returns the resulting xmlDtdPtr or NULL in case of error.
11875
 */
11876
11877
xmlDtdPtr
11878
xmlSAXParseDTD(xmlSAXHandlerPtr sax, const xmlChar *ExternalID,
11879
0
                          const xmlChar *SystemID) {
11880
0
    xmlDtdPtr ret = NULL;
11881
0
    xmlParserCtxtPtr ctxt;
11882
0
    xmlParserInputPtr input = NULL;
11883
0
    xmlChar* systemIdCanonic;
11884
11885
0
    if ((ExternalID == NULL) && (SystemID == NULL)) return(NULL);
11886
11887
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
11888
0
    if (ctxt == NULL) {
11889
0
  return(NULL);
11890
0
    }
11891
0
    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
11892
11893
    /*
11894
     * Canonicalise the system ID
11895
     */
11896
0
    systemIdCanonic = xmlCanonicPath(SystemID);
11897
0
    if ((SystemID != NULL) && (systemIdCanonic == NULL)) {
11898
0
  xmlFreeParserCtxt(ctxt);
11899
0
  return(NULL);
11900
0
    }
11901
11902
    /*
11903
     * Ask the Entity resolver to load the damn thing
11904
     */
11905
11906
0
    if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL))
11907
0
  input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID,
11908
0
                                   systemIdCanonic);
11909
0
    if (input == NULL) {
11910
0
  xmlFreeParserCtxt(ctxt);
11911
0
  if (systemIdCanonic != NULL)
11912
0
      xmlFree(systemIdCanonic);
11913
0
  return(NULL);
11914
0
    }
11915
11916
0
    if (input->filename == NULL)
11917
0
  input->filename = (char *) systemIdCanonic;
11918
0
    else
11919
0
  xmlFree(systemIdCanonic);
11920
11921
0
    ret = xmlCtxtParseDtd(ctxt, input, ExternalID, SystemID);
11922
11923
0
    xmlFreeParserCtxt(ctxt);
11924
0
    return(ret);
11925
0
}
11926
11927
11928
/**
11929
 * xmlParseDTD:
11930
 * @ExternalID:  a NAME* containing the External ID of the DTD
11931
 * @SystemID:  a NAME* containing the URL to the DTD
11932
 *
11933
 * Load and parse an external subset.
11934
 *
11935
 * Returns the resulting xmlDtdPtr or NULL in case of error.
11936
 */
11937
11938
xmlDtdPtr
11939
0
xmlParseDTD(const xmlChar *ExternalID, const xmlChar *SystemID) {
11940
0
    return(xmlSAXParseDTD(NULL, ExternalID, SystemID));
11941
0
}
11942
#endif /* LIBXML_VALID_ENABLED */
11943
11944
/************************************************************************
11945
 *                  *
11946
 *    Front ends when parsing an Entity     *
11947
 *                  *
11948
 ************************************************************************/
11949
11950
static xmlNodePtr
11951
xmlCtxtParseContentInternal(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
11952
0
                            int hasTextDecl, int buildTree) {
11953
0
    xmlNodePtr root = NULL;
11954
0
    xmlNodePtr list = NULL;
11955
0
    xmlChar *rootName = BAD_CAST "#root";
11956
0
    int result;
11957
11958
0
    if (buildTree) {
11959
0
        root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL);
11960
0
        if (root == NULL) {
11961
0
            xmlErrMemory(ctxt);
11962
0
            goto error;
11963
0
        }
11964
0
    }
11965
11966
0
    if (xmlCtxtPushInput(ctxt, input) < 0)
11967
0
        goto error;
11968
11969
0
    nameNsPush(ctxt, rootName, NULL, NULL, 0, 0);
11970
0
    spacePush(ctxt, -1);
11971
11972
0
    if (buildTree)
11973
0
        nodePush(ctxt, root);
11974
11975
0
    if (hasTextDecl) {
11976
0
        xmlDetectEncoding(ctxt);
11977
11978
        /*
11979
         * Parse a possible text declaration first
11980
         */
11981
0
        if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
11982
0
            (IS_BLANK_CH(NXT(5)))) {
11983
0
            xmlParseTextDecl(ctxt);
11984
            /*
11985
             * An XML-1.0 document can't reference an entity not XML-1.0
11986
             */
11987
0
            if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
11988
0
                (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
11989
0
                xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
11990
0
                               "Version mismatch between document and "
11991
0
                               "entity\n");
11992
0
            }
11993
0
        }
11994
0
    }
11995
11996
0
    xmlParseContentInternal(ctxt);
11997
11998
0
    if (ctxt->input->cur < ctxt->input->end)
11999
0
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
12000
12001
0
    if ((ctxt->wellFormed) ||
12002
0
        ((ctxt->recovery) && (!xmlCtxtIsCatastrophicError(ctxt)))) {
12003
0
        if (root != NULL) {
12004
0
            xmlNodePtr cur;
12005
12006
            /*
12007
             * Unlink newly created node list.
12008
             */
12009
0
            list = root->children;
12010
0
            root->children = NULL;
12011
0
            root->last = NULL;
12012
0
            for (cur = list; cur != NULL; cur = cur->next)
12013
0
                cur->parent = NULL;
12014
0
        }
12015
0
    }
12016
12017
    /*
12018
     * Read the rest of the stream in case of errors. We want
12019
     * to account for the whole entity size.
12020
     */
12021
0
    do {
12022
0
        ctxt->input->cur = ctxt->input->end;
12023
0
        xmlParserShrink(ctxt);
12024
0
        result = xmlParserGrow(ctxt);
12025
0
    } while (result > 0);
12026
12027
0
    if (buildTree)
12028
0
        nodePop(ctxt);
12029
12030
0
    namePop(ctxt);
12031
0
    spacePop(ctxt);
12032
12033
0
    xmlCtxtPopInput(ctxt);
12034
12035
0
error:
12036
0
    xmlFreeNode(root);
12037
12038
0
    return(list);
12039
0
}
12040
12041
static void
12042
0
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
12043
0
    xmlParserInputPtr input;
12044
0
    xmlNodePtr list;
12045
0
    unsigned long consumed;
12046
0
    int isExternal;
12047
0
    int buildTree;
12048
0
    int oldMinNsIndex;
12049
0
    int oldNodelen, oldNodemem;
12050
12051
0
    isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY);
12052
0
    buildTree = (ctxt->node != NULL);
12053
12054
    /*
12055
     * Recursion check
12056
     */
12057
0
    if (ent->flags & XML_ENT_EXPANDING) {
12058
0
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
12059
0
        xmlHaltParser(ctxt);
12060
0
        goto error;
12061
0
    }
12062
12063
    /*
12064
     * Load entity
12065
     */
12066
0
    input = xmlNewEntityInputStream(ctxt, ent);
12067
0
    if (input == NULL)
12068
0
        goto error;
12069
12070
    /*
12071
     * When building a tree, we need to limit the scope of namespace
12072
     * declarations, so that entities don't reference xmlNs structs
12073
     * from the parent of a reference.
12074
     */
12075
0
    oldMinNsIndex = ctxt->nsdb->minNsIndex;
12076
0
    if (buildTree)
12077
0
        ctxt->nsdb->minNsIndex = ctxt->nsNr;
12078
12079
0
    oldNodelen = ctxt->nodelen;
12080
0
    oldNodemem = ctxt->nodemem;
12081
0
    ctxt->nodelen = 0;
12082
0
    ctxt->nodemem = 0;
12083
12084
    /*
12085
     * Parse content
12086
     *
12087
     * This initiates a recursive call chain:
12088
     *
12089
     * - xmlCtxtParseContentInternal
12090
     * - xmlParseContentInternal
12091
     * - xmlParseReference
12092
     * - xmlCtxtParseEntity
12093
     *
12094
     * The nesting depth is limited by the maximum number of inputs,
12095
     * see xmlCtxtPushInput.
12096
     *
12097
     * It's possible to make this non-recursive (minNsIndex must be
12098
     * stored in the input struct) at the expense of code readability.
12099
     */
12100
12101
0
    ent->flags |= XML_ENT_EXPANDING;
12102
12103
0
    list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree);
12104
12105
0
    ent->flags &= ~XML_ENT_EXPANDING;
12106
12107
0
    ctxt->nsdb->minNsIndex = oldMinNsIndex;
12108
0
    ctxt->nodelen = oldNodelen;
12109
0
    ctxt->nodemem = oldNodemem;
12110
12111
    /*
12112
     * Entity size accounting
12113
     */
12114
0
    consumed = input->consumed;
12115
0
    xmlSaturatedAddSizeT(&consumed, input->end - input->base);
12116
12117
0
    if ((ent->flags & XML_ENT_CHECKED) == 0)
12118
0
        xmlSaturatedAdd(&ent->expandedSize, consumed);
12119
12120
0
    if ((ent->flags & XML_ENT_PARSED) == 0) {
12121
0
        if (isExternal)
12122
0
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
12123
12124
0
        ent->children = list;
12125
12126
0
        while (list != NULL) {
12127
0
            list->parent = (xmlNodePtr) ent;
12128
12129
            /*
12130
             * Downstream code like the nginx xslt module can set
12131
             * ctxt->myDoc->extSubset to a separate DTD, so the entity
12132
             * might have a different or a NULL document.
12133
             */
12134
0
            if (list->doc != ent->doc)
12135
0
                xmlSetTreeDoc(list, ent->doc);
12136
12137
0
            if (list->next == NULL)
12138
0
                ent->last = list;
12139
0
            list = list->next;
12140
0
        }
12141
0
    } else {
12142
0
        xmlFreeNodeList(list);
12143
0
    }
12144
12145
0
    xmlFreeInputStream(input);
12146
12147
0
error:
12148
0
    ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED;
12149
0
}
12150
12151
/**
12152
 * xmlParseCtxtExternalEntity:
12153
 * @ctxt:  the existing parsing context
12154
 * @URL:  the URL for the entity to load
12155
 * @ID:  the System ID for the entity to load
12156
 * @listOut:  the return value for the set of parsed nodes
12157
 *
12158
 * Parse an external general entity within an existing parsing context
12159
 * An external general parsed entity is well-formed if it matches the
12160
 * production labeled extParsedEnt.
12161
 *
12162
 * [78] extParsedEnt ::= TextDecl? content
12163
 *
12164
 * Returns 0 if the entity is well formed, -1 in case of args problem and
12165
 *    the parser error code otherwise
12166
 */
12167
12168
int
12169
xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctxt, const xmlChar *URL,
12170
0
                           const xmlChar *ID, xmlNodePtr *listOut) {
12171
0
    xmlParserInputPtr input;
12172
0
    xmlNodePtr list;
12173
12174
0
    if (listOut != NULL)
12175
0
        *listOut = NULL;
12176
12177
0
    if (ctxt == NULL)
12178
0
        return(XML_ERR_ARGUMENT);
12179
12180
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
12181
0
                            XML_RESOURCE_GENERAL_ENTITY);
12182
0
    if (input == NULL)
12183
0
        return(ctxt->errNo);
12184
12185
0
    xmlCtxtInitializeLate(ctxt);
12186
12187
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 1, 1);
12188
0
    if (listOut != NULL)
12189
0
        *listOut = list;
12190
0
    else
12191
0
        xmlFreeNodeList(list);
12192
12193
0
    xmlFreeInputStream(input);
12194
0
    return(ctxt->errNo);
12195
0
}
12196
12197
#ifdef LIBXML_SAX1_ENABLED
12198
/**
12199
 * xmlParseExternalEntity:
12200
 * @doc:  the document the chunk pertains to
12201
 * @sax:  the SAX handler block (possibly NULL)
12202
 * @user_data:  The user data returned on SAX callbacks (possibly NULL)
12203
 * @depth:  Used for loop detection, use 0
12204
 * @URL:  the URL for the entity to load
12205
 * @ID:  the System ID for the entity to load
12206
 * @list:  the return value for the set of parsed nodes
12207
 *
12208
 * DEPRECATED: Use xmlParseCtxtExternalEntity.
12209
 *
12210
 * Parse an external general entity
12211
 * An external general parsed entity is well-formed if it matches the
12212
 * production labeled extParsedEnt.
12213
 *
12214
 * [78] extParsedEnt ::= TextDecl? content
12215
 *
12216
 * Returns 0 if the entity is well formed, -1 in case of args problem and
12217
 *    the parser error code otherwise
12218
 */
12219
12220
int
12221
xmlParseExternalEntity(xmlDocPtr doc, xmlSAXHandlerPtr sax, void *user_data,
12222
0
    int depth, const xmlChar *URL, const xmlChar *ID, xmlNodePtr *list) {
12223
0
    xmlParserCtxtPtr ctxt;
12224
0
    int ret;
12225
12226
0
    if (list != NULL)
12227
0
        *list = NULL;
12228
12229
0
    if (doc == NULL)
12230
0
        return(XML_ERR_ARGUMENT);
12231
12232
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
12233
0
    if (ctxt == NULL)
12234
0
        return(XML_ERR_NO_MEMORY);
12235
12236
0
    ctxt->depth = depth;
12237
0
    ctxt->myDoc = doc;
12238
0
    ret = xmlParseCtxtExternalEntity(ctxt, URL, ID, list);
12239
12240
0
    xmlFreeParserCtxt(ctxt);
12241
0
    return(ret);
12242
0
}
12243
12244
/**
12245
 * xmlParseBalancedChunkMemory:
12246
 * @doc:  the document the chunk pertains to (must not be NULL)
12247
 * @sax:  the SAX handler block (possibly NULL)
12248
 * @user_data:  The user data returned on SAX callbacks (possibly NULL)
12249
 * @depth:  Used for loop detection, use 0
12250
 * @string:  the input string in UTF8 or ISO-Latin (zero terminated)
12251
 * @lst:  the return value for the set of parsed nodes
12252
 *
12253
 * Parse a well-balanced chunk of an XML document
12254
 * called by the parser
12255
 * The allowed sequence for the Well Balanced Chunk is the one defined by
12256
 * the content production in the XML grammar:
12257
 *
12258
 * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)*
12259
 *
12260
 * Returns 0 if the chunk is well balanced, -1 in case of args problem and
12261
 *    the parser error code otherwise
12262
 */
12263
12264
int
12265
xmlParseBalancedChunkMemory(xmlDocPtr doc, xmlSAXHandlerPtr sax,
12266
0
     void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst) {
12267
0
    return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data,
12268
0
                                                depth, string, lst, 0 );
12269
0
}
12270
#endif /* LIBXML_SAX1_ENABLED */
12271
12272
/**
12273
 * xmlCtxtParseContent:
12274
 * @ctxt:  parser context
12275
 * @input:  parser input
12276
 * @node:  target node or document
12277
 * @hasTextDecl:  whether to parse text declaration
12278
 *
12279
 * Parse a well-balanced chunk of XML matching the 'content' production.
12280
 *
12281
 * Namespaces in scope of @node and entities of @node's document are
12282
 * recognized. When validating, the DTD of @node's document is used.
12283
 *
12284
 * Always consumes @input even in error case.
12285
 *
12286
 * Available since 2.14.0.
12287
 *
12288
 * Returns a node list or NULL in case of error.
12289
 */
12290
xmlNodePtr
12291
xmlCtxtParseContent(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
12292
0
                    xmlNodePtr node, int hasTextDecl) {
12293
0
    xmlDocPtr doc;
12294
0
    xmlNodePtr cur, list = NULL;
12295
0
    int nsnr = 0;
12296
0
    xmlDictPtr oldDict;
12297
0
    int oldOptions, oldDictNames, oldLoadSubset;
12298
12299
0
    if ((ctxt == NULL) || (input == NULL) || (node == NULL)) {
12300
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12301
0
        goto exit;
12302
0
    }
12303
12304
0
    doc = node->doc;
12305
0
    if (doc == NULL) {
12306
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12307
0
        goto exit;
12308
0
    }
12309
12310
0
    switch (node->type) {
12311
0
        case XML_ELEMENT_NODE:
12312
0
        case XML_DOCUMENT_NODE:
12313
0
        case XML_HTML_DOCUMENT_NODE:
12314
0
            break;
12315
12316
0
        case XML_ATTRIBUTE_NODE:
12317
0
        case XML_TEXT_NODE:
12318
0
        case XML_CDATA_SECTION_NODE:
12319
0
        case XML_ENTITY_REF_NODE:
12320
0
        case XML_PI_NODE:
12321
0
        case XML_COMMENT_NODE:
12322
0
            for (cur = node->parent; cur != NULL; cur = node->parent) {
12323
0
                if ((cur->type == XML_ELEMENT_NODE) ||
12324
0
                    (cur->type == XML_DOCUMENT_NODE) ||
12325
0
                    (cur->type == XML_HTML_DOCUMENT_NODE)) {
12326
0
                    node = cur;
12327
0
                    break;
12328
0
                }
12329
0
            }
12330
0
            break;
12331
12332
0
        default:
12333
0
            xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12334
0
            goto exit;
12335
0
    }
12336
12337
0
#ifdef LIBXML_HTML_ENABLED
12338
0
    if (ctxt->html)
12339
0
        htmlCtxtReset(ctxt);
12340
0
    else
12341
0
#endif
12342
0
        xmlCtxtReset(ctxt);
12343
12344
0
    oldDict = ctxt->dict;
12345
0
    oldOptions = ctxt->options;
12346
0
    oldDictNames = ctxt->dictNames;
12347
0
    oldLoadSubset = ctxt->loadsubset;
12348
12349
    /*
12350
     * Use input doc's dict if present, else assure XML_PARSE_NODICT is set.
12351
     */
12352
0
    if (doc->dict != NULL) {
12353
0
        ctxt->dict = doc->dict;
12354
0
    } else {
12355
0
        ctxt->options |= XML_PARSE_NODICT;
12356
0
        ctxt->dictNames = 0;
12357
0
    }
12358
12359
    /*
12360
     * Disable IDs
12361
     */
12362
0
    ctxt->loadsubset |= XML_SKIP_IDS;
12363
12364
0
    ctxt->myDoc = doc;
12365
12366
0
#ifdef LIBXML_HTML_ENABLED
12367
0
    if (ctxt->html) {
12368
        /*
12369
         * When parsing in context, it makes no sense to add implied
12370
         * elements like html/body/etc...
12371
         */
12372
0
        ctxt->options |= HTML_PARSE_NOIMPLIED;
12373
12374
0
        list = htmlCtxtParseContentInternal(ctxt, input);
12375
0
    } else
12376
0
#endif
12377
0
    {
12378
0
        xmlCtxtInitializeLate(ctxt);
12379
12380
        /*
12381
         * initialize the SAX2 namespaces stack
12382
         */
12383
0
        cur = node;
12384
0
        while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) {
12385
0
            xmlNsPtr ns = cur->nsDef;
12386
0
            xmlHashedString hprefix, huri;
12387
12388
0
            while (ns != NULL) {
12389
0
                hprefix = xmlDictLookupHashed(ctxt->dict, ns->prefix, -1);
12390
0
                huri = xmlDictLookupHashed(ctxt->dict, ns->href, -1);
12391
0
                if (xmlParserNsPush(ctxt, &hprefix, &huri, ns, 1) > 0)
12392
0
                    nsnr++;
12393
0
                ns = ns->next;
12394
0
            }
12395
0
            cur = cur->parent;
12396
0
        }
12397
12398
0
        list = xmlCtxtParseContentInternal(ctxt, input, hasTextDecl, 1);
12399
12400
0
        if (nsnr > 0)
12401
0
            xmlParserNsPop(ctxt, nsnr);
12402
0
    }
12403
12404
0
    ctxt->dict = oldDict;
12405
0
    ctxt->options = oldOptions;
12406
0
    ctxt->dictNames = oldDictNames;
12407
0
    ctxt->loadsubset = oldLoadSubset;
12408
0
    ctxt->myDoc = NULL;
12409
0
    ctxt->node = NULL;
12410
12411
0
exit:
12412
0
    xmlFreeInputStream(input);
12413
0
    return(list);
12414
0
}
12415
12416
/**
12417
 * xmlParseInNodeContext:
12418
 * @node:  the context node
12419
 * @data:  the input string
12420
 * @datalen:  the input string length in bytes
12421
 * @options:  a combination of xmlParserOption
12422
 * @listOut:  the return value for the set of parsed nodes
12423
 *
12424
 * Parse a well-balanced chunk of an XML document
12425
 * within the context (DTD, namespaces, etc ...) of the given node.
12426
 *
12427
 * The allowed sequence for the data is a Well Balanced Chunk defined by
12428
 * the content production in the XML grammar:
12429
 *
12430
 * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)*
12431
 *
12432
 * This function assumes the encoding of @node's document which is
12433
 * typically not what you want. A better alternative is
12434
 * xmlCtxtParseContent.
12435
 *
12436
 * Returns XML_ERR_OK if the chunk is well balanced, and the parser
12437
 * error code otherwise
12438
 */
12439
xmlParserErrors
12440
xmlParseInNodeContext(xmlNodePtr node, const char *data, int datalen,
12441
0
                      int options, xmlNodePtr *listOut) {
12442
0
    xmlParserCtxtPtr ctxt;
12443
0
    xmlParserInputPtr input;
12444
0
    xmlDocPtr doc;
12445
0
    xmlNodePtr list;
12446
0
    xmlParserErrors ret;
12447
12448
0
    if (listOut == NULL)
12449
0
        return(XML_ERR_INTERNAL_ERROR);
12450
0
    *listOut = NULL;
12451
12452
0
    if ((node == NULL) || (data == NULL) || (datalen < 0))
12453
0
        return(XML_ERR_INTERNAL_ERROR);
12454
12455
0
    doc = node->doc;
12456
0
    if (doc == NULL)
12457
0
        return(XML_ERR_INTERNAL_ERROR);
12458
12459
0
#ifdef LIBXML_HTML_ENABLED
12460
0
    if (doc->type == XML_HTML_DOCUMENT_NODE) {
12461
0
        ctxt = htmlNewParserCtxt();
12462
0
    }
12463
0
    else
12464
0
#endif
12465
0
        ctxt = xmlNewParserCtxt();
12466
12467
0
    if (ctxt == NULL)
12468
0
        return(XML_ERR_NO_MEMORY);
12469
12470
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, data, datalen,
12471
0
                                      (const char *) doc->encoding,
12472
0
                                      XML_INPUT_BUF_STATIC);
12473
0
    if (input == NULL) {
12474
0
        xmlFreeParserCtxt(ctxt);
12475
0
        return(XML_ERR_NO_MEMORY);
12476
0
    }
12477
12478
0
    xmlCtxtUseOptions(ctxt, options);
12479
12480
0
    list = xmlCtxtParseContent(ctxt, input, node, /* hasTextDecl */ 0);
12481
12482
0
    if (list == NULL) {
12483
0
        ret = ctxt->errNo;
12484
0
        if (ret == XML_ERR_ARGUMENT)
12485
0
            ret = XML_ERR_INTERNAL_ERROR;
12486
0
    } else {
12487
0
        ret = XML_ERR_OK;
12488
0
        *listOut = list;
12489
0
    }
12490
12491
0
    xmlFreeParserCtxt(ctxt);
12492
12493
0
    return(ret);
12494
0
}
12495
12496
#ifdef LIBXML_SAX1_ENABLED
12497
/**
12498
 * xmlParseBalancedChunkMemoryRecover:
12499
 * @doc:  the document the chunk pertains to (must not be NULL)
12500
 * @sax:  the SAX handler block (possibly NULL)
12501
 * @user_data:  The user data returned on SAX callbacks (possibly NULL)
12502
 * @depth:  Used for loop detection, use 0
12503
 * @string:  the input string in UTF8 or ISO-Latin (zero terminated)
12504
 * @listOut:  the return value for the set of parsed nodes
12505
 * @recover: return nodes even if the data is broken (use 0)
12506
 *
12507
 * Parse a well-balanced chunk of an XML document
12508
 *
12509
 * The allowed sequence for the Well Balanced Chunk is the one defined by
12510
 * the content production in the XML grammar:
12511
 *
12512
 * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)*
12513
 *
12514
 * Returns 0 if the chunk is well balanced, or thehe parser error code
12515
 * otherwise.
12516
 *
12517
 * In case recover is set to 1, the nodelist will not be empty even if
12518
 * the parsed chunk is not well balanced, assuming the parsing succeeded to
12519
 * some extent.
12520
 */
12521
int
12522
xmlParseBalancedChunkMemoryRecover(xmlDocPtr doc, xmlSAXHandlerPtr sax,
12523
     void *user_data, int depth, const xmlChar *string, xmlNodePtr *listOut,
12524
0
     int recover) {
12525
0
    xmlParserCtxtPtr ctxt;
12526
0
    xmlParserInputPtr input;
12527
0
    xmlNodePtr list;
12528
0
    int ret;
12529
12530
0
    if (listOut != NULL)
12531
0
        *listOut = NULL;
12532
12533
0
    if (string == NULL)
12534
0
        return(XML_ERR_ARGUMENT);
12535
12536
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
12537
0
    if (ctxt == NULL)
12538
0
        return(XML_ERR_NO_MEMORY);
12539
12540
0
    xmlCtxtInitializeLate(ctxt);
12541
12542
0
    ctxt->depth = depth;
12543
0
    ctxt->myDoc = doc;
12544
0
    if (recover) {
12545
0
        ctxt->options |= XML_PARSE_RECOVER;
12546
0
        ctxt->recovery = 1;
12547
0
    }
12548
12549
0
    input = xmlNewStringInputStream(ctxt, string);
12550
0
    if (input == NULL) {
12551
0
        ret = ctxt->errNo;
12552
0
        goto error;
12553
0
    }
12554
12555
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 0, 1);
12556
0
    if (listOut != NULL)
12557
0
        *listOut = list;
12558
0
    else
12559
0
        xmlFreeNodeList(list);
12560
12561
0
    if (!ctxt->wellFormed)
12562
0
        ret = ctxt->errNo;
12563
0
    else
12564
0
        ret = XML_ERR_OK;
12565
12566
0
error:
12567
0
    xmlFreeInputStream(input);
12568
0
    xmlFreeParserCtxt(ctxt);
12569
0
    return(ret);
12570
0
}
12571
12572
/**
12573
 * xmlSAXParseEntity:
12574
 * @sax:  the SAX handler block
12575
 * @filename:  the filename
12576
 *
12577
 * DEPRECATED: Don't use.
12578
 *
12579
 * parse an XML external entity out of context and build a tree.
12580
 * It use the given SAX function block to handle the parsing callback.
12581
 * If sax is NULL, fallback to the default DOM tree building routines.
12582
 *
12583
 * [78] extParsedEnt ::= TextDecl? content
12584
 *
12585
 * This correspond to a "Well Balanced" chunk
12586
 *
12587
 * Returns the resulting document tree
12588
 */
12589
12590
xmlDocPtr
12591
0
xmlSAXParseEntity(xmlSAXHandlerPtr sax, const char *filename) {
12592
0
    xmlDocPtr ret;
12593
0
    xmlParserCtxtPtr ctxt;
12594
12595
0
    ctxt = xmlCreateFileParserCtxt(filename);
12596
0
    if (ctxt == NULL) {
12597
0
  return(NULL);
12598
0
    }
12599
0
    if (sax != NULL) {
12600
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12601
0
            *ctxt->sax = *sax;
12602
0
        } else {
12603
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12604
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12605
0
        }
12606
0
        ctxt->userData = NULL;
12607
0
    }
12608
12609
0
    xmlParseExtParsedEnt(ctxt);
12610
12611
0
    if (ctxt->wellFormed) {
12612
0
  ret = ctxt->myDoc;
12613
0
    } else {
12614
0
        ret = NULL;
12615
0
        xmlFreeDoc(ctxt->myDoc);
12616
0
    }
12617
12618
0
    xmlFreeParserCtxt(ctxt);
12619
12620
0
    return(ret);
12621
0
}
12622
12623
/**
12624
 * xmlParseEntity:
12625
 * @filename:  the filename
12626
 *
12627
 * parse an XML external entity out of context and build a tree.
12628
 *
12629
 * [78] extParsedEnt ::= TextDecl? content
12630
 *
12631
 * This correspond to a "Well Balanced" chunk
12632
 *
12633
 * Returns the resulting document tree
12634
 */
12635
12636
xmlDocPtr
12637
0
xmlParseEntity(const char *filename) {
12638
0
    return(xmlSAXParseEntity(NULL, filename));
12639
0
}
12640
#endif /* LIBXML_SAX1_ENABLED */
12641
12642
/**
12643
 * xmlCreateEntityParserCtxt:
12644
 * @URL:  the entity URL
12645
 * @ID:  the entity PUBLIC ID
12646
 * @base:  a possible base for the target URI
12647
 *
12648
 * DEPRECATED: Don't use.
12649
 *
12650
 * Create a parser context for an external entity
12651
 * Automatic support for ZLIB/Compress compressed document is provided
12652
 * by default if found at compile-time.
12653
 *
12654
 * Returns the new parser context or NULL
12655
 */
12656
xmlParserCtxtPtr
12657
xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID,
12658
0
                    const xmlChar *base) {
12659
0
    xmlParserCtxtPtr ctxt;
12660
0
    xmlParserInputPtr input;
12661
0
    xmlChar *uri = NULL;
12662
12663
0
    ctxt = xmlNewParserCtxt();
12664
0
    if (ctxt == NULL)
12665
0
  return(NULL);
12666
12667
0
    if (base != NULL) {
12668
0
        if (xmlBuildURISafe(URL, base, &uri) < 0)
12669
0
            goto error;
12670
0
        if (uri != NULL)
12671
0
            URL = uri;
12672
0
    }
12673
12674
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
12675
0
                            XML_RESOURCE_UNKNOWN);
12676
0
    if (input == NULL)
12677
0
        goto error;
12678
12679
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12680
0
        xmlFreeInputStream(input);
12681
0
        goto error;
12682
0
    }
12683
12684
0
    xmlFree(uri);
12685
0
    return(ctxt);
12686
12687
0
error:
12688
0
    xmlFree(uri);
12689
0
    xmlFreeParserCtxt(ctxt);
12690
0
    return(NULL);
12691
0
}
12692
12693
/************************************************************************
12694
 *                  *
12695
 *    Front ends when parsing from a file     *
12696
 *                  *
12697
 ************************************************************************/
12698
12699
/**
12700
 * xmlCreateURLParserCtxt:
12701
 * @filename:  the filename or URL
12702
 * @options:  a combination of xmlParserOption
12703
 *
12704
 * DEPRECATED: Use xmlNewParserCtxt and xmlCtxtReadFile.
12705
 *
12706
 * Create a parser context for a file or URL content.
12707
 * Automatic support for ZLIB/Compress compressed document is provided
12708
 * by default if found at compile-time and for file accesses
12709
 *
12710
 * Returns the new parser context or NULL
12711
 */
12712
xmlParserCtxtPtr
12713
xmlCreateURLParserCtxt(const char *filename, int options)
12714
0
{
12715
0
    xmlParserCtxtPtr ctxt;
12716
0
    xmlParserInputPtr input;
12717
12718
0
    ctxt = xmlNewParserCtxt();
12719
0
    if (ctxt == NULL)
12720
0
  return(NULL);
12721
12722
0
    options |= XML_PARSE_UNZIP;
12723
12724
0
    xmlCtxtUseOptions(ctxt, options);
12725
0
    ctxt->linenumbers = 1;
12726
12727
0
    input = xmlLoadResource(ctxt, filename, NULL, XML_RESOURCE_MAIN_DOCUMENT);
12728
0
    if (input == NULL) {
12729
0
  xmlFreeParserCtxt(ctxt);
12730
0
  return(NULL);
12731
0
    }
12732
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12733
0
        xmlFreeInputStream(input);
12734
0
        xmlFreeParserCtxt(ctxt);
12735
0
        return(NULL);
12736
0
    }
12737
12738
0
    return(ctxt);
12739
0
}
12740
12741
/**
12742
 * xmlCreateFileParserCtxt:
12743
 * @filename:  the filename
12744
 *
12745
 * DEPRECATED: Use xmlNewParserCtxt and xmlCtxtReadFile.
12746
 *
12747
 * Create a parser context for a file content.
12748
 * Automatic support for ZLIB/Compress compressed document is provided
12749
 * by default if found at compile-time.
12750
 *
12751
 * Returns the new parser context or NULL
12752
 */
12753
xmlParserCtxtPtr
12754
xmlCreateFileParserCtxt(const char *filename)
12755
0
{
12756
0
    return(xmlCreateURLParserCtxt(filename, 0));
12757
0
}
12758
12759
#ifdef LIBXML_SAX1_ENABLED
12760
/**
12761
 * xmlSAXParseFileWithData:
12762
 * @sax:  the SAX handler block
12763
 * @filename:  the filename
12764
 * @recovery:  work in recovery mode, i.e. tries to read no Well Formed
12765
 *             documents
12766
 * @data:  the userdata
12767
 *
12768
 * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile.
12769
 *
12770
 * parse an XML file and build a tree. Automatic support for ZLIB/Compress
12771
 * compressed document is provided by default if found at compile-time.
12772
 * It use the given SAX function block to handle the parsing callback.
12773
 * If sax is NULL, fallback to the default DOM tree building routines.
12774
 *
12775
 * User data (void *) is stored within the parser context in the
12776
 * context's _private member, so it is available nearly everywhere in libxml
12777
 *
12778
 * Returns the resulting document tree
12779
 */
12780
12781
xmlDocPtr
12782
xmlSAXParseFileWithData(xmlSAXHandlerPtr sax, const char *filename,
12783
0
                        int recovery, void *data) {
12784
0
    xmlDocPtr ret = NULL;
12785
0
    xmlParserCtxtPtr ctxt;
12786
0
    xmlParserInputPtr input;
12787
12788
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12789
0
    if (ctxt == NULL)
12790
0
  return(NULL);
12791
12792
0
    if (data != NULL)
12793
0
  ctxt->_private = data;
12794
12795
0
    if (recovery) {
12796
0
        ctxt->options |= XML_PARSE_RECOVER;
12797
0
        ctxt->recovery = 1;
12798
0
    }
12799
12800
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
12801
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, NULL, 0);
12802
0
    else
12803
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, NULL, 0);
12804
12805
0
    if (input != NULL)
12806
0
        ret = xmlCtxtParseDocument(ctxt, input);
12807
12808
0
    xmlFreeParserCtxt(ctxt);
12809
0
    return(ret);
12810
0
}
12811
12812
/**
12813
 * xmlSAXParseFile:
12814
 * @sax:  the SAX handler block
12815
 * @filename:  the filename
12816
 * @recovery:  work in recovery mode, i.e. tries to read no Well Formed
12817
 *             documents
12818
 *
12819
 * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile.
12820
 *
12821
 * parse an XML file and build a tree. Automatic support for ZLIB/Compress
12822
 * compressed document is provided by default if found at compile-time.
12823
 * It use the given SAX function block to handle the parsing callback.
12824
 * If sax is NULL, fallback to the default DOM tree building routines.
12825
 *
12826
 * Returns the resulting document tree
12827
 */
12828
12829
xmlDocPtr
12830
xmlSAXParseFile(xmlSAXHandlerPtr sax, const char *filename,
12831
0
                          int recovery) {
12832
0
    return(xmlSAXParseFileWithData(sax,filename,recovery,NULL));
12833
0
}
12834
12835
/**
12836
 * xmlRecoverDoc:
12837
 * @cur:  a pointer to an array of xmlChar
12838
 *
12839
 * DEPRECATED: Use xmlReadDoc with XML_PARSE_RECOVER.
12840
 *
12841
 * parse an XML in-memory document and build a tree.
12842
 * In the case the document is not Well Formed, a attempt to build a
12843
 * tree is tried anyway
12844
 *
12845
 * Returns the resulting document tree or NULL in case of failure
12846
 */
12847
12848
xmlDocPtr
12849
0
xmlRecoverDoc(const xmlChar *cur) {
12850
0
    return(xmlSAXParseDoc(NULL, cur, 1));
12851
0
}
12852
12853
/**
12854
 * xmlParseFile:
12855
 * @filename:  the filename
12856
 *
12857
 * DEPRECATED: Use xmlReadFile.
12858
 *
12859
 * parse an XML file and build a tree. Automatic support for ZLIB/Compress
12860
 * compressed document is provided by default if found at compile-time.
12861
 *
12862
 * Returns the resulting document tree if the file was wellformed,
12863
 * NULL otherwise.
12864
 */
12865
12866
xmlDocPtr
12867
0
xmlParseFile(const char *filename) {
12868
0
    return(xmlSAXParseFile(NULL, filename, 0));
12869
0
}
12870
12871
/**
12872
 * xmlRecoverFile:
12873
 * @filename:  the filename
12874
 *
12875
 * DEPRECATED: Use xmlReadFile with XML_PARSE_RECOVER.
12876
 *
12877
 * parse an XML file and build a tree. Automatic support for ZLIB/Compress
12878
 * compressed document is provided by default if found at compile-time.
12879
 * In the case the document is not Well Formed, it attempts to build
12880
 * a tree anyway
12881
 *
12882
 * Returns the resulting document tree or NULL in case of failure
12883
 */
12884
12885
xmlDocPtr
12886
0
xmlRecoverFile(const char *filename) {
12887
0
    return(xmlSAXParseFile(NULL, filename, 1));
12888
0
}
12889
12890
12891
/**
12892
 * xmlSetupParserForBuffer:
12893
 * @ctxt:  an XML parser context
12894
 * @buffer:  a xmlChar * buffer
12895
 * @filename:  a file name
12896
 *
12897
 * DEPRECATED: Don't use.
12898
 *
12899
 * Setup the parser context to parse a new buffer; Clears any prior
12900
 * contents from the parser context. The buffer parameter must not be
12901
 * NULL, but the filename parameter can be
12902
 */
12903
void
12904
xmlSetupParserForBuffer(xmlParserCtxtPtr ctxt, const xmlChar* buffer,
12905
                             const char* filename)
12906
0
{
12907
0
    xmlParserInputPtr input;
12908
12909
0
    if ((ctxt == NULL) || (buffer == NULL))
12910
0
        return;
12911
12912
0
    xmlClearParserCtxt(ctxt);
12913
12914
0
    input = xmlCtxtNewInputFromString(ctxt, filename, (const char *) buffer,
12915
0
                                      NULL, 0);
12916
0
    if (input == NULL)
12917
0
        return;
12918
0
    if (xmlCtxtPushInput(ctxt, input) < 0)
12919
0
        xmlFreeInputStream(input);
12920
0
}
12921
12922
/**
12923
 * xmlSAXUserParseFile:
12924
 * @sax:  a SAX handler
12925
 * @user_data:  The user data returned on SAX callbacks
12926
 * @filename:  a file name
12927
 *
12928
 * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile.
12929
 *
12930
 * parse an XML file and call the given SAX handler routines.
12931
 * Automatic support for ZLIB/Compress compressed document is provided
12932
 *
12933
 * Returns 0 in case of success or a error number otherwise
12934
 */
12935
int
12936
xmlSAXUserParseFile(xmlSAXHandlerPtr sax, void *user_data,
12937
0
                    const char *filename) {
12938
0
    int ret = 0;
12939
0
    xmlParserCtxtPtr ctxt;
12940
12941
0
    ctxt = xmlCreateFileParserCtxt(filename);
12942
0
    if (ctxt == NULL) return -1;
12943
0
    if (sax != NULL) {
12944
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12945
0
            *ctxt->sax = *sax;
12946
0
        } else {
12947
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12948
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12949
0
        }
12950
0
  ctxt->userData = user_data;
12951
0
    }
12952
12953
0
    xmlParseDocument(ctxt);
12954
12955
0
    if (ctxt->wellFormed)
12956
0
  ret = 0;
12957
0
    else {
12958
0
        if (ctxt->errNo != 0)
12959
0
      ret = ctxt->errNo;
12960
0
  else
12961
0
      ret = -1;
12962
0
    }
12963
0
    if (ctxt->myDoc != NULL) {
12964
0
        xmlFreeDoc(ctxt->myDoc);
12965
0
  ctxt->myDoc = NULL;
12966
0
    }
12967
0
    xmlFreeParserCtxt(ctxt);
12968
12969
0
    return ret;
12970
0
}
12971
#endif /* LIBXML_SAX1_ENABLED */
12972
12973
/************************************************************************
12974
 *                  *
12975
 *    Front ends when parsing from memory     *
12976
 *                  *
12977
 ************************************************************************/
12978
12979
/**
12980
 * xmlCreateMemoryParserCtxt:
12981
 * @buffer:  a pointer to a char array
12982
 * @size:  the size of the array
12983
 *
12984
 * Create a parser context for an XML in-memory document. The input buffer
12985
 * must not contain a terminating null byte.
12986
 *
12987
 * Returns the new parser context or NULL
12988
 */
12989
xmlParserCtxtPtr
12990
0
xmlCreateMemoryParserCtxt(const char *buffer, int size) {
12991
0
    xmlParserCtxtPtr ctxt;
12992
0
    xmlParserInputPtr input;
12993
12994
0
    if (size < 0)
12995
0
  return(NULL);
12996
12997
0
    ctxt = xmlNewParserCtxt();
12998
0
    if (ctxt == NULL)
12999
0
  return(NULL);
13000
13001
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, 0);
13002
0
    if (input == NULL) {
13003
0
  xmlFreeParserCtxt(ctxt);
13004
0
  return(NULL);
13005
0
    }
13006
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13007
0
        xmlFreeInputStream(input);
13008
0
        xmlFreeParserCtxt(ctxt);
13009
0
        return(NULL);
13010
0
    }
13011
13012
0
    return(ctxt);
13013
0
}
13014
13015
#ifdef LIBXML_SAX1_ENABLED
13016
/**
13017
 * xmlSAXParseMemoryWithData:
13018
 * @sax:  the SAX handler block
13019
 * @buffer:  an pointer to a char array
13020
 * @size:  the size of the array
13021
 * @recovery:  work in recovery mode, i.e. tries to read no Well Formed
13022
 *             documents
13023
 * @data:  the userdata
13024
 *
13025
 * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory.
13026
 *
13027
 * parse an XML in-memory block and use the given SAX function block
13028
 * to handle the parsing callback. If sax is NULL, fallback to the default
13029
 * DOM tree building routines.
13030
 *
13031
 * User data (void *) is stored within the parser context in the
13032
 * context's _private member, so it is available nearly everywhere in libxml
13033
 *
13034
 * Returns the resulting document tree
13035
 */
13036
13037
xmlDocPtr
13038
xmlSAXParseMemoryWithData(xmlSAXHandlerPtr sax, const char *buffer,
13039
0
                          int size, int recovery, void *data) {
13040
0
    xmlDocPtr ret = NULL;
13041
0
    xmlParserCtxtPtr ctxt;
13042
0
    xmlParserInputPtr input;
13043
13044
0
    if (size < 0)
13045
0
        return(NULL);
13046
13047
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
13048
0
    if (ctxt == NULL)
13049
0
        return(NULL);
13050
13051
0
    if (data != NULL)
13052
0
  ctxt->_private=data;
13053
13054
0
    if (recovery) {
13055
0
        ctxt->options |= XML_PARSE_RECOVER;
13056
0
        ctxt->recovery = 1;
13057
0
    }
13058
13059
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL,
13060
0
                                      XML_INPUT_BUF_STATIC);
13061
13062
0
    if (input != NULL)
13063
0
        ret = xmlCtxtParseDocument(ctxt, input);
13064
13065
0
    xmlFreeParserCtxt(ctxt);
13066
0
    return(ret);
13067
0
}
13068
13069
/**
13070
 * xmlSAXParseMemory:
13071
 * @sax:  the SAX handler block
13072
 * @buffer:  an pointer to a char array
13073
 * @size:  the size of the array
13074
 * @recovery:  work in recovery mode, i.e. tries to read not Well Formed
13075
 *             documents
13076
 *
13077
 * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory.
13078
 *
13079
 * parse an XML in-memory block and use the given SAX function block
13080
 * to handle the parsing callback. If sax is NULL, fallback to the default
13081
 * DOM tree building routines.
13082
 *
13083
 * Returns the resulting document tree
13084
 */
13085
xmlDocPtr
13086
xmlSAXParseMemory(xmlSAXHandlerPtr sax, const char *buffer,
13087
0
            int size, int recovery) {
13088
0
    return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL);
13089
0
}
13090
13091
/**
13092
 * xmlParseMemory:
13093
 * @buffer:  an pointer to a char array
13094
 * @size:  the size of the array
13095
 *
13096
 * DEPRECATED: Use xmlReadMemory.
13097
 *
13098
 * parse an XML in-memory block and build a tree.
13099
 *
13100
 * Returns the resulting document tree
13101
 */
13102
13103
0
xmlDocPtr xmlParseMemory(const char *buffer, int size) {
13104
0
   return(xmlSAXParseMemory(NULL, buffer, size, 0));
13105
0
}
13106
13107
/**
13108
 * xmlRecoverMemory:
13109
 * @buffer:  an pointer to a char array
13110
 * @size:  the size of the array
13111
 *
13112
 * DEPRECATED: Use xmlReadMemory with XML_PARSE_RECOVER.
13113
 *
13114
 * parse an XML in-memory block and build a tree.
13115
 * In the case the document is not Well Formed, an attempt to
13116
 * build a tree is tried anyway
13117
 *
13118
 * Returns the resulting document tree or NULL in case of error
13119
 */
13120
13121
0
xmlDocPtr xmlRecoverMemory(const char *buffer, int size) {
13122
0
   return(xmlSAXParseMemory(NULL, buffer, size, 1));
13123
0
}
13124
13125
/**
13126
 * xmlSAXUserParseMemory:
13127
 * @sax:  a SAX handler
13128
 * @user_data:  The user data returned on SAX callbacks
13129
 * @buffer:  an in-memory XML document input
13130
 * @size:  the length of the XML document in bytes
13131
 *
13132
 * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory.
13133
 *
13134
 * parse an XML in-memory buffer and call the given SAX handler routines.
13135
 *
13136
 * Returns 0 in case of success or a error number otherwise
13137
 */
13138
int xmlSAXUserParseMemory(xmlSAXHandlerPtr sax, void *user_data,
13139
0
        const char *buffer, int size) {
13140
0
    int ret = 0;
13141
0
    xmlParserCtxtPtr ctxt;
13142
13143
0
    ctxt = xmlCreateMemoryParserCtxt(buffer, size);
13144
0
    if (ctxt == NULL) return -1;
13145
0
    if (sax != NULL) {
13146
0
        if (sax->initialized == XML_SAX2_MAGIC) {
13147
0
            *ctxt->sax = *sax;
13148
0
        } else {
13149
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
13150
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
13151
0
        }
13152
0
  ctxt->userData = user_data;
13153
0
    }
13154
13155
0
    xmlParseDocument(ctxt);
13156
13157
0
    if (ctxt->wellFormed)
13158
0
  ret = 0;
13159
0
    else {
13160
0
        if (ctxt->errNo != 0)
13161
0
      ret = ctxt->errNo;
13162
0
  else
13163
0
      ret = -1;
13164
0
    }
13165
0
    if (ctxt->myDoc != NULL) {
13166
0
        xmlFreeDoc(ctxt->myDoc);
13167
0
  ctxt->myDoc = NULL;
13168
0
    }
13169
0
    xmlFreeParserCtxt(ctxt);
13170
13171
0
    return ret;
13172
0
}
13173
#endif /* LIBXML_SAX1_ENABLED */
13174
13175
/**
13176
 * xmlCreateDocParserCtxt:
13177
 * @str:  a pointer to an array of xmlChar
13178
 *
13179
 * Creates a parser context for an XML in-memory document.
13180
 *
13181
 * Returns the new parser context or NULL
13182
 */
13183
xmlParserCtxtPtr
13184
0
xmlCreateDocParserCtxt(const xmlChar *str) {
13185
0
    xmlParserCtxtPtr ctxt;
13186
0
    xmlParserInputPtr input;
13187
13188
0
    ctxt = xmlNewParserCtxt();
13189
0
    if (ctxt == NULL)
13190
0
  return(NULL);
13191
13192
0
    input = xmlCtxtNewInputFromString(ctxt, NULL, (const char *) str, NULL, 0);
13193
0
    if (input == NULL) {
13194
0
  xmlFreeParserCtxt(ctxt);
13195
0
  return(NULL);
13196
0
    }
13197
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13198
0
        xmlFreeInputStream(input);
13199
0
        xmlFreeParserCtxt(ctxt);
13200
0
        return(NULL);
13201
0
    }
13202
13203
0
    return(ctxt);
13204
0
}
13205
13206
#ifdef LIBXML_SAX1_ENABLED
13207
/**
13208
 * xmlSAXParseDoc:
13209
 * @sax:  the SAX handler block
13210
 * @cur:  a pointer to an array of xmlChar
13211
 * @recovery:  work in recovery mode, i.e. tries to read no Well Formed
13212
 *             documents
13213
 *
13214
 * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadDoc.
13215
 *
13216
 * parse an XML in-memory document and build a tree.
13217
 * It use the given SAX function block to handle the parsing callback.
13218
 * If sax is NULL, fallback to the default DOM tree building routines.
13219
 *
13220
 * Returns the resulting document tree
13221
 */
13222
13223
xmlDocPtr
13224
0
xmlSAXParseDoc(xmlSAXHandlerPtr sax, const xmlChar *cur, int recovery) {
13225
0
    xmlDocPtr ret;
13226
0
    xmlParserCtxtPtr ctxt;
13227
0
    xmlSAXHandlerPtr oldsax = NULL;
13228
13229
0
    if (cur == NULL) return(NULL);
13230
13231
13232
0
    ctxt = xmlCreateDocParserCtxt(cur);
13233
0
    if (ctxt == NULL) return(NULL);
13234
0
    if (sax != NULL) {
13235
0
        oldsax = ctxt->sax;
13236
0
        ctxt->sax = sax;
13237
0
        ctxt->userData = NULL;
13238
0
    }
13239
13240
0
    xmlParseDocument(ctxt);
13241
0
    if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc;
13242
0
    else {
13243
0
       ret = NULL;
13244
0
       xmlFreeDoc(ctxt->myDoc);
13245
0
       ctxt->myDoc = NULL;
13246
0
    }
13247
0
    if (sax != NULL)
13248
0
  ctxt->sax = oldsax;
13249
0
    xmlFreeParserCtxt(ctxt);
13250
13251
0
    return(ret);
13252
0
}
13253
13254
/**
13255
 * xmlParseDoc:
13256
 * @cur:  a pointer to an array of xmlChar
13257
 *
13258
 * DEPRECATED: Use xmlReadDoc.
13259
 *
13260
 * parse an XML in-memory document and build a tree.
13261
 *
13262
 * Returns the resulting document tree
13263
 */
13264
13265
xmlDocPtr
13266
0
xmlParseDoc(const xmlChar *cur) {
13267
0
    return(xmlSAXParseDoc(NULL, cur, 0));
13268
0
}
13269
#endif /* LIBXML_SAX1_ENABLED */
13270
13271
/************************************************************************
13272
 *                  *
13273
 *  New set (2.6.0) of simpler and more flexible APIs   *
13274
 *                  *
13275
 ************************************************************************/
13276
13277
/**
13278
 * DICT_FREE:
13279
 * @str:  a string
13280
 *
13281
 * Free a string if it is not owned by the "dict" dictionary in the
13282
 * current scope
13283
 */
13284
#define DICT_FREE(str)            \
13285
27.1k
  if ((str) && ((!dict) ||       \
13286
0
      (xmlDictOwns(dict, (const xmlChar *)(str)) == 0))) \
13287
27.1k
      xmlFree((char *)(str));
13288
13289
/**
13290
 * xmlCtxtReset:
13291
 * @ctxt: an XML parser context
13292
 *
13293
 * Reset a parser context
13294
 */
13295
void
13296
xmlCtxtReset(xmlParserCtxtPtr ctxt)
13297
6.77k
{
13298
6.77k
    xmlParserInputPtr input;
13299
6.77k
    xmlDictPtr dict;
13300
13301
6.77k
    if (ctxt == NULL)
13302
0
        return;
13303
13304
6.77k
    dict = ctxt->dict;
13305
13306
6.77k
    while ((input = xmlCtxtPopInput(ctxt)) != NULL) { /* Non consuming */
13307
0
        xmlFreeInputStream(input);
13308
0
    }
13309
6.77k
    ctxt->inputNr = 0;
13310
6.77k
    ctxt->input = NULL;
13311
13312
6.77k
    ctxt->spaceNr = 0;
13313
6.77k
    if (ctxt->spaceTab != NULL) {
13314
6.77k
  ctxt->spaceTab[0] = -1;
13315
6.77k
  ctxt->space = &ctxt->spaceTab[0];
13316
6.77k
    } else {
13317
0
        ctxt->space = NULL;
13318
0
    }
13319
13320
13321
6.77k
    ctxt->nodeNr = 0;
13322
6.77k
    ctxt->node = NULL;
13323
13324
6.77k
    ctxt->nameNr = 0;
13325
6.77k
    ctxt->name = NULL;
13326
13327
6.77k
    ctxt->nsNr = 0;
13328
6.77k
    xmlParserNsReset(ctxt->nsdb);
13329
13330
6.77k
    DICT_FREE(ctxt->version);
13331
6.77k
    ctxt->version = NULL;
13332
6.77k
    DICT_FREE(ctxt->encoding);
13333
6.77k
    ctxt->encoding = NULL;
13334
6.77k
    DICT_FREE(ctxt->extSubURI);
13335
6.77k
    ctxt->extSubURI = NULL;
13336
6.77k
    DICT_FREE(ctxt->extSubSystem);
13337
6.77k
    ctxt->extSubSystem = NULL;
13338
13339
6.77k
    if (ctxt->directory != NULL) {
13340
0
        xmlFree(ctxt->directory);
13341
0
        ctxt->directory = NULL;
13342
0
    }
13343
13344
6.77k
    if (ctxt->myDoc != NULL)
13345
0
        xmlFreeDoc(ctxt->myDoc);
13346
6.77k
    ctxt->myDoc = NULL;
13347
13348
6.77k
    ctxt->standalone = -1;
13349
6.77k
    ctxt->hasExternalSubset = 0;
13350
6.77k
    ctxt->hasPErefs = 0;
13351
6.77k
    ctxt->html = 0;
13352
6.77k
    ctxt->instate = XML_PARSER_START;
13353
13354
6.77k
    ctxt->wellFormed = 1;
13355
6.77k
    ctxt->nsWellFormed = 1;
13356
6.77k
    ctxt->disableSAX = 0;
13357
6.77k
    ctxt->valid = 1;
13358
6.77k
    ctxt->record_info = 0;
13359
6.77k
    ctxt->checkIndex = 0;
13360
6.77k
    ctxt->endCheckState = 0;
13361
6.77k
    ctxt->inSubset = 0;
13362
6.77k
    ctxt->errNo = XML_ERR_OK;
13363
6.77k
    ctxt->depth = 0;
13364
6.77k
    ctxt->catalogs = NULL;
13365
6.77k
    ctxt->sizeentities = 0;
13366
6.77k
    ctxt->sizeentcopy = 0;
13367
6.77k
    xmlInitNodeInfoSeq(&ctxt->node_seq);
13368
13369
6.77k
    if (ctxt->attsDefault != NULL) {
13370
0
        xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator);
13371
0
        ctxt->attsDefault = NULL;
13372
0
    }
13373
6.77k
    if (ctxt->attsSpecial != NULL) {
13374
0
        xmlHashFree(ctxt->attsSpecial, NULL);
13375
0
        ctxt->attsSpecial = NULL;
13376
0
    }
13377
13378
6.77k
#ifdef LIBXML_CATALOG_ENABLED
13379
6.77k
    if (ctxt->catalogs != NULL)
13380
0
  xmlCatalogFreeLocal(ctxt->catalogs);
13381
6.77k
#endif
13382
6.77k
    ctxt->nbErrors = 0;
13383
6.77k
    ctxt->nbWarnings = 0;
13384
6.77k
    if (ctxt->lastError.code != XML_ERR_OK)
13385
0
        xmlResetError(&ctxt->lastError);
13386
6.77k
}
13387
13388
/**
13389
 * xmlCtxtResetPush:
13390
 * @ctxt: an XML parser context
13391
 * @chunk:  a pointer to an array of chars
13392
 * @size:  number of chars in the array
13393
 * @filename:  an optional file name or URI
13394
 * @encoding:  the document encoding, or NULL
13395
 *
13396
 * Reset a push parser context
13397
 *
13398
 * Returns 0 in case of success and 1 in case of error
13399
 */
13400
int
13401
xmlCtxtResetPush(xmlParserCtxtPtr ctxt, const char *chunk,
13402
                 int size, const char *filename, const char *encoding)
13403
0
{
13404
0
    xmlParserInputPtr input;
13405
13406
0
    if (ctxt == NULL)
13407
0
        return(1);
13408
13409
0
    xmlCtxtReset(ctxt);
13410
13411
0
    input = xmlNewPushInput(filename, chunk, size);
13412
0
    if (input == NULL)
13413
0
        return(1);
13414
13415
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13416
0
        xmlFreeInputStream(input);
13417
0
        return(1);
13418
0
    }
13419
13420
0
    if (encoding != NULL)
13421
0
        xmlSwitchEncodingName(ctxt, encoding);
13422
13423
0
    return(0);
13424
0
}
13425
13426
static int
13427
xmlCtxtSetOptionsInternal(xmlParserCtxtPtr ctxt, int options, int keepMask)
13428
272k
{
13429
272k
    int allMask;
13430
13431
272k
    if (ctxt == NULL)
13432
0
        return(-1);
13433
13434
    /*
13435
     * XInclude options aren't handled by the parser.
13436
     *
13437
     * XML_PARSE_XINCLUDE
13438
     * XML_PARSE_NOXINCNODE
13439
     * XML_PARSE_NOBASEFIX
13440
     */
13441
272k
    allMask = XML_PARSE_RECOVER |
13442
272k
              XML_PARSE_NOENT |
13443
272k
              XML_PARSE_DTDLOAD |
13444
272k
              XML_PARSE_DTDATTR |
13445
272k
              XML_PARSE_DTDVALID |
13446
272k
              XML_PARSE_NOERROR |
13447
272k
              XML_PARSE_NOWARNING |
13448
272k
              XML_PARSE_PEDANTIC |
13449
272k
              XML_PARSE_NOBLANKS |
13450
272k
#ifdef LIBXML_SAX1_ENABLED
13451
272k
              XML_PARSE_SAX1 |
13452
272k
#endif
13453
272k
              XML_PARSE_NONET |
13454
272k
              XML_PARSE_NODICT |
13455
272k
              XML_PARSE_NSCLEAN |
13456
272k
              XML_PARSE_NOCDATA |
13457
272k
              XML_PARSE_COMPACT |
13458
272k
              XML_PARSE_OLD10 |
13459
272k
              XML_PARSE_HUGE |
13460
272k
              XML_PARSE_OLDSAX |
13461
272k
              XML_PARSE_IGNORE_ENC |
13462
272k
              XML_PARSE_BIG_LINES |
13463
272k
              XML_PARSE_NO_XXE |
13464
272k
              XML_PARSE_UNZIP |
13465
272k
              XML_PARSE_NO_SYS_CATALOG |
13466
272k
              XML_PARSE_CATALOG_PI;
13467
13468
272k
    ctxt->options = (ctxt->options & keepMask) | (options & allMask);
13469
13470
    /*
13471
     * For some options, struct members are historically the source
13472
     * of truth. The values are initalized from global variables and
13473
     * old code could also modify them directly. Several older API
13474
     * functions that don't take an options argument rely on these
13475
     * deprecated mechanisms.
13476
     *
13477
     * Once public access to struct members and the globals are
13478
     * disabled, we can use the options bitmask as source of
13479
     * truth, making all these struct members obsolete.
13480
     *
13481
     * The XML_DETECT_IDS flags is misnamed. It simply enables
13482
     * loading of the external subset.
13483
     */
13484
272k
    ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0;
13485
272k
    ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0;
13486
272k
    ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0;
13487
272k
    ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0;
13488
272k
    ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0;
13489
272k
    ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0;
13490
272k
    ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1;
13491
272k
    ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1;
13492
13493
272k
    if (options & XML_PARSE_HUGE) {
13494
265k
        if (ctxt->dict != NULL)
13495
265k
            xmlDictSetLimit(ctxt->dict, 0);
13496
265k
    }
13497
13498
272k
    ctxt->linenumbers = 1;
13499
13500
272k
    return(options & ~allMask);
13501
272k
}
13502
13503
/**
13504
 * xmlCtxtSetOptions:
13505
 * @ctxt: an XML parser context
13506
 * @options:  a bitmask of xmlParserOption values
13507
 *
13508
 * Applies the options to the parser context. Unset options are
13509
 * cleared.
13510
 *
13511
 * Available since 2.13.0. With older versions, you can use
13512
 * xmlCtxtUseOptions.
13513
 *
13514
 * XML_PARSE_RECOVER
13515
 *
13516
 * Enable "recovery" mode which allows non-wellformed documents.
13517
 * How this mode behaves exactly is unspecified and may change
13518
 * without further notice. Use of this feature is DISCOURAGED.
13519
 *
13520
 * Not supported by the push parser.
13521
 *
13522
 * XML_PARSE_NOENT
13523
 *
13524
 * Despite the confusing name, this option enables substitution
13525
 * of entities. The resulting tree won't contain any entity
13526
 * reference nodes.
13527
 *
13528
 * This option also enables loading of external entities (both
13529
 * general and parameter entities) which is dangerous. If you
13530
 * process untrusted data, it's recommended to set the
13531
 * XML_PARSE_NO_XXE option to disable loading of external
13532
 * entities.
13533
 *
13534
 * XML_PARSE_DTDLOAD
13535
 *
13536
 * Enables loading of an external DTD and the loading and
13537
 * substitution of external parameter entities. Has no effect
13538
 * if XML_PARSE_NO_XXE is set.
13539
 *
13540
 * XML_PARSE_DTDATTR
13541
 *
13542
 * Adds default attributes from the DTD to the result document.
13543
 *
13544
 * Implies XML_PARSE_DTDLOAD, but loading of external content
13545
 * can be disabled with XML_PARSE_NO_XXE.
13546
 *
13547
 * XML_PARSE_DTDVALID
13548
 *
13549
 * This option enables DTD validation which requires to load
13550
 * external DTDs and external entities (both general and
13551
 * parameter entities) unless XML_PARSE_NO_XXE was set.
13552
 *
13553
 * XML_PARSE_NO_XXE
13554
 *
13555
 * Disables loading of external DTDs or entities.
13556
 *
13557
 * Available since 2.13.0.
13558
 *
13559
 * XML_PARSE_NOERROR
13560
 *
13561
 * Disable error and warning reports to the error handlers.
13562
 * Errors are still accessible with xmlCtxtGetLastError.
13563
 *
13564
 * XML_PARSE_NOWARNING
13565
 *
13566
 * Disable warning reports.
13567
 *
13568
 * XML_PARSE_PEDANTIC
13569
 *
13570
 * Enable some pedantic warnings.
13571
 *
13572
 * XML_PARSE_NOBLANKS
13573
 *
13574
 * Remove some whitespace from the result document. Where to
13575
 * remove whitespace depends on DTD element declarations or a
13576
 * broken heuristic with unfixable bugs. Use of this option is
13577
 * DISCOURAGED.
13578
 *
13579
 * Not supported by the push parser.
13580
 *
13581
 * XML_PARSE_SAX1
13582
 *
13583
 * Always invoke the deprecated SAX1 startElement and endElement
13584
 * handlers. This option is DEPRECATED.
13585
 *
13586
 * XML_PARSE_NONET
13587
 *
13588
 * Disable network access with the builtin HTTP client.
13589
 *
13590
 * XML_PARSE_NODICT
13591
 *
13592
 * Create a document without interned strings, making all
13593
 * strings separate memory allocations.
13594
 *
13595
 * XML_PARSE_NSCLEAN
13596
 *
13597
 * Remove redundant namespace declarations from the result
13598
 * document.
13599
 *
13600
 * XML_PARSE_NOCDATA
13601
 *
13602
 * Output normal text nodes instead of CDATA nodes.
13603
 *
13604
 * XML_PARSE_COMPACT
13605
 *
13606
 * Store small strings directly in the node struct to save
13607
 * memory.
13608
 *
13609
 * XML_PARSE_OLD10
13610
 *
13611
 * Use old Name productions from before XML 1.0 Fifth Edition.
13612
 * This options is DEPRECATED.
13613
 *
13614
 * XML_PARSE_HUGE
13615
 *
13616
 * Relax some internal limits.
13617
 *
13618
 * Maximum size of text nodes, tags, comments, processing instructions,
13619
 * CDATA sections, entity values
13620
 *
13621
 * normal: 10M
13622
 * huge:    1B
13623
 *
13624
 * Maximum size of names, system literals, pubid literals
13625
 *
13626
 * normal: 50K
13627
 * huge:   10M
13628
 *
13629
 * Maximum nesting depth of elements
13630
 *
13631
 * normal:  256
13632
 * huge:   2048
13633
 *
13634
 * Maximum nesting depth of entities
13635
 *
13636
 * normal: 20
13637
 * huge:   40
13638
 *
13639
 * XML_PARSE_OLDSAX
13640
 *
13641
 * Enable an unspecified legacy mode for SAX parsers. This
13642
 * option is DEPRECATED.
13643
 *
13644
 * XML_PARSE_IGNORE_ENC
13645
 *
13646
 * Ignore the encoding in the XML declaration. This option is
13647
 * mostly unneeded these days. The only effect is to enforce
13648
 * UTF-8 decoding of ASCII-like data.
13649
 *
13650
 * XML_PARSE_BIG_LINES
13651
 *
13652
 * Enable reporting of line numbers larger than 65535.
13653
 *
13654
 * XML_PARSE_UNZIP
13655
 *
13656
 * Enable input decompression. Setting this option is discouraged
13657
 * to avoid zip bombs.
13658
 *
13659
 * Available since 2.14.0.
13660
 *
13661
 * XML_PARSE_NO_SYS_CATALOG
13662
 *
13663
 * Disables the global system XML catalog.
13664
 *
13665
 * Available since 2.14.0.
13666
 *
13667
 * XML_PARSE_CATALOG_PI
13668
 *
13669
 * Enable XML catalog processing instructions.
13670
 *
13671
 * Available since 2.14.0.
13672
 *
13673
 * Returns 0 in case of success, the set of unknown or unimplemented options
13674
 *         in case of error.
13675
 */
13676
int
13677
xmlCtxtSetOptions(xmlParserCtxtPtr ctxt, int options)
13678
0
{
13679
0
#ifdef LIBXML_HTML_ENABLED
13680
0
    if ((ctxt != NULL) && (ctxt->html))
13681
0
        return(htmlCtxtSetOptions(ctxt, options));
13682
0
#endif
13683
13684
0
    return(xmlCtxtSetOptionsInternal(ctxt, options, 0));
13685
0
}
13686
13687
/**
13688
 * xmlCtxtGetOptions:
13689
 * @ctxt: an XML parser context
13690
 *
13691
 * Get the current options of the parser context.
13692
 *
13693
 * Available since 2.14.0.
13694
 *
13695
 * Returns the current options set in the parser context, or -1 if ctxt is NULL.
13696
 */
13697
int
13698
xmlCtxtGetOptions(xmlParserCtxtPtr ctxt)
13699
0
{
13700
0
    if (ctxt == NULL)
13701
0
        return(-1);
13702
13703
0
    return(ctxt->options);
13704
0
}
13705
13706
/**
13707
 * xmlCtxtUseOptions:
13708
 * @ctxt: an XML parser context
13709
 * @options:  a combination of xmlParserOption
13710
 *
13711
 * DEPRECATED: Use xmlCtxtSetOptions.
13712
 *
13713
 * Applies the options to the parser context. The following options
13714
 * are never cleared and can only be enabled:
13715
 *
13716
 * XML_PARSE_NOERROR
13717
 * XML_PARSE_NOWARNING
13718
 * XML_PARSE_NONET
13719
 * XML_PARSE_NSCLEAN
13720
 * XML_PARSE_NOCDATA
13721
 * XML_PARSE_COMPACT
13722
 * XML_PARSE_OLD10
13723
 * XML_PARSE_HUGE
13724
 * XML_PARSE_OLDSAX
13725
 * XML_PARSE_IGNORE_ENC
13726
 * XML_PARSE_BIG_LINES
13727
 *
13728
 * Returns 0 in case of success, the set of unknown or unimplemented options
13729
 *         in case of error.
13730
 */
13731
int
13732
xmlCtxtUseOptions(xmlParserCtxtPtr ctxt, int options)
13733
272k
{
13734
272k
    int keepMask;
13735
13736
272k
#ifdef LIBXML_HTML_ENABLED
13737
272k
    if ((ctxt != NULL) && (ctxt->html))
13738
0
        return(htmlCtxtUseOptions(ctxt, options));
13739
272k
#endif
13740
13741
    /*
13742
     * For historic reasons, some options can only be enabled.
13743
     */
13744
272k
    keepMask = XML_PARSE_NOERROR |
13745
272k
               XML_PARSE_NOWARNING |
13746
272k
               XML_PARSE_NONET |
13747
272k
               XML_PARSE_NSCLEAN |
13748
272k
               XML_PARSE_NOCDATA |
13749
272k
               XML_PARSE_COMPACT |
13750
272k
               XML_PARSE_OLD10 |
13751
272k
               XML_PARSE_HUGE |
13752
272k
               XML_PARSE_OLDSAX |
13753
272k
               XML_PARSE_IGNORE_ENC |
13754
272k
               XML_PARSE_BIG_LINES;
13755
13756
272k
    return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask));
13757
272k
}
13758
13759
/**
13760
 * xmlCtxtSetMaxAmplification:
13761
 * @ctxt: an XML parser context
13762
 * @maxAmpl:  maximum amplification factor
13763
 *
13764
 * To protect against exponential entity expansion ("billion laughs"), the
13765
 * size of serialized output is (roughly) limited to the input size
13766
 * multiplied by this factor. The default value is 5.
13767
 *
13768
 * When working with documents making heavy use of entity expansion, it can
13769
 * be necessary to increase the value. For security reasons, this should only
13770
 * be considered when processing trusted input.
13771
 */
13772
void
13773
xmlCtxtSetMaxAmplification(xmlParserCtxtPtr ctxt, unsigned maxAmpl)
13774
0
{
13775
0
    ctxt->maxAmpl = maxAmpl;
13776
0
}
13777
13778
/**
13779
 * xmlCtxtParseDocument:
13780
 * @ctxt:  an XML parser context
13781
 * @input:  parser input
13782
 *
13783
 * Parse an XML document and return the resulting document tree.
13784
 * Takes ownership of the input object.
13785
 *
13786
 * Available since 2.13.0.
13787
 *
13788
 * Returns the resulting document tree or NULL
13789
 */
13790
xmlDocPtr
13791
xmlCtxtParseDocument(xmlParserCtxtPtr ctxt, xmlParserInputPtr input)
13792
6.77k
{
13793
6.77k
    xmlDocPtr ret = NULL;
13794
13795
6.77k
    if ((ctxt == NULL) || (input == NULL)) {
13796
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
13797
0
        xmlFreeInputStream(input);
13798
0
        return(NULL);
13799
0
    }
13800
13801
    /* assert(ctxt->inputNr == 0); */
13802
6.77k
    while (ctxt->inputNr > 0)
13803
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13804
13805
6.77k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13806
0
        xmlFreeInputStream(input);
13807
0
        return(NULL);
13808
0
    }
13809
13810
6.77k
    xmlParseDocument(ctxt);
13811
13812
6.77k
    ret = xmlCtxtGetDocument(ctxt);
13813
13814
    /* assert(ctxt->inputNr == 1); */
13815
13.5k
    while (ctxt->inputNr > 0)
13816
6.77k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13817
13818
6.77k
    return(ret);
13819
6.77k
}
13820
13821
/**
13822
 * xmlReadDoc:
13823
 * @cur:  a pointer to a zero terminated string
13824
 * @URL:  base URL (optional)
13825
 * @encoding:  the document encoding (optional)
13826
 * @options:  a combination of xmlParserOption
13827
 *
13828
 * Convenience function to parse an XML document from a
13829
 * zero-terminated string.
13830
 *
13831
 * See xmlCtxtReadDoc for details.
13832
 *
13833
 * Returns the resulting document tree
13834
 */
13835
xmlDocPtr
13836
xmlReadDoc(const xmlChar *cur, const char *URL, const char *encoding,
13837
           int options)
13838
0
{
13839
0
    xmlParserCtxtPtr ctxt;
13840
0
    xmlParserInputPtr input;
13841
0
    xmlDocPtr doc = NULL;
13842
13843
0
    ctxt = xmlNewParserCtxt();
13844
0
    if (ctxt == NULL)
13845
0
        return(NULL);
13846
13847
0
    xmlCtxtUseOptions(ctxt, options);
13848
13849
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) cur, encoding,
13850
0
                                      XML_INPUT_BUF_STATIC);
13851
13852
0
    if (input != NULL)
13853
0
        doc = xmlCtxtParseDocument(ctxt, input);
13854
13855
0
    xmlFreeParserCtxt(ctxt);
13856
0
    return(doc);
13857
0
}
13858
13859
/**
13860
 * xmlReadFile:
13861
 * @filename:  a file or URL
13862
 * @encoding:  the document encoding (optional)
13863
 * @options:  a combination of xmlParserOption
13864
 *
13865
 * Convenience function to parse an XML file from the filesystem,
13866
 * the network or a global user-define resource loader.
13867
 *
13868
 * This function always enables the XML_PARSE_UNZIP option for
13869
 * backward compatibility. If a "-" filename is passed, it will
13870
 * read from stdin. Both of these features are potentially
13871
 * insecure and might be removed from later versions.
13872
 *
13873
 * See xmlCtxtReadFile for details.
13874
 *
13875
 * Returns the resulting document tree
13876
 */
13877
xmlDocPtr
13878
xmlReadFile(const char *filename, const char *encoding, int options)
13879
0
{
13880
0
    xmlParserCtxtPtr ctxt;
13881
0
    xmlParserInputPtr input;
13882
0
    xmlDocPtr doc = NULL;
13883
13884
0
    ctxt = xmlNewParserCtxt();
13885
0
    if (ctxt == NULL)
13886
0
        return(NULL);
13887
13888
0
    options |= XML_PARSE_UNZIP;
13889
13890
0
    xmlCtxtUseOptions(ctxt, options);
13891
13892
    /*
13893
     * Backward compatibility for users of command line utilities like
13894
     * xmlstarlet expecting "-" to mean stdin. This is dangerous and
13895
     * should be removed at some point.
13896
     */
13897
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
13898
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO,
13899
0
                                      encoding, 0);
13900
0
    else
13901
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13902
13903
0
    if (input != NULL)
13904
0
        doc = xmlCtxtParseDocument(ctxt, input);
13905
13906
0
    xmlFreeParserCtxt(ctxt);
13907
0
    return(doc);
13908
0
}
13909
13910
/**
13911
 * xmlReadMemory:
13912
 * @buffer:  a pointer to a char array
13913
 * @size:  the size of the array
13914
 * @url:  base URL (optional)
13915
 * @encoding:  the document encoding (optional)
13916
 * @options:  a combination of xmlParserOption
13917
 *
13918
 * Parse an XML in-memory document and build a tree. The input buffer must
13919
 * not contain a terminating null byte.
13920
 *
13921
 * See xmlCtxtReadMemory for details.
13922
 *
13923
 * Returns the resulting document tree
13924
 */
13925
xmlDocPtr
13926
xmlReadMemory(const char *buffer, int size, const char *url,
13927
              const char *encoding, int options)
13928
0
{
13929
0
    xmlParserCtxtPtr ctxt;
13930
0
    xmlParserInputPtr input;
13931
0
    xmlDocPtr doc = NULL;
13932
13933
0
    if (size < 0)
13934
0
  return(NULL);
13935
13936
0
    ctxt = xmlNewParserCtxt();
13937
0
    if (ctxt == NULL)
13938
0
        return(NULL);
13939
13940
0
    xmlCtxtUseOptions(ctxt, options);
13941
13942
0
    input = xmlCtxtNewInputFromMemory(ctxt, url, buffer, size, encoding,
13943
0
                                      XML_INPUT_BUF_STATIC);
13944
13945
0
    if (input != NULL)
13946
0
        doc = xmlCtxtParseDocument(ctxt, input);
13947
13948
0
    xmlFreeParserCtxt(ctxt);
13949
0
    return(doc);
13950
0
}
13951
13952
/**
13953
 * xmlReadFd:
13954
 * @fd:  an open file descriptor
13955
 * @URL:  base URL (optional)
13956
 * @encoding:  the document encoding (optional)
13957
 * @options:  a combination of xmlParserOption
13958
 *
13959
 * Parse an XML from a file descriptor and build a tree.
13960
 *
13961
 * See xmlCtxtReadFd for details.
13962
 *
13963
 * NOTE that the file descriptor will not be closed when the
13964
 * context is freed or reset.
13965
 *
13966
 * Returns the resulting document tree
13967
 */
13968
xmlDocPtr
13969
xmlReadFd(int fd, const char *URL, const char *encoding, int options)
13970
0
{
13971
0
    xmlParserCtxtPtr ctxt;
13972
0
    xmlParserInputPtr input;
13973
0
    xmlDocPtr doc = NULL;
13974
13975
0
    ctxt = xmlNewParserCtxt();
13976
0
    if (ctxt == NULL)
13977
0
        return(NULL);
13978
13979
0
    xmlCtxtUseOptions(ctxt, options);
13980
13981
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
13982
13983
0
    if (input != NULL)
13984
0
        doc = xmlCtxtParseDocument(ctxt, input);
13985
13986
0
    xmlFreeParserCtxt(ctxt);
13987
0
    return(doc);
13988
0
}
13989
13990
/**
13991
 * xmlReadIO:
13992
 * @ioread:  an I/O read function
13993
 * @ioclose:  an I/O close function (optional)
13994
 * @ioctx:  an I/O handler
13995
 * @URL:  base URL (optional)
13996
 * @encoding:  the document encoding (optional)
13997
 * @options:  a combination of xmlParserOption
13998
 *
13999
 * Parse an XML document from I/O functions and context and build a tree.
14000
 *
14001
 * See xmlCtxtReadIO for details.
14002
 *
14003
 * Returns the resulting document tree
14004
 */
14005
xmlDocPtr
14006
xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose,
14007
          void *ioctx, const char *URL, const char *encoding, int options)
14008
0
{
14009
0
    xmlParserCtxtPtr ctxt;
14010
0
    xmlParserInputPtr input;
14011
0
    xmlDocPtr doc = NULL;
14012
14013
0
    ctxt = xmlNewParserCtxt();
14014
0
    if (ctxt == NULL)
14015
0
        return(NULL);
14016
14017
0
    xmlCtxtUseOptions(ctxt, options);
14018
14019
0
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
14020
0
                                  encoding, 0);
14021
14022
0
    if (input != NULL)
14023
0
        doc = xmlCtxtParseDocument(ctxt, input);
14024
14025
0
    xmlFreeParserCtxt(ctxt);
14026
0
    return(doc);
14027
0
}
14028
14029
/**
14030
 * xmlCtxtReadDoc:
14031
 * @ctxt:  an XML parser context
14032
 * @str:  a pointer to a zero terminated string
14033
 * @URL:  base URL (optional)
14034
 * @encoding:  the document encoding (optional)
14035
 * @options:  a combination of xmlParserOption
14036
 *
14037
 * Parse an XML in-memory document and build a tree.
14038
 *
14039
 * @URL is used as base to resolve external entities and for error
14040
 * reporting.
14041
 *
14042
 * See xmlCtxtUseOptions for details.
14043
 *
14044
 * Returns the resulting document tree
14045
 */
14046
xmlDocPtr
14047
xmlCtxtReadDoc(xmlParserCtxtPtr ctxt, const xmlChar *str,
14048
               const char *URL, const char *encoding, int options)
14049
0
{
14050
0
    xmlParserInputPtr input;
14051
14052
0
    if (ctxt == NULL)
14053
0
        return(NULL);
14054
14055
0
    xmlCtxtReset(ctxt);
14056
0
    xmlCtxtUseOptions(ctxt, options);
14057
14058
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) str, encoding,
14059
0
                                      XML_INPUT_BUF_STATIC);
14060
0
    if (input == NULL)
14061
0
        return(NULL);
14062
14063
0
    return(xmlCtxtParseDocument(ctxt, input));
14064
0
}
14065
14066
/**
14067
 * xmlCtxtReadFile:
14068
 * @ctxt:  an XML parser context
14069
 * @filename:  a file or URL
14070
 * @encoding:  the document encoding (optional)
14071
 * @options:  a combination of xmlParserOption
14072
 *
14073
 * Parse an XML file from the filesystem, the network or a user-defined
14074
 * resource loader.
14075
 *
14076
 * This function always enables the XML_PARSE_UNZIP option for
14077
 * backward compatibility. This feature is potentially insecure
14078
 * and might be removed from later versions.
14079
 *
14080
 * Returns the resulting document tree
14081
 */
14082
xmlDocPtr
14083
xmlCtxtReadFile(xmlParserCtxtPtr ctxt, const char *filename,
14084
                const char *encoding, int options)
14085
0
{
14086
0
    xmlParserInputPtr input;
14087
14088
0
    if (ctxt == NULL)
14089
0
        return(NULL);
14090
14091
0
    options |= XML_PARSE_UNZIP;
14092
14093
0
    xmlCtxtReset(ctxt);
14094
0
    xmlCtxtUseOptions(ctxt, options);
14095
14096
0
    input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
14097
0
    if (input == NULL)
14098
0
        return(NULL);
14099
14100
0
    return(xmlCtxtParseDocument(ctxt, input));
14101
0
}
14102
14103
/**
14104
 * xmlCtxtReadMemory:
14105
 * @ctxt:  an XML parser context
14106
 * @buffer:  a pointer to a char array
14107
 * @size:  the size of the array
14108
 * @URL:  base URL (optional)
14109
 * @encoding:  the document encoding (optional)
14110
 * @options:  a combination of xmlParserOption
14111
 *
14112
 * Parse an XML in-memory document and build a tree. The input buffer must
14113
 * not contain a terminating null byte.
14114
 *
14115
 * @URL is used as base to resolve external entities and for error
14116
 * reporting.
14117
 *
14118
 * See xmlCtxtUseOptions for details.
14119
 *
14120
 * Returns the resulting document tree
14121
 */
14122
xmlDocPtr
14123
xmlCtxtReadMemory(xmlParserCtxtPtr ctxt, const char *buffer, int size,
14124
                  const char *URL, const char *encoding, int options)
14125
0
{
14126
0
    xmlParserInputPtr input;
14127
14128
0
    if ((ctxt == NULL) || (size < 0))
14129
0
        return(NULL);
14130
14131
0
    xmlCtxtReset(ctxt);
14132
0
    xmlCtxtUseOptions(ctxt, options);
14133
14134
0
    input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding,
14135
0
                                      XML_INPUT_BUF_STATIC);
14136
0
    if (input == NULL)
14137
0
        return(NULL);
14138
14139
0
    return(xmlCtxtParseDocument(ctxt, input));
14140
0
}
14141
14142
/**
14143
 * xmlCtxtReadFd:
14144
 * @ctxt:  an XML parser context
14145
 * @fd:  an open file descriptor
14146
 * @URL:  base URL (optional)
14147
 * @encoding:  the document encoding (optional)
14148
 * @options:  a combination of xmlParserOption
14149
 *
14150
 * Parse an XML document from a file descriptor and build a tree.
14151
 *
14152
 * NOTE that the file descriptor will not be closed when the
14153
 * context is freed or reset.
14154
 *
14155
 * @URL is used as base to resolve external entities and for error
14156
 * reporting.
14157
 *
14158
 * See xmlCtxtUseOptions for details.
14159
 *
14160
 * Returns the resulting document tree
14161
 */
14162
xmlDocPtr
14163
xmlCtxtReadFd(xmlParserCtxtPtr ctxt, int fd,
14164
              const char *URL, const char *encoding, int options)
14165
0
{
14166
0
    xmlParserInputPtr input;
14167
14168
0
    if (ctxt == NULL)
14169
0
        return(NULL);
14170
14171
0
    xmlCtxtReset(ctxt);
14172
0
    xmlCtxtUseOptions(ctxt, options);
14173
14174
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
14175
0
    if (input == NULL)
14176
0
        return(NULL);
14177
14178
0
    return(xmlCtxtParseDocument(ctxt, input));
14179
0
}
14180
14181
/**
14182
 * xmlCtxtReadIO:
14183
 * @ctxt:  an XML parser context
14184
 * @ioread:  an I/O read function
14185
 * @ioclose:  an I/O close function
14186
 * @ioctx:  an I/O handler
14187
 * @URL:  the base URL to use for the document
14188
 * @encoding:  the document encoding, or NULL
14189
 * @options:  a combination of xmlParserOption
14190
 *
14191
 * parse an XML document from I/O functions and source and build a tree.
14192
 * This reuses the existing @ctxt parser context
14193
 *
14194
 * @URL is used as base to resolve external entities and for error
14195
 * reporting.
14196
 *
14197
 * See xmlCtxtUseOptions for details.
14198
 *
14199
 * Returns the resulting document tree
14200
 */
14201
xmlDocPtr
14202
xmlCtxtReadIO(xmlParserCtxtPtr ctxt, xmlInputReadCallback ioread,
14203
              xmlInputCloseCallback ioclose, void *ioctx,
14204
        const char *URL,
14205
              const char *encoding, int options)
14206
6.77k
{
14207
6.77k
    xmlParserInputPtr input;
14208
14209
6.77k
    if (ctxt == NULL)
14210
0
        return(NULL);
14211
14212
6.77k
    xmlCtxtReset(ctxt);
14213
6.77k
    xmlCtxtUseOptions(ctxt, options);
14214
14215
6.77k
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
14216
6.77k
                                  encoding, 0);
14217
6.77k
    if (input == NULL)
14218
0
        return(NULL);
14219
14220
6.77k
    return(xmlCtxtParseDocument(ctxt, input));
14221
6.77k
}
14222