Coverage Report

Created: 2026-07-30 06:08

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/libxml2/parser.c
Line
Count
Source
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
 * Author: Daniel Veillard
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
#include "private/tree.h"
79
80
516k
#define NS_INDEX_EMPTY  INT_MAX
81
34.7k
#define NS_INDEX_XML    (INT_MAX - 1)
82
358k
#define URI_HASH_EMPTY  0xD943A04E
83
13.0k
#define URI_HASH_XML    0xF0451F02
84
85
#ifndef STDIN_FILENO
86
0
  #define STDIN_FILENO 0
87
#endif
88
89
#ifndef SIZE_MAX
90
  #define SIZE_MAX ((size_t) -1)
91
#endif
92
93
370k
#define XML_MAX_ATTRS 100000000 /* 100 million */
94
95
354k
#define XML_SPECIAL_EXTERNAL    (1 << 20)
96
309k
#define XML_SPECIAL_TYPE_MASK   (XML_SPECIAL_EXTERNAL - 1)
97
98
316k
#define XML_ATTVAL_ALLOC        (1 << 0)
99
701k
#define XML_ATTVAL_NORM_CHANGE  (1 << 1)
100
101
struct _xmlStartTag {
102
    const xmlChar *prefix;
103
    const xmlChar *URI;
104
    int line;
105
    int nsNr;
106
};
107
108
typedef struct {
109
    void *saxData;
110
    unsigned prefixHashValue;
111
    unsigned uriHashValue;
112
    unsigned elementId;
113
    int oldIndex;
114
} xmlParserNsExtra;
115
116
typedef struct {
117
    unsigned hashValue;
118
    int index;
119
} xmlParserNsBucket;
120
121
struct _xmlParserNsData {
122
    xmlParserNsExtra *extra;
123
124
    unsigned hashSize;
125
    unsigned hashElems;
126
    xmlParserNsBucket *hash;
127
128
    unsigned elementId;
129
    int defaultNsIndex;
130
    int minNsIndex;
131
};
132
133
static int
134
xmlParseElementStart(xmlParserCtxtPtr ctxt);
135
136
static void
137
xmlParseElementEnd(xmlParserCtxtPtr ctxt);
138
139
static xmlEntityPtr
140
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr);
141
142
static const xmlChar *
143
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt);
144
145
/************************************************************************
146
 *                  *
147
 *  Arbitrary limits set in the parser. See XML_PARSE_HUGE    *
148
 *                  *
149
 ************************************************************************/
150
151
#define XML_PARSER_BIG_ENTITY 1000
152
#define XML_PARSER_LOT_ENTITY 5000
153
154
/*
155
 * Constants for protection against abusive entity expansion
156
 * ("billion laughs").
157
 */
158
159
/*
160
 * A certain amount of entity expansion which is always allowed.
161
 */
162
2.22M
#define XML_PARSER_ALLOWED_EXPANSION 1000000
163
164
/*
165
 * Fixed cost for each entity reference. This crudely models processing time
166
 * as well to protect, for example, against exponential expansion of empty
167
 * or very short entities.
168
 */
169
2.23M
#define XML_ENT_FIXED_COST 20
170
171
171M
#define XML_PARSER_BIG_BUFFER_SIZE 300
172
911k
#define XML_PARSER_BUFFER_SIZE 100
173
122k
#define SAX_COMPAT_MODE BAD_CAST "SAX compatibility mode document"
174
175
/**
176
 * XML_PARSER_CHUNK_SIZE
177
 *
178
 * When calling GROW that's the minimal amount of data
179
 * the parser expected to have received. It is not a hard
180
 * limit but an optimization when reading strings like Names
181
 * It is not strictly needed as long as inputs available characters
182
 * are followed by 0, which should be provided by the I/O level
183
 */
184
#define XML_PARSER_CHUNK_SIZE 100
185
186
/**
187
 * Constant string describing the version of the library used at
188
 * run-time.
189
 */
190
const char *const
191
xmlParserVersion = LIBXML_VERSION_STRING LIBXML_VERSION_EXTRA;
192
193
/*
194
 * List of XML prefixed PI allowed by W3C specs
195
 */
196
197
static const char* const xmlW3CPIs[] = {
198
    "xml-stylesheet",
199
    "xml-model",
200
    NULL
201
};
202
203
204
/* DEPR void xmlParserHandleReference(xmlParserCtxtPtr ctxt); */
205
static xmlEntityPtr xmlParseStringPEReference(xmlParserCtxtPtr ctxt,
206
                                              const xmlChar **str);
207
208
static void
209
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent);
210
211
static int
212
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity);
213
214
static void
215
xmlParsePERefInternal(xmlParserCtxt *ctxt, int markupDecl);
216
217
/************************************************************************
218
 *                  *
219
 *    Some factorized error routines        *
220
 *                  *
221
 ************************************************************************/
222
223
static void
224
3.96k
xmlErrMemory(xmlParserCtxtPtr ctxt) {
225
3.96k
    xmlCtxtErrMemory(ctxt);
226
3.96k
}
227
228
/**
229
 * Handle a redefinition of attribute error
230
 *
231
 * @param ctxt  an XML parser context
232
 * @param prefix  the attribute prefix
233
 * @param localname  the attribute localname
234
 */
235
static void
236
xmlErrAttributeDup(xmlParserCtxtPtr ctxt, const xmlChar * prefix,
237
                   const xmlChar * localname)
238
32.1k
{
239
32.1k
    if (prefix == NULL)
240
27.9k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
241
27.9k
                   XML_ERR_FATAL, localname, NULL, NULL, 0,
242
27.9k
                   "Attribute %s redefined\n", localname);
243
4.15k
    else
244
4.15k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
245
4.15k
                   XML_ERR_FATAL, prefix, localname, NULL, 0,
246
4.15k
                   "Attribute %s:%s redefined\n", prefix, localname);
247
32.1k
}
248
249
/**
250
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
251
 *
252
 * @param ctxt  an XML parser context
253
 * @param error  the error number
254
 * @param msg  the error message
255
 */
256
static void LIBXML_ATTR_FORMAT(3,0)
257
xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
258
               const char *msg)
259
89.0M
{
260
89.0M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
261
89.0M
               NULL, NULL, NULL, 0, "%s", msg);
262
89.0M
}
263
264
/**
265
 * Handle a warning.
266
 *
267
 * @param ctxt  an XML parser context
268
 * @param error  the error number
269
 * @param msg  the error message
270
 * @param str1  extra data
271
 * @param str2  extra data
272
 */
273
void LIBXML_ATTR_FORMAT(3,0)
274
xmlWarningMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
275
              const char *msg, const xmlChar *str1, const xmlChar *str2)
276
15.2k
{
277
15.2k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING,
278
15.2k
               str1, str2, NULL, 0, msg, str1, str2);
279
15.2k
}
280
281
#ifdef LIBXML_VALID_ENABLED
282
/**
283
 * Handle a validity error.
284
 *
285
 * @param ctxt  an XML parser context
286
 * @param error  the error number
287
 * @param msg  the error message
288
 * @param str1  extra data
289
 * @param str2  extra data
290
 */
291
static void LIBXML_ATTR_FORMAT(3,0)
292
xmlValidityError(xmlParserCtxtPtr ctxt, xmlParserErrors error,
293
              const char *msg, const xmlChar *str1, const xmlChar *str2)
294
19.2k
{
295
19.2k
    ctxt->valid = 0;
296
297
19.2k
    xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR,
298
19.2k
               str1, str2, NULL, 0, msg, str1, str2);
299
19.2k
}
300
#endif
301
302
/**
303
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
304
 *
305
 * @param ctxt  an XML parser context
306
 * @param error  the error number
307
 * @param msg  the error message
308
 * @param val  an integer value
309
 */
310
static void LIBXML_ATTR_FORMAT(3,0)
311
xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error,
312
                  const char *msg, int val)
313
16.5M
{
314
16.5M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
315
16.5M
               NULL, NULL, NULL, val, msg, val);
316
16.5M
}
317
318
/**
319
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
320
 *
321
 * @param ctxt  an XML parser context
322
 * @param error  the error number
323
 * @param msg  the error message
324
 * @param str1  an string info
325
 * @param val  an integer value
326
 * @param str2  an string info
327
 */
328
static void LIBXML_ATTR_FORMAT(3,0)
329
xmlFatalErrMsgStrIntStr(xmlParserCtxtPtr ctxt, xmlParserErrors error,
330
                  const char *msg, const xmlChar *str1, int val,
331
      const xmlChar *str2)
332
437k
{
333
437k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
334
437k
               str1, str2, NULL, val, msg, str1, val, str2);
335
437k
}
336
337
/**
338
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
339
 *
340
 * @param ctxt  an XML parser context
341
 * @param error  the error number
342
 * @param msg  the error message
343
 * @param val  a string value
344
 */
345
static void LIBXML_ATTR_FORMAT(3,0)
346
xmlFatalErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error,
347
                  const char *msg, const xmlChar * val)
348
988k
{
349
988k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
350
988k
               val, NULL, NULL, 0, msg, val);
351
988k
}
352
353
/**
354
 * Handle a non fatal parser error
355
 *
356
 * @param ctxt  an XML parser context
357
 * @param error  the error number
358
 * @param msg  the error message
359
 * @param val  a string value
360
 */
361
static void LIBXML_ATTR_FORMAT(3,0)
362
xmlErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error,
363
                  const char *msg, const xmlChar * val)
364
7.89k
{
365
7.89k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR,
366
7.89k
               val, NULL, NULL, 0, msg, val);
367
7.89k
}
368
369
/**
370
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
371
 *
372
 * @param ctxt  an XML parser context
373
 * @param error  the error number
374
 * @param msg  the message
375
 * @param info1  extra information string
376
 * @param info2  extra information string
377
 * @param info3  extra information string
378
 */
379
static void LIBXML_ATTR_FORMAT(3,0)
380
xmlNsErr(xmlParserCtxtPtr ctxt, xmlParserErrors error,
381
         const char *msg,
382
         const xmlChar * info1, const xmlChar * info2,
383
         const xmlChar * info3)
384
120k
{
385
120k
    ctxt->nsWellFormed = 0;
386
387
120k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR,
388
120k
               info1, info2, info3, 0, msg, info1, info2, info3);
389
120k
}
390
391
/**
392
 * Handle a namespace warning error
393
 *
394
 * @param ctxt  an XML parser context
395
 * @param error  the error number
396
 * @param msg  the message
397
 * @param info1  extra information string
398
 * @param info2  extra information string
399
 * @param info3  extra information string
400
 */
401
static void LIBXML_ATTR_FORMAT(3,0)
402
xmlNsWarn(xmlParserCtxtPtr ctxt, xmlParserErrors error,
403
         const char *msg,
404
         const xmlChar * info1, const xmlChar * info2,
405
         const xmlChar * info3)
406
5.05k
{
407
5.05k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING,
408
5.05k
               info1, info2, info3, 0, msg, info1, info2, info3);
409
5.05k
}
410
411
/**
412
 * Check for non-linear entity expansion behaviour.
413
 *
414
 * In some cases like xmlExpandEntityInAttValue, this function is called
415
 * for each, possibly nested entity and its unexpanded content length.
416
 *
417
 * In other cases like #xmlParseReference, it's only called for each
418
 * top-level entity with its unexpanded content length plus the sum of
419
 * the unexpanded content lengths (plus fixed cost) of all nested
420
 * entities.
421
 *
422
 * Summing the unexpanded lengths also adds the length of the reference.
423
 * This is by design. Taking the length of the entity name into account
424
 * discourages attacks that try to waste CPU time with abusively long
425
 * entity names. See test/recurse/lol6.xml for example. Each call also
426
 * adds some fixed cost XML_ENT_FIXED_COST to discourage attacks with
427
 * short entities.
428
 *
429
 * @param ctxt  parser context
430
 * @param extra  sum of unexpanded entity sizes
431
 * @returns 1 on error, 0 on success.
432
 */
433
static int
434
xmlParserEntityCheck(xmlParserCtxtPtr ctxt, unsigned long extra)
435
2.44M
{
436
2.44M
    unsigned long consumed;
437
2.44M
    unsigned long *expandedSize;
438
2.44M
    xmlParserInputPtr input = ctxt->input;
439
2.44M
    xmlEntityPtr entity = input->entity;
440
441
2.44M
    if ((entity) && (entity->flags & XML_ENT_CHECKED))
442
222k
        return(0);
443
444
    /*
445
     * Compute total consumed bytes so far, including input streams of
446
     * external entities.
447
     */
448
2.22M
    consumed = input->consumed;
449
2.22M
    xmlSaturatedAddSizeT(&consumed, input->cur - input->base);
450
2.22M
    xmlSaturatedAdd(&consumed, ctxt->sizeentities);
451
452
2.22M
    if (entity)
453
54.3k
        expandedSize = &entity->expandedSize;
454
2.16M
    else
455
2.16M
        expandedSize = &ctxt->sizeentcopy;
456
457
    /*
458
     * Add extra cost and some fixed cost.
459
     */
460
2.22M
    xmlSaturatedAdd(expandedSize, extra);
461
2.22M
    xmlSaturatedAdd(expandedSize, XML_ENT_FIXED_COST);
462
463
    /*
464
     * It's important to always use saturation arithmetic when tracking
465
     * entity sizes to make the size checks reliable. If "sizeentcopy"
466
     * overflows, we have to abort.
467
     */
468
2.22M
    if ((ctxt->maxAmpl > 0) &&
469
2.22M
        (*expandedSize > XML_PARSER_ALLOWED_EXPANSION) &&
470
201k
        ((*expandedSize >= ULONG_MAX) ||
471
201k
         (*expandedSize / ctxt->maxAmpl > consumed))) {
472
1.25k
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
473
1.25k
                       "Maximum entity amplification factor exceeded, see "
474
1.25k
                       "xmlCtxtSetMaxAmplification.\n");
475
1.25k
        return(1);
476
1.25k
    }
477
478
2.22M
    return(0);
479
2.22M
}
480
481
/************************************************************************
482
 *                  *
483
 *    Library wide options          *
484
 *                  *
485
 ************************************************************************/
486
487
/**
488
 * Examines if the library has been compiled with a given feature.
489
 *
490
 * @param feature  the feature to be examined
491
 * @returns zero (0) if the feature does not exist or an unknown
492
 * feature is requested, non-zero otherwise.
493
 */
494
int
495
xmlHasFeature(xmlFeature feature)
496
0
{
497
0
    switch (feature) {
498
0
  case XML_WITH_THREAD:
499
0
#ifdef LIBXML_THREAD_ENABLED
500
0
      return(1);
501
#else
502
      return(0);
503
#endif
504
0
        case XML_WITH_TREE:
505
0
            return(1);
506
0
        case XML_WITH_OUTPUT:
507
0
#ifdef LIBXML_OUTPUT_ENABLED
508
0
            return(1);
509
#else
510
            return(0);
511
#endif
512
0
        case XML_WITH_PUSH:
513
0
#ifdef LIBXML_PUSH_ENABLED
514
0
            return(1);
515
#else
516
            return(0);
517
#endif
518
0
        case XML_WITH_READER:
519
0
#ifdef LIBXML_READER_ENABLED
520
0
            return(1);
521
#else
522
            return(0);
523
#endif
524
0
        case XML_WITH_PATTERN:
525
0
#ifdef LIBXML_PATTERN_ENABLED
526
0
            return(1);
527
#else
528
            return(0);
529
#endif
530
0
        case XML_WITH_WRITER:
531
0
#ifdef LIBXML_WRITER_ENABLED
532
0
            return(1);
533
#else
534
            return(0);
535
#endif
536
0
        case XML_WITH_SAX1:
537
0
#ifdef LIBXML_SAX1_ENABLED
538
0
            return(1);
539
#else
540
            return(0);
541
#endif
542
0
        case XML_WITH_HTTP:
543
0
            return(0);
544
0
        case XML_WITH_VALID:
545
0
#ifdef LIBXML_VALID_ENABLED
546
0
            return(1);
547
#else
548
            return(0);
549
#endif
550
0
        case XML_WITH_HTML:
551
0
#ifdef LIBXML_HTML_ENABLED
552
0
            return(1);
553
#else
554
            return(0);
555
#endif
556
0
        case XML_WITH_LEGACY:
557
0
            return(0);
558
0
        case XML_WITH_C14N:
559
0
#ifdef LIBXML_C14N_ENABLED
560
0
            return(1);
561
#else
562
            return(0);
563
#endif
564
0
        case XML_WITH_CATALOG:
565
0
#ifdef LIBXML_CATALOG_ENABLED
566
0
            return(1);
567
#else
568
            return(0);
569
#endif
570
0
        case XML_WITH_XPATH:
571
0
#ifdef LIBXML_XPATH_ENABLED
572
0
            return(1);
573
#else
574
            return(0);
575
#endif
576
0
        case XML_WITH_XPTR:
577
0
#ifdef LIBXML_XPTR_ENABLED
578
0
            return(1);
579
#else
580
            return(0);
581
#endif
582
0
        case XML_WITH_XINCLUDE:
583
0
#ifdef LIBXML_XINCLUDE_ENABLED
584
0
            return(1);
585
#else
586
            return(0);
587
#endif
588
0
        case XML_WITH_ICONV:
589
0
#ifdef LIBXML_ICONV_ENABLED
590
0
            return(1);
591
#else
592
            return(0);
593
#endif
594
0
        case XML_WITH_ISO8859X:
595
0
#ifdef LIBXML_ISO8859X_ENABLED
596
0
            return(1);
597
#else
598
            return(0);
599
#endif
600
0
        case XML_WITH_UNICODE:
601
0
            return(0);
602
0
        case XML_WITH_REGEXP:
603
0
#ifdef LIBXML_REGEXP_ENABLED
604
0
            return(1);
605
#else
606
            return(0);
607
#endif
608
0
        case XML_WITH_AUTOMATA:
609
0
#ifdef LIBXML_REGEXP_ENABLED
610
0
            return(1);
611
#else
612
            return(0);
613
#endif
614
0
        case XML_WITH_EXPR:
615
0
            return(0);
616
0
        case XML_WITH_RELAXNG:
617
0
#ifdef LIBXML_RELAXNG_ENABLED
618
0
            return(1);
619
#else
620
            return(0);
621
#endif
622
0
        case XML_WITH_SCHEMAS:
623
0
#ifdef LIBXML_SCHEMAS_ENABLED
624
0
            return(1);
625
#else
626
            return(0);
627
#endif
628
0
        case XML_WITH_SCHEMATRON:
629
#ifdef LIBXML_SCHEMATRON_ENABLED
630
            return(1);
631
#else
632
0
            return(0);
633
0
#endif
634
0
        case XML_WITH_MODULES:
635
0
#ifdef LIBXML_MODULES_ENABLED
636
0
            return(1);
637
#else
638
            return(0);
639
#endif
640
0
        case XML_WITH_DEBUG:
641
#ifdef LIBXML_DEBUG_ENABLED
642
            return(1);
643
#else
644
0
            return(0);
645
0
#endif
646
0
        case XML_WITH_DEBUG_MEM:
647
0
            return(0);
648
0
        case XML_WITH_ZLIB:
649
0
#ifdef LIBXML_ZLIB_ENABLED
650
0
            return(1);
651
#else
652
            return(0);
653
#endif
654
0
        case XML_WITH_LZMA:
655
0
            return(0);
656
0
        case XML_WITH_ICU:
657
#ifdef LIBXML_ICU_ENABLED
658
            return(1);
659
#else
660
0
            return(0);
661
0
#endif
662
0
        default:
663
0
      break;
664
0
     }
665
0
     return(0);
666
0
}
667
668
/************************************************************************
669
 *                  *
670
 *      Simple string buffer        *
671
 *                  *
672
 ************************************************************************/
673
674
typedef struct {
675
    xmlChar *mem;
676
    unsigned size;
677
    unsigned cap; /* size < cap */
678
    unsigned max; /* size <= max */
679
    xmlParserErrors code;
680
} xmlSBuf;
681
682
static void
683
404k
xmlSBufInit(xmlSBuf *buf, unsigned max) {
684
404k
    buf->mem = NULL;
685
404k
    buf->size = 0;
686
404k
    buf->cap = 0;
687
404k
    buf->max = max;
688
404k
    buf->code = XML_ERR_OK;
689
404k
}
690
691
static int
692
300k
xmlSBufGrow(xmlSBuf *buf, unsigned len) {
693
300k
    xmlChar *mem;
694
300k
    unsigned cap;
695
696
300k
    if (len >= UINT_MAX / 2 - buf->size) {
697
0
        if (buf->code == XML_ERR_OK)
698
0
            buf->code = XML_ERR_RESOURCE_LIMIT;
699
0
        return(-1);
700
0
    }
701
702
300k
    cap = (buf->size + len) * 2;
703
300k
    if (cap < 240)
704
227k
        cap = 240;
705
706
300k
    mem = xmlRealloc(buf->mem, cap);
707
300k
    if (mem == NULL) {
708
642
        buf->code = XML_ERR_NO_MEMORY;
709
642
        return(-1);
710
642
    }
711
712
300k
    buf->mem = mem;
713
300k
    buf->cap = cap;
714
715
300k
    return(0);
716
300k
}
717
718
static void
719
395M
xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) {
720
395M
    if (buf->max - buf->size < len) {
721
2.69M
        if (buf->code == XML_ERR_OK)
722
623
            buf->code = XML_ERR_RESOURCE_LIMIT;
723
2.69M
        return;
724
2.69M
    }
725
726
393M
    if (buf->cap - buf->size <= len) {
727
292k
        if (xmlSBufGrow(buf, len) < 0)
728
563
            return;
729
292k
    }
730
731
393M
    if (len > 0)
732
393M
        memcpy(buf->mem + buf->size, str, len);
733
393M
    buf->size += len;
734
393M
}
735
736
static void
737
390M
xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) {
738
390M
    xmlSBufAddString(buf, (const xmlChar *) str, len);
739
390M
}
740
741
static void
742
599k
xmlSBufAddChar(xmlSBuf *buf, int c) {
743
599k
    xmlChar *end;
744
745
599k
    if (buf->max - buf->size < 4) {
746
22.1k
        if (buf->code == XML_ERR_OK)
747
16
            buf->code = XML_ERR_RESOURCE_LIMIT;
748
22.1k
        return;
749
22.1k
    }
750
751
577k
    if (buf->cap - buf->size <= 4) {
752
8.53k
        if (xmlSBufGrow(buf, 4) < 0)
753
79
            return;
754
8.53k
    }
755
756
577k
    end = buf->mem + buf->size;
757
758
577k
    if (c < 0x80) {
759
565k
        *end = (xmlChar) c;
760
565k
        buf->size += 1;
761
565k
    } else {
762
11.6k
        buf->size += xmlCopyCharMultiByte(end, c);
763
11.6k
    }
764
577k
}
765
766
static void
767
284M
xmlSBufAddReplChar(xmlSBuf *buf) {
768
284M
    xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3);
769
284M
}
770
771
static void
772
1.34k
xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
773
1.34k
    if (buf->code == XML_ERR_NO_MEMORY)
774
706
        xmlCtxtErrMemory(ctxt);
775
639
    else
776
639
        xmlFatalErr(ctxt, buf->code, errMsg);
777
1.34k
}
778
779
static xmlChar *
780
xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt,
781
252k
              const char *errMsg) {
782
252k
    if (buf->mem == NULL) {
783
31.2k
        buf->mem = xmlMalloc(1);
784
31.2k
        if (buf->mem == NULL) {
785
64
            buf->code = XML_ERR_NO_MEMORY;
786
31.2k
        } else {
787
31.2k
            buf->mem[0] = 0;
788
31.2k
        }
789
221k
    } else {
790
221k
        buf->mem[buf->size] = 0;
791
221k
    }
792
793
252k
    if (buf->code == XML_ERR_OK) {
794
251k
        if (sizeOut != NULL)
795
29.4k
            *sizeOut = buf->size;
796
251k
        return(buf->mem);
797
251k
    }
798
799
790
    xmlSBufReportError(buf, ctxt, errMsg);
800
801
790
    xmlFree(buf->mem);
802
803
790
    if (sizeOut != NULL)
804
119
        *sizeOut = 0;
805
790
    return(NULL);
806
252k
}
807
808
static void
809
144k
xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
810
144k
    if (buf->code != XML_ERR_OK)
811
555
        xmlSBufReportError(buf, ctxt, errMsg);
812
813
144k
    xmlFree(buf->mem);
814
144k
}
815
816
static int
817
xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str,
818
2.07G
                    const char *errMsg) {
819
2.07G
    int c = str[0];
820
2.07G
    int c1 = str[1];
821
822
2.07G
    if ((c1 & 0xC0) != 0x80)
823
113M
        goto encoding_error;
824
825
1.96G
    if (c < 0xE0) {
826
        /* 2-byte sequence */
827
97.5M
        if (c < 0xC2)
828
82.6M
            goto encoding_error;
829
830
14.8M
        return(2);
831
1.86G
    } else {
832
1.86G
        int c2 = str[2];
833
834
1.86G
        if ((c2 & 0xC0) != 0x80)
835
16.6k
            goto encoding_error;
836
837
1.86G
        if (c < 0xF0) {
838
            /* 3-byte sequence */
839
1.86G
            if (c == 0xE0) {
840
                /* overlong */
841
2.46k
                if (c1 < 0xA0)
842
239
                    goto encoding_error;
843
1.86G
            } else if (c == 0xED) {
844
                /* surrogate */
845
2.55k
                if (c1 >= 0xA0)
846
400
                    goto encoding_error;
847
1.86G
            } else if (c == 0xEF) {
848
                /* U+FFFE and U+FFFF are invalid Chars */
849
1.42G
                if ((c1 == 0xBF) && (c2 >= 0xBE))
850
1.63k
                    xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg);
851
1.42G
            }
852
853
1.86G
            return(3);
854
1.86G
        } else {
855
            /* 4-byte sequence */
856
448k
            if ((str[3] & 0xC0) != 0x80)
857
2.61k
                goto encoding_error;
858
445k
            if (c == 0xF0) {
859
                /* overlong */
860
56.6k
                if (c1 < 0x90)
861
563
                    goto encoding_error;
862
389k
            } else if (c >= 0xF4) {
863
                /* greater than 0x10FFFF */
864
10.4k
                if ((c > 0xF4) || (c1 >= 0x90))
865
10.1k
                    goto encoding_error;
866
10.4k
            }
867
868
435k
            return(4);
869
445k
        }
870
1.86G
    }
871
872
196M
encoding_error:
873
    /* Only report the first error */
874
196M
    if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) {
875
27.3k
        xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL);
876
27.3k
        ctxt->input->flags |= XML_INPUT_ENCODING_ERROR;
877
27.3k
    }
878
879
196M
    return(0);
880
1.96G
}
881
882
/************************************************************************
883
 *                  *
884
 *    SAX2 defaulted attributes handling      *
885
 *                  *
886
 ************************************************************************/
887
888
/**
889
 * Final initialization of the parser context before starting to parse.
890
 *
891
 * This accounts for users modifying struct members of parser context
892
 * directly.
893
 *
894
 * @param ctxt  an XML parser context
895
 */
896
static void
897
93.2k
xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) {
898
93.2k
    xmlSAXHandlerPtr sax;
899
900
    /* Avoid unused variable warning if features are disabled. */
901
93.2k
    (void) sax;
902
903
    /*
904
     * Changing the SAX struct directly is still widespread practice
905
     * in internal and external code.
906
     */
907
93.2k
    if (ctxt == NULL) return;
908
93.2k
    sax = ctxt->sax;
909
93.2k
#ifdef LIBXML_SAX1_ENABLED
910
    /*
911
     * Only enable SAX2 if there SAX2 element handlers, except when there
912
     * are no element handlers at all.
913
     */
914
93.2k
    if (((ctxt->options & XML_PARSE_SAX1) == 0) &&
915
57.3k
        (sax) &&
916
57.3k
        (sax->initialized == XML_SAX2_MAGIC) &&
917
57.3k
        ((sax->startElementNs != NULL) ||
918
0
         (sax->endElementNs != NULL) ||
919
0
         ((sax->startElement == NULL) && (sax->endElement == NULL))))
920
57.3k
        ctxt->sax2 = 1;
921
#else
922
    ctxt->sax2 = 1;
923
#endif /* LIBXML_SAX1_ENABLED */
924
925
    /*
926
     * Some users replace the dictionary directly in the context struct.
927
     * We really need an API function to do that cleanly.
928
     */
929
93.2k
    ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3);
930
93.2k
    ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5);
931
93.2k
    ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36);
932
93.2k
    if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) ||
933
92.4k
    (ctxt->str_xml_ns == NULL)) {
934
724
        xmlErrMemory(ctxt);
935
724
    }
936
937
93.2k
    xmlDictSetLimit(ctxt->dict,
938
93.2k
                    (ctxt->options & XML_PARSE_HUGE) ?
939
42.0k
                        0 :
940
93.2k
                        XML_MAX_DICTIONARY_LIMIT);
941
942
93.2k
#ifdef LIBXML_VALID_ENABLED
943
93.2k
    if (ctxt->validate)
944
62.3k
        ctxt->vctxt.flags |= XML_VCTXT_VALIDATE;
945
30.8k
    else
946
30.8k
        ctxt->vctxt.flags &= ~XML_VCTXT_VALIDATE;
947
93.2k
#endif /* LIBXML_VALID_ENABLED */
948
93.2k
}
949
950
typedef struct {
951
    xmlHashedString prefix;
952
    xmlHashedString name;
953
    xmlHashedString value;
954
    const xmlChar *valueEnd;
955
    int external;
956
    int expandedSize;
957
} xmlDefAttr;
958
959
typedef struct _xmlDefAttrs xmlDefAttrs;
960
typedef xmlDefAttrs *xmlDefAttrsPtr;
961
struct _xmlDefAttrs {
962
    int nbAttrs;  /* number of defaulted attributes on that element */
963
    int maxAttrs;       /* the size of the array */
964
#if __STDC_VERSION__ >= 199901L
965
    /* Using a C99 flexible array member avoids UBSan errors. */
966
    xmlDefAttr attrs[] ATTRIBUTE_COUNTED_BY(maxAttrs);
967
#else
968
    xmlDefAttr attrs[1];
969
#endif
970
};
971
972
/**
973
 * Normalize the space in non CDATA attribute values:
974
 * If the attribute type is not CDATA, then the XML processor MUST further
975
 * process the normalized attribute value by discarding any leading and
976
 * trailing space (\#x20) characters, and by replacing sequences of space
977
 * (\#x20) characters by a single space (\#x20) character.
978
 * Note that the size of dst need to be at least src, and if one doesn't need
979
 * to preserve dst (and it doesn't come from a dictionary or read-only) then
980
 * passing src as dst is just fine.
981
 *
982
 * @param src  the source string
983
 * @param dst  the target string
984
 * @returns a pointer to the normalized value (dst) or NULL if no conversion
985
 *         is needed.
986
 */
987
xmlChar *
988
xmlAttrNormalizeSpace(const xmlChar *src, xmlChar *dst)
989
65.8k
{
990
65.8k
    if ((src == NULL) || (dst == NULL))
991
0
        return(NULL);
992
993
67.0k
    while (*src == 0x20) src++;
994
208M
    while (*src != 0) {
995
208M
  if (*src == 0x20) {
996
12.4M
      while (*src == 0x20) src++;
997
16.5k
      if (*src != 0)
998
15.9k
    *dst++ = 0x20;
999
208M
  } else {
1000
208M
      *dst++ = *src++;
1001
208M
  }
1002
208M
    }
1003
65.8k
    *dst = 0;
1004
65.8k
    if (dst == src)
1005
64.3k
       return(NULL);
1006
1.53k
    return(dst);
1007
65.8k
}
1008
1009
/**
1010
 * TODO: This function should also remove leading and trailing
1011
 *       whitespaces, and also group whitespaces together, but right
1012
 *       now the parse doesn't do that with XML_PARSE_NOENT, so this
1013
 *       replacement is consistent with the parser normalization
1014
 *
1015
 *       Maybe merge with xmlAttrNormalizeSpace, or do something
1016
 *       similar
1017
 *
1018
 * Normalize attritube entity values
1019
 * Replaces any space character with 0x20
1020
 * https://www.w3.org/TR/REC-xml/#AVNormalize
1021
 */
1022
xmlChar *
1023
xmlAttrNormalize(xmlChar *src)
1024
0
{
1025
0
    xmlChar *out = NULL;
1026
0
    xmlChar *dst = NULL;
1027
1028
0
    if (src == NULL)
1029
0
        return(NULL);
1030
1031
0
    out = src;
1032
0
    dst = out;
1033
0
    while (*src != 0) {
1034
0
        if (*src < 0x20) {
1035
0
            src++;
1036
0
            *dst++ = 0x20;
1037
0
        } else {
1038
0
            *dst++ = *src++;
1039
0
        }
1040
0
    }
1041
0
    *dst = 0;
1042
0
    return(out);
1043
0
}
1044
1045
/**
1046
 * Add a defaulted attribute for an element
1047
 *
1048
 * @param ctxt  an XML parser context
1049
 * @param fullname  the element fullname
1050
 * @param fullattr  the attribute fullname
1051
 * @param value  the attribute value
1052
 */
1053
static void
1054
xmlAddDefAttrs(xmlParserCtxtPtr ctxt,
1055
               const xmlChar *fullname,
1056
               const xmlChar *fullattr,
1057
56.8k
               const xmlChar *value) {
1058
56.8k
    xmlDefAttrsPtr defaults;
1059
56.8k
    xmlDefAttr *attr;
1060
56.8k
    int len, expandedSize;
1061
56.8k
    xmlHashedString name;
1062
56.8k
    xmlHashedString prefix;
1063
56.8k
    xmlHashedString hvalue;
1064
56.8k
    const xmlChar *localname;
1065
1066
    /*
1067
     * Allows to detect attribute redefinitions
1068
     */
1069
56.8k
    if (ctxt->attsSpecial != NULL) {
1070
50.0k
        if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL)
1071
28.4k
      return;
1072
50.0k
    }
1073
1074
28.3k
    if (ctxt->attsDefault == NULL) {
1075
7.04k
        ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict);
1076
7.04k
  if (ctxt->attsDefault == NULL)
1077
30
      goto mem_error;
1078
7.04k
    }
1079
1080
    /*
1081
     * split the element name into prefix:localname , the string found
1082
     * are within the DTD and then not associated to namespace names.
1083
     */
1084
28.3k
    localname = xmlSplitQName3(fullname, &len);
1085
28.3k
    if (localname == NULL) {
1086
27.6k
        name = xmlDictLookupHashed(ctxt->dict, fullname, -1);
1087
27.6k
  prefix.name = NULL;
1088
27.6k
    } else {
1089
682
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1090
682
  prefix = xmlDictLookupHashed(ctxt->dict, fullname, len);
1091
682
        if (prefix.name == NULL)
1092
6
            goto mem_error;
1093
682
    }
1094
28.3k
    if (name.name == NULL)
1095
6
        goto mem_error;
1096
1097
    /*
1098
     * make sure there is some storage
1099
     */
1100
28.3k
    defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name);
1101
28.3k
    if ((defaults == NULL) ||
1102
20.2k
        (defaults->nbAttrs >= defaults->maxAttrs)) {
1103
9.48k
        xmlDefAttrsPtr temp;
1104
9.48k
        int newSize;
1105
1106
9.48k
        if (defaults == NULL) {
1107
8.06k
            newSize = 4;
1108
8.06k
        } else {
1109
1.41k
            if ((defaults->maxAttrs >= XML_MAX_ATTRS) ||
1110
1.41k
                ((size_t) defaults->maxAttrs >
1111
1.41k
                     SIZE_MAX / 2 / sizeof(temp[0]) - sizeof(*defaults)))
1112
0
                goto mem_error;
1113
1114
1.41k
            if (defaults->maxAttrs > XML_MAX_ATTRS / 2)
1115
0
                newSize = XML_MAX_ATTRS;
1116
1.41k
            else
1117
1.41k
                newSize = defaults->maxAttrs * 2;
1118
1.41k
        }
1119
9.48k
        temp = xmlRealloc(defaults,
1120
9.48k
                          sizeof(*defaults) + newSize * sizeof(xmlDefAttr));
1121
9.48k
  if (temp == NULL)
1122
18
      goto mem_error;
1123
9.46k
        if (defaults == NULL)
1124
8.04k
            temp->nbAttrs = 0;
1125
9.46k
  temp->maxAttrs = newSize;
1126
9.46k
        defaults = temp;
1127
9.46k
  if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name,
1128
9.46k
                          defaults, NULL) < 0) {
1129
3
      xmlFree(defaults);
1130
3
      goto mem_error;
1131
3
  }
1132
9.46k
    }
1133
1134
    /*
1135
     * Split the attribute name into prefix:localname , the string found
1136
     * are within the DTD and hen not associated to namespace names.
1137
     */
1138
28.3k
    localname = xmlSplitQName3(fullattr, &len);
1139
28.3k
    if (localname == NULL) {
1140
23.0k
        name = xmlDictLookupHashed(ctxt->dict, fullattr, -1);
1141
23.0k
  prefix.name = NULL;
1142
23.0k
    } else {
1143
5.26k
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1144
5.26k
  prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len);
1145
5.26k
        if (prefix.name == NULL)
1146
6
            goto mem_error;
1147
5.26k
    }
1148
28.3k
    if (name.name == NULL)
1149
6
        goto mem_error;
1150
1151
    /* intern the string and precompute the end */
1152
28.3k
    len = strlen((const char *) value);
1153
28.3k
    hvalue = xmlDictLookupHashed(ctxt->dict, value, len);
1154
28.3k
    if (hvalue.name == NULL)
1155
11
        goto mem_error;
1156
1157
28.3k
    expandedSize = strlen((const char *) name.name);
1158
28.3k
    if (prefix.name != NULL)
1159
5.24k
        expandedSize += strlen((const char *) prefix.name);
1160
28.3k
    expandedSize += len;
1161
1162
28.3k
    attr = &defaults->attrs[defaults->nbAttrs++];
1163
28.3k
    attr->name = name;
1164
28.3k
    attr->prefix = prefix;
1165
28.3k
    attr->value = hvalue;
1166
28.3k
    attr->valueEnd = hvalue.name + len;
1167
28.3k
    attr->external = PARSER_EXTERNAL(ctxt);
1168
28.3k
    attr->expandedSize = expandedSize;
1169
1170
28.3k
    return;
1171
1172
86
mem_error:
1173
86
    xmlErrMemory(ctxt);
1174
86
}
1175
1176
/**
1177
 * Register this attribute type
1178
 *
1179
 * @param ctxt  an XML parser context
1180
 * @param fullname  the element fullname
1181
 * @param fullattr  the attribute fullname
1182
 * @param type  the attribute type
1183
 */
1184
static void
1185
xmlAddSpecialAttr(xmlParserCtxtPtr ctxt,
1186
      const xmlChar *fullname,
1187
      const xmlChar *fullattr,
1188
      int type)
1189
87.4k
{
1190
87.4k
    if (ctxt->attsSpecial == NULL) {
1191
9.90k
        ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict);
1192
9.90k
  if (ctxt->attsSpecial == NULL)
1193
43
      goto mem_error;
1194
9.90k
    }
1195
1196
87.4k
    if (PARSER_EXTERNAL(ctxt))
1197
44.8k
        type |= XML_SPECIAL_EXTERNAL;
1198
1199
87.4k
    if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr,
1200
87.4k
                    XML_INT_TO_PTR(type)) < 0)
1201
7
        goto mem_error;
1202
87.4k
    return;
1203
1204
87.4k
mem_error:
1205
50
    xmlErrMemory(ctxt);
1206
50
}
1207
1208
/**
1209
 * Removes CDATA attributes from the special attribute table
1210
 */
1211
static void
1212
xmlCleanSpecialAttrCallback(void *payload, void *data,
1213
                            const xmlChar *fullname, const xmlChar *fullattr,
1214
54.8k
                            const xmlChar *unused ATTRIBUTE_UNUSED) {
1215
54.8k
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data;
1216
1217
54.8k
    if (XML_PTR_TO_INT(payload) == XML_ATTRIBUTE_CDATA) {
1218
3.02k
        xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL);
1219
3.02k
    }
1220
54.8k
}
1221
1222
/**
1223
 * Trim the list of attributes defined to remove all those of type
1224
 * CDATA as they are not special. This call should be done when finishing
1225
 * to parse the DTD and before starting to parse the document root.
1226
 *
1227
 * @param ctxt  an XML parser context
1228
 */
1229
static void
1230
xmlCleanSpecialAttr(xmlParserCtxtPtr ctxt)
1231
51.8k
{
1232
51.8k
    if (ctxt->attsSpecial == NULL)
1233
42.0k
        return;
1234
1235
9.81k
    xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt);
1236
1237
9.81k
    if (xmlHashSize(ctxt->attsSpecial) == 0) {
1238
464
        xmlHashFree(ctxt->attsSpecial, NULL);
1239
464
        ctxt->attsSpecial = NULL;
1240
464
    }
1241
9.81k
}
1242
1243
/**
1244
 * Checks that the value conforms to the LanguageID production:
1245
 *
1246
 * @deprecated Internal function, do not use.
1247
 *
1248
 * NOTE: this is somewhat deprecated, those productions were removed from
1249
 * the XML Second edition.
1250
 *
1251
 *     [33] LanguageID ::= Langcode ('-' Subcode)*
1252
 *     [34] Langcode ::= ISO639Code |  IanaCode |  UserCode
1253
 *     [35] ISO639Code ::= ([a-z] | [A-Z]) ([a-z] | [A-Z])
1254
 *     [36] IanaCode ::= ('i' | 'I') '-' ([a-z] | [A-Z])+
1255
 *     [37] UserCode ::= ('x' | 'X') '-' ([a-z] | [A-Z])+
1256
 *     [38] Subcode ::= ([a-z] | [A-Z])+
1257
 *
1258
 * The current REC reference the successors of RFC 1766, currently 5646
1259
 *
1260
 * http://www.rfc-editor.org/rfc/rfc5646.txt
1261
 *
1262
 *     langtag       = language
1263
 *                     ["-" script]
1264
 *                     ["-" region]
1265
 *                     *("-" variant)
1266
 *                     *("-" extension)
1267
 *                     ["-" privateuse]
1268
 *     language      = 2*3ALPHA            ; shortest ISO 639 code
1269
 *                     ["-" extlang]       ; sometimes followed by
1270
 *                                         ; extended language subtags
1271
 *                   / 4ALPHA              ; or reserved for future use
1272
 *                   / 5*8ALPHA            ; or registered language subtag
1273
 *
1274
 *     extlang       = 3ALPHA              ; selected ISO 639 codes
1275
 *                     *2("-" 3ALPHA)      ; permanently reserved
1276
 *
1277
 *     script        = 4ALPHA              ; ISO 15924 code
1278
 *
1279
 *     region        = 2ALPHA              ; ISO 3166-1 code
1280
 *                   / 3DIGIT              ; UN M.49 code
1281
 *
1282
 *     variant       = 5*8alphanum         ; registered variants
1283
 *                   / (DIGIT 3alphanum)
1284
 *
1285
 *     extension     = singleton 1*("-" (2*8alphanum))
1286
 *
1287
 *                                         ; Single alphanumerics
1288
 *                                         ; "x" reserved for private use
1289
 *     singleton     = DIGIT               ; 0 - 9
1290
 *                   / %x41-57             ; A - W
1291
 *                   / %x59-5A             ; Y - Z
1292
 *                   / %x61-77             ; a - w
1293
 *                   / %x79-7A             ; y - z
1294
 *
1295
 * it sounds right to still allow Irregular i-xxx IANA and user codes too
1296
 * The parser below doesn't try to cope with extension or privateuse
1297
 * that could be added but that's not interoperable anyway
1298
 *
1299
 * @param lang  pointer to the string value
1300
 * @returns 1 if correct 0 otherwise
1301
 **/
1302
int
1303
xmlCheckLanguageID(const xmlChar * lang)
1304
10.6k
{
1305
10.6k
    const xmlChar *cur = lang, *nxt;
1306
1307
10.6k
    if (cur == NULL)
1308
282
        return (0);
1309
10.3k
    if (((cur[0] == 'i') && (cur[1] == '-')) ||
1310
10.1k
        ((cur[0] == 'I') && (cur[1] == '-')) ||
1311
9.84k
        ((cur[0] == 'x') && (cur[1] == '-')) ||
1312
9.60k
        ((cur[0] == 'X') && (cur[1] == '-'))) {
1313
        /*
1314
         * Still allow IANA code and user code which were coming
1315
         * from the previous version of the XML-1.0 specification
1316
         * it's deprecated but we should not fail
1317
         */
1318
1.13k
        cur += 2;
1319
2.27k
        while (((cur[0] >= 'A') && (cur[0] <= 'Z')) ||
1320
1.62k
               ((cur[0] >= 'a') && (cur[0] <= 'z')))
1321
1.14k
            cur++;
1322
1.13k
        return(cur[0] == 0);
1323
1.13k
    }
1324
9.18k
    nxt = cur;
1325
33.8k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1326
19.0k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1327
24.6k
           nxt++;
1328
9.18k
    if (nxt - cur >= 4) {
1329
        /*
1330
         * Reserved
1331
         */
1332
708
        if ((nxt - cur > 8) || (nxt[0] != 0))
1333
502
            return(0);
1334
206
        return(1);
1335
708
    }
1336
8.47k
    if (nxt - cur < 2)
1337
355
        return(0);
1338
    /* we got an ISO 639 code */
1339
8.11k
    if (nxt[0] == 0)
1340
471
        return(1);
1341
7.64k
    if (nxt[0] != '-')
1342
340
        return(0);
1343
1344
7.30k
    nxt++;
1345
7.30k
    cur = nxt;
1346
    /* now we can have extlang or script or region or variant */
1347
7.30k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1348
718
        goto region_m49;
1349
1350
29.3k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1351
19.4k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1352
22.8k
           nxt++;
1353
6.58k
    if (nxt - cur == 4)
1354
1.51k
        goto script;
1355
5.07k
    if (nxt - cur == 2)
1356
1.16k
        goto region;
1357
3.90k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1358
304
        goto variant;
1359
3.60k
    if (nxt - cur != 3)
1360
743
        return(0);
1361
    /* we parsed an extlang */
1362
2.85k
    if (nxt[0] == 0)
1363
229
        return(1);
1364
2.62k
    if (nxt[0] != '-')
1365
349
        return(0);
1366
1367
2.28k
    nxt++;
1368
2.28k
    cur = nxt;
1369
    /* now we can have script or region or variant */
1370
2.28k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1371
275
        goto region_m49;
1372
1373
11.1k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1374
4.27k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1375
9.18k
           nxt++;
1376
2.00k
    if (nxt - cur == 2)
1377
216
        goto region;
1378
1.78k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1379
414
        goto variant;
1380
1.37k
    if (nxt - cur != 4)
1381
877
        return(0);
1382
    /* we parsed a script */
1383
2.01k
script:
1384
2.01k
    if (nxt[0] == 0)
1385
230
        return(1);
1386
1.78k
    if (nxt[0] != '-')
1387
231
        return(0);
1388
1389
1.55k
    nxt++;
1390
1.55k
    cur = nxt;
1391
    /* now we can have region or variant */
1392
1.55k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1393
447
        goto region_m49;
1394
1395
5.29k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1396
3.94k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1397
4.19k
           nxt++;
1398
1399
1.10k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1400
216
        goto variant;
1401
889
    if (nxt - cur != 2)
1402
542
        return(0);
1403
    /* we parsed a region */
1404
1.96k
region:
1405
1.96k
    if (nxt[0] == 0)
1406
301
        return(1);
1407
1.66k
    if (nxt[0] != '-')
1408
831
        return(0);
1409
1410
830
    nxt++;
1411
830
    cur = nxt;
1412
    /* now we can just have a variant */
1413
4.37k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1414
2.98k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1415
3.54k
           nxt++;
1416
1417
830
    if ((nxt - cur < 5) || (nxt - cur > 8))
1418
530
        return(0);
1419
1420
    /* we parsed a variant */
1421
1.23k
variant:
1422
1.23k
    if (nxt[0] == 0)
1423
281
        return(1);
1424
953
    if (nxt[0] != '-')
1425
671
        return(0);
1426
    /* extensions and private use subtags not checked */
1427
282
    return (1);
1428
1429
1.44k
region_m49:
1430
1.44k
    if (((nxt[1] >= '0') && (nxt[1] <= '9')) &&
1431
651
        ((nxt[2] >= '0') && (nxt[2] <= '9'))) {
1432
231
        nxt += 3;
1433
231
        goto region;
1434
231
    }
1435
1.20k
    return(0);
1436
1.44k
}
1437
1438
/************************************************************************
1439
 *                  *
1440
 *    Parser stacks related functions and macros    *
1441
 *                  *
1442
 ************************************************************************/
1443
1444
static xmlChar *
1445
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar **str);
1446
1447
/**
1448
 * Create a new namespace database.
1449
 *
1450
 * @returns the new obejct.
1451
 */
1452
xmlParserNsData *
1453
88.2k
xmlParserNsCreate(void) {
1454
88.2k
    xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb));
1455
1456
88.2k
    if (nsdb == NULL)
1457
6
        return(NULL);
1458
88.2k
    memset(nsdb, 0, sizeof(*nsdb));
1459
88.2k
    nsdb->defaultNsIndex = INT_MAX;
1460
1461
88.2k
    return(nsdb);
1462
88.2k
}
1463
1464
/**
1465
 * Free a namespace database.
1466
 *
1467
 * @param nsdb  namespace database
1468
 */
1469
void
1470
88.2k
xmlParserNsFree(xmlParserNsData *nsdb) {
1471
88.2k
    if (nsdb == NULL)
1472
0
        return;
1473
1474
88.2k
    xmlFree(nsdb->extra);
1475
88.2k
    xmlFree(nsdb->hash);
1476
88.2k
    xmlFree(nsdb);
1477
88.2k
}
1478
1479
/**
1480
 * Reset a namespace database.
1481
 *
1482
 * @param nsdb  namespace database
1483
 */
1484
static void
1485
75.7k
xmlParserNsReset(xmlParserNsData *nsdb) {
1486
75.7k
    if (nsdb == NULL)
1487
0
        return;
1488
1489
75.7k
    nsdb->hashElems = 0;
1490
75.7k
    nsdb->elementId = 0;
1491
75.7k
    nsdb->defaultNsIndex = INT_MAX;
1492
1493
75.7k
    if (nsdb->hash)
1494
897
        memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0]));
1495
75.7k
}
1496
1497
/**
1498
 * Signal that a new element has started.
1499
 *
1500
 * @param nsdb  namespace database
1501
 * @returns 0 on success, -1 if the element counter overflowed.
1502
 */
1503
static int
1504
820k
xmlParserNsStartElement(xmlParserNsData *nsdb) {
1505
820k
    if (nsdb->elementId == UINT_MAX)
1506
0
        return(-1);
1507
820k
    nsdb->elementId++;
1508
1509
820k
    return(0);
1510
820k
}
1511
1512
/**
1513
 * Lookup namespace with given prefix. If `bucketPtr` is non-NULL, it will
1514
 * be set to the matching bucket, or the first empty bucket if no match
1515
 * was found.
1516
 *
1517
 * @param ctxt  parser context
1518
 * @param prefix  namespace prefix
1519
 * @param bucketPtr  optional bucket (return value)
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
2.49M
                  xmlParserNsBucket **bucketPtr) {
1526
2.49M
    xmlParserNsBucket *bucket, *tombstone;
1527
2.49M
    unsigned index, hashValue;
1528
1529
2.49M
    if (prefix->name == NULL)
1530
766k
        return(ctxt->nsdb->defaultNsIndex);
1531
1532
1.72M
    if (ctxt->nsdb->hashSize == 0)
1533
47.0k
        return(INT_MAX);
1534
1535
1.67M
    hashValue = prefix->hashValue;
1536
1.67M
    index = hashValue & (ctxt->nsdb->hashSize - 1);
1537
1.67M
    bucket = &ctxt->nsdb->hash[index];
1538
1.67M
    tombstone = NULL;
1539
1540
2.21M
    while (bucket->hashValue) {
1541
1.90M
        if (bucket->index == INT_MAX) {
1542
288k
            if (tombstone == NULL)
1543
274k
                tombstone = bucket;
1544
1.61M
        } else if (bucket->hashValue == hashValue) {
1545
1.37M
            if (ctxt->nsTab[bucket->index * 2] == prefix->name) {
1546
1.37M
                if (bucketPtr != NULL)
1547
1.00M
                    *bucketPtr = bucket;
1548
1.37M
                return(bucket->index);
1549
1.37M
            }
1550
1.37M
        }
1551
1552
533k
        index++;
1553
533k
        bucket++;
1554
533k
        if (index == ctxt->nsdb->hashSize) {
1555
60.3k
            index = 0;
1556
60.3k
            bucket = ctxt->nsdb->hash;
1557
60.3k
        }
1558
533k
    }
1559
1560
305k
    if (bucketPtr != NULL)
1561
268k
        *bucketPtr = tombstone ? tombstone : bucket;
1562
305k
    return(INT_MAX);
1563
1.67M
}
1564
1565
/**
1566
 * Lookup namespace URI with given prefix.
1567
 *
1568
 * @param ctxt  parser context
1569
 * @param prefix  namespace prefix
1570
 * @returns the namespace URI on success, NULL if no namespace was found.
1571
 */
1572
static const xmlChar *
1573
741k
xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) {
1574
741k
    const xmlChar *ret;
1575
741k
    int nsIndex;
1576
1577
741k
    if (prefix->name == ctxt->str_xml)
1578
1.13k
        return(ctxt->str_xml_ns);
1579
1580
    /*
1581
     * minNsIndex is used when building an entity tree. We must
1582
     * ignore namespaces declared outside the entity.
1583
     */
1584
740k
    nsIndex = xmlParserNsLookup(ctxt, prefix, NULL);
1585
740k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1586
644k
        return(NULL);
1587
1588
95.6k
    ret = ctxt->nsTab[nsIndex * 2 + 1];
1589
95.6k
    if (ret[0] == 0)
1590
12.8k
        ret = NULL;
1591
95.6k
    return(ret);
1592
740k
}
1593
1594
/**
1595
 * Lookup extra data for the given prefix. This returns data stored
1596
 * with xmlParserNsUdpateSax.
1597
 *
1598
 * @param ctxt  parser context
1599
 * @param prefix  namespace prefix
1600
 * @returns the data on success, NULL if no namespace was found.
1601
 */
1602
void *
1603
60.7k
xmlParserNsLookupSax(xmlParserCtxt *ctxt, const xmlChar *prefix) {
1604
60.7k
    xmlHashedString hprefix;
1605
60.7k
    int nsIndex;
1606
1607
60.7k
    if (prefix == ctxt->str_xml)
1608
11.9k
        return(NULL);
1609
1610
48.8k
    hprefix.name = prefix;
1611
48.8k
    if (prefix != NULL)
1612
12.2k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1613
36.6k
    else
1614
36.6k
        hprefix.hashValue = 0;
1615
48.8k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1616
48.8k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1617
0
        return(NULL);
1618
1619
48.8k
    return(ctxt->nsdb->extra[nsIndex].saxData);
1620
48.8k
}
1621
1622
/**
1623
 * Sets or updates extra data for the given prefix. This value will be
1624
 * returned by xmlParserNsLookupSax as long as the namespace with the
1625
 * given prefix is in scope.
1626
 *
1627
 * @param ctxt  parser context
1628
 * @param prefix  namespace prefix
1629
 * @param saxData  extra data for SAX handler
1630
 * @returns the data on success, NULL if no namespace was found.
1631
 */
1632
int
1633
xmlParserNsUpdateSax(xmlParserCtxt *ctxt, const xmlChar *prefix,
1634
345k
                     void *saxData) {
1635
345k
    xmlHashedString hprefix;
1636
345k
    int nsIndex;
1637
1638
345k
    if (prefix == ctxt->str_xml)
1639
0
        return(-1);
1640
1641
345k
    hprefix.name = prefix;
1642
345k
    if (prefix != NULL)
1643
329k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1644
16.8k
    else
1645
16.8k
        hprefix.hashValue = 0;
1646
345k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1647
345k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1648
0
        return(-1);
1649
1650
345k
    ctxt->nsdb->extra[nsIndex].saxData = saxData;
1651
345k
    return(0);
1652
345k
}
1653
1654
/**
1655
 * Grows the namespace tables.
1656
 *
1657
 * @param ctxt  parser context
1658
 * @returns 0 on success, -1 if a memory allocation failed.
1659
 */
1660
static int
1661
20.7k
xmlParserNsGrow(xmlParserCtxtPtr ctxt) {
1662
20.7k
    const xmlChar **table;
1663
20.7k
    xmlParserNsExtra *extra;
1664
20.7k
    int newSize;
1665
1666
20.7k
    newSize = xmlGrowCapacity(ctxt->nsMax,
1667
20.7k
                              sizeof(table[0]) + sizeof(extra[0]),
1668
20.7k
                              16, XML_MAX_ITEMS);
1669
20.7k
    if (newSize < 0)
1670
0
        goto error;
1671
1672
20.7k
    table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0]));
1673
20.7k
    if (table == NULL)
1674
56
        goto error;
1675
20.6k
    ctxt->nsTab = table;
1676
1677
20.6k
    extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0]));
1678
20.6k
    if (extra == NULL)
1679
59
        goto error;
1680
20.6k
    ctxt->nsdb->extra = extra;
1681
1682
20.6k
    ctxt->nsMax = newSize;
1683
20.6k
    return(0);
1684
1685
115
error:
1686
115
    xmlErrMemory(ctxt);
1687
115
    return(-1);
1688
20.6k
}
1689
1690
/**
1691
 * Push a new namespace on the table.
1692
 *
1693
 * @param ctxt  parser context
1694
 * @param prefix  prefix with hash value
1695
 * @param uri  uri with hash value
1696
 * @param saxData  extra data for SAX handler
1697
 * @param defAttr  whether the namespace comes from a default attribute
1698
 * @returns 1 if the namespace was pushed, 0 if the namespace was ignored,
1699
 * -1 if a memory allocation failed.
1700
 */
1701
static int
1702
xmlParserNsPush(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix,
1703
695k
                const xmlHashedString *uri, void *saxData, int defAttr) {
1704
695k
    xmlParserNsBucket *bucket = NULL;
1705
695k
    xmlParserNsExtra *extra;
1706
695k
    const xmlChar **ns;
1707
695k
    unsigned hashValue, nsIndex, oldIndex;
1708
1709
695k
    if ((prefix != NULL) && (prefix->name == ctxt->str_xml))
1710
205
        return(0);
1711
1712
695k
    if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) {
1713
115
        xmlErrMemory(ctxt);
1714
115
        return(-1);
1715
115
    }
1716
1717
    /*
1718
     * Default namespace and 'xml' namespace
1719
     */
1720
695k
    if ((prefix == NULL) || (prefix->name == NULL)) {
1721
39.3k
        oldIndex = ctxt->nsdb->defaultNsIndex;
1722
1723
39.3k
        if (oldIndex != INT_MAX) {
1724
33.3k
            extra = &ctxt->nsdb->extra[oldIndex];
1725
1726
33.3k
            if (extra->elementId == ctxt->nsdb->elementId) {
1727
1.85k
                if (defAttr == 0)
1728
1.64k
                    xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns");
1729
1.85k
                return(0);
1730
1.85k
            }
1731
1732
31.4k
            if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1733
11.8k
                (uri->name == ctxt->nsTab[oldIndex * 2 + 1]))
1734
9.09k
                return(0);
1735
31.4k
        }
1736
1737
28.3k
        ctxt->nsdb->defaultNsIndex = ctxt->nsNr;
1738
28.3k
        goto populate_entry;
1739
39.3k
    }
1740
1741
    /*
1742
     * Hash table lookup
1743
     */
1744
655k
    oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket);
1745
655k
    if (oldIndex != INT_MAX) {
1746
382k
        extra = &ctxt->nsdb->extra[oldIndex];
1747
1748
        /*
1749
         * Check for duplicate definitions on the same element.
1750
         */
1751
382k
        if (extra->elementId == ctxt->nsdb->elementId) {
1752
810
            if (defAttr == 0)
1753
577
                xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name);
1754
810
            return(0);
1755
810
        }
1756
1757
381k
        if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1758
14.0k
            (uri->name == ctxt->nsTab[bucket->index * 2 + 1]))
1759
11.5k
            return(0);
1760
1761
370k
        bucket->index = ctxt->nsNr;
1762
370k
        goto populate_entry;
1763
381k
    }
1764
1765
    /*
1766
     * Insert new bucket
1767
     */
1768
1769
273k
    hashValue = prefix->hashValue;
1770
1771
    /*
1772
     * Grow hash table, 50% fill factor
1773
     */
1774
273k
    if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) {
1775
6.01k
        xmlParserNsBucket *newHash;
1776
6.01k
        unsigned newSize, i, index;
1777
1778
6.01k
        if (ctxt->nsdb->hashSize > UINT_MAX / 2) {
1779
0
            xmlErrMemory(ctxt);
1780
0
            return(-1);
1781
0
        }
1782
6.01k
        newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16;
1783
6.01k
        newHash = xmlMalloc(newSize * sizeof(newHash[0]));
1784
6.01k
        if (newHash == NULL) {
1785
16
            xmlErrMemory(ctxt);
1786
16
            return(-1);
1787
16
        }
1788
5.99k
        memset(newHash, 0, newSize * sizeof(newHash[0]));
1789
1790
912k
        for (i = 0; i < ctxt->nsdb->hashSize; i++) {
1791
906k
            unsigned hv = ctxt->nsdb->hash[i].hashValue;
1792
906k
            unsigned newIndex;
1793
1794
906k
            if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX))
1795
903k
                continue;
1796
3.34k
            newIndex = hv & (newSize - 1);
1797
1798
4.74k
            while (newHash[newIndex].hashValue != 0) {
1799
1.39k
                newIndex++;
1800
1.39k
                if (newIndex == newSize)
1801
416
                    newIndex = 0;
1802
1.39k
            }
1803
1804
3.34k
            newHash[newIndex] = ctxt->nsdb->hash[i];
1805
3.34k
        }
1806
1807
5.99k
        xmlFree(ctxt->nsdb->hash);
1808
5.99k
        ctxt->nsdb->hash = newHash;
1809
5.99k
        ctxt->nsdb->hashSize = newSize;
1810
1811
        /*
1812
         * Relookup
1813
         */
1814
5.99k
        index = hashValue & (newSize - 1);
1815
1816
6.81k
        while (newHash[index].hashValue != 0) {
1817
817
            index++;
1818
817
            if (index == newSize)
1819
269
                index = 0;
1820
817
        }
1821
1822
5.99k
        bucket = &newHash[index];
1823
5.99k
    }
1824
1825
273k
    bucket->hashValue = hashValue;
1826
273k
    bucket->index = ctxt->nsNr;
1827
273k
    ctxt->nsdb->hashElems++;
1828
273k
    oldIndex = INT_MAX;
1829
1830
671k
populate_entry:
1831
671k
    nsIndex = ctxt->nsNr;
1832
1833
671k
    ns = &ctxt->nsTab[nsIndex * 2];
1834
671k
    ns[0] = prefix ? prefix->name : NULL;
1835
671k
    ns[1] = uri->name;
1836
1837
671k
    extra = &ctxt->nsdb->extra[nsIndex];
1838
671k
    extra->saxData = saxData;
1839
671k
    extra->prefixHashValue = prefix ? prefix->hashValue : 0;
1840
671k
    extra->uriHashValue = uri->hashValue;
1841
671k
    extra->elementId = ctxt->nsdb->elementId;
1842
671k
    extra->oldIndex = oldIndex;
1843
1844
671k
    ctxt->nsNr++;
1845
1846
671k
    return(1);
1847
273k
}
1848
1849
/**
1850
 * Pops the top `nr` namespaces and restores the hash table.
1851
 *
1852
 * @param ctxt  an XML parser context
1853
 * @param nr  the number to pop
1854
 * @returns the number of namespaces popped.
1855
 */
1856
static int
1857
xmlParserNsPop(xmlParserCtxtPtr ctxt, int nr)
1858
361k
{
1859
361k
    int i;
1860
1861
    /* assert(nr <= ctxt->nsNr); */
1862
1863
1.01M
    for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) {
1864
651k
        const xmlChar *prefix = ctxt->nsTab[i * 2];
1865
651k
        xmlParserNsExtra *extra = &ctxt->nsdb->extra[i];
1866
1867
651k
        if (prefix == NULL) {
1868
26.3k
            ctxt->nsdb->defaultNsIndex = extra->oldIndex;
1869
625k
        } else {
1870
625k
            xmlHashedString hprefix;
1871
625k
            xmlParserNsBucket *bucket = NULL;
1872
1873
625k
            hprefix.name = prefix;
1874
625k
            hprefix.hashValue = extra->prefixHashValue;
1875
625k
            xmlParserNsLookup(ctxt, &hprefix, &bucket);
1876
            /* assert(bucket && bucket->hashValue); */
1877
625k
            bucket->index = extra->oldIndex;
1878
625k
        }
1879
651k
    }
1880
1881
361k
    ctxt->nsNr -= nr;
1882
361k
    return(nr);
1883
361k
}
1884
1885
static int
1886
15.7k
xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt) {
1887
15.7k
    const xmlChar **atts;
1888
15.7k
    unsigned *attallocs;
1889
15.7k
    int newSize;
1890
1891
15.7k
    newSize = xmlGrowCapacity(ctxt->maxatts / 5,
1892
15.7k
                              sizeof(atts[0]) * 5 + sizeof(attallocs[0]),
1893
15.7k
                              10, XML_MAX_ATTRS);
1894
15.7k
    if (newSize < 0) {
1895
0
        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
1896
0
                    "Maximum number of attributes exceeded");
1897
0
        return(-1);
1898
0
    }
1899
1900
15.7k
    atts = xmlRealloc(ctxt->atts, newSize * sizeof(atts[0]) * 5);
1901
15.7k
    if (atts == NULL)
1902
50
        goto mem_error;
1903
15.6k
    ctxt->atts = atts;
1904
1905
15.6k
    attallocs = xmlRealloc(ctxt->attallocs,
1906
15.6k
                           newSize * sizeof(attallocs[0]));
1907
15.6k
    if (attallocs == NULL)
1908
44
        goto mem_error;
1909
15.6k
    ctxt->attallocs = attallocs;
1910
1911
15.6k
    ctxt->maxatts = newSize * 5;
1912
1913
15.6k
    return(0);
1914
1915
94
mem_error:
1916
94
    xmlErrMemory(ctxt);
1917
94
    return(-1);
1918
15.6k
}
1919
1920
/**
1921
 * Pushes a new parser input on top of the input stack
1922
 *
1923
 * @param ctxt  an XML parser context
1924
 * @param value  the parser input
1925
 * @returns -1 in case of error, the index in the stack otherwise
1926
 */
1927
int
1928
xmlCtxtPushInput(xmlParserCtxt *ctxt, xmlParserInput *value)
1929
250k
{
1930
250k
    char *directory = NULL;
1931
250k
    int maxDepth;
1932
1933
250k
    if ((ctxt == NULL) || (value == NULL))
1934
3.01k
        return(-1);
1935
1936
247k
    maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
1937
1938
247k
    if (ctxt->inputNr >= ctxt->inputMax) {
1939
15.9k
        xmlParserInputPtr *tmp;
1940
15.9k
        int newSize;
1941
1942
15.9k
        newSize = xmlGrowCapacity(ctxt->inputMax, sizeof(tmp[0]),
1943
15.9k
                                  5, maxDepth);
1944
15.9k
        if (newSize < 0) {
1945
6
            xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
1946
6
                           "Maximum entity nesting depth exceeded");
1947
6
            return(-1);
1948
6
        }
1949
15.9k
        tmp = xmlRealloc(ctxt->inputTab, newSize * sizeof(tmp[0]));
1950
15.9k
        if (tmp == NULL) {
1951
84
            xmlErrMemory(ctxt);
1952
84
            return(-1);
1953
84
        }
1954
15.8k
        ctxt->inputTab = tmp;
1955
15.8k
        ctxt->inputMax = newSize;
1956
15.8k
    }
1957
1958
246k
    if ((ctxt->inputNr == 0) && (value->filename != NULL)) {
1959
125k
        directory = xmlParserGetDirectory(value->filename);
1960
125k
        if (directory == NULL) {
1961
51
            xmlErrMemory(ctxt);
1962
51
            return(-1);
1963
51
        }
1964
125k
    }
1965
1966
246k
    if (ctxt->input_id >= INT_MAX) {
1967
0
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, "Input ID overflow\n");
1968
0
        return(-1);
1969
0
    }
1970
1971
246k
    ctxt->inputTab[ctxt->inputNr] = value;
1972
246k
    ctxt->input = value;
1973
1974
246k
    if (ctxt->inputNr == 0) {
1975
125k
        xmlFree(ctxt->directory);
1976
125k
        ctxt->directory = directory;
1977
125k
    }
1978
1979
    /*
1980
     * The input ID is unused internally, but there are entity
1981
     * loaders in downstream code that detect the main document
1982
     * by checking for "input_id == 1".
1983
     */
1984
246k
    value->id = ctxt->input_id++;
1985
1986
246k
    return(ctxt->inputNr++);
1987
246k
}
1988
1989
/**
1990
 * Pops the top parser input from the input stack
1991
 *
1992
 * @param ctxt  an XML parser context
1993
 * @returns the input just removed
1994
 */
1995
xmlParserInput *
1996
xmlCtxtPopInput(xmlParserCtxt *ctxt)
1997
496k
{
1998
496k
    xmlParserInputPtr ret;
1999
2000
496k
    if (ctxt == NULL)
2001
0
        return(NULL);
2002
496k
    if (ctxt->inputNr <= 0)
2003
252k
        return (NULL);
2004
243k
    ctxt->inputNr--;
2005
243k
    if (ctxt->inputNr > 0)
2006
121k
        ctxt->input = ctxt->inputTab[ctxt->inputNr - 1];
2007
122k
    else
2008
122k
        ctxt->input = NULL;
2009
243k
    ret = ctxt->inputTab[ctxt->inputNr];
2010
243k
    ctxt->inputTab[ctxt->inputNr] = NULL;
2011
243k
    return (ret);
2012
496k
}
2013
2014
/**
2015
 * Pushes a new element node on top of the node stack
2016
 *
2017
 * @deprecated Internal function, do not use.
2018
 *
2019
 * @param ctxt  an XML parser context
2020
 * @param value  the element node
2021
 * @returns -1 in case of error, the index in the stack otherwise
2022
 */
2023
int
2024
nodePush(xmlParserCtxt *ctxt, xmlNode *value)
2025
653k
{
2026
653k
    if (ctxt == NULL)
2027
0
        return(0);
2028
2029
653k
    if (ctxt->nodeNr >= ctxt->nodeMax) {
2030
72.9k
        int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
2031
72.9k
        xmlNodePtr *tmp;
2032
72.9k
        int newSize;
2033
2034
72.9k
        newSize = xmlGrowCapacity(ctxt->nodeMax, sizeof(tmp[0]),
2035
72.9k
                                  10, maxDepth);
2036
72.9k
        if (newSize < 0) {
2037
31
            xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
2038
31
                    "Excessive depth in document: %d,"
2039
31
                    " use XML_PARSE_HUGE option\n",
2040
31
                    ctxt->nodeNr);
2041
31
            return(-1);
2042
31
        }
2043
2044
72.9k
  tmp = xmlRealloc(ctxt->nodeTab, newSize * sizeof(tmp[0]));
2045
72.9k
        if (tmp == NULL) {
2046
115
            xmlErrMemory(ctxt);
2047
115
            return (-1);
2048
115
        }
2049
72.7k
        ctxt->nodeTab = tmp;
2050
72.7k
  ctxt->nodeMax = newSize;
2051
72.7k
    }
2052
2053
653k
    ctxt->nodeTab[ctxt->nodeNr] = value;
2054
653k
    ctxt->node = value;
2055
653k
    return (ctxt->nodeNr++);
2056
653k
}
2057
2058
/**
2059
 * Pops the top element node from the node stack
2060
 *
2061
 * @deprecated Internal function, do not use.
2062
 *
2063
 * @param ctxt  an XML parser context
2064
 * @returns the node just removed
2065
 */
2066
xmlNode *
2067
nodePop(xmlParserCtxt *ctxt)
2068
830k
{
2069
830k
    xmlNodePtr ret;
2070
2071
830k
    if (ctxt == NULL) return(NULL);
2072
830k
    if (ctxt->nodeNr <= 0)
2073
262k
        return (NULL);
2074
567k
    ctxt->nodeNr--;
2075
567k
    if (ctxt->nodeNr > 0)
2076
551k
        ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1];
2077
16.6k
    else
2078
16.6k
        ctxt->node = NULL;
2079
567k
    ret = ctxt->nodeTab[ctxt->nodeNr];
2080
567k
    ctxt->nodeTab[ctxt->nodeNr] = NULL;
2081
567k
    return (ret);
2082
830k
}
2083
2084
/**
2085
 * Pushes a new element name/prefix/URL on top of the name stack
2086
 *
2087
 * @param ctxt  an XML parser context
2088
 * @param value  the element name
2089
 * @param prefix  the element prefix
2090
 * @param URI  the element namespace name
2091
 * @param line  the current line number for error messages
2092
 * @param nsNr  the number of namespaces pushed on the namespace table
2093
 * @returns -1 in case of error, the index in the stack otherwise
2094
 */
2095
static int
2096
nameNsPush(xmlParserCtxtPtr ctxt, const xmlChar * value,
2097
           const xmlChar *prefix, const xmlChar *URI, int line, int nsNr)
2098
941k
{
2099
941k
    xmlStartTag *tag;
2100
2101
941k
    if (ctxt->nameNr >= ctxt->nameMax) {
2102
76.3k
        const xmlChar **tmp;
2103
76.3k
        xmlStartTag *tmp2;
2104
76.3k
        int newSize;
2105
2106
76.3k
        newSize = xmlGrowCapacity(ctxt->nameMax,
2107
76.3k
                                  sizeof(tmp[0]) + sizeof(tmp2[0]),
2108
76.3k
                                  10, XML_MAX_ITEMS);
2109
76.3k
        if (newSize < 0)
2110
0
            goto mem_error;
2111
2112
76.3k
        tmp = xmlRealloc(ctxt->nameTab, newSize * sizeof(tmp[0]));
2113
76.3k
        if (tmp == NULL)
2114
72
      goto mem_error;
2115
76.2k
  ctxt->nameTab = tmp;
2116
2117
76.2k
        tmp2 = xmlRealloc(ctxt->pushTab, newSize * sizeof(tmp2[0]));
2118
76.2k
        if (tmp2 == NULL)
2119
73
      goto mem_error;
2120
76.2k
  ctxt->pushTab = tmp2;
2121
2122
76.2k
        ctxt->nameMax = newSize;
2123
865k
    } else if (ctxt->pushTab == NULL) {
2124
46.4k
        ctxt->pushTab = xmlMalloc(ctxt->nameMax * sizeof(ctxt->pushTab[0]));
2125
46.4k
        if (ctxt->pushTab == NULL)
2126
323
            goto mem_error;
2127
46.4k
    }
2128
941k
    ctxt->nameTab[ctxt->nameNr] = value;
2129
941k
    ctxt->name = value;
2130
941k
    tag = &ctxt->pushTab[ctxt->nameNr];
2131
941k
    tag->prefix = prefix;
2132
941k
    tag->URI = URI;
2133
941k
    tag->line = line;
2134
941k
    tag->nsNr = nsNr;
2135
941k
    return (ctxt->nameNr++);
2136
468
mem_error:
2137
468
    xmlErrMemory(ctxt);
2138
468
    return (-1);
2139
941k
}
2140
#ifdef LIBXML_PUSH_ENABLED
2141
/**
2142
 * Pops the top element/prefix/URI name from the name stack
2143
 *
2144
 * @param ctxt  an XML parser context
2145
 * @returns the name just removed
2146
 */
2147
static const xmlChar *
2148
nameNsPop(xmlParserCtxtPtr ctxt)
2149
7.00k
{
2150
7.00k
    const xmlChar *ret;
2151
2152
7.00k
    if (ctxt->nameNr <= 0)
2153
0
        return (NULL);
2154
7.00k
    ctxt->nameNr--;
2155
7.00k
    if (ctxt->nameNr > 0)
2156
6.68k
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2157
326
    else
2158
326
        ctxt->name = NULL;
2159
7.00k
    ret = ctxt->nameTab[ctxt->nameNr];
2160
7.00k
    ctxt->nameTab[ctxt->nameNr] = NULL;
2161
7.00k
    return (ret);
2162
7.00k
}
2163
#endif /* LIBXML_PUSH_ENABLED */
2164
2165
/**
2166
 * Pops the top element name from the name stack
2167
 *
2168
 * @deprecated Internal function, do not use.
2169
 *
2170
 * @param ctxt  an XML parser context
2171
 * @returns the name just removed
2172
 */
2173
static const xmlChar *
2174
namePop(xmlParserCtxtPtr ctxt)
2175
847k
{
2176
847k
    const xmlChar *ret;
2177
2178
847k
    if ((ctxt == NULL) || (ctxt->nameNr <= 0))
2179
133
        return (NULL);
2180
847k
    ctxt->nameNr--;
2181
847k
    if (ctxt->nameNr > 0)
2182
831k
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2183
16.0k
    else
2184
16.0k
        ctxt->name = NULL;
2185
847k
    ret = ctxt->nameTab[ctxt->nameNr];
2186
847k
    ctxt->nameTab[ctxt->nameNr] = NULL;
2187
847k
    return (ret);
2188
847k
}
2189
2190
1.09M
static int spacePush(xmlParserCtxtPtr ctxt, int val) {
2191
1.09M
    if (ctxt->spaceNr >= ctxt->spaceMax) {
2192
96.4k
        int *tmp;
2193
96.4k
        int newSize;
2194
2195
96.4k
        newSize = xmlGrowCapacity(ctxt->spaceMax, sizeof(tmp[0]),
2196
96.4k
                                  10, XML_MAX_ITEMS);
2197
96.4k
        if (newSize < 0) {
2198
0
      xmlErrMemory(ctxt);
2199
0
      return(-1);
2200
0
        }
2201
2202
96.4k
        tmp = xmlRealloc(ctxt->spaceTab, newSize * sizeof(tmp[0]));
2203
96.4k
        if (tmp == NULL) {
2204
189
      xmlErrMemory(ctxt);
2205
189
      return(-1);
2206
189
  }
2207
96.2k
  ctxt->spaceTab = tmp;
2208
2209
96.2k
        ctxt->spaceMax = newSize;
2210
96.2k
    }
2211
1.09M
    ctxt->spaceTab[ctxt->spaceNr] = val;
2212
1.09M
    ctxt->space = &ctxt->spaceTab[ctxt->spaceNr];
2213
1.09M
    return(ctxt->spaceNr++);
2214
1.09M
}
2215
2216
1.00M
static int spacePop(xmlParserCtxtPtr ctxt) {
2217
1.00M
    int ret;
2218
1.00M
    if (ctxt->spaceNr <= 0) return(0);
2219
1.00M
    ctxt->spaceNr--;
2220
1.00M
    if (ctxt->spaceNr > 0)
2221
988k
  ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1];
2222
16.3k
    else
2223
16.3k
        ctxt->space = &ctxt->spaceTab[0];
2224
1.00M
    ret = ctxt->spaceTab[ctxt->spaceNr];
2225
1.00M
    ctxt->spaceTab[ctxt->spaceNr] = -1;
2226
1.00M
    return(ret);
2227
1.00M
}
2228
2229
/*
2230
 * Macros for accessing the content. Those should be used only by the parser,
2231
 * and not exported.
2232
 *
2233
 * Dirty macros, i.e. one often need to make assumption on the context to
2234
 * use them
2235
 *
2236
 *   CUR_PTR return the current pointer to the xmlChar to be parsed.
2237
 *           To be used with extreme caution since operations consuming
2238
 *           characters may move the input buffer to a different location !
2239
 *   CUR     returns the current xmlChar value, i.e. a 8 bit value if compiled
2240
 *           This should be used internally by the parser
2241
 *           only to compare to ASCII values otherwise it would break when
2242
 *           running with UTF-8 encoding.
2243
 *   RAW     same as CUR but in the input buffer, bypass any token
2244
 *           extraction that may have been done
2245
 *   NXT(n)  returns the n'th next xmlChar. Same as CUR is should be used only
2246
 *           to compare on ASCII based substring.
2247
 *   SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined
2248
 *           strings without newlines within the parser.
2249
 *   NEXT1(l) Skip 1 xmlChar, and must also be used only to skip 1 non-newline ASCII
2250
 *           defined char within the parser.
2251
 * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding
2252
 *
2253
 *   NEXT    Skip to the next character, this does the proper decoding
2254
 *           in UTF-8 mode. It also pop-up unfinished entities on the fly.
2255
 *   NEXTL(l) Skip the current unicode character of l xmlChars long.
2256
 *   COPY_BUF  copy the current unicode char to the target buffer, increment
2257
 *            the index
2258
 *   GROW, SHRINK  handling of input buffers
2259
 */
2260
2261
26.3M
#define RAW (*ctxt->input->cur)
2262
2.18G
#define CUR (*ctxt->input->cur)
2263
9.48M
#define NXT(val) ctxt->input->cur[(val)]
2264
2.97G
#define CUR_PTR ctxt->input->cur
2265
2.01M
#define BASE_PTR ctxt->input->base
2266
2267
#define CMP4( s, c1, c2, c3, c4 ) \
2268
35.6M
  ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \
2269
17.9M
    ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 )
2270
#define CMP5( s, c1, c2, c3, c4, c5 ) \
2271
34.9M
  ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 )
2272
#define CMP6( s, c1, c2, c3, c4, c5, c6 ) \
2273
33.9M
  ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 )
2274
#define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \
2275
33.1M
  ( CMP6( s, c1, c2, c3, c4, c5, c6 ) && ((unsigned char *) s)[ 6 ] == c7 )
2276
#define CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) \
2277
32.5M
  ( CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) && ((unsigned char *) s)[ 7 ] == c8 )
2278
#define CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) \
2279
16.1M
  ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \
2280
16.1M
    ((unsigned char *) s)[ 8 ] == c9 )
2281
#define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \
2282
4.23k
  ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \
2283
4.23k
    ((unsigned char *) s)[ 9 ] == c10 )
2284
2285
3.00M
#define SKIP(val) do {             \
2286
3.00M
    ctxt->input->cur += (val),ctxt->input->col+=(val);      \
2287
3.00M
    if (*ctxt->input->cur == 0)           \
2288
3.00M
        xmlParserGrow(ctxt);           \
2289
3.00M
  } while (0)
2290
2291
#define SKIPL(val) do {             \
2292
    int skipl;                \
2293
    for(skipl=0; skipl<val; skipl++) {          \
2294
  if (*(ctxt->input->cur) == '\n') {        \
2295
  ctxt->input->line++; ctxt->input->col = 1;      \
2296
  } else ctxt->input->col++;          \
2297
  ctxt->input->cur++;           \
2298
    }                 \
2299
    if (*ctxt->input->cur == 0)           \
2300
        xmlParserGrow(ctxt);            \
2301
  } while (0)
2302
2303
#define SHRINK \
2304
36.3M
    if (!PARSER_PROGRESSIVE(ctxt)) \
2305
36.3M
  xmlParserShrink(ctxt);
2306
2307
#define GROW \
2308
66.7M
    if ((!PARSER_PROGRESSIVE(ctxt)) && \
2309
66.7M
        (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \
2310
4.83M
  xmlParserGrow(ctxt);
2311
2312
7.77M
#define SKIP_BLANKS xmlSkipBlankChars(ctxt)
2313
2314
1.11M
#define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt)
2315
2316
911M
#define NEXT xmlNextChar(ctxt)
2317
2318
1.50M
#define NEXT1 {               \
2319
1.50M
  ctxt->input->col++;           \
2320
1.50M
  ctxt->input->cur++;           \
2321
1.50M
  if (*ctxt->input->cur == 0)         \
2322
1.50M
      xmlParserGrow(ctxt);           \
2323
1.50M
    }
2324
2325
1.64G
#define NEXTL(l) do {             \
2326
1.64G
    if (*(ctxt->input->cur) == '\n') {         \
2327
51.4M
  ctxt->input->line++; ctxt->input->col = 1;      \
2328
1.59G
    } else ctxt->input->col++;           \
2329
1.64G
    ctxt->input->cur += l;        \
2330
1.64G
  } while (0)
2331
2332
#define COPY_BUF(b, i, v)           \
2333
249M
    if (v < 0x80) b[i++] = v;           \
2334
249M
    else i += xmlCopyCharMultiByte(&b[i],v)
2335
2336
static int
2337
253M
xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) {
2338
253M
    int c = xmlCurrentChar(ctxt, len);
2339
2340
253M
    if (c == XML_INVALID_CHAR)
2341
50.4M
        c = 0xFFFD; /* replacement character */
2342
2343
253M
    return(c);
2344
253M
}
2345
2346
/**
2347
 * Skip whitespace in the input stream.
2348
 *
2349
 * @deprecated Internal function, do not use.
2350
 *
2351
 * @param ctxt  the XML parser context
2352
 * @returns the number of space chars skipped
2353
 */
2354
int
2355
8.15M
xmlSkipBlankChars(xmlParserCtxt *ctxt) {
2356
8.15M
    const xmlChar *cur;
2357
8.15M
    int res = 0;
2358
2359
8.15M
    cur = ctxt->input->cur;
2360
8.15M
    while (IS_BLANK_CH(*cur)) {
2361
2.20M
        if (*cur == '\n') {
2362
528k
            ctxt->input->line++; ctxt->input->col = 1;
2363
1.67M
        } else {
2364
1.67M
            ctxt->input->col++;
2365
1.67M
        }
2366
2.20M
        cur++;
2367
2.20M
        if (res < INT_MAX)
2368
2.20M
            res++;
2369
2.20M
        if (*cur == 0) {
2370
28.5k
            ctxt->input->cur = cur;
2371
28.5k
            xmlParserGrow(ctxt);
2372
28.5k
            cur = ctxt->input->cur;
2373
28.5k
        }
2374
2.20M
    }
2375
8.15M
    ctxt->input->cur = cur;
2376
2377
8.15M
    if (res > 4)
2378
19.2k
        GROW;
2379
2380
8.15M
    return(res);
2381
8.15M
}
2382
2383
static void
2384
110k
xmlPopPE(xmlParserCtxtPtr ctxt) {
2385
110k
    unsigned long consumed;
2386
110k
    xmlEntityPtr ent;
2387
2388
110k
    ent = ctxt->input->entity;
2389
2390
110k
    ent->flags &= ~XML_ENT_EXPANDING;
2391
2392
110k
    if ((ent->flags & XML_ENT_CHECKED) == 0) {
2393
10.9k
        int result;
2394
2395
        /*
2396
         * Read the rest of the stream in case of errors. We want
2397
         * to account for the whole entity size.
2398
         */
2399
11.2k
        do {
2400
11.2k
            ctxt->input->cur = ctxt->input->end;
2401
11.2k
            xmlParserShrink(ctxt);
2402
11.2k
            result = xmlParserGrow(ctxt);
2403
11.2k
        } while (result > 0);
2404
2405
10.9k
        consumed = ctxt->input->consumed;
2406
10.9k
        xmlSaturatedAddSizeT(&consumed,
2407
10.9k
                             ctxt->input->end - ctxt->input->base);
2408
2409
10.9k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
2410
2411
        /*
2412
         * Add to sizeentities when parsing an external entity
2413
         * for the first time.
2414
         */
2415
10.9k
        if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
2416
6.27k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
2417
6.27k
        }
2418
2419
10.9k
        ent->flags |= XML_ENT_CHECKED;
2420
10.9k
    }
2421
2422
110k
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
2423
2424
110k
    xmlParserEntityCheck(ctxt, ent->expandedSize);
2425
2426
110k
    GROW;
2427
110k
}
2428
2429
/**
2430
 * Skip whitespace in the input stream, also handling parameter
2431
 * entities.
2432
 *
2433
 * @param ctxt  the XML parser context
2434
 * @returns the number of space chars skipped
2435
 */
2436
static int
2437
1.11M
xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) {
2438
1.11M
    int res = 0;
2439
1.11M
    int inParam;
2440
1.11M
    int expandParam;
2441
2442
1.11M
    inParam = PARSER_IN_PE(ctxt);
2443
1.11M
    expandParam = PARSER_EXTERNAL(ctxt);
2444
2445
1.11M
    if (!inParam && !expandParam)
2446
380k
        return(xmlSkipBlankChars(ctxt));
2447
2448
    /*
2449
     * It's Okay to use CUR/NEXT here since all the blanks are on
2450
     * the ASCII range.
2451
     */
2452
2.26M
    while (PARSER_STOPPED(ctxt) == 0) {
2453
2.26M
        if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */
2454
1.50M
            NEXT;
2455
1.50M
        } else if (CUR == '%') {
2456
46.8k
            if ((expandParam == 0) ||
2457
46.6k
                (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0))
2458
28.6k
                break;
2459
2460
            /*
2461
             * Expand parameter entity. We continue to consume
2462
             * whitespace at the start of the entity and possible
2463
             * even consume the whole entity and pop it. We might
2464
             * even pop multiple PEs in this loop.
2465
             */
2466
18.2k
            xmlParsePERefInternal(ctxt, 0);
2467
2468
18.2k
            inParam = PARSER_IN_PE(ctxt);
2469
18.2k
            expandParam = PARSER_EXTERNAL(ctxt);
2470
712k
        } else if (CUR == 0) {
2471
42.3k
            if (inParam == 0)
2472
333
                break;
2473
2474
            /*
2475
             * Don't pop parameter entities that start a markup
2476
             * declaration to detect Well-formedness constraint:
2477
             * PE Between Declarations.
2478
             */
2479
42.0k
            if (ctxt->input->flags & XML_INPUT_MARKUP_DECL)
2480
30.0k
                break;
2481
2482
11.9k
            xmlPopPE(ctxt);
2483
2484
11.9k
            inParam = PARSER_IN_PE(ctxt);
2485
11.9k
            expandParam = PARSER_EXTERNAL(ctxt);
2486
669k
        } else {
2487
669k
            break;
2488
669k
        }
2489
2490
        /*
2491
         * Also increase the counter when entering or exiting a PERef.
2492
         * The spec says: "When a parameter-entity reference is recognized
2493
         * in the DTD and included, its replacement text MUST be enlarged
2494
         * by the attachment of one leading and one following space (#x20)
2495
         * character."
2496
         */
2497
1.53M
        if (res < INT_MAX)
2498
1.53M
            res++;
2499
1.53M
    }
2500
2501
730k
    return(res);
2502
1.11M
}
2503
2504
/************************************************************************
2505
 *                  *
2506
 *    Commodity functions to handle entities      *
2507
 *                  *
2508
 ************************************************************************/
2509
2510
/**
2511
 * @deprecated Internal function, don't use.
2512
 *
2513
 * @param ctxt  an XML parser context
2514
 * @returns the current xmlChar in the parser context
2515
 */
2516
xmlChar
2517
0
xmlPopInput(xmlParserCtxt *ctxt) {
2518
0
    xmlParserInputPtr input;
2519
2520
0
    if ((ctxt == NULL) || (ctxt->inputNr <= 1)) return(0);
2521
0
    input = xmlCtxtPopInput(ctxt);
2522
0
    xmlFreeInputStream(input);
2523
0
    if (*ctxt->input->cur == 0)
2524
0
        xmlParserGrow(ctxt);
2525
0
    return(CUR);
2526
0
}
2527
2528
/**
2529
 * Push an input stream onto the stack.
2530
 *
2531
 * @deprecated Internal function, don't use.
2532
 *
2533
 * @param ctxt  an XML parser context
2534
 * @param input  an XML parser input fragment (entity, XML fragment ...).
2535
 * @returns -1 in case of error or the index in the input stack
2536
 */
2537
int
2538
0
xmlPushInput(xmlParserCtxt *ctxt, xmlParserInput *input) {
2539
0
    int ret;
2540
2541
0
    if ((ctxt == NULL) || (input == NULL))
2542
0
        return(-1);
2543
2544
0
    ret = xmlCtxtPushInput(ctxt, input);
2545
0
    if (ret >= 0)
2546
0
        GROW;
2547
0
    return(ret);
2548
0
}
2549
2550
/**
2551
 * Parse a numeric character reference. Always consumes '&'.
2552
 *
2553
 * @deprecated Internal function, don't use.
2554
 *
2555
 *     [66] CharRef ::= '&#' [0-9]+ ';' |
2556
 *                      '&#x' [0-9a-fA-F]+ ';'
2557
 *
2558
 * [ WFC: Legal Character ]
2559
 * Characters referred to using character references must match the
2560
 * production for Char.
2561
 *
2562
 * @param ctxt  an XML parser context
2563
 * @returns the value parsed (as an int), 0 in case of error
2564
 */
2565
int
2566
101k
xmlParseCharRef(xmlParserCtxt *ctxt) {
2567
101k
    int val = 0;
2568
101k
    int count = 0;
2569
2570
    /*
2571
     * Using RAW/CUR/NEXT is okay since we are working on ASCII range here
2572
     */
2573
101k
    if ((RAW == '&') && (NXT(1) == '#') &&
2574
101k
        (NXT(2) == 'x')) {
2575
39.8k
  SKIP(3);
2576
39.8k
  GROW;
2577
259k
  while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) {
2578
229k
      if (count++ > 20) {
2579
11.8k
    count = 0;
2580
11.8k
    GROW;
2581
11.8k
      }
2582
229k
      if ((RAW >= '0') && (RAW <= '9'))
2583
146k
          val = val * 16 + (CUR - '0');
2584
83.1k
      else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20))
2585
22.3k
          val = val * 16 + (CUR - 'a') + 10;
2586
60.8k
      else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20))
2587
50.4k
          val = val * 16 + (CUR - 'A') + 10;
2588
10.3k
      else {
2589
10.3k
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2590
10.3k
    val = 0;
2591
10.3k
    break;
2592
10.3k
      }
2593
219k
      if (val > 0x110000)
2594
135k
          val = 0x110000;
2595
2596
219k
      NEXT;
2597
219k
      count++;
2598
219k
  }
2599
39.8k
  if (RAW == ';') {
2600
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2601
29.4k
      ctxt->input->col++;
2602
29.4k
      ctxt->input->cur++;
2603
29.4k
  }
2604
61.3k
    } else if  ((RAW == '&') && (NXT(1) == '#')) {
2605
61.3k
  SKIP(2);
2606
61.3k
  GROW;
2607
191k
  while (RAW != ';') { /* loop blocked by count */
2608
141k
      if (count++ > 20) {
2609
2.01k
    count = 0;
2610
2.01k
    GROW;
2611
2.01k
      }
2612
141k
      if ((RAW >= '0') && (RAW <= '9'))
2613
130k
          val = val * 10 + (CUR - '0');
2614
10.5k
      else {
2615
10.5k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2616
10.5k
    val = 0;
2617
10.5k
    break;
2618
10.5k
      }
2619
130k
      if (val > 0x110000)
2620
12.0k
          val = 0x110000;
2621
2622
130k
      NEXT;
2623
130k
      count++;
2624
130k
  }
2625
61.3k
  if (RAW == ';') {
2626
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2627
50.7k
      ctxt->input->col++;
2628
50.7k
      ctxt->input->cur++;
2629
50.7k
  }
2630
61.3k
    } else {
2631
0
        if (RAW == '&')
2632
0
            SKIP(1);
2633
0
        xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2634
0
    }
2635
2636
    /*
2637
     * [ WFC: Legal Character ]
2638
     * Characters referred to using character references must match the
2639
     * production for Char.
2640
     */
2641
101k
    if (val >= 0x110000) {
2642
1.57k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2643
1.57k
                "xmlParseCharRef: character reference out of bounds\n",
2644
1.57k
          val);
2645
1.57k
        val = 0xFFFD;
2646
99.6k
    } else if (!IS_CHAR(val)) {
2647
28.5k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2648
28.5k
                          "xmlParseCharRef: invalid xmlChar value %d\n",
2649
28.5k
                    val);
2650
28.5k
    }
2651
101k
    return(val);
2652
101k
}
2653
2654
/**
2655
 * Parse Reference declarations, variant parsing from a string rather
2656
 * than an an input flow.
2657
 *
2658
 *     [66] CharRef ::= '&#' [0-9]+ ';' |
2659
 *                      '&#x' [0-9a-fA-F]+ ';'
2660
 *
2661
 * [ WFC: Legal Character ]
2662
 * Characters referred to using character references must match the
2663
 * production for Char.
2664
 *
2665
 * @param ctxt  an XML parser context
2666
 * @param str  a pointer to an index in the string
2667
 * @returns the value parsed (as an int), 0 in case of error, str will be
2668
 *         updated to the current value of the index
2669
 */
2670
static int
2671
571k
xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) {
2672
571k
    const xmlChar *ptr;
2673
571k
    xmlChar cur;
2674
571k
    int val = 0;
2675
2676
571k
    if ((str == NULL) || (*str == NULL)) return(0);
2677
571k
    ptr = *str;
2678
571k
    cur = *ptr;
2679
571k
    if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) {
2680
8.12k
  ptr += 3;
2681
8.12k
  cur = *ptr;
2682
29.0k
  while (cur != ';') { /* Non input consuming loop */
2683
21.6k
      if ((cur >= '0') && (cur <= '9'))
2684
11.4k
          val = val * 16 + (cur - '0');
2685
10.2k
      else if ((cur >= 'a') && (cur <= 'f'))
2686
2.43k
          val = val * 16 + (cur - 'a') + 10;
2687
7.76k
      else if ((cur >= 'A') && (cur <= 'F'))
2688
7.05k
          val = val * 16 + (cur - 'A') + 10;
2689
713
      else {
2690
713
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2691
713
    val = 0;
2692
713
    break;
2693
713
      }
2694
20.9k
      if (val > 0x110000)
2695
749
          val = 0x110000;
2696
2697
20.9k
      ptr++;
2698
20.9k
      cur = *ptr;
2699
20.9k
  }
2700
8.12k
  if (cur == ';')
2701
7.41k
      ptr++;
2702
562k
    } else if  ((cur == '&') && (ptr[1] == '#')){
2703
562k
  ptr += 2;
2704
562k
  cur = *ptr;
2705
1.70M
  while (cur != ';') { /* Non input consuming loops */
2706
1.14M
      if ((cur >= '0') && (cur <= '9'))
2707
1.14M
          val = val * 10 + (cur - '0');
2708
2.07k
      else {
2709
2.07k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2710
2.07k
    val = 0;
2711
2.07k
    break;
2712
2.07k
      }
2713
1.14M
      if (val > 0x110000)
2714
2.07k
          val = 0x110000;
2715
2716
1.14M
      ptr++;
2717
1.14M
      cur = *ptr;
2718
1.14M
  }
2719
562k
  if (cur == ';')
2720
560k
      ptr++;
2721
562k
    } else {
2722
0
  xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2723
0
  return(0);
2724
0
    }
2725
571k
    *str = ptr;
2726
2727
    /*
2728
     * [ WFC: Legal Character ]
2729
     * Characters referred to using character references must match the
2730
     * production for Char.
2731
     */
2732
571k
    if (val >= 0x110000) {
2733
503
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2734
503
                "xmlParseStringCharRef: character reference out of bounds\n",
2735
503
                val);
2736
570k
    } else if (IS_CHAR(val)) {
2737
566k
        return(val);
2738
566k
    } else {
2739
3.69k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2740
3.69k
        "xmlParseStringCharRef: invalid xmlChar value %d\n",
2741
3.69k
        val);
2742
3.69k
    }
2743
4.20k
    return(0);
2744
571k
}
2745
2746
/**
2747
 *     [69] PEReference ::= '%' Name ';'
2748
 *
2749
 * @deprecated Internal function, do not use.
2750
 *
2751
 * [ WFC: No Recursion ]
2752
 * A parsed entity must not contain a recursive
2753
 * reference to itself, either directly or indirectly.
2754
 *
2755
 * [ WFC: Entity Declared ]
2756
 * In a document without any DTD, a document with only an internal DTD
2757
 * subset which contains no parameter entity references, or a document
2758
 * with "standalone='yes'", ...  ... The declaration of a parameter
2759
 * entity must precede any reference to it...
2760
 *
2761
 * [ VC: Entity Declared ]
2762
 * In a document with an external subset or external parameter entities
2763
 * with "standalone='no'", ...  ... The declaration of a parameter entity
2764
 * must precede any reference to it...
2765
 *
2766
 * [ WFC: In DTD ]
2767
 * Parameter-entity references may only appear in the DTD.
2768
 * NOTE: misleading but this is handled.
2769
 *
2770
 * A PEReference may have been detected in the current input stream
2771
 * the handling is done accordingly to
2772
 *      http://www.w3.org/TR/REC-xml#entproc
2773
 * i.e.
2774
 *   - Included in literal in entity values
2775
 *   - Included as Parameter Entity reference within DTDs
2776
 * @param ctxt  the parser context
2777
 */
2778
void
2779
0
xmlParserHandlePEReference(xmlParserCtxt *ctxt) {
2780
0
    xmlParsePERefInternal(ctxt, 0);
2781
0
}
2782
2783
/**
2784
 * @deprecated Internal function, don't use.
2785
 *
2786
 * @param ctxt  the parser context
2787
 * @param str  the input string
2788
 * @param len  the string length
2789
 * @param what  combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF
2790
 * @param end  an end marker xmlChar, 0 if none
2791
 * @param end2  an end marker xmlChar, 0 if none
2792
 * @param end3  an end marker xmlChar, 0 if none
2793
 * @returns A newly allocated string with the substitution done. The caller
2794
 *      must deallocate it !
2795
 */
2796
xmlChar *
2797
xmlStringLenDecodeEntities(xmlParserCtxt *ctxt, const xmlChar *str, int len,
2798
                           int what ATTRIBUTE_UNUSED,
2799
0
                           xmlChar end, xmlChar end2, xmlChar end3) {
2800
0
    if ((ctxt == NULL) || (str == NULL) || (len < 0))
2801
0
        return(NULL);
2802
2803
0
    if ((str[len] != 0) ||
2804
0
        (end != 0) || (end2 != 0) || (end3 != 0))
2805
0
        return(NULL);
2806
2807
0
    return(xmlExpandEntitiesInAttValue(ctxt, str, 0));
2808
0
}
2809
2810
/**
2811
 * @deprecated Internal function, don't use.
2812
 *
2813
 * @param ctxt  the parser context
2814
 * @param str  the input string
2815
 * @param what  combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF
2816
 * @param end  an end marker xmlChar, 0 if none
2817
 * @param end2  an end marker xmlChar, 0 if none
2818
 * @param end3  an end marker xmlChar, 0 if none
2819
 * @returns A newly allocated string with the substitution done. The caller
2820
 *      must deallocate it !
2821
 */
2822
xmlChar *
2823
xmlStringDecodeEntities(xmlParserCtxt *ctxt, const xmlChar *str,
2824
                        int what ATTRIBUTE_UNUSED,
2825
0
            xmlChar end, xmlChar  end2, xmlChar end3) {
2826
0
    if ((ctxt == NULL) || (str == NULL))
2827
0
        return(NULL);
2828
2829
0
    if ((end != 0) || (end2 != 0) || (end3 != 0))
2830
0
        return(NULL);
2831
2832
0
    return(xmlExpandEntitiesInAttValue(ctxt, str, 0));
2833
0
}
2834
2835
/************************************************************************
2836
 *                  *
2837
 *    Commodity functions, cleanup needed ?     *
2838
 *                  *
2839
 ************************************************************************/
2840
2841
/**
2842
 * Is this a sequence of blank chars that one can ignore ?
2843
 *
2844
 * @param ctxt  an XML parser context
2845
 * @param str  a xmlChar *
2846
 * @param len  the size of `str`
2847
 * @param blank_chars  we know the chars are blanks
2848
 * @returns 1 if ignorable 0 otherwise.
2849
 */
2850
2851
static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len,
2852
637k
                     int blank_chars) {
2853
637k
    int i;
2854
637k
    xmlNodePtr lastChild;
2855
2856
    /*
2857
     * Check for xml:space value.
2858
     */
2859
637k
    if ((ctxt->space == NULL) || (*(ctxt->space) == 1) ||
2860
636k
        (*(ctxt->space) == -2))
2861
579k
  return(0);
2862
2863
    /*
2864
     * Check that the string is made of blanks
2865
     */
2866
57.2k
    if (blank_chars == 0) {
2867
115k
  for (i = 0;i < len;i++)
2868
107k
      if (!(IS_BLANK_CH(str[i]))) return(0);
2869
49.6k
    }
2870
2871
    /*
2872
     * Look if the element is mixed content in the DTD if available
2873
     */
2874
15.5k
    if (ctxt->node == NULL) return(0);
2875
15.5k
    if (ctxt->myDoc != NULL) {
2876
15.5k
        xmlElementPtr elemDecl = NULL;
2877
15.5k
        xmlDocPtr doc = ctxt->myDoc;
2878
15.5k
        const xmlChar *prefix = NULL;
2879
2880
15.5k
        if (ctxt->node->ns)
2881
1.40k
            prefix = ctxt->node->ns->prefix;
2882
15.5k
        if (doc->intSubset != NULL)
2883
8.70k
            elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name,
2884
8.70k
                                      prefix);
2885
15.5k
        if ((elemDecl == NULL) && (doc->extSubset != NULL))
2886
4.43k
            elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name,
2887
4.43k
                                      prefix);
2888
15.5k
        if (elemDecl != NULL) {
2889
4.55k
            if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)
2890
3.55k
                return(1);
2891
1.00k
            if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) ||
2892
792
                (elemDecl->etype == XML_ELEMENT_TYPE_MIXED))
2893
691
                return(0);
2894
1.00k
        }
2895
15.5k
    }
2896
2897
    /*
2898
     * Otherwise, heuristic :-\
2899
     *
2900
     * When push parsing, we could be at the end of a chunk.
2901
     * This makes the look-ahead and consequently the NOBLANKS
2902
     * option unreliable.
2903
     */
2904
11.3k
    if ((RAW != '<') && (RAW != 0xD)) return(0);
2905
5.96k
    if ((ctxt->node->children == NULL) &&
2906
4.08k
  (RAW == '<') && (NXT(1) == '/')) return(0);
2907
2908
5.74k
    lastChild = xmlGetLastChild(ctxt->node);
2909
5.74k
    if (lastChild == NULL) {
2910
3.86k
        if ((ctxt->node->type != XML_ELEMENT_NODE) &&
2911
0
            (ctxt->node->content != NULL)) return(0);
2912
3.86k
    } else if (xmlNodeIsText(lastChild))
2913
208
        return(0);
2914
1.67k
    else if ((ctxt->node->children != NULL) &&
2915
1.67k
             (xmlNodeIsText(ctxt->node->children)))
2916
246
        return(0);
2917
5.29k
    return(1);
2918
5.74k
}
2919
2920
/************************************************************************
2921
 *                  *
2922
 *    Extra stuff for namespace support     *
2923
 *  Relates to http://www.w3.org/TR/WD-xml-names      *
2924
 *                  *
2925
 ************************************************************************/
2926
2927
/**
2928
 * Parse an UTF8 encoded XML qualified name string
2929
 *
2930
 * @deprecated Don't use.
2931
 *
2932
 * @param ctxt  an XML parser context
2933
 * @param name  an XML parser context
2934
 * @param prefixOut  a xmlChar **
2935
 * @returns the local part, and prefix is updated
2936
 *   to get the Prefix if any.
2937
 */
2938
2939
xmlChar *
2940
0
xmlSplitQName(xmlParserCtxt *ctxt, const xmlChar *name, xmlChar **prefixOut) {
2941
0
    xmlChar *ret;
2942
0
    const xmlChar *localname;
2943
2944
0
    localname = xmlSplitQName4(name, prefixOut);
2945
0
    if (localname == NULL) {
2946
0
        xmlCtxtErrMemory(ctxt);
2947
0
        return(NULL);
2948
0
    }
2949
2950
0
    ret = xmlStrdup(localname);
2951
0
    if (ret == NULL) {
2952
0
        xmlCtxtErrMemory(ctxt);
2953
0
        xmlFree(*prefixOut);
2954
0
    }
2955
2956
0
    return(ret);
2957
0
}
2958
2959
/************************************************************************
2960
 *                  *
2961
 *      The parser itself       *
2962
 *  Relates to http://www.w3.org/TR/REC-xml       *
2963
 *                  *
2964
 ************************************************************************/
2965
2966
/************************************************************************
2967
 *                  *
2968
 *  Routines to parse Name, NCName and NmToken      *
2969
 *                  *
2970
 ************************************************************************/
2971
2972
/*
2973
 * The two following functions are related to the change of accepted
2974
 * characters for Name and NmToken in the Revision 5 of XML-1.0
2975
 * They correspond to the modified production [4] and the new production [4a]
2976
 * changes in that revision. Also note that the macros used for the
2977
 * productions Letter, Digit, CombiningChar and Extender are not needed
2978
 * anymore.
2979
 * We still keep compatibility to pre-revision5 parsing semantic if the
2980
 * new XML_PARSE_OLD10 option is given to the parser.
2981
 */
2982
2983
static int
2984
1.57M
xmlIsNameStartCharNew(int c) {
2985
    /*
2986
     * Use the new checks of production [4] [4a] amd [5] of the
2987
     * Update 5 of XML-1.0
2988
     */
2989
1.57M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2990
1.55M
        (((c >= 'a') && (c <= 'z')) ||
2991
289k
         ((c >= 'A') && (c <= 'Z')) ||
2992
280k
         (c == '_') || (c == ':') ||
2993
272k
         ((c >= 0xC0) && (c <= 0xD6)) ||
2994
271k
         ((c >= 0xD8) && (c <= 0xF6)) ||
2995
270k
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2996
268k
         ((c >= 0x370) && (c <= 0x37D)) ||
2997
268k
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2998
252k
         ((c >= 0x200C) && (c <= 0x200D)) ||
2999
252k
         ((c >= 0x2070) && (c <= 0x218F)) ||
3000
250k
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
3001
250k
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
3002
249k
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
3003
248k
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
3004
233k
         ((c >= 0x10000) && (c <= 0xEFFFF))))
3005
1.33M
        return(1);
3006
244k
    return(0);
3007
1.57M
}
3008
3009
static int
3010
141M
xmlIsNameCharNew(int c) {
3011
    /*
3012
     * Use the new checks of production [4] [4a] amd [5] of the
3013
     * Update 5 of XML-1.0
3014
     */
3015
141M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3016
141M
        (((c >= 'a') && (c <= 'z')) ||
3017
127M
         ((c >= 'A') && (c <= 'Z')) ||
3018
127M
         ((c >= '0') && (c <= '9')) || /* !start */
3019
126M
         (c == '_') || (c == ':') ||
3020
126M
         (c == '-') || (c == '.') || (c == 0xB7) || /* !start */
3021
126M
         ((c >= 0xC0) && (c <= 0xD6)) ||
3022
126M
         ((c >= 0xD8) && (c <= 0xF6)) ||
3023
125M
         ((c >= 0xF8) && (c <= 0x2FF)) ||
3024
125M
         ((c >= 0x300) && (c <= 0x36F)) || /* !start */
3025
125M
         ((c >= 0x370) && (c <= 0x37D)) ||
3026
125M
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
3027
125M
         ((c >= 0x200C) && (c <= 0x200D)) ||
3028
125M
         ((c >= 0x203F) && (c <= 0x2040)) || /* !start */
3029
125M
         ((c >= 0x2070) && (c <= 0x218F)) ||
3030
13.1M
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
3031
13.1M
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
3032
13.1M
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
3033
13.1M
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
3034
1.31M
         ((c >= 0x10000) && (c <= 0xEFFFF))))
3035
140M
         return(1);
3036
1.33M
    return(0);
3037
141M
}
3038
3039
static int
3040
1.43M
xmlIsNameStartCharOld(int c) {
3041
1.43M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3042
1.43M
        ((IS_LETTER(c) || (c == '_') || (c == ':'))))
3043
1.06M
        return(1);
3044
369k
    return(0);
3045
1.43M
}
3046
3047
static int
3048
10.0M
xmlIsNameCharOld(int c) {
3049
10.0M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3050
10.0M
        ((IS_LETTER(c)) || (IS_DIGIT(c)) ||
3051
1.09M
         (c == '.') || (c == '-') ||
3052
1.08M
         (c == '_') || (c == ':') ||
3053
1.07M
         (IS_COMBINING(c)) ||
3054
1.06M
         (IS_EXTENDER(c))))
3055
8.93M
        return(1);
3056
1.07M
    return(0);
3057
10.0M
}
3058
3059
static int
3060
3.01M
xmlIsNameStartChar(int c, int old10) {
3061
3.01M
    if (!old10)
3062
1.57M
        return(xmlIsNameStartCharNew(c));
3063
1.43M
    else
3064
1.43M
        return(xmlIsNameStartCharOld(c));
3065
3.01M
}
3066
3067
static int
3068
151M
xmlIsNameChar(int c, int old10) {
3069
151M
    if (!old10)
3070
141M
        return(xmlIsNameCharNew(c));
3071
10.0M
    else
3072
10.0M
        return(xmlIsNameCharOld(c));
3073
151M
}
3074
3075
/*
3076
 * Scan an XML Name, NCName or Nmtoken.
3077
 *
3078
 * Returns a pointer to the end of the name on success. If the
3079
 * name is invalid, returns `ptr`. If the name is longer than
3080
 * `maxSize` bytes, returns NULL.
3081
 *
3082
 * @param ptr  pointer to the start of the name
3083
 * @param maxSize  maximum size in bytes
3084
 * @param flags  XML_SCAN_* flags
3085
 * @returns a pointer to the end of the name or NULL
3086
 */
3087
const xmlChar *
3088
2.42M
xmlScanName(const xmlChar *ptr, size_t maxSize, int flags) {
3089
2.42M
    int stop = flags & XML_SCAN_NC ? ':' : 0;
3090
2.42M
    int old10 = flags & XML_SCAN_OLD10 ? 1 : 0;
3091
3092
20.1M
    while (1) {
3093
20.1M
        int c, len;
3094
3095
20.1M
        c = *ptr;
3096
20.1M
        if (c < 0x80) {
3097
9.42M
            if (c == stop)
3098
146k
                break;
3099
9.28M
            len = 1;
3100
10.7M
        } else {
3101
10.7M
            len = 4;
3102
10.7M
            c = xmlGetUTF8Char(ptr, &len);
3103
10.7M
            if (c < 0)
3104
4.49k
                break;
3105
10.7M
        }
3106
3107
20.0M
        if (flags & XML_SCAN_NMTOKEN ?
3108
17.6M
                !xmlIsNameChar(c, old10) :
3109
20.0M
                !xmlIsNameStartChar(c, old10))
3110
2.27M
            break;
3111
3112
17.7M
        if ((size_t) len > maxSize)
3113
199
            return(NULL);
3114
17.7M
        ptr += len;
3115
17.7M
        maxSize -= len;
3116
17.7M
        flags |= XML_SCAN_NMTOKEN;
3117
17.7M
    }
3118
3119
2.42M
    return(ptr);
3120
2.42M
}
3121
3122
static const xmlChar *
3123
264k
xmlParseNameComplex(xmlParserCtxtPtr ctxt) {
3124
264k
    const xmlChar *ret;
3125
264k
    int len = 0, l;
3126
264k
    int c;
3127
264k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3128
116k
                    XML_MAX_TEXT_LENGTH :
3129
264k
                    XML_MAX_NAME_LENGTH;
3130
264k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3131
3132
    /*
3133
     * Handler for more complex cases
3134
     */
3135
264k
    c = xmlCurrentChar(ctxt, &l);
3136
264k
    if (!xmlIsNameStartChar(c, old10))
3137
193k
        return(NULL);
3138
71.2k
    len += l;
3139
71.2k
    NEXTL(l);
3140
71.2k
    c = xmlCurrentChar(ctxt, &l);
3141
29.8M
    while (xmlIsNameChar(c, old10)) {
3142
29.7M
        if (len <= INT_MAX - l)
3143
29.7M
            len += l;
3144
29.7M
        NEXTL(l);
3145
29.7M
        c = xmlCurrentChar(ctxt, &l);
3146
29.7M
    }
3147
71.2k
    if (len > maxLength) {
3148
474
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3149
474
        return(NULL);
3150
474
    }
3151
70.8k
    if (ctxt->input->cur - ctxt->input->base < len) {
3152
        /*
3153
         * There were a couple of bugs where PERefs lead to to a change
3154
         * of the buffer. Check the buffer size to avoid passing an invalid
3155
         * pointer to xmlDictLookup.
3156
         */
3157
0
        xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR,
3158
0
                    "unexpected change of input buffer");
3159
0
        return (NULL);
3160
0
    }
3161
70.8k
    if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r'))
3162
266
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len);
3163
70.5k
    else
3164
70.5k
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len);
3165
70.8k
    if (ret == NULL)
3166
9
        xmlErrMemory(ctxt);
3167
70.8k
    return(ret);
3168
70.8k
}
3169
3170
/**
3171
 * Parse an XML name.
3172
 *
3173
 * @deprecated Internal function, don't use.
3174
 *
3175
 *     [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' |
3176
 *                      CombiningChar | Extender
3177
 *
3178
 *     [5] Name ::= (Letter | '_' | ':') (NameChar)*
3179
 *
3180
 *     [6] Names ::= Name (#x20 Name)*
3181
 *
3182
 * @param ctxt  an XML parser context
3183
 * @returns the Name parsed or NULL
3184
 */
3185
3186
const xmlChar *
3187
3.45M
xmlParseName(xmlParserCtxt *ctxt) {
3188
3.45M
    const xmlChar *in;
3189
3.45M
    const xmlChar *ret;
3190
3.45M
    size_t count = 0;
3191
3.45M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3192
1.82M
                       XML_MAX_TEXT_LENGTH :
3193
3.45M
                       XML_MAX_NAME_LENGTH;
3194
3195
3.45M
    GROW;
3196
3197
    /*
3198
     * Accelerator for simple ASCII names
3199
     */
3200
3.45M
    in = ctxt->input->cur;
3201
3.45M
    if (((*in >= 0x61) && (*in <= 0x7A)) ||
3202
880k
  ((*in >= 0x41) && (*in <= 0x5A)) ||
3203
3.23M
  (*in == '_') || (*in == ':')) {
3204
3.23M
  in++;
3205
33.9M
  while (((*in >= 0x61) && (*in <= 0x7A)) ||
3206
7.36M
         ((*in >= 0x41) && (*in <= 0x5A)) ||
3207
4.86M
         ((*in >= 0x30) && (*in <= 0x39)) ||
3208
4.37M
         (*in == '_') || (*in == '-') ||
3209
4.21M
         (*in == ':') || (*in == '.'))
3210
30.7M
      in++;
3211
3.23M
  if ((*in > 0) && (*in < 0x80)) {
3212
3.18M
      count = in - ctxt->input->cur;
3213
3.18M
            if (count > maxLength) {
3214
98
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3215
98
                return(NULL);
3216
98
            }
3217
3.18M
      ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count);
3218
3.18M
      ctxt->input->cur = in;
3219
3.18M
      ctxt->input->col += count;
3220
3.18M
      if (ret == NULL)
3221
14
          xmlErrMemory(ctxt);
3222
3.18M
      return(ret);
3223
3.18M
  }
3224
3.23M
    }
3225
    /* accelerator for special cases */
3226
264k
    return(xmlParseNameComplex(ctxt));
3227
3.45M
}
3228
3229
static xmlHashedString
3230
443k
xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) {
3231
443k
    xmlHashedString ret;
3232
443k
    int len = 0, l;
3233
443k
    int c;
3234
443k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3235
334k
                    XML_MAX_TEXT_LENGTH :
3236
443k
                    XML_MAX_NAME_LENGTH;
3237
443k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3238
443k
    size_t startPosition = 0;
3239
3240
443k
    ret.name = NULL;
3241
443k
    ret.hashValue = 0;
3242
3243
    /*
3244
     * Handler for more complex cases
3245
     */
3246
443k
    startPosition = CUR_PTR - BASE_PTR;
3247
443k
    c = xmlCurrentChar(ctxt, &l);
3248
443k
    if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */
3249
432k
  (!xmlIsNameStartChar(c, old10) || (c == ':'))) {
3250
404k
  return(ret);
3251
404k
    }
3252
3253
91.4M
    while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */
3254
91.4M
     (xmlIsNameChar(c, old10) && (c != ':'))) {
3255
91.3M
        if (len <= INT_MAX - l)
3256
91.3M
      len += l;
3257
91.3M
  NEXTL(l);
3258
91.3M
  c = xmlCurrentChar(ctxt, &l);
3259
91.3M
    }
3260
39.1k
    if (len > maxLength) {
3261
173
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3262
173
        return(ret);
3263
173
    }
3264
38.9k
    ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len);
3265
38.9k
    if (ret.name == NULL)
3266
9
        xmlErrMemory(ctxt);
3267
38.9k
    return(ret);
3268
39.1k
}
3269
3270
/**
3271
 * Parse an XML name.
3272
 *
3273
 *     [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' |
3274
 *                          CombiningChar | Extender
3275
 *
3276
 *     [5NS] NCName ::= (Letter | '_') (NCNameChar)*
3277
 *
3278
 * @param ctxt  an XML parser context
3279
 * @returns the Name parsed or NULL
3280
 */
3281
3282
static xmlHashedString
3283
1.43M
xmlParseNCName(xmlParserCtxtPtr ctxt) {
3284
1.43M
    const xmlChar *in, *e;
3285
1.43M
    xmlHashedString ret;
3286
1.43M
    size_t count = 0;
3287
1.43M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3288
959k
                       XML_MAX_TEXT_LENGTH :
3289
1.43M
                       XML_MAX_NAME_LENGTH;
3290
3291
1.43M
    ret.name = NULL;
3292
3293
    /*
3294
     * Accelerator for simple ASCII names
3295
     */
3296
1.43M
    in = ctxt->input->cur;
3297
1.43M
    e = ctxt->input->end;
3298
1.43M
    if ((((*in >= 0x61) && (*in <= 0x7A)) ||
3299
464k
   ((*in >= 0x41) && (*in <= 0x5A)) ||
3300
1.01M
   (*in == '_')) && (in < e)) {
3301
1.01M
  in++;
3302
9.99M
  while ((((*in >= 0x61) && (*in <= 0x7A)) ||
3303
1.15M
          ((*in >= 0x41) && (*in <= 0x5A)) ||
3304
1.06M
          ((*in >= 0x30) && (*in <= 0x39)) ||
3305
1.03M
          (*in == '_') || (*in == '-') ||
3306
8.97M
          (*in == '.')) && (in < e))
3307
8.97M
      in++;
3308
1.01M
  if (in >= e)
3309
4.39k
      goto complex;
3310
1.00M
  if ((*in > 0) && (*in < 0x80)) {
3311
991k
      count = in - ctxt->input->cur;
3312
991k
            if (count > maxLength) {
3313
76
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3314
76
                return(ret);
3315
76
            }
3316
991k
      ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count);
3317
991k
      ctxt->input->cur = in;
3318
991k
      ctxt->input->col += count;
3319
991k
      if (ret.name == NULL) {
3320
16
          xmlErrMemory(ctxt);
3321
16
      }
3322
991k
      return(ret);
3323
991k
  }
3324
1.00M
    }
3325
443k
complex:
3326
443k
    return(xmlParseNCNameComplex(ctxt));
3327
1.43M
}
3328
3329
/**
3330
 * Parse an XML name and compares for match
3331
 * (specialized for endtag parsing)
3332
 *
3333
 * @param ctxt  an XML parser context
3334
 * @param other  the name to compare with
3335
 * @returns NULL for an illegal name, (xmlChar*) 1 for success
3336
 * and the name for mismatch
3337
 */
3338
3339
static const xmlChar *
3340
51.3k
xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) {
3341
51.3k
    register const xmlChar *cmp = other;
3342
51.3k
    register const xmlChar *in;
3343
51.3k
    const xmlChar *ret;
3344
3345
51.3k
    GROW;
3346
3347
51.3k
    in = ctxt->input->cur;
3348
148k
    while (*in != 0 && *in == *cmp) {
3349
97.1k
  ++in;
3350
97.1k
  ++cmp;
3351
97.1k
    }
3352
51.3k
    if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
3353
  /* success */
3354
38.3k
  ctxt->input->col += in - ctxt->input->cur;
3355
38.3k
  ctxt->input->cur = in;
3356
38.3k
  return (const xmlChar*) 1;
3357
38.3k
    }
3358
    /* failure (or end of input buffer), check with full function */
3359
12.9k
    ret = xmlParseName (ctxt);
3360
    /* strings coming from the dictionary direct compare possible */
3361
12.9k
    if (ret == other) {
3362
1.16k
  return (const xmlChar*) 1;
3363
1.16k
    }
3364
11.7k
    return ret;
3365
12.9k
}
3366
3367
/**
3368
 * Parse an XML name.
3369
 *
3370
 * @param ctxt  an XML parser context
3371
 * @param str  a pointer to the string pointer (IN/OUT)
3372
 * @returns the Name parsed or NULL. The `str` pointer
3373
 * is updated to the current location in the string.
3374
 */
3375
3376
static xmlChar *
3377
2.24M
xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) {
3378
2.24M
    xmlChar *ret;
3379
2.24M
    const xmlChar *cur = *str;
3380
2.24M
    int flags = 0;
3381
2.24M
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3382
1.39M
                    XML_MAX_TEXT_LENGTH :
3383
2.24M
                    XML_MAX_NAME_LENGTH;
3384
3385
2.24M
    if (ctxt->options & XML_PARSE_OLD10)
3386
1.02M
        flags |= XML_SCAN_OLD10;
3387
3388
2.24M
    cur = xmlScanName(*str, maxLength, flags);
3389
2.24M
    if (cur == NULL) {
3390
199
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3391
199
        return(NULL);
3392
199
    }
3393
2.24M
    if (cur == *str)
3394
12.8k
        return(NULL);
3395
3396
2.23M
    ret = xmlStrndup(*str, cur - *str);
3397
2.23M
    if (ret == NULL)
3398
138
        xmlErrMemory(ctxt);
3399
2.23M
    *str = cur;
3400
2.23M
    return(ret);
3401
2.24M
}
3402
3403
/**
3404
 * Parse an XML Nmtoken.
3405
 *
3406
 * @deprecated Internal function, don't use.
3407
 *
3408
 *     [7] Nmtoken ::= (NameChar)+
3409
 *
3410
 *     [8] Nmtokens ::= Nmtoken (#x20 Nmtoken)*
3411
 *
3412
 * @param ctxt  an XML parser context
3413
 * @returns the Nmtoken parsed or NULL
3414
 */
3415
3416
xmlChar *
3417
57.6k
xmlParseNmtoken(xmlParserCtxt *ctxt) {
3418
57.6k
    xmlChar buf[XML_MAX_NAMELEN + 5];
3419
57.6k
    xmlChar *ret;
3420
57.6k
    int len = 0, l;
3421
57.6k
    int c;
3422
57.6k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3423
26.4k
                    XML_MAX_TEXT_LENGTH :
3424
57.6k
                    XML_MAX_NAME_LENGTH;
3425
57.6k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3426
3427
57.6k
    c = xmlCurrentChar(ctxt, &l);
3428
3429
238k
    while (xmlIsNameChar(c, old10)) {
3430
181k
  COPY_BUF(buf, len, c);
3431
181k
  NEXTL(l);
3432
181k
  c = xmlCurrentChar(ctxt, &l);
3433
181k
  if (len >= XML_MAX_NAMELEN) {
3434
      /*
3435
       * Okay someone managed to make a huge token, so he's ready to pay
3436
       * for the processing speed.
3437
       */
3438
1.27k
      xmlChar *buffer;
3439
1.27k
      int max = len * 2;
3440
3441
1.27k
      buffer = xmlMalloc(max);
3442
1.27k
      if (buffer == NULL) {
3443
8
          xmlErrMemory(ctxt);
3444
8
    return(NULL);
3445
8
      }
3446
1.27k
      memcpy(buffer, buf, len);
3447
12.5M
      while (xmlIsNameChar(c, old10)) {
3448
12.5M
    if (len + 10 > max) {
3449
8.40k
        xmlChar *tmp;
3450
8.40k
                    int newSize;
3451
3452
8.40k
                    newSize = xmlGrowCapacity(max, 1, 1, maxLength);
3453
8.40k
                    if (newSize < 0) {
3454
327
                        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3455
327
                        xmlFree(buffer);
3456
327
                        return(NULL);
3457
327
                    }
3458
8.07k
        tmp = xmlRealloc(buffer, newSize);
3459
8.07k
        if (tmp == NULL) {
3460
9
      xmlErrMemory(ctxt);
3461
9
      xmlFree(buffer);
3462
9
      return(NULL);
3463
9
        }
3464
8.06k
        buffer = tmp;
3465
8.06k
                    max = newSize;
3466
8.06k
    }
3467
12.5M
    COPY_BUF(buffer, len, c);
3468
12.5M
    NEXTL(l);
3469
12.5M
    c = xmlCurrentChar(ctxt, &l);
3470
12.5M
      }
3471
934
      buffer[len] = 0;
3472
934
      return(buffer);
3473
1.27k
  }
3474
181k
    }
3475
56.3k
    if (len == 0)
3476
5.63k
        return(NULL);
3477
50.7k
    if (len > maxLength) {
3478
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3479
0
        return(NULL);
3480
0
    }
3481
50.7k
    ret = xmlStrndup(buf, len);
3482
50.7k
    if (ret == NULL)
3483
31
        xmlErrMemory(ctxt);
3484
50.7k
    return(ret);
3485
50.7k
}
3486
3487
/**
3488
 * Validate an entity value and expand parameter entities.
3489
 *
3490
 * @param ctxt  parser context
3491
 * @param buf  string buffer
3492
 * @param str  entity value
3493
 * @param length  size of entity value
3494
 * @param depth  nesting depth
3495
 */
3496
static void
3497
xmlExpandPEsInEntityValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf,
3498
277k
                          const xmlChar *str, int length, int depth) {
3499
277k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3500
277k
    const xmlChar *end, *chunk;
3501
277k
    int c, l;
3502
3503
277k
    if (str == NULL)
3504
38.0k
        return;
3505
3506
239k
    depth += 1;
3507
239k
    if (depth > maxDepth) {
3508
4
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3509
4
                       "Maximum entity nesting depth exceeded");
3510
4
  return;
3511
4
    }
3512
3513
239k
    end = str + length;
3514
239k
    chunk = str;
3515
3516
1.71G
    while ((str < end) && (!PARSER_STOPPED(ctxt))) {
3517
1.71G
        c = *str;
3518
3519
1.71G
        if (c >= 0x80) {
3520
1.62G
            l = xmlUTF8MultibyteLen(ctxt, str,
3521
1.62G
                    "invalid character in entity value\n");
3522
1.62G
            if (l == 0) {
3523
71.8M
                if (chunk < str)
3524
84.9k
                    xmlSBufAddString(buf, chunk, str - chunk);
3525
71.8M
                xmlSBufAddReplChar(buf);
3526
71.8M
                str += 1;
3527
71.8M
                chunk = str;
3528
1.55G
            } else {
3529
1.55G
                str += l;
3530
1.55G
            }
3531
1.62G
        } else if (c == '&') {
3532
329k
            if (str[1] == '#') {
3533
261k
                if (chunk < str)
3534
251k
                    xmlSBufAddString(buf, chunk, str - chunk);
3535
3536
261k
                c = xmlParseStringCharRef(ctxt, &str);
3537
261k
                if (c == 0)
3538
4.15k
                    return;
3539
3540
256k
                xmlSBufAddChar(buf, c);
3541
3542
256k
                chunk = str;
3543
256k
            } else {
3544
68.2k
                xmlChar *name;
3545
3546
                /*
3547
                 * General entity references are checked for
3548
                 * syntactic validity.
3549
                 */
3550
68.2k
                str++;
3551
68.2k
                name = xmlParseStringName(ctxt, &str);
3552
3553
68.2k
                if ((name == NULL) || (*str++ != ';')) {
3554
4.44k
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR,
3555
4.44k
                            "EntityValue: '&' forbidden except for entities "
3556
4.44k
                            "references\n");
3557
4.44k
                    xmlFree(name);
3558
4.44k
                    return;
3559
4.44k
                }
3560
3561
63.7k
                xmlFree(name);
3562
63.7k
            }
3563
92.7M
        } else if (c == '%') {
3564
229k
            xmlEntityPtr ent;
3565
3566
229k
            if (chunk < str)
3567
30.3k
                xmlSBufAddString(buf, chunk, str - chunk);
3568
3569
229k
            ent = xmlParseStringPEReference(ctxt, &str);
3570
229k
            if (ent == NULL)
3571
17.4k
                return;
3572
3573
211k
            if (!PARSER_EXTERNAL(ctxt)) {
3574
287
                xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL);
3575
287
                return;
3576
287
            }
3577
3578
211k
            if (ent->content == NULL) {
3579
                /*
3580
                 * Note: external parsed entities will not be loaded,
3581
                 * it is not required for a non-validating parser to
3582
                 * complete external PEReferences coming from the
3583
                 * internal subset
3584
                 */
3585
38.4k
                if (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
3586
38.2k
                    ((ctxt->replaceEntities) ||
3587
35.0k
                     (ctxt->validate))) {
3588
35.0k
                    xmlLoadEntityContent(ctxt, ent);
3589
35.0k
                } else {
3590
3.34k
                    xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING,
3591
3.34k
                                  "not validating will not read content for "
3592
3.34k
                                  "PE entity %s\n", ent->name, NULL);
3593
3.34k
                }
3594
38.4k
            }
3595
3596
            /*
3597
             * TODO: Skip if ent->content is still NULL.
3598
             */
3599
3600
211k
            if (xmlParserEntityCheck(ctxt, ent->length))
3601
15
                return;
3602
3603
211k
            if (ent->flags & XML_ENT_EXPANDING) {
3604
255
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3605
255
                return;
3606
255
            }
3607
3608
211k
            ent->flags |= XML_ENT_EXPANDING;
3609
211k
            xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length,
3610
211k
                                      depth);
3611
211k
            ent->flags &= ~XML_ENT_EXPANDING;
3612
3613
211k
            chunk = str;
3614
92.4M
        } else {
3615
            /* Normal ASCII char */
3616
92.4M
            if (!IS_BYTE_CHAR(c)) {
3617
11.5M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3618
11.5M
                        "invalid character in entity value\n");
3619
11.5M
                if (chunk < str)
3620
17.8k
                    xmlSBufAddString(buf, chunk, str - chunk);
3621
11.5M
                xmlSBufAddReplChar(buf);
3622
11.5M
                str += 1;
3623
11.5M
                chunk = str;
3624
80.9M
            } else {
3625
80.9M
                str += 1;
3626
80.9M
            }
3627
92.4M
        }
3628
1.71G
    }
3629
3630
212k
    if (chunk < str)
3631
197k
        xmlSBufAddString(buf, chunk, str - chunk);
3632
212k
}
3633
3634
/**
3635
 * Parse a value for ENTITY declarations
3636
 *
3637
 * @deprecated Internal function, don't use.
3638
 *
3639
 *     [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' |
3640
 *                         "'" ([^%&'] | PEReference | Reference)* "'"
3641
 *
3642
 * @param ctxt  an XML parser context
3643
 * @param orig  if non-NULL store a copy of the original entity value
3644
 * @returns the EntityValue parsed with reference substituted or NULL
3645
 */
3646
xmlChar *
3647
67.9k
xmlParseEntityValue(xmlParserCtxt *ctxt, xmlChar **orig) {
3648
67.9k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3649
32.5k
                         XML_MAX_HUGE_LENGTH :
3650
67.9k
                         XML_MAX_TEXT_LENGTH;
3651
67.9k
    xmlSBuf buf;
3652
67.9k
    const xmlChar *start;
3653
67.9k
    int quote, length;
3654
3655
67.9k
    xmlSBufInit(&buf, maxLength);
3656
3657
67.9k
    GROW;
3658
3659
67.9k
    quote = CUR;
3660
67.9k
    if ((quote != '"') && (quote != '\'')) {
3661
0
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
3662
0
  return(NULL);
3663
0
    }
3664
67.9k
    CUR_PTR++;
3665
3666
67.9k
    length = 0;
3667
3668
    /*
3669
     * Copy raw content of the entity into a buffer
3670
     */
3671
699M
    while (1) {
3672
699M
        int c;
3673
3674
699M
        if (PARSER_STOPPED(ctxt))
3675
23
            goto error;
3676
3677
699M
        if (CUR_PTR >= ctxt->input->end) {
3678
1.97k
            xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL);
3679
1.97k
            goto error;
3680
1.97k
        }
3681
3682
699M
        c = CUR;
3683
3684
699M
        if (c == 0) {
3685
225
            xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3686
225
                    "invalid character in entity value\n");
3687
225
            goto error;
3688
225
        }
3689
699M
        if (c == quote)
3690
65.7k
            break;
3691
699M
        NEXTL(1);
3692
699M
        length += 1;
3693
3694
        /*
3695
         * TODO: Check growth threshold
3696
         */
3697
699M
        if (ctxt->input->end - CUR_PTR < 10)
3698
117k
            GROW;
3699
699M
    }
3700
3701
65.7k
    start = CUR_PTR - length;
3702
3703
65.7k
    if (orig != NULL) {
3704
65.7k
        *orig = xmlStrndup(start, length);
3705
65.7k
        if (*orig == NULL)
3706
96
            xmlErrMemory(ctxt);
3707
65.7k
    }
3708
3709
65.7k
    xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr);
3710
3711
65.7k
    NEXTL(1);
3712
3713
65.7k
    return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long"));
3714
3715
2.22k
error:
3716
2.22k
    xmlSBufCleanup(&buf, ctxt, "entity length too long");
3717
2.22k
    return(NULL);
3718
67.9k
}
3719
3720
/**
3721
 * Check an entity reference in an attribute value for validity
3722
 * without expanding it.
3723
 *
3724
 * @param ctxt  parser context
3725
 * @param pent  entity
3726
 * @param depth  nesting depth
3727
 */
3728
static void
3729
9.46k
xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) {
3730
9.46k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3731
9.46k
    const xmlChar *str;
3732
9.46k
    unsigned long expandedSize = pent->length;
3733
9.46k
    int c, flags;
3734
3735
9.46k
    depth += 1;
3736
9.46k
    if (depth > maxDepth) {
3737
6
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3738
6
                       "Maximum entity nesting depth exceeded");
3739
6
  return;
3740
6
    }
3741
3742
9.45k
    if (pent->flags & XML_ENT_EXPANDING) {
3743
33
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3744
33
        return;
3745
33
    }
3746
3747
    /*
3748
     * If we're parsing a default attribute value in DTD content,
3749
     * the entity might reference other entities which weren't
3750
     * defined yet, so the check isn't reliable.
3751
     */
3752
9.42k
    if (ctxt->inSubset == 0)
3753
9.24k
        flags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
3754
177
    else
3755
177
        flags = XML_ENT_VALIDATED;
3756
3757
9.42k
    str = pent->content;
3758
9.42k
    if (str == NULL)
3759
93
        goto done;
3760
3761
    /*
3762
     * Note that entity values are already validated. We only check
3763
     * for illegal less-than signs and compute the expanded size
3764
     * of the entity. No special handling for multi-byte characters
3765
     * is needed.
3766
     */
3767
84.9M
    while (!PARSER_STOPPED(ctxt)) {
3768
84.9M
        c = *str;
3769
3770
84.9M
  if (c != '&') {
3771
84.9M
            if (c == 0)
3772
8.98k
                break;
3773
3774
84.9M
            if (c == '<')
3775
3.28k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3776
3.28k
                        "'<' in entity '%s' is not allowed in attributes "
3777
3.28k
                        "values\n", pent->name);
3778
3779
84.9M
            str += 1;
3780
84.9M
        } else if (str[1] == '#') {
3781
1.27k
            int val;
3782
3783
1.27k
      val = xmlParseStringCharRef(ctxt, &str);
3784
1.27k
      if (val == 0) {
3785
27
                pent->content[0] = 0;
3786
27
                break;
3787
27
            }
3788
17.3k
  } else {
3789
17.3k
            xmlChar *name;
3790
17.3k
            xmlEntityPtr ent;
3791
3792
17.3k
      name = xmlParseStringEntityRef(ctxt, &str);
3793
17.3k
      if (name == NULL) {
3794
55
                pent->content[0] = 0;
3795
55
                break;
3796
55
            }
3797
3798
17.3k
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3799
17.3k
            xmlFree(name);
3800
3801
17.3k
            if ((ent != NULL) &&
3802
15.9k
                (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) {
3803
15.3k
                if ((ent->flags & flags) != flags) {
3804
7.36k
                    pent->flags |= XML_ENT_EXPANDING;
3805
7.36k
                    xmlCheckEntityInAttValue(ctxt, ent, depth);
3806
7.36k
                    pent->flags &= ~XML_ENT_EXPANDING;
3807
7.36k
                }
3808
3809
15.3k
                xmlSaturatedAdd(&expandedSize, ent->expandedSize);
3810
15.3k
                xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST);
3811
15.3k
            }
3812
17.3k
        }
3813
84.9M
    }
3814
3815
9.42k
done:
3816
9.42k
    if (ctxt->inSubset == 0)
3817
9.24k
        pent->expandedSize = expandedSize;
3818
3819
9.42k
    pent->flags |= flags;
3820
9.42k
}
3821
3822
/**
3823
 * Expand general entity references in an entity or attribute value.
3824
 * Perform attribute value normalization.
3825
 *
3826
 * @param ctxt  parser context
3827
 * @param buf  string buffer
3828
 * @param str  entity or attribute value
3829
 * @param pent  entity for entity value, NULL for attribute values
3830
 * @param normalize  whether to collapse whitespace
3831
 * @param inSpace  whitespace state
3832
 * @param depth  nesting depth
3833
 * @param check  whether to check for amplification
3834
 * @returns  whether there was a normalization change
3835
 */
3836
static int
3837
xmlExpandEntityInAttValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf,
3838
                          const xmlChar *str, xmlEntityPtr pent, int normalize,
3839
1.01M
                          int *inSpace, int depth, int check) {
3840
1.01M
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3841
1.01M
    int c, chunkSize;
3842
1.01M
    int normChange = 0;
3843
3844
1.01M
    if (str == NULL)
3845
202
        return(0);
3846
3847
1.01M
    depth += 1;
3848
1.01M
    if (depth > maxDepth) {
3849
6
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3850
6
                       "Maximum entity nesting depth exceeded");
3851
6
  return(0);
3852
6
    }
3853
3854
1.01M
    if (pent != NULL) {
3855
991k
        if (pent->flags & XML_ENT_EXPANDING) {
3856
14
            xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3857
14
            return(0);
3858
14
        }
3859
3860
991k
        if (check) {
3861
990k
            if (xmlParserEntityCheck(ctxt, pent->length))
3862
543
                return(0);
3863
990k
        }
3864
991k
    }
3865
3866
1.01M
    chunkSize = 0;
3867
3868
    /*
3869
     * Note that entity values are already validated. No special
3870
     * handling for multi-byte characters is needed.
3871
     */
3872
6.33G
    while (!PARSER_STOPPED(ctxt)) {
3873
6.33G
        c = *str;
3874
3875
6.33G
  if (c != '&') {
3876
6.33G
            if (c == 0)
3877
913k
                break;
3878
3879
            /*
3880
             * If this function is called without an entity, it is used to
3881
             * expand entities in an attribute content where less-than was
3882
             * already unscaped and is allowed.
3883
             */
3884
6.32G
            if ((pent != NULL) && (c == '<')) {
3885
104k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3886
104k
                        "'<' in entity '%s' is not allowed in attributes "
3887
104k
                        "values\n", pent->name);
3888
104k
                break;
3889
104k
            }
3890
3891
6.32G
            if (c <= 0x20) {
3892
88.3M
                if ((normalize) && (*inSpace)) {
3893
                    /* Skip char */
3894
551k
                    if (chunkSize > 0) {
3895
101k
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3896
101k
                        chunkSize = 0;
3897
101k
                    }
3898
551k
                    normChange = 1;
3899
87.7M
                } else if (c < 0x20) {
3900
85.4M
                    if (chunkSize > 0) {
3901
457k
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3902
457k
                        chunkSize = 0;
3903
457k
                    }
3904
3905
85.4M
                    xmlSBufAddCString(buf, " ", 1);
3906
85.4M
                } else {
3907
2.25M
                    chunkSize += 1;
3908
2.25M
                }
3909
3910
88.3M
                *inSpace = 1;
3911
6.24G
            } else {
3912
6.24G
                chunkSize += 1;
3913
6.24G
                *inSpace = 0;
3914
6.24G
            }
3915
3916
6.32G
            str += 1;
3917
6.32G
        } else if (str[1] == '#') {
3918
308k
            int val;
3919
3920
308k
            if (chunkSize > 0) {
3921
306k
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3922
306k
                chunkSize = 0;
3923
306k
            }
3924
3925
308k
      val = xmlParseStringCharRef(ctxt, &str);
3926
308k
      if (val == 0) {
3927
21
                if (pent != NULL)
3928
21
                    pent->content[0] = 0;
3929
21
                break;
3930
21
            }
3931
3932
308k
            if (val == ' ') {
3933
3.62k
                if ((normalize) && (*inSpace))
3934
201
                    normChange = 1;
3935
3.41k
                else
3936
3.41k
                    xmlSBufAddCString(buf, " ", 1);
3937
3.62k
                *inSpace = 1;
3938
305k
            } else {
3939
305k
                xmlSBufAddChar(buf, val);
3940
305k
                *inSpace = 0;
3941
305k
            }
3942
1.92M
  } else {
3943
1.92M
            xmlChar *name;
3944
1.92M
            xmlEntityPtr ent;
3945
3946
1.92M
            if (chunkSize > 0) {
3947
1.22M
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3948
1.22M
                chunkSize = 0;
3949
1.22M
            }
3950
3951
1.92M
      name = xmlParseStringEntityRef(ctxt, &str);
3952
1.92M
            if (name == NULL) {
3953
119
                if (pent != NULL)
3954
109
                    pent->content[0] = 0;
3955
119
                break;
3956
119
            }
3957
3958
1.92M
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3959
1.92M
            xmlFree(name);
3960
3961
1.92M
      if ((ent != NULL) &&
3962
1.73M
    (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
3963
917k
    if (ent->content == NULL) {
3964
0
        xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
3965
0
          "predefined entity has no content\n");
3966
0
                    break;
3967
0
                }
3968
3969
917k
                xmlSBufAddString(buf, ent->content, ent->length);
3970
3971
917k
                *inSpace = 0;
3972
1.01M
      } else if ((ent != NULL) && (ent->content != NULL)) {
3973
817k
                if (pent != NULL)
3974
816k
                    pent->flags |= XML_ENT_EXPANDING;
3975
817k
    normChange |= xmlExpandEntityInAttValue(ctxt, buf,
3976
817k
                        ent->content, ent, normalize, inSpace, depth, check);
3977
817k
                if (pent != NULL)
3978
816k
                    pent->flags &= ~XML_ENT_EXPANDING;
3979
817k
      }
3980
1.92M
        }
3981
6.33G
    }
3982
3983
1.01M
    if (chunkSize > 0)
3984
621k
        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3985
3986
1.01M
    return(normChange);
3987
1.01M
}
3988
3989
/**
3990
 * Expand general entity references in an entity or attribute value.
3991
 * Perform attribute value normalization.
3992
 *
3993
 * @param ctxt  parser context
3994
 * @param str  entity or attribute value
3995
 * @param normalize  whether to collapse whitespace
3996
 * @returns the expanded attribtue value.
3997
 */
3998
xmlChar *
3999
xmlExpandEntitiesInAttValue(xmlParserCtxt *ctxt, const xmlChar *str,
4000
27.7k
                            int normalize) {
4001
27.7k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4002
12.6k
                         XML_MAX_HUGE_LENGTH :
4003
27.7k
                         XML_MAX_TEXT_LENGTH;
4004
27.7k
    xmlSBuf buf;
4005
27.7k
    int inSpace = 1;
4006
4007
27.7k
    xmlSBufInit(&buf, maxLength);
4008
4009
27.7k
    xmlExpandEntityInAttValue(ctxt, &buf, str, NULL, normalize, &inSpace,
4010
27.7k
                              ctxt->inputNr, /* check */ 0);
4011
4012
27.7k
    if ((normalize) && (inSpace) && (buf.size > 0))
4013
0
        buf.size--;
4014
4015
27.7k
    return(xmlSBufFinish(&buf, NULL, ctxt, "AttValue length too long"));
4016
27.7k
}
4017
4018
/**
4019
 * Parse a value for an attribute.
4020
 *
4021
 * NOTE: if no normalization is needed, the routine will return pointers
4022
 * directly from the data buffer.
4023
 *
4024
 * 3.3.3 Attribute-Value Normalization:
4025
 *
4026
 * Before the value of an attribute is passed to the application or
4027
 * checked for validity, the XML processor must normalize it as follows:
4028
 *
4029
 * - a character reference is processed by appending the referenced
4030
 *   character to the attribute value
4031
 * - an entity reference is processed by recursively processing the
4032
 *   replacement text of the entity
4033
 * - a whitespace character (\#x20, \#xD, \#xA, \#x9) is processed by
4034
 *   appending \#x20 to the normalized value, except that only a single
4035
 *   \#x20 is appended for a "#xD#xA" sequence that is part of an external
4036
 *   parsed entity or the literal entity value of an internal parsed entity
4037
 * - other characters are processed by appending them to the normalized value
4038
 *
4039
 * If the declared value is not CDATA, then the XML processor must further
4040
 * process the normalized attribute value by discarding any leading and
4041
 * trailing space (\#x20) characters, and by replacing sequences of space
4042
 * (\#x20) characters by a single space (\#x20) character.
4043
 * All attributes for which no declaration has been read should be treated
4044
 * by a non-validating parser as if declared CDATA.
4045
 *
4046
 * @param ctxt  an XML parser context
4047
 * @param attlen  attribute len result
4048
 * @param outFlags  resulting XML_ATTVAL_* flags
4049
 * @param special  value from attsSpecial
4050
 * @param isNamespace  whether this is a namespace declaration
4051
 * @returns the AttValue parsed or NULL. The value has to be freed by the
4052
 *     caller if it was copied, this can be detected by val[*len] == 0.
4053
 */
4054
static xmlChar *
4055
xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *attlen, int *outFlags,
4056
309k
                         int special, int isNamespace) {
4057
309k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4058
124k
                         XML_MAX_HUGE_LENGTH :
4059
309k
                         XML_MAX_TEXT_LENGTH;
4060
309k
    xmlSBuf buf;
4061
309k
    xmlChar *ret;
4062
309k
    int c, l, quote, entFlags, chunkSize;
4063
309k
    int inSpace = 1;
4064
309k
    int replaceEntities;
4065
309k
    int normalize = (special & XML_SPECIAL_TYPE_MASK) > XML_ATTRIBUTE_CDATA;
4066
309k
    int attvalFlags = 0;
4067
4068
    /* Always expand namespace URIs */
4069
309k
    replaceEntities = (ctxt->replaceEntities) || (isNamespace);
4070
4071
309k
    xmlSBufInit(&buf, maxLength);
4072
4073
309k
    GROW;
4074
4075
309k
    quote = CUR;
4076
309k
    if ((quote != '"') && (quote != '\'')) {
4077
8.18k
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
4078
8.18k
  return(NULL);
4079
8.18k
    }
4080
300k
    NEXTL(1);
4081
4082
300k
    if (ctxt->inSubset == 0)
4083
223k
        entFlags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
4084
77.6k
    else
4085
77.6k
        entFlags = XML_ENT_VALIDATED;
4086
4087
300k
    inSpace = 1;
4088
300k
    chunkSize = 0;
4089
4090
559M
    while (1) {
4091
559M
        if (PARSER_STOPPED(ctxt))
4092
957
            goto error;
4093
4094
559M
        if (CUR_PTR >= ctxt->input->end) {
4095
10.7k
            xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED,
4096
10.7k
                           "AttValue: ' expected\n");
4097
10.7k
            goto error;
4098
10.7k
        }
4099
4100
        /*
4101
         * TODO: Check growth threshold
4102
         */
4103
559M
        if (ctxt->input->end - CUR_PTR < 10)
4104
235k
            GROW;
4105
4106
559M
        c = CUR;
4107
4108
559M
        if (c >= 0x80) {
4109
453M
            l = xmlUTF8MultibyteLen(ctxt, CUR_PTR,
4110
453M
                    "invalid character in attribute value\n");
4111
453M
            if (l == 0) {
4112
124M
                if (chunkSize > 0) {
4113
153k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4114
153k
                    chunkSize = 0;
4115
153k
                }
4116
124M
                xmlSBufAddReplChar(&buf);
4117
124M
                NEXTL(1);
4118
329M
            } else {
4119
329M
                chunkSize += l;
4120
329M
                NEXTL(l);
4121
329M
            }
4122
4123
453M
            inSpace = 0;
4124
453M
        } else if (c != '&') {
4125
105M
            if (c > 0x20) {
4126
6.98M
                if (c == quote)
4127
287k
                    break;
4128
4129
6.69M
                if (c == '<')
4130
217k
                    xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL);
4131
4132
6.69M
                chunkSize += 1;
4133
6.69M
                inSpace = 0;
4134
98.3M
            } else if (!IS_BYTE_CHAR(c)) {
4135
76.6M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
4136
76.6M
                        "invalid character in attribute value\n");
4137
76.6M
                if (chunkSize > 0) {
4138
97.7k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4139
97.7k
                    chunkSize = 0;
4140
97.7k
                }
4141
76.6M
                xmlSBufAddReplChar(&buf);
4142
76.6M
                inSpace = 0;
4143
76.6M
            } else {
4144
                /* Whitespace */
4145
21.6M
                if ((normalize) && (inSpace)) {
4146
                    /* Skip char */
4147
650k
                    if (chunkSize > 0) {
4148
2.75k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4149
2.75k
                        chunkSize = 0;
4150
2.75k
                    }
4151
650k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4152
21.0M
                } else if (c < 0x20) {
4153
                    /* Convert to space */
4154
20.6M
                    if (chunkSize > 0) {
4155
87.5k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4156
87.5k
                        chunkSize = 0;
4157
87.5k
                    }
4158
4159
20.6M
                    xmlSBufAddCString(&buf, " ", 1);
4160
20.6M
                } else {
4161
367k
                    chunkSize += 1;
4162
367k
                }
4163
4164
21.6M
                inSpace = 1;
4165
4166
21.6M
                if ((c == 0xD) && (NXT(1) == 0xA))
4167
4.03k
                    CUR_PTR++;
4168
21.6M
            }
4169
4170
105M
            NEXTL(1);
4171
105M
        } else if (NXT(1) == '#') {
4172
50.5k
            int val;
4173
4174
50.5k
            if (chunkSize > 0) {
4175
31.6k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4176
31.6k
                chunkSize = 0;
4177
31.6k
            }
4178
4179
50.5k
            val = xmlParseCharRef(ctxt);
4180
50.5k
            if (val == 0)
4181
1.45k
                goto error;
4182
4183
49.1k
            if ((val == '&') && (!replaceEntities)) {
4184
                /*
4185
                 * The reparsing will be done in xmlNodeParseContent()
4186
                 * called from SAX2.c
4187
                 */
4188
3.51k
                xmlSBufAddCString(&buf, "&#38;", 5);
4189
3.51k
                inSpace = 0;
4190
45.6k
            } else if (val == ' ') {
4191
8.17k
                if ((normalize) && (inSpace))
4192
623
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4193
7.55k
                else
4194
7.55k
                    xmlSBufAddCString(&buf, " ", 1);
4195
8.17k
                inSpace = 1;
4196
37.4k
            } else {
4197
37.4k
                xmlSBufAddChar(&buf, val);
4198
37.4k
                inSpace = 0;
4199
37.4k
            }
4200
393k
        } else {
4201
393k
            const xmlChar *name;
4202
393k
            xmlEntityPtr ent;
4203
4204
393k
            if (chunkSize > 0) {
4205
96.0k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4206
96.0k
                chunkSize = 0;
4207
96.0k
            }
4208
4209
393k
            name = xmlParseEntityRefInternal(ctxt);
4210
393k
            if (name == NULL) {
4211
                /*
4212
                 * Probably a literal '&' which wasn't escaped.
4213
                 * TODO: Handle gracefully in recovery mode.
4214
                 */
4215
62.8k
                continue;
4216
62.8k
            }
4217
4218
330k
            ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1);
4219
330k
            if (ent == NULL)
4220
22.6k
                continue;
4221
4222
307k
            if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) {
4223
26.5k
                if ((ent->content[0] == '&') && (!replaceEntities))
4224
3.42k
                    xmlSBufAddCString(&buf, "&#38;", 5);
4225
23.1k
                else
4226
23.1k
                    xmlSBufAddString(&buf, ent->content, ent->length);
4227
26.5k
                inSpace = 0;
4228
281k
            } else if (replaceEntities) {
4229
174k
                if (xmlExpandEntityInAttValue(ctxt, &buf,
4230
174k
                        ent->content, ent, normalize, &inSpace, ctxt->inputNr,
4231
174k
                        /* check */ 1) > 0)
4232
48.8k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4233
174k
            } else {
4234
107k
                if ((ent->flags & entFlags) != entFlags)
4235
2.09k
                    xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr);
4236
4237
107k
                if (xmlParserEntityCheck(ctxt, ent->expandedSize)) {
4238
278
                    ent->content[0] = 0;
4239
278
                    goto error;
4240
278
                }
4241
4242
                /*
4243
                 * Just output the reference
4244
                 */
4245
106k
                xmlSBufAddCString(&buf, "&", 1);
4246
106k
                xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name));
4247
106k
                xmlSBufAddCString(&buf, ";", 1);
4248
4249
106k
                inSpace = 0;
4250
106k
            }
4251
307k
  }
4252
559M
    }
4253
4254
287k
    if ((buf.mem == NULL) && (outFlags != NULL)) {
4255
128k
        ret = (xmlChar *) CUR_PTR - chunkSize;
4256
4257
128k
        if (attlen != NULL)
4258
128k
            *attlen = chunkSize;
4259
128k
        if ((normalize) && (inSpace) && (chunkSize > 0)) {
4260
214
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4261
214
            *attlen -= 1;
4262
214
        }
4263
4264
        /* Report potential error */
4265
128k
        xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4266
158k
    } else {
4267
158k
        if (chunkSize > 0)
4268
116k
            xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4269
4270
158k
        if ((normalize) && (inSpace) && (buf.size > 0)) {
4271
1.03k
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4272
1.03k
            buf.size--;
4273
1.03k
        }
4274
4275
158k
        ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long");
4276
158k
        attvalFlags |= XML_ATTVAL_ALLOC;
4277
4278
158k
        if (ret != NULL) {
4279
158k
            if (attlen != NULL)
4280
29.4k
                *attlen = buf.size;
4281
158k
        }
4282
158k
    }
4283
4284
287k
    if (outFlags != NULL)
4285
158k
        *outFlags = attvalFlags;
4286
4287
287k
    NEXTL(1);
4288
4289
287k
    return(ret);
4290
4291
13.4k
error:
4292
13.4k
    xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4293
13.4k
    return(NULL);
4294
300k
}
4295
4296
/**
4297
 * Parse a value for an attribute
4298
 * Note: the parser won't do substitution of entities here, this
4299
 * will be handled later in #xmlStringGetNodeList
4300
 *
4301
 * @deprecated Internal function, don't use.
4302
 *
4303
 *     [10] AttValue ::= '"' ([^<&"] | Reference)* '"' |
4304
 *                       "'" ([^<&'] | Reference)* "'"
4305
 *
4306
 * 3.3.3 Attribute-Value Normalization:
4307
 *
4308
 * Before the value of an attribute is passed to the application or
4309
 * checked for validity, the XML processor must normalize it as follows:
4310
 *
4311
 * - a character reference is processed by appending the referenced
4312
 *   character to the attribute value
4313
 * - an entity reference is processed by recursively processing the
4314
 *   replacement text of the entity
4315
 * - a whitespace character (\#x20, \#xD, \#xA, \#x9) is processed by
4316
 *   appending \#x20 to the normalized value, except that only a single
4317
 *   \#x20 is appended for a "#xD#xA" sequence that is part of an external
4318
 *   parsed entity or the literal entity value of an internal parsed entity
4319
 * - other characters are processed by appending them to the normalized value
4320
 *
4321
 * If the declared value is not CDATA, then the XML processor must further
4322
 * process the normalized attribute value by discarding any leading and
4323
 * trailing space (\#x20) characters, and by replacing sequences of space
4324
 * (\#x20) characters by a single space (\#x20) character.
4325
 * All attributes for which no declaration has been read should be treated
4326
 * by a non-validating parser as if declared CDATA.
4327
 *
4328
 * @param ctxt  an XML parser context
4329
 * @returns the AttValue parsed or NULL. The value has to be freed by the
4330
 * caller.
4331
 */
4332
xmlChar *
4333
144k
xmlParseAttValue(xmlParserCtxt *ctxt) {
4334
144k
    if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL);
4335
144k
    return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0));
4336
144k
}
4337
4338
/**
4339
 * Parse an XML Literal
4340
 *
4341
 * @deprecated Internal function, don't use.
4342
 *
4343
 *     [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'")
4344
 *
4345
 * @param ctxt  an XML parser context
4346
 * @returns the SystemLiteral parsed or NULL
4347
 */
4348
4349
xmlChar *
4350
33.5k
xmlParseSystemLiteral(xmlParserCtxt *ctxt) {
4351
33.5k
    xmlChar *buf = NULL;
4352
33.5k
    int len = 0;
4353
33.5k
    int size = XML_PARSER_BUFFER_SIZE;
4354
33.5k
    int cur, l;
4355
33.5k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4356
14.0k
                    XML_MAX_TEXT_LENGTH :
4357
33.5k
                    XML_MAX_NAME_LENGTH;
4358
33.5k
    xmlChar stop;
4359
4360
33.5k
    if (RAW == '"') {
4361
20.0k
        NEXT;
4362
20.0k
  stop = '"';
4363
20.0k
    } else if (RAW == '\'') {
4364
11.8k
        NEXT;
4365
11.8k
  stop = '\'';
4366
11.8k
    } else {
4367
1.62k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4368
1.62k
  return(NULL);
4369
1.62k
    }
4370
4371
31.9k
    buf = xmlMalloc(size);
4372
31.9k
    if (buf == NULL) {
4373
42
        xmlErrMemory(ctxt);
4374
42
  return(NULL);
4375
42
    }
4376
31.8k
    cur = xmlCurrentCharRecover(ctxt, &l);
4377
41.6M
    while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */
4378
41.5M
  if (len + 5 >= size) {
4379
6.92k
      xmlChar *tmp;
4380
6.92k
            int newSize;
4381
4382
6.92k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4383
6.92k
            if (newSize < 0) {
4384
6
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral");
4385
6
                xmlFree(buf);
4386
6
                return(NULL);
4387
6
            }
4388
6.91k
      tmp = xmlRealloc(buf, newSize);
4389
6.91k
      if (tmp == NULL) {
4390
9
          xmlFree(buf);
4391
9
    xmlErrMemory(ctxt);
4392
9
    return(NULL);
4393
9
      }
4394
6.90k
      buf = tmp;
4395
6.90k
            size = newSize;
4396
6.90k
  }
4397
41.5M
  COPY_BUF(buf, len, cur);
4398
41.5M
  NEXTL(l);
4399
41.5M
  cur = xmlCurrentCharRecover(ctxt, &l);
4400
41.5M
    }
4401
31.8k
    buf[len] = 0;
4402
31.8k
    if (!IS_CHAR(cur)) {
4403
3.73k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4404
28.1k
    } else {
4405
28.1k
  NEXT;
4406
28.1k
    }
4407
31.8k
    return(buf);
4408
31.8k
}
4409
4410
/**
4411
 * Parse an XML public literal
4412
 *
4413
 * @deprecated Internal function, don't use.
4414
 *
4415
 *     [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'"
4416
 *
4417
 * @param ctxt  an XML parser context
4418
 * @returns the PubidLiteral parsed or NULL.
4419
 */
4420
4421
xmlChar *
4422
14.5k
xmlParsePubidLiteral(xmlParserCtxt *ctxt) {
4423
14.5k
    xmlChar *buf = NULL;
4424
14.5k
    int len = 0;
4425
14.5k
    int size = XML_PARSER_BUFFER_SIZE;
4426
14.5k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4427
5.14k
                    XML_MAX_TEXT_LENGTH :
4428
14.5k
                    XML_MAX_NAME_LENGTH;
4429
14.5k
    xmlChar cur;
4430
14.5k
    xmlChar stop;
4431
4432
14.5k
    if (RAW == '"') {
4433
1.10k
        NEXT;
4434
1.10k
  stop = '"';
4435
13.4k
    } else if (RAW == '\'') {
4436
12.3k
        NEXT;
4437
12.3k
  stop = '\'';
4438
12.3k
    } else {
4439
1.04k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4440
1.04k
  return(NULL);
4441
1.04k
    }
4442
13.4k
    buf = xmlMalloc(size);
4443
13.4k
    if (buf == NULL) {
4444
22
  xmlErrMemory(ctxt);
4445
22
  return(NULL);
4446
22
    }
4447
13.4k
    cur = CUR;
4448
1.28M
    while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) &&
4449
1.27M
           (PARSER_STOPPED(ctxt) == 0)) { /* checked */
4450
1.27M
  if (len + 1 >= size) {
4451
759
      xmlChar *tmp;
4452
759
            int newSize;
4453
4454
759
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4455
759
            if (newSize < 0) {
4456
6
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID");
4457
6
                xmlFree(buf);
4458
6
                return(NULL);
4459
6
            }
4460
753
      tmp = xmlRealloc(buf, newSize);
4461
753
      if (tmp == NULL) {
4462
6
    xmlErrMemory(ctxt);
4463
6
    xmlFree(buf);
4464
6
    return(NULL);
4465
6
      }
4466
747
      buf = tmp;
4467
747
            size = newSize;
4468
747
  }
4469
1.27M
  buf[len++] = cur;
4470
1.27M
  NEXT;
4471
1.27M
  cur = CUR;
4472
1.27M
    }
4473
13.4k
    buf[len] = 0;
4474
13.4k
    if (cur != stop) {
4475
3.79k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4476
9.66k
    } else {
4477
9.66k
  NEXTL(1);
4478
9.66k
    }
4479
13.4k
    return(buf);
4480
13.4k
}
4481
4482
static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial);
4483
4484
/*
4485
 * used for the test in the inner loop of the char data testing
4486
 */
4487
static const unsigned char test_char_data[256] = {
4488
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4489
    0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */
4490
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4491
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4492
    0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */
4493
    0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F,
4494
    0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37,
4495
    0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */
4496
    0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47,
4497
    0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F,
4498
    0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57,
4499
    0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */
4500
    0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67,
4501
    0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F,
4502
    0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77,
4503
    0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F,
4504
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */
4505
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4506
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4507
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4508
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4509
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4510
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4511
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4512
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4513
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4514
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4515
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4516
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4517
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4518
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4519
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
4520
};
4521
4522
static void
4523
xmlCharacters(xmlParserCtxtPtr ctxt, const xmlChar *buf, int size,
4524
2.03M
              int isBlank) {
4525
2.03M
    int checkBlanks;
4526
4527
2.03M
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
4528
278k
        return;
4529
4530
1.75M
    checkBlanks = (!ctxt->keepBlanks) ||
4531
1.12M
                  (ctxt->sax->ignorableWhitespace != ctxt->sax->characters);
4532
4533
    /*
4534
     * Calling areBlanks with only parts of a text node
4535
     * is fundamentally broken, making the NOBLANKS option
4536
     * essentially unusable.
4537
     */
4538
1.75M
    if ((checkBlanks) &&
4539
637k
        (areBlanks(ctxt, buf, size, isBlank))) {
4540
8.84k
        if ((ctxt->sax->ignorableWhitespace != NULL) &&
4541
8.84k
            (ctxt->keepBlanks))
4542
0
            ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size);
4543
1.75M
    } else {
4544
1.75M
        if (ctxt->sax->characters != NULL)
4545
1.75M
            ctxt->sax->characters(ctxt->userData, buf, size);
4546
4547
        /*
4548
         * The old code used to update this value for "complex" data
4549
         * even if checkBlanks was false. This was probably a bug.
4550
         */
4551
1.75M
        if ((checkBlanks) && (*ctxt->space == -1))
4552
47.6k
            *ctxt->space = -2;
4553
1.75M
    }
4554
1.75M
}
4555
4556
/**
4557
 * Parse character data. Always makes progress if the first char isn't
4558
 * '<' or '&'.
4559
 *
4560
 * The right angle bracket (>) may be represented using the string "&gt;",
4561
 * and must, for compatibility, be escaped using "&gt;" or a character
4562
 * reference when it appears in the string "]]>" in content, when that
4563
 * string is not marking the end of a CDATA section.
4564
 *
4565
 *     [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*)
4566
 * @param ctxt  an XML parser context
4567
 * @param partial  buffer may contain partial UTF-8 sequences
4568
 */
4569
static void
4570
16.8M
xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) {
4571
16.8M
    const xmlChar *in;
4572
16.8M
    int line = ctxt->input->line;
4573
16.8M
    int col = ctxt->input->col;
4574
16.8M
    int ccol;
4575
16.8M
    int terminate = 0;
4576
4577
16.8M
    GROW;
4578
    /*
4579
     * Accelerated common case where input don't need to be
4580
     * modified before passing it to the handler.
4581
     */
4582
16.8M
    in = ctxt->input->cur;
4583
16.9M
    do {
4584
16.9M
get_more_space:
4585
17.0M
        while (*in == 0x20) { in++; ctxt->input->col++; }
4586
16.9M
        if (*in == 0xA) {
4587
12.2M
            do {
4588
12.2M
                ctxt->input->line++; ctxt->input->col = 1;
4589
12.2M
                in++;
4590
12.2M
            } while (*in == 0xA);
4591
75.7k
            goto get_more_space;
4592
75.7k
        }
4593
16.9M
        if (*in == '<') {
4594
68.1k
            while (in > ctxt->input->cur) {
4595
34.0k
                const xmlChar *tmp = ctxt->input->cur;
4596
34.0k
                size_t nbchar = in - tmp;
4597
4598
34.0k
                if (nbchar > XML_MAX_ITEMS)
4599
0
                    nbchar = XML_MAX_ITEMS;
4600
34.0k
                ctxt->input->cur += nbchar;
4601
4602
34.0k
                xmlCharacters(ctxt, tmp, nbchar, 1);
4603
34.0k
            }
4604
34.0k
            return;
4605
34.0k
        }
4606
4607
17.3M
get_more:
4608
17.3M
        ccol = ctxt->input->col;
4609
32.1M
        while (test_char_data[*in]) {
4610
14.8M
            in++;
4611
14.8M
            ccol++;
4612
14.8M
        }
4613
17.3M
        ctxt->input->col = ccol;
4614
17.3M
        if (*in == 0xA) {
4615
5.74M
            do {
4616
5.74M
                ctxt->input->line++; ctxt->input->col = 1;
4617
5.74M
                in++;
4618
5.74M
            } while (*in == 0xA);
4619
63.5k
            goto get_more;
4620
63.5k
        }
4621
17.2M
        if (*in == ']') {
4622
407k
            size_t avail = ctxt->input->end - in;
4623
4624
407k
            if (partial && avail < 2) {
4625
66
                terminate = 1;
4626
66
                goto invoke_callback;
4627
66
            }
4628
407k
            if (in[1] == ']') {
4629
354k
                if (partial && avail < 3) {
4630
279
                    terminate = 1;
4631
279
                    goto invoke_callback;
4632
279
                }
4633
354k
                if (in[2] == '>')
4634
30.1k
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4635
354k
            }
4636
4637
407k
            in++;
4638
407k
            ctxt->input->col++;
4639
407k
            goto get_more;
4640
407k
        }
4641
4642
16.8M
invoke_callback:
4643
17.2M
        while (in > ctxt->input->cur) {
4644
386k
            const xmlChar *tmp = ctxt->input->cur;
4645
386k
            size_t nbchar = in - tmp;
4646
4647
386k
            if (nbchar > XML_MAX_ITEMS)
4648
0
                nbchar = XML_MAX_ITEMS;
4649
386k
            ctxt->input->cur += nbchar;
4650
4651
386k
            xmlCharacters(ctxt, tmp, nbchar, 0);
4652
4653
386k
            line = ctxt->input->line;
4654
386k
            col = ctxt->input->col;
4655
386k
        }
4656
16.8M
        ctxt->input->cur = in;
4657
16.8M
        if (*in == 0xD) {
4658
8.71k
            in++;
4659
8.71k
            if (*in == 0xA) {
4660
5.79k
                ctxt->input->cur = in;
4661
5.79k
                in++;
4662
5.79k
                ctxt->input->line++; ctxt->input->col = 1;
4663
5.79k
                continue; /* while */
4664
5.79k
            }
4665
2.92k
            in--;
4666
2.92k
        }
4667
16.8M
        if (*in == '<') {
4668
189k
            return;
4669
189k
        }
4670
16.6M
        if (*in == '&') {
4671
60.4k
            return;
4672
60.4k
        }
4673
16.6M
        if (terminate) {
4674
345
            return;
4675
345
        }
4676
16.6M
        SHRINK;
4677
16.6M
        GROW;
4678
16.6M
        in = ctxt->input->cur;
4679
16.6M
    } while (((*in >= 0x20) && (*in <= 0x7F)) ||
4680
16.6M
             (*in == 0x09) || (*in == 0x0a));
4681
16.6M
    ctxt->input->line = line;
4682
16.6M
    ctxt->input->col = col;
4683
16.6M
    xmlParseCharDataComplex(ctxt, partial);
4684
16.6M
}
4685
4686
/**
4687
 * Always makes progress if the first char isn't '<' or '&'.
4688
 *
4689
 * parse a CharData section.this is the fallback function
4690
 * of #xmlParseCharData when the parsing requires handling
4691
 * of non-ASCII characters.
4692
 *
4693
 * @param ctxt  an XML parser context
4694
 * @param partial  whether the input can end with truncated UTF-8
4695
 */
4696
static void
4697
16.6M
xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) {
4698
16.6M
    xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5];
4699
16.6M
    int nbchar = 0;
4700
16.6M
    int cur, l;
4701
4702
16.6M
    cur = xmlCurrentCharRecover(ctxt, &l);
4703
186M
    while ((cur != '<') && /* checked */
4704
186M
           (cur != '&') &&
4705
186M
     (IS_CHAR(cur))) {
4706
169M
        if (cur == ']') {
4707
685k
            size_t avail = ctxt->input->end - ctxt->input->cur;
4708
4709
685k
            if (partial && avail < 2)
4710
216
                break;
4711
685k
            if (NXT(1) == ']') {
4712
491k
                if (partial && avail < 3)
4713
430
                    break;
4714
491k
                if (NXT(2) == '>')
4715
156k
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4716
491k
            }
4717
685k
        }
4718
4719
169M
  COPY_BUF(buf, nbchar, cur);
4720
  /* move current position before possible calling of ctxt->sax->characters */
4721
169M
  NEXTL(l);
4722
169M
  if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) {
4723
1.44M
      buf[nbchar] = 0;
4724
4725
1.44M
            xmlCharacters(ctxt, buf, nbchar, 0);
4726
1.44M
      nbchar = 0;
4727
1.44M
            SHRINK;
4728
1.44M
  }
4729
169M
  cur = xmlCurrentCharRecover(ctxt, &l);
4730
169M
    }
4731
16.6M
    if (nbchar != 0) {
4732
173k
        buf[nbchar] = 0;
4733
4734
173k
        xmlCharacters(ctxt, buf, nbchar, 0);
4735
173k
    }
4736
    /*
4737
     * cur == 0 can mean
4738
     *
4739
     * - End of buffer.
4740
     * - An actual 0 character.
4741
     * - An incomplete UTF-8 sequence. This is allowed if partial is set.
4742
     */
4743
16.6M
    if (ctxt->input->cur < ctxt->input->end) {
4744
16.5M
        if ((cur == 0) && (CUR != 0)) {
4745
2.25k
            if (partial == 0) {
4746
1.56k
                xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4747
1.56k
                        "Incomplete UTF-8 sequence starting with %02X\n", CUR);
4748
1.56k
                NEXTL(1);
4749
1.56k
            }
4750
16.5M
        } else if ((cur != '<') && (cur != '&') && (cur != ']')) {
4751
            /* Generate the error and skip the offending character */
4752
16.4M
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4753
16.4M
                              "PCDATA invalid Char value %d\n", cur);
4754
16.4M
            NEXTL(l);
4755
16.4M
        }
4756
16.5M
    }
4757
16.6M
}
4758
4759
/**
4760
 * @deprecated Internal function, don't use.
4761
 * @param ctxt  an XML parser context
4762
 * @param cdata  unused
4763
 */
4764
void
4765
0
xmlParseCharData(xmlParserCtxt *ctxt, ATTRIBUTE_UNUSED int cdata) {
4766
0
    xmlParseCharDataInternal(ctxt, 0);
4767
0
}
4768
4769
/**
4770
 * Parse an External ID or a Public ID
4771
 *
4772
 * @deprecated Internal function, don't use.
4773
 *
4774
 * NOTE: Productions [75] and [83] interact badly since [75] can generate
4775
 * `'PUBLIC' S PubidLiteral S SystemLiteral`
4776
 *
4777
 *     [75] ExternalID ::= 'SYSTEM' S SystemLiteral
4778
 *                       | 'PUBLIC' S PubidLiteral S SystemLiteral
4779
 *
4780
 *     [83] PublicID ::= 'PUBLIC' S PubidLiteral
4781
 *
4782
 * @param ctxt  an XML parser context
4783
 * @param publicId  a xmlChar** receiving PubidLiteral
4784
 * @param strict  indicate whether we should restrict parsing to only
4785
 *          production [75], see NOTE below
4786
 * @returns the function returns SystemLiteral and in the second
4787
 *                case publicID receives PubidLiteral, is strict is off
4788
 *                it is possible to return NULL and have publicID set.
4789
 */
4790
4791
xmlChar *
4792
85.1k
xmlParseExternalID(xmlParserCtxt *ctxt, xmlChar **publicId, int strict) {
4793
85.1k
    xmlChar *URI = NULL;
4794
4795
85.1k
    *publicId = NULL;
4796
85.1k
    if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) {
4797
24.0k
        SKIP(6);
4798
24.0k
  if (SKIP_BLANKS == 0) {
4799
2.19k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4800
2.19k
                     "Space required after 'SYSTEM'\n");
4801
2.19k
  }
4802
24.0k
  URI = xmlParseSystemLiteral(ctxt);
4803
24.0k
  if (URI == NULL) {
4804
327
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4805
327
        }
4806
61.0k
    } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) {
4807
14.5k
        SKIP(6);
4808
14.5k
  if (SKIP_BLANKS == 0) {
4809
2.19k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4810
2.19k
        "Space required after 'PUBLIC'\n");
4811
2.19k
  }
4812
14.5k
  *publicId = xmlParsePubidLiteral(ctxt);
4813
14.5k
  if (*publicId == NULL) {
4814
1.08k
      xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL);
4815
1.08k
  }
4816
14.5k
  if (strict) {
4817
      /*
4818
       * We don't handle [83] so "S SystemLiteral" is required.
4819
       */
4820
9.15k
      if (SKIP_BLANKS == 0) {
4821
2.59k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4822
2.59k
      "Space required after the Public Identifier\n");
4823
2.59k
      }
4824
9.15k
  } else {
4825
      /*
4826
       * We handle [83] so we return immediately, if
4827
       * "S SystemLiteral" is not detected. We skip blanks if no
4828
             * system literal was found, but this is harmless since we must
4829
             * be at the end of a NotationDecl.
4830
       */
4831
5.39k
      if (SKIP_BLANKS == 0) return(NULL);
4832
544
      if ((CUR != '\'') && (CUR != '"')) return(NULL);
4833
544
  }
4834
9.47k
  URI = xmlParseSystemLiteral(ctxt);
4835
9.47k
  if (URI == NULL) {
4836
1.35k
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4837
1.35k
        }
4838
9.47k
    }
4839
80.0k
    return(URI);
4840
85.1k
}
4841
4842
/**
4843
 * Skip an XML (SGML) comment <!-- .... -->
4844
 *  The spec says that "For compatibility, the string "--" (double-hyphen)
4845
 *  must not occur within comments. "
4846
 * This is the slow routine in case the accelerator for ascii didn't work
4847
 *
4848
 *     [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->'
4849
 * @param ctxt  an XML parser context
4850
 * @param buf  the already parsed part of the buffer
4851
 * @param len  number of bytes in the buffer
4852
 * @param size  allocated size of the buffer
4853
 */
4854
static void
4855
xmlParseCommentComplex(xmlParserCtxtPtr ctxt, xmlChar *buf,
4856
74.7k
                       size_t len, size_t size) {
4857
74.7k
    int q, ql;
4858
74.7k
    int r, rl;
4859
74.7k
    int cur, l;
4860
74.7k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4861
15.1k
                    XML_MAX_HUGE_LENGTH :
4862
74.7k
                    XML_MAX_TEXT_LENGTH;
4863
4864
74.7k
    if (buf == NULL) {
4865
14.7k
        len = 0;
4866
14.7k
  size = XML_PARSER_BUFFER_SIZE;
4867
14.7k
  buf = xmlMalloc(size);
4868
14.7k
  if (buf == NULL) {
4869
93
      xmlErrMemory(ctxt);
4870
93
      return;
4871
93
  }
4872
14.7k
    }
4873
74.6k
    q = xmlCurrentCharRecover(ctxt, &ql);
4874
74.6k
    if (q == 0)
4875
7.74k
        goto not_terminated;
4876
66.9k
    if (!IS_CHAR(q)) {
4877
1.19k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4878
1.19k
                          "xmlParseComment: invalid xmlChar value %d\n",
4879
1.19k
                    q);
4880
1.19k
  xmlFree (buf);
4881
1.19k
  return;
4882
1.19k
    }
4883
65.7k
    NEXTL(ql);
4884
65.7k
    r = xmlCurrentCharRecover(ctxt, &rl);
4885
65.7k
    if (r == 0)
4886
445
        goto not_terminated;
4887
65.2k
    if (!IS_CHAR(r)) {
4888
1.02k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4889
1.02k
                          "xmlParseComment: invalid xmlChar value %d\n",
4890
1.02k
                    r);
4891
1.02k
  xmlFree (buf);
4892
1.02k
  return;
4893
1.02k
    }
4894
64.2k
    NEXTL(rl);
4895
64.2k
    cur = xmlCurrentCharRecover(ctxt, &l);
4896
64.2k
    if (cur == 0)
4897
1.68k
        goto not_terminated;
4898
18.7M
    while (IS_CHAR(cur) && /* checked */
4899
18.7M
           ((cur != '>') ||
4900
18.6M
      (r != '-') || (q != '-'))) {
4901
18.6M
  if ((r == '-') && (q == '-')) {
4902
337k
      xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL);
4903
337k
  }
4904
18.6M
  if (len + 5 >= size) {
4905
29.0k
      xmlChar *tmp;
4906
29.0k
            int newSize;
4907
4908
29.0k
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4909
29.0k
            if (newSize < 0) {
4910
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4911
0
                             "Comment too big found", NULL);
4912
0
                xmlFree (buf);
4913
0
                return;
4914
0
            }
4915
29.0k
      tmp = xmlRealloc(buf, newSize);
4916
29.0k
      if (tmp == NULL) {
4917
25
    xmlErrMemory(ctxt);
4918
25
    xmlFree(buf);
4919
25
    return;
4920
25
      }
4921
28.9k
      buf = tmp;
4922
28.9k
            size = newSize;
4923
28.9k
  }
4924
18.6M
  COPY_BUF(buf, len, q);
4925
4926
18.6M
  q = r;
4927
18.6M
  ql = rl;
4928
18.6M
  r = cur;
4929
18.6M
  rl = l;
4930
4931
18.6M
  NEXTL(l);
4932
18.6M
  cur = xmlCurrentCharRecover(ctxt, &l);
4933
4934
18.6M
    }
4935
62.5k
    buf[len] = 0;
4936
62.5k
    if (cur == 0) {
4937
9.94k
  xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4938
9.94k
                       "Comment not terminated \n<!--%.50s\n", buf);
4939
52.6k
    } else if (!IS_CHAR(cur)) {
4940
1.07k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4941
1.07k
                          "xmlParseComment: invalid xmlChar value %d\n",
4942
1.07k
                    cur);
4943
51.5k
    } else {
4944
51.5k
        NEXT;
4945
51.5k
  if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
4946
51.5k
      (!ctxt->disableSAX))
4947
51.2k
      ctxt->sax->comment(ctxt->userData, buf);
4948
51.5k
    }
4949
62.5k
    xmlFree(buf);
4950
62.5k
    return;
4951
9.86k
not_terminated:
4952
9.86k
    xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4953
9.86k
       "Comment not terminated\n", NULL);
4954
9.86k
    xmlFree(buf);
4955
9.86k
}
4956
4957
/**
4958
 * Parse an XML (SGML) comment. Always consumes '<!'.
4959
 *
4960
 * @deprecated Internal function, don't use.
4961
 *
4962
 *  The spec says that "For compatibility, the string "--" (double-hyphen)
4963
 *  must not occur within comments. "
4964
 *
4965
 *     [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->'
4966
 * @param ctxt  an XML parser context
4967
 */
4968
void
4969
590k
xmlParseComment(xmlParserCtxt *ctxt) {
4970
590k
    xmlChar *buf = NULL;
4971
590k
    size_t size = XML_PARSER_BUFFER_SIZE;
4972
590k
    size_t len = 0;
4973
590k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4974
36.5k
                       XML_MAX_HUGE_LENGTH :
4975
590k
                       XML_MAX_TEXT_LENGTH;
4976
590k
    const xmlChar *in;
4977
590k
    size_t nbchar = 0;
4978
590k
    int ccol;
4979
4980
    /*
4981
     * Check that there is a comment right here.
4982
     */
4983
590k
    if ((RAW != '<') || (NXT(1) != '!'))
4984
0
        return;
4985
590k
    SKIP(2);
4986
590k
    if ((RAW != '-') || (NXT(1) != '-'))
4987
64
        return;
4988
590k
    SKIP(2);
4989
590k
    GROW;
4990
4991
    /*
4992
     * Accelerated common case where input don't need to be
4993
     * modified before passing it to the handler.
4994
     */
4995
590k
    in = ctxt->input->cur;
4996
591k
    do {
4997
591k
  if (*in == 0xA) {
4998
357k
      do {
4999
357k
    ctxt->input->line++; ctxt->input->col = 1;
5000
357k
    in++;
5001
357k
      } while (*in == 0xA);
5002
2.59k
  }
5003
1.48M
get_more:
5004
1.48M
        ccol = ctxt->input->col;
5005
9.68M
  while (((*in > '-') && (*in <= 0x7F)) ||
5006
3.13M
         ((*in >= 0x20) && (*in < '-')) ||
5007
8.20M
         (*in == 0x09)) {
5008
8.20M
        in++;
5009
8.20M
        ccol++;
5010
8.20M
  }
5011
1.48M
  ctxt->input->col = ccol;
5012
1.48M
  if (*in == 0xA) {
5013
849k
      do {
5014
849k
    ctxt->input->line++; ctxt->input->col = 1;
5015
849k
    in++;
5016
849k
      } while (*in == 0xA);
5017
21.0k
      goto get_more;
5018
21.0k
  }
5019
1.46M
  nbchar = in - ctxt->input->cur;
5020
  /*
5021
   * save current set of data
5022
   */
5023
1.46M
  if (nbchar > 0) {
5024
947k
            if (nbchar > maxLength - len) {
5025
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
5026
0
                                  "Comment too big found", NULL);
5027
0
                xmlFree(buf);
5028
0
                return;
5029
0
            }
5030
947k
            if (buf == NULL) {
5031
155k
                if ((*in == '-') && (in[1] == '-'))
5032
78.2k
                    size = nbchar + 1;
5033
77.4k
                else
5034
77.4k
                    size = XML_PARSER_BUFFER_SIZE + nbchar;
5035
155k
                buf = xmlMalloc(size);
5036
155k
                if (buf == NULL) {
5037
41
                    xmlErrMemory(ctxt);
5038
41
                    return;
5039
41
                }
5040
155k
                len = 0;
5041
791k
            } else if (len + nbchar + 1 >= size) {
5042
72.2k
                xmlChar *new_buf;
5043
72.2k
                size += len + nbchar + XML_PARSER_BUFFER_SIZE;
5044
72.2k
                new_buf = xmlRealloc(buf, size);
5045
72.2k
                if (new_buf == NULL) {
5046
8
                    xmlErrMemory(ctxt);
5047
8
                    xmlFree(buf);
5048
8
                    return;
5049
8
                }
5050
72.2k
                buf = new_buf;
5051
72.2k
            }
5052
947k
            memcpy(&buf[len], ctxt->input->cur, nbchar);
5053
947k
            len += nbchar;
5054
947k
            buf[len] = 0;
5055
947k
  }
5056
1.46M
  ctxt->input->cur = in;
5057
1.46M
  if (*in == 0xA) {
5058
0
      in++;
5059
0
      ctxt->input->line++; ctxt->input->col = 1;
5060
0
  }
5061
1.46M
  if (*in == 0xD) {
5062
10.1k
      in++;
5063
10.1k
      if (*in == 0xA) {
5064
8.88k
    ctxt->input->cur = in;
5065
8.88k
    in++;
5066
8.88k
    ctxt->input->line++; ctxt->input->col = 1;
5067
8.88k
    goto get_more;
5068
8.88k
      }
5069
1.28k
      in--;
5070
1.28k
  }
5071
1.45M
  SHRINK;
5072
1.45M
  GROW;
5073
1.45M
  in = ctxt->input->cur;
5074
1.45M
  if (*in == '-') {
5075
1.37M
      if (in[1] == '-') {
5076
967k
          if (in[2] == '>') {
5077
515k
        SKIP(3);
5078
515k
        if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
5079
515k
            (!ctxt->disableSAX)) {
5080
510k
      if (buf != NULL)
5081
95.0k
          ctxt->sax->comment(ctxt->userData, buf);
5082
415k
      else
5083
415k
          ctxt->sax->comment(ctxt->userData, BAD_CAST "");
5084
510k
        }
5085
515k
        if (buf != NULL)
5086
95.5k
            xmlFree(buf);
5087
515k
        return;
5088
515k
    }
5089
452k
    if (buf != NULL) {
5090
416k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5091
416k
                          "Double hyphen within comment: "
5092
416k
                                      "<!--%.50s\n",
5093
416k
              buf);
5094
416k
    } else
5095
36.2k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5096
36.2k
                          "Double hyphen within comment\n", NULL);
5097
452k
    in++;
5098
452k
    ctxt->input->col++;
5099
452k
      }
5100
860k
      in++;
5101
860k
      ctxt->input->col++;
5102
860k
      goto get_more;
5103
1.37M
  }
5104
1.45M
    } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a));
5105
74.7k
    xmlParseCommentComplex(ctxt, buf, len, size);
5106
74.7k
}
5107
5108
5109
/**
5110
 * Parse the name of a PI
5111
 *
5112
 * @deprecated Internal function, don't use.
5113
 *
5114
 *     [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l'))
5115
 *
5116
 * @param ctxt  an XML parser context
5117
 * @returns the PITarget name or NULL
5118
 */
5119
5120
const xmlChar *
5121
62.6k
xmlParsePITarget(xmlParserCtxt *ctxt) {
5122
62.6k
    const xmlChar *name;
5123
5124
62.6k
    name = xmlParseName(ctxt);
5125
62.6k
    if ((name != NULL) &&
5126
53.5k
        ((name[0] == 'x') || (name[0] == 'X')) &&
5127
44.1k
        ((name[1] == 'm') || (name[1] == 'M')) &&
5128
39.7k
        ((name[2] == 'l') || (name[2] == 'L'))) {
5129
2.82k
  int i;
5130
2.82k
  if ((name[0] == 'x') && (name[1] == 'm') &&
5131
2.37k
      (name[2] == 'l') && (name[3] == 0)) {
5132
1.22k
      xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5133
1.22k
     "XML declaration allowed only at the start of the document\n");
5134
1.22k
      return(name);
5135
1.59k
  } else if (name[3] == 0) {
5136
599
      xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL);
5137
599
      return(name);
5138
599
  }
5139
2.73k
  for (i = 0;;i++) {
5140
2.73k
      if (xmlW3CPIs[i] == NULL) break;
5141
1.97k
      if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i]))
5142
231
          return(name);
5143
1.97k
  }
5144
768
  xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5145
768
          "xmlParsePITarget: invalid name prefix 'xml'\n",
5146
768
          NULL, NULL);
5147
768
    }
5148
60.5k
    if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) {
5149
768
  xmlNsErr(ctxt, XML_NS_ERR_COLON,
5150
768
     "colons are forbidden from PI names '%s'\n", name, NULL, NULL);
5151
768
    }
5152
60.5k
    return(name);
5153
62.6k
}
5154
5155
#ifdef LIBXML_CATALOG_ENABLED
5156
/**
5157
 * Parse an XML Catalog Processing Instruction.
5158
 *
5159
 * <?oasis-xml-catalog catalog="http://example.com/catalog.xml"?>
5160
 *
5161
 * Occurs only if allowed by the user and if happening in the Misc
5162
 * part of the document before any doctype information
5163
 * This will add the given catalog to the parsing context in order
5164
 * to be used if there is a resolution need further down in the document
5165
 *
5166
 * @param ctxt  an XML parser context
5167
 * @param catalog  the PI value string
5168
 */
5169
5170
static void
5171
0
xmlParseCatalogPI(xmlParserCtxtPtr ctxt, const xmlChar *catalog) {
5172
0
    xmlChar *URL = NULL;
5173
0
    const xmlChar *tmp, *base;
5174
0
    xmlChar marker;
5175
5176
0
    tmp = catalog;
5177
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5178
0
    if (xmlStrncmp(tmp, BAD_CAST"catalog", 7))
5179
0
  goto error;
5180
0
    tmp += 7;
5181
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5182
0
    if (*tmp != '=') {
5183
0
  return;
5184
0
    }
5185
0
    tmp++;
5186
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5187
0
    marker = *tmp;
5188
0
    if ((marker != '\'') && (marker != '"'))
5189
0
  goto error;
5190
0
    tmp++;
5191
0
    base = tmp;
5192
0
    while ((*tmp != 0) && (*tmp != marker)) tmp++;
5193
0
    if (*tmp == 0)
5194
0
  goto error;
5195
0
    URL = xmlStrndup(base, tmp - base);
5196
0
    tmp++;
5197
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5198
0
    if (*tmp != 0)
5199
0
  goto error;
5200
5201
0
    if (URL != NULL) {
5202
        /*
5203
         * Unfortunately, the catalog API doesn't report OOM errors.
5204
         * xmlGetLastError isn't very helpful since we don't know
5205
         * where the last error came from. We'd have to reset it
5206
         * before this call and restore it afterwards.
5207
         */
5208
0
  ctxt->catalogs = xmlCatalogAddLocal(ctxt->catalogs, URL);
5209
0
  xmlFree(URL);
5210
0
    }
5211
0
    return;
5212
5213
0
error:
5214
0
    xmlWarningMsg(ctxt, XML_WAR_CATALOG_PI,
5215
0
            "Catalog PI syntax error: %s\n",
5216
0
      catalog, NULL);
5217
0
    if (URL != NULL)
5218
0
  xmlFree(URL);
5219
0
}
5220
#endif
5221
5222
/**
5223
 * Parse an XML Processing Instruction.
5224
 *
5225
 * @deprecated Internal function, don't use.
5226
 *
5227
 *     [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>'
5228
 *
5229
 * The processing is transferred to SAX once parsed.
5230
 *
5231
 * @param ctxt  an XML parser context
5232
 */
5233
5234
void
5235
62.6k
xmlParsePI(xmlParserCtxt *ctxt) {
5236
62.6k
    xmlChar *buf = NULL;
5237
62.6k
    size_t len = 0;
5238
62.6k
    size_t size = XML_PARSER_BUFFER_SIZE;
5239
62.6k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
5240
27.3k
                       XML_MAX_HUGE_LENGTH :
5241
62.6k
                       XML_MAX_TEXT_LENGTH;
5242
62.6k
    int cur, l;
5243
62.6k
    const xmlChar *target;
5244
5245
62.6k
    if ((RAW == '<') && (NXT(1) == '?')) {
5246
  /*
5247
   * this is a Processing Instruction.
5248
   */
5249
62.6k
  SKIP(2);
5250
5251
  /*
5252
   * Parse the target name and check for special support like
5253
   * namespace.
5254
   */
5255
62.6k
        target = xmlParsePITarget(ctxt);
5256
62.6k
  if (target != NULL) {
5257
53.5k
      if ((RAW == '?') && (NXT(1) == '>')) {
5258
33.2k
    SKIP(2);
5259
5260
    /*
5261
     * SAX: PI detected.
5262
     */
5263
33.2k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5264
32.8k
        (ctxt->sax->processingInstruction != NULL))
5265
32.8k
        ctxt->sax->processingInstruction(ctxt->userData,
5266
32.8k
                                         target, NULL);
5267
33.2k
    return;
5268
33.2k
      }
5269
20.2k
      buf = xmlMalloc(size);
5270
20.2k
      if (buf == NULL) {
5271
67
    xmlErrMemory(ctxt);
5272
67
    return;
5273
67
      }
5274
20.2k
      if (SKIP_BLANKS == 0) {
5275
15.8k
    xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED,
5276
15.8k
        "ParsePI: PI %s space expected\n", target);
5277
15.8k
      }
5278
20.2k
      cur = xmlCurrentCharRecover(ctxt, &l);
5279
4.69M
      while (IS_CHAR(cur) && /* checked */
5280
4.68M
       ((cur != '?') || (NXT(1) != '>'))) {
5281
4.67M
    if (len + 5 >= size) {
5282
3.81k
        xmlChar *tmp;
5283
3.81k
                    int newSize;
5284
5285
3.81k
                    newSize = xmlGrowCapacity(size, 1, 1, maxLength);
5286
3.81k
                    if (newSize < 0) {
5287
0
                        xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5288
0
                                          "PI %s too big found", target);
5289
0
                        xmlFree(buf);
5290
0
                        return;
5291
0
                    }
5292
3.81k
        tmp = xmlRealloc(buf, newSize);
5293
3.81k
        if (tmp == NULL) {
5294
8
      xmlErrMemory(ctxt);
5295
8
      xmlFree(buf);
5296
8
      return;
5297
8
        }
5298
3.80k
        buf = tmp;
5299
3.80k
                    size = newSize;
5300
3.80k
    }
5301
4.67M
    COPY_BUF(buf, len, cur);
5302
4.67M
    NEXTL(l);
5303
4.67M
    cur = xmlCurrentCharRecover(ctxt, &l);
5304
4.67M
      }
5305
20.2k
      buf[len] = 0;
5306
20.2k
      if (cur != '?') {
5307
9.02k
    xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5308
9.02k
          "ParsePI: PI %s never end ...\n", target);
5309
11.1k
      } else {
5310
11.1k
    SKIP(2);
5311
5312
11.1k
#ifdef LIBXML_CATALOG_ENABLED
5313
11.1k
    if ((ctxt->inSubset == 0) &&
5314
9.96k
        (xmlStrEqual(target, XML_CATALOG_PI))) {
5315
417
        xmlCatalogAllow allow = xmlCatalogGetDefaults();
5316
5317
417
        if ((ctxt->options & XML_PARSE_CATALOG_PI) &&
5318
207
                        ((allow == XML_CATA_ALLOW_DOCUMENT) ||
5319
207
       (allow == XML_CATA_ALLOW_ALL)))
5320
0
      xmlParseCatalogPI(ctxt, buf);
5321
417
    }
5322
11.1k
#endif
5323
5324
    /*
5325
     * SAX: PI detected.
5326
     */
5327
11.1k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5328
10.8k
        (ctxt->sax->processingInstruction != NULL))
5329
10.8k
        ctxt->sax->processingInstruction(ctxt->userData,
5330
10.8k
                                         target, buf);
5331
11.1k
      }
5332
20.2k
      xmlFree(buf);
5333
20.2k
  } else {
5334
9.09k
      xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL);
5335
9.09k
  }
5336
62.6k
    }
5337
62.6k
}
5338
5339
/**
5340
 * Parse a notation declaration. Always consumes '<!'.
5341
 *
5342
 * @deprecated Internal function, don't use.
5343
 *
5344
 *     [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID |  PublicID)
5345
 *                           S? '>'
5346
 *
5347
 * Hence there is actually 3 choices:
5348
 *
5349
 *     'PUBLIC' S PubidLiteral
5350
 *     'PUBLIC' S PubidLiteral S SystemLiteral
5351
 *     'SYSTEM' S SystemLiteral
5352
 *
5353
 * See the NOTE on #xmlParseExternalID.
5354
 *
5355
 * @param ctxt  an XML parser context
5356
 */
5357
5358
void
5359
8.10k
xmlParseNotationDecl(xmlParserCtxt *ctxt) {
5360
8.10k
    const xmlChar *name;
5361
8.10k
    xmlChar *Pubid;
5362
8.10k
    xmlChar *Systemid;
5363
5364
8.10k
    if ((CUR != '<') || (NXT(1) != '!'))
5365
0
        return;
5366
8.10k
    SKIP(2);
5367
5368
8.10k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5369
7.99k
#ifdef LIBXML_VALID_ENABLED
5370
7.99k
  int oldInputNr = ctxt->inputNr;
5371
7.99k
#endif
5372
5373
7.99k
  SKIP(8);
5374
7.99k
  if (SKIP_BLANKS_PE == 0) {
5375
237
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5376
237
         "Space required after '<!NOTATION'\n");
5377
237
      return;
5378
237
  }
5379
5380
7.75k
        name = xmlParseName(ctxt);
5381
7.75k
  if (name == NULL) {
5382
232
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5383
232
      return;
5384
232
  }
5385
7.52k
  if (xmlStrchr(name, ':') != NULL) {
5386
416
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5387
416
         "colons are forbidden from notation names '%s'\n",
5388
416
         name, NULL, NULL);
5389
416
  }
5390
7.52k
  if (SKIP_BLANKS_PE == 0) {
5391
612
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5392
612
         "Space required after the NOTATION name'\n");
5393
612
      return;
5394
612
  }
5395
5396
  /*
5397
   * Parse the IDs.
5398
   */
5399
6.91k
  Systemid = xmlParseExternalID(ctxt, &Pubid, 0);
5400
6.91k
  SKIP_BLANKS_PE;
5401
5402
6.91k
  if (RAW == '>') {
5403
3.05k
#ifdef LIBXML_VALID_ENABLED
5404
3.05k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
5405
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
5406
0
                           "Notation declaration doesn't start and stop"
5407
0
                                 " in the same entity\n",
5408
0
                                 NULL, NULL);
5409
0
      }
5410
3.05k
#endif
5411
3.05k
      NEXT;
5412
3.05k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5413
2.76k
    (ctxt->sax->notationDecl != NULL))
5414
2.76k
    ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid);
5415
3.85k
  } else {
5416
3.85k
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5417
3.85k
  }
5418
6.91k
  if (Systemid != NULL) xmlFree(Systemid);
5419
6.91k
  if (Pubid != NULL) xmlFree(Pubid);
5420
6.91k
    }
5421
8.10k
}
5422
5423
/**
5424
 * Parse an entity declaration. Always consumes '<!'.
5425
 *
5426
 * @deprecated Internal function, don't use.
5427
 *
5428
 *     [70] EntityDecl ::= GEDecl | PEDecl
5429
 *
5430
 *     [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>'
5431
 *
5432
 *     [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
5433
 *
5434
 *     [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?)
5435
 *
5436
 *     [74] PEDef ::= EntityValue | ExternalID
5437
 *
5438
 *     [76] NDataDecl ::= S 'NDATA' S Name
5439
 *
5440
 * [ VC: Notation Declared ]
5441
 * The Name must match the declared name of a notation.
5442
 *
5443
 * @param ctxt  an XML parser context
5444
 */
5445
5446
void
5447
95.1k
xmlParseEntityDecl(xmlParserCtxt *ctxt) {
5448
95.1k
    const xmlChar *name = NULL;
5449
95.1k
    xmlChar *value = NULL;
5450
95.1k
    xmlChar *URI = NULL, *literal = NULL;
5451
95.1k
    const xmlChar *ndata = NULL;
5452
95.1k
    int isParameter = 0;
5453
95.1k
    xmlChar *orig = NULL;
5454
5455
95.1k
    if ((CUR != '<') || (NXT(1) != '!'))
5456
0
        return;
5457
95.1k
    SKIP(2);
5458
5459
    /* GROW; done in the caller */
5460
95.1k
    if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5461
95.0k
#ifdef LIBXML_VALID_ENABLED
5462
95.0k
  int oldInputNr = ctxt->inputNr;
5463
95.0k
#endif
5464
5465
95.0k
  SKIP(6);
5466
95.0k
  if (SKIP_BLANKS_PE == 0) {
5467
7.75k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5468
7.75k
         "Space required after '<!ENTITY'\n");
5469
7.75k
  }
5470
5471
95.0k
  if (RAW == '%') {
5472
40.4k
      NEXT;
5473
40.4k
      if (SKIP_BLANKS_PE == 0) {
5474
3.55k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5475
3.55k
             "Space required after '%%'\n");
5476
3.55k
      }
5477
40.4k
      isParameter = 1;
5478
40.4k
  }
5479
5480
95.0k
        name = xmlParseName(ctxt);
5481
95.0k
  if (name == NULL) {
5482
1.53k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5483
1.53k
                     "xmlParseEntityDecl: no name\n");
5484
1.53k
            return;
5485
1.53k
  }
5486
93.5k
  if (xmlStrchr(name, ':') != NULL) {
5487
306
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5488
306
         "colons are forbidden from entities names '%s'\n",
5489
306
         name, NULL, NULL);
5490
306
  }
5491
93.5k
  if (SKIP_BLANKS_PE == 0) {
5492
9.71k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5493
9.71k
         "Space required after the entity name\n");
5494
9.71k
  }
5495
5496
  /*
5497
   * handle the various case of definitions...
5498
   */
5499
93.5k
  if (isParameter) {
5500
40.1k
      if ((RAW == '"') || (RAW == '\'')) {
5501
27.7k
          value = xmlParseEntityValue(ctxt, &orig);
5502
27.7k
    if (value) {
5503
27.1k
        if ((ctxt->sax != NULL) &&
5504
27.1k
      (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5505
23.0k
      ctxt->sax->entityDecl(ctxt->userData, name,
5506
23.0k
                        XML_INTERNAL_PARAMETER_ENTITY,
5507
23.0k
            NULL, NULL, value);
5508
27.1k
    }
5509
27.7k
      } else {
5510
12.4k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5511
12.4k
    if ((URI == NULL) && (literal == NULL)) {
5512
741
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5513
741
    }
5514
12.4k
    if (URI) {
5515
11.3k
                    if (xmlStrchr(URI, '#')) {
5516
292
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5517
11.0k
                    } else {
5518
11.0k
                        if ((ctxt->sax != NULL) &&
5519
11.0k
                            (!ctxt->disableSAX) &&
5520
10.1k
                            (ctxt->sax->entityDecl != NULL))
5521
10.1k
                            ctxt->sax->entityDecl(ctxt->userData, name,
5522
10.1k
                                        XML_EXTERNAL_PARAMETER_ENTITY,
5523
10.1k
                                        literal, URI, NULL);
5524
11.0k
                    }
5525
11.3k
    }
5526
12.4k
      }
5527
53.3k
  } else {
5528
53.3k
      if ((RAW == '"') || (RAW == '\'')) {
5529
40.2k
          value = xmlParseEntityValue(ctxt, &orig);
5530
40.2k
    if ((ctxt->sax != NULL) &&
5531
40.2k
        (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5532
32.9k
        ctxt->sax->entityDecl(ctxt->userData, name,
5533
32.9k
        XML_INTERNAL_GENERAL_ENTITY,
5534
32.9k
        NULL, NULL, value);
5535
    /*
5536
     * For expat compatibility in SAX mode.
5537
     */
5538
40.2k
    if ((ctxt->myDoc == NULL) ||
5539
39.7k
        (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) {
5540
5.25k
        if (ctxt->myDoc == NULL) {
5541
496
      ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5542
496
      if (ctxt->myDoc == NULL) {
5543
3
          xmlErrMemory(ctxt);
5544
3
          goto done;
5545
3
      }
5546
493
      ctxt->myDoc->properties = XML_DOC_INTERNAL;
5547
493
        }
5548
5.24k
        if (ctxt->myDoc->intSubset == NULL) {
5549
493
      ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5550
493
              BAD_CAST "fake", NULL, NULL);
5551
493
                        if (ctxt->myDoc->intSubset == NULL) {
5552
3
                            xmlErrMemory(ctxt);
5553
3
                            goto done;
5554
3
                        }
5555
493
                    }
5556
5557
5.24k
        xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY,
5558
5.24k
                    NULL, NULL, value);
5559
5.24k
    }
5560
40.2k
      } else {
5561
13.1k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5562
13.1k
    if ((URI == NULL) && (literal == NULL)) {
5563
3.38k
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5564
3.38k
    }
5565
13.1k
    if (URI) {
5566
9.10k
                    if (xmlStrchr(URI, '#')) {
5567
445
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5568
445
                    }
5569
9.10k
    }
5570
13.1k
    if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) {
5571
4.08k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5572
4.08k
           "Space required before 'NDATA'\n");
5573
4.08k
    }
5574
13.1k
    if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) {
5575
1.30k
        SKIP(5);
5576
1.30k
        if (SKIP_BLANKS_PE == 0) {
5577
318
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5578
318
               "Space required after 'NDATA'\n");
5579
318
        }
5580
1.30k
        ndata = xmlParseName(ctxt);
5581
1.30k
        if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5582
818
            (ctxt->sax->unparsedEntityDecl != NULL))
5583
818
      ctxt->sax->unparsedEntityDecl(ctxt->userData, name,
5584
818
            literal, URI, ndata);
5585
11.8k
    } else {
5586
11.8k
        if ((ctxt->sax != NULL) &&
5587
11.8k
            (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5588
11.1k
      ctxt->sax->entityDecl(ctxt->userData, name,
5589
11.1k
            XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5590
11.1k
            literal, URI, NULL);
5591
        /*
5592
         * For expat compatibility in SAX mode.
5593
         * assuming the entity replacement was asked for
5594
         */
5595
11.8k
        if ((ctxt->replaceEntities != 0) &&
5596
8.21k
      ((ctxt->myDoc == NULL) ||
5597
8.18k
      (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) {
5598
225
      if (ctxt->myDoc == NULL) {
5599
31
          ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5600
31
          if (ctxt->myDoc == NULL) {
5601
4
              xmlErrMemory(ctxt);
5602
4
        goto done;
5603
4
          }
5604
27
          ctxt->myDoc->properties = XML_DOC_INTERNAL;
5605
27
      }
5606
5607
221
      if (ctxt->myDoc->intSubset == NULL) {
5608
27
          ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5609
27
            BAD_CAST "fake", NULL, NULL);
5610
27
                            if (ctxt->myDoc->intSubset == NULL) {
5611
3
                                xmlErrMemory(ctxt);
5612
3
                                goto done;
5613
3
                            }
5614
27
                        }
5615
218
      xmlSAX2EntityDecl(ctxt, name,
5616
218
                  XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5617
218
                  literal, URI, NULL);
5618
218
        }
5619
11.8k
    }
5620
13.1k
      }
5621
53.3k
  }
5622
93.5k
  SKIP_BLANKS_PE;
5623
93.5k
  if (RAW != '>') {
5624
14.9k
      xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED,
5625
14.9k
              "xmlParseEntityDecl: entity %s not terminated\n", name);
5626
78.5k
  } else {
5627
78.5k
#ifdef LIBXML_VALID_ENABLED
5628
78.5k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
5629
72
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
5630
72
                           "Entity declaration doesn't start and stop in"
5631
72
                                 " the same entity\n",
5632
72
                                 NULL, NULL);
5633
72
      }
5634
78.5k
#endif
5635
78.5k
      NEXT;
5636
78.5k
  }
5637
93.5k
  if (orig != NULL) {
5638
      /*
5639
       * Ugly mechanism to save the raw entity value.
5640
       */
5641
65.6k
      xmlEntityPtr cur = NULL;
5642
5643
65.6k
      if (isParameter) {
5644
27.3k
          if ((ctxt->sax != NULL) &&
5645
27.3k
        (ctxt->sax->getParameterEntity != NULL))
5646
27.3k
        cur = ctxt->sax->getParameterEntity(ctxt->userData, name);
5647
38.2k
      } else {
5648
38.2k
          if ((ctxt->sax != NULL) &&
5649
38.2k
        (ctxt->sax->getEntity != NULL))
5650
38.2k
        cur = ctxt->sax->getEntity(ctxt->userData, name);
5651
38.2k
    if ((cur == NULL) && (ctxt->userData==ctxt)) {
5652
1.87k
        cur = xmlSAX2GetEntity(ctxt, name);
5653
1.87k
    }
5654
38.2k
      }
5655
65.6k
            if ((cur != NULL) && (cur->orig == NULL)) {
5656
35.2k
    cur->orig = orig;
5657
35.2k
                orig = NULL;
5658
35.2k
      }
5659
65.6k
  }
5660
5661
93.5k
done:
5662
93.5k
  if (value != NULL) xmlFree(value);
5663
93.5k
  if (URI != NULL) xmlFree(URI);
5664
93.5k
  if (literal != NULL) xmlFree(literal);
5665
93.5k
        if (orig != NULL) xmlFree(orig);
5666
93.5k
    }
5667
95.1k
}
5668
5669
/**
5670
 * Parse an attribute default declaration
5671
 *
5672
 * @deprecated Internal function, don't use.
5673
 *
5674
 *     [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue)
5675
 *
5676
 * [ VC: Required Attribute ]
5677
 * if the default declaration is the keyword \#REQUIRED, then the
5678
 * attribute must be specified for all elements of the type in the
5679
 * attribute-list declaration.
5680
 *
5681
 * [ VC: Attribute Default Legal ]
5682
 * The declared default value must meet the lexical constraints of
5683
 * the declared attribute type c.f. #xmlValidateAttributeDecl
5684
 *
5685
 * [ VC: Fixed Attribute Default ]
5686
 * if an attribute has a default value declared with the \#FIXED
5687
 * keyword, instances of that attribute must match the default value.
5688
 *
5689
 * [ WFC: No < in Attribute Values ]
5690
 * handled in #xmlParseAttValue
5691
 *
5692
 * @param ctxt  an XML parser context
5693
 * @param value  Receive a possible fixed default value for the attribute
5694
 * @returns XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED
5695
 *          or XML_ATTRIBUTE_FIXED.
5696
 */
5697
5698
int
5699
121k
xmlParseDefaultDecl(xmlParserCtxt *ctxt, xmlChar **value) {
5700
121k
    int val;
5701
121k
    xmlChar *ret;
5702
5703
121k
    *value = NULL;
5704
121k
    if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) {
5705
5.01k
  SKIP(9);
5706
5.01k
  return(XML_ATTRIBUTE_REQUIRED);
5707
5.01k
    }
5708
116k
    if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) {
5709
36.2k
  SKIP(8);
5710
36.2k
  return(XML_ATTRIBUTE_IMPLIED);
5711
36.2k
    }
5712
79.8k
    val = XML_ATTRIBUTE_NONE;
5713
79.8k
    if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) {
5714
4.43k
  SKIP(6);
5715
4.43k
  val = XML_ATTRIBUTE_FIXED;
5716
4.43k
  if (SKIP_BLANKS_PE == 0) {
5717
297
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5718
297
         "Space required after '#FIXED'\n");
5719
297
  }
5720
4.43k
    }
5721
79.8k
    ret = xmlParseAttValue(ctxt);
5722
79.8k
    if (ret == NULL) {
5723
9.86k
  xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo,
5724
9.86k
           "Attribute default value declaration error\n");
5725
9.86k
    } else
5726
69.9k
        *value = ret;
5727
79.8k
    return(val);
5728
116k
}
5729
5730
/**
5731
 * Parse an Notation attribute type.
5732
 *
5733
 * @deprecated Internal function, don't use.
5734
 *
5735
 * Note: the leading 'NOTATION' S part has already being parsed...
5736
 *
5737
 *     [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
5738
 *
5739
 * [ VC: Notation Attributes ]
5740
 * Values of this type must match one of the notation names included
5741
 * in the declaration; all notation names in the declaration must be declared.
5742
 *
5743
 * @param ctxt  an XML parser context
5744
 * @returns the notation attribute tree built while parsing
5745
 */
5746
5747
xmlEnumeration *
5748
1.82k
xmlParseNotationType(xmlParserCtxt *ctxt) {
5749
1.82k
    const xmlChar *name;
5750
1.82k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5751
5752
1.82k
    if (RAW != '(') {
5753
238
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5754
238
  return(NULL);
5755
238
    }
5756
3.13k
    do {
5757
3.13k
        NEXT;
5758
3.13k
  SKIP_BLANKS_PE;
5759
3.13k
        name = xmlParseName(ctxt);
5760
3.13k
  if (name == NULL) {
5761
216
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5762
216
         "Name expected in NOTATION declaration\n");
5763
216
            xmlFreeEnumeration(ret);
5764
216
      return(NULL);
5765
216
  }
5766
2.91k
        tmp = NULL;
5767
2.91k
#ifdef LIBXML_VALID_ENABLED
5768
2.91k
        if (ctxt->validate) {
5769
1.93k
            tmp = ret;
5770
5.55k
            while (tmp != NULL) {
5771
4.27k
                if (xmlStrEqual(name, tmp->name)) {
5772
651
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5773
651
              "standalone: attribute notation value token %s duplicated\n",
5774
651
                                     name, NULL);
5775
651
                    if (!xmlDictOwns(ctxt->dict, name))
5776
0
                        xmlFree((xmlChar *) name);
5777
651
                    break;
5778
651
                }
5779
3.61k
                tmp = tmp->next;
5780
3.61k
            }
5781
1.93k
        }
5782
2.91k
#endif /* LIBXML_VALID_ENABLED */
5783
2.91k
  if (tmp == NULL) {
5784
2.26k
      cur = xmlCreateEnumeration(name);
5785
2.26k
      if (cur == NULL) {
5786
7
                xmlErrMemory(ctxt);
5787
7
                xmlFreeEnumeration(ret);
5788
7
                return(NULL);
5789
7
            }
5790
2.26k
      if (last == NULL) ret = last = cur;
5791
895
      else {
5792
895
    last->next = cur;
5793
895
    last = cur;
5794
895
      }
5795
2.26k
  }
5796
2.91k
  SKIP_BLANKS_PE;
5797
2.91k
    } while (RAW == '|');
5798
1.36k
    if (RAW != ')') {
5799
352
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5800
352
        xmlFreeEnumeration(ret);
5801
352
  return(NULL);
5802
352
    }
5803
1.00k
    NEXT;
5804
1.00k
    return(ret);
5805
1.36k
}
5806
5807
/**
5808
 * Parse an Enumeration attribute type.
5809
 *
5810
 * @deprecated Internal function, don't use.
5811
 *
5812
 *     [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')'
5813
 *
5814
 * [ VC: Enumeration ]
5815
 * Values of this type must match one of the Nmtoken tokens in
5816
 * the declaration
5817
 *
5818
 * @param ctxt  an XML parser context
5819
 * @returns the enumeration attribute tree built while parsing
5820
 */
5821
5822
xmlEnumeration *
5823
31.1k
xmlParseEnumerationType(xmlParserCtxt *ctxt) {
5824
31.1k
    xmlChar *name;
5825
31.1k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5826
5827
31.1k
    if (RAW != '(') {
5828
1.27k
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL);
5829
1.27k
  return(NULL);
5830
1.27k
    }
5831
41.4k
    do {
5832
41.4k
        NEXT;
5833
41.4k
  SKIP_BLANKS_PE;
5834
41.4k
        name = xmlParseNmtoken(ctxt);
5835
41.4k
  if (name == NULL) {
5836
284
      xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL);
5837
284
      return(ret);
5838
284
  }
5839
41.1k
        tmp = NULL;
5840
41.1k
#ifdef LIBXML_VALID_ENABLED
5841
41.1k
        if (ctxt->validate) {
5842
27.5k
            tmp = ret;
5843
47.1k
            while (tmp != NULL) {
5844
20.0k
                if (xmlStrEqual(name, tmp->name)) {
5845
484
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5846
484
              "standalone: attribute enumeration value token %s duplicated\n",
5847
484
                                     name, NULL);
5848
484
                    if (!xmlDictOwns(ctxt->dict, name))
5849
484
                        xmlFree(name);
5850
484
                    break;
5851
484
                }
5852
19.5k
                tmp = tmp->next;
5853
19.5k
            }
5854
27.5k
        }
5855
41.1k
#endif /* LIBXML_VALID_ENABLED */
5856
41.1k
  if (tmp == NULL) {
5857
40.6k
      cur = xmlCreateEnumeration(name);
5858
40.6k
      if (!xmlDictOwns(ctxt->dict, name))
5859
40.6k
    xmlFree(name);
5860
40.6k
      if (cur == NULL) {
5861
39
                xmlErrMemory(ctxt);
5862
39
                xmlFreeEnumeration(ret);
5863
39
                return(NULL);
5864
39
            }
5865
40.6k
      if (last == NULL) ret = last = cur;
5866
10.9k
      else {
5867
10.9k
    last->next = cur;
5868
10.9k
    last = cur;
5869
10.9k
      }
5870
40.6k
  }
5871
41.1k
  SKIP_BLANKS_PE;
5872
41.1k
    } while (RAW == '|');
5873
29.5k
    if (RAW != ')') {
5874
1.21k
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL);
5875
1.21k
  return(ret);
5876
1.21k
    }
5877
28.3k
    NEXT;
5878
28.3k
    return(ret);
5879
29.5k
}
5880
5881
/**
5882
 * Parse an Enumerated attribute type.
5883
 *
5884
 * @deprecated Internal function, don't use.
5885
 *
5886
 *     [57] EnumeratedType ::= NotationType | Enumeration
5887
 *
5888
 *     [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
5889
 *
5890
 * @param ctxt  an XML parser context
5891
 * @param tree  the enumeration tree built while parsing
5892
 * @returns XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION
5893
 */
5894
5895
int
5896
33.1k
xmlParseEnumeratedType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5897
33.1k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5898
2.03k
  SKIP(8);
5899
2.03k
  if (SKIP_BLANKS_PE == 0) {
5900
216
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5901
216
         "Space required after 'NOTATION'\n");
5902
216
      return(0);
5903
216
  }
5904
1.82k
  *tree = xmlParseNotationType(ctxt);
5905
1.82k
  if (*tree == NULL) return(0);
5906
1.00k
  return(XML_ATTRIBUTE_NOTATION);
5907
1.82k
    }
5908
31.1k
    *tree = xmlParseEnumerationType(ctxt);
5909
31.1k
    if (*tree == NULL) return(0);
5910
29.6k
    return(XML_ATTRIBUTE_ENUMERATION);
5911
31.1k
}
5912
5913
/**
5914
 * Parse the Attribute list def for an element
5915
 *
5916
 * @deprecated Internal function, don't use.
5917
 *
5918
 *     [54] AttType ::= StringType | TokenizedType | EnumeratedType
5919
 *
5920
 *     [55] StringType ::= 'CDATA'
5921
 *
5922
 *     [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' |
5923
 *                            'ENTITIES' | 'NMTOKEN' | 'NMTOKENS'
5924
 *
5925
 * Validity constraints for attribute values syntax are checked in
5926
 * #xmlValidateAttributeValue
5927
 *
5928
 * [ VC: ID ]
5929
 * Values of type ID must match the Name production. A name must not
5930
 * appear more than once in an XML document as a value of this type;
5931
 * i.e., ID values must uniquely identify the elements which bear them.
5932
 *
5933
 * [ VC: One ID per Element Type ]
5934
 * No element type may have more than one ID attribute specified.
5935
 *
5936
 * [ VC: ID Attribute Default ]
5937
 * An ID attribute must have a declared default of \#IMPLIED or \#REQUIRED.
5938
 *
5939
 * [ VC: IDREF ]
5940
 * Values of type IDREF must match the Name production, and values
5941
 * of type IDREFS must match Names; each IDREF Name must match the value
5942
 * of an ID attribute on some element in the XML document; i.e. IDREF
5943
 * values must match the value of some ID attribute.
5944
 *
5945
 * [ VC: Entity Name ]
5946
 * Values of type ENTITY must match the Name production, values
5947
 * of type ENTITIES must match Names; each Entity Name must match the
5948
 * name of an unparsed entity declared in the DTD.
5949
 *
5950
 * [ VC: Name Token ]
5951
 * Values of type NMTOKEN must match the Nmtoken production; values
5952
 * of type NMTOKENS must match Nmtokens.
5953
 *
5954
 * @param ctxt  an XML parser context
5955
 * @param tree  the enumeration tree built while parsing
5956
 * @returns the attribute type
5957
 */
5958
int
5959
127k
xmlParseAttributeType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5960
127k
    if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) {
5961
17.7k
  SKIP(5);
5962
17.7k
  return(XML_ATTRIBUTE_CDATA);
5963
109k
     } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) {
5964
24.2k
  SKIP(6);
5965
24.2k
  return(XML_ATTRIBUTE_IDREFS);
5966
85.0k
     } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) {
5967
1.72k
  SKIP(5);
5968
1.72k
  return(XML_ATTRIBUTE_IDREF);
5969
83.2k
     } else if ((RAW == 'I') && (NXT(1) == 'D')) {
5970
33.9k
        SKIP(2);
5971
33.9k
  return(XML_ATTRIBUTE_ID);
5972
49.3k
     } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5973
2.07k
  SKIP(6);
5974
2.07k
  return(XML_ATTRIBUTE_ENTITY);
5975
47.3k
     } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) {
5976
3.24k
  SKIP(8);
5977
3.24k
  return(XML_ATTRIBUTE_ENTITIES);
5978
44.0k
     } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) {
5979
3.74k
  SKIP(8);
5980
3.74k
  return(XML_ATTRIBUTE_NMTOKENS);
5981
40.3k
     } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) {
5982
7.12k
  SKIP(7);
5983
7.12k
  return(XML_ATTRIBUTE_NMTOKEN);
5984
7.12k
     }
5985
33.1k
     return(xmlParseEnumeratedType(ctxt, tree));
5986
127k
}
5987
5988
/**
5989
 * Parse an attribute list declaration for an element. Always consumes '<!'.
5990
 *
5991
 * @deprecated Internal function, don't use.
5992
 *
5993
 *     [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>'
5994
 *
5995
 *     [53] AttDef ::= S Name S AttType S DefaultDecl
5996
 * @param ctxt  an XML parser context
5997
 */
5998
void
5999
90.8k
xmlParseAttributeListDecl(xmlParserCtxt *ctxt) {
6000
90.8k
    const xmlChar *elemName;
6001
90.8k
    const xmlChar *attrName;
6002
90.8k
    xmlEnumerationPtr tree;
6003
6004
90.8k
    if ((CUR != '<') || (NXT(1) != '!'))
6005
0
        return;
6006
90.8k
    SKIP(2);
6007
6008
90.8k
    if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) {
6009
90.6k
#ifdef LIBXML_VALID_ENABLED
6010
90.6k
  int oldInputNr = ctxt->inputNr;
6011
90.6k
#endif
6012
6013
90.6k
  SKIP(7);
6014
90.6k
  if (SKIP_BLANKS_PE == 0) {
6015
15.1k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6016
15.1k
                     "Space required after '<!ATTLIST'\n");
6017
15.1k
  }
6018
90.6k
        elemName = xmlParseName(ctxt);
6019
90.6k
  if (elemName == NULL) {
6020
2.48k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6021
2.48k
         "ATTLIST: no name for Element\n");
6022
2.48k
      return;
6023
2.48k
  }
6024
88.1k
  SKIP_BLANKS_PE;
6025
88.1k
  GROW;
6026
197k
  while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) {
6027
144k
      int type;
6028
144k
      int def;
6029
144k
      xmlChar *defaultValue = NULL;
6030
6031
144k
      GROW;
6032
144k
            tree = NULL;
6033
144k
      attrName = xmlParseName(ctxt);
6034
144k
      if (attrName == NULL) {
6035
8.25k
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6036
8.25k
             "ATTLIST: no name for Attribute\n");
6037
8.25k
    break;
6038
8.25k
      }
6039
136k
      GROW;
6040
136k
      if (SKIP_BLANKS_PE == 0) {
6041
9.40k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6042
9.40k
            "Space required after the attribute name\n");
6043
9.40k
    break;
6044
9.40k
      }
6045
6046
127k
      type = xmlParseAttributeType(ctxt, &tree);
6047
127k
      if (type <= 0) {
6048
2.48k
          break;
6049
2.48k
      }
6050
6051
124k
      GROW;
6052
124k
      if (SKIP_BLANKS_PE == 0) {
6053
3.51k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6054
3.51k
             "Space required after the attribute type\n");
6055
3.51k
          if (tree != NULL)
6056
1.41k
        xmlFreeEnumeration(tree);
6057
3.51k
    break;
6058
3.51k
      }
6059
6060
121k
      def = xmlParseDefaultDecl(ctxt, &defaultValue);
6061
121k
      if (def <= 0) {
6062
0
                if (defaultValue != NULL)
6063
0
        xmlFree(defaultValue);
6064
0
          if (tree != NULL)
6065
0
        xmlFreeEnumeration(tree);
6066
0
          break;
6067
0
      }
6068
121k
      if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL))
6069
65.8k
          xmlAttrNormalizeSpace(defaultValue, defaultValue);
6070
6071
121k
      GROW;
6072
121k
            if (RAW != '>') {
6073
82.8k
    if (SKIP_BLANKS_PE == 0) {
6074
12.0k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6075
12.0k
      "Space required after the attribute default value\n");
6076
12.0k
        if (defaultValue != NULL)
6077
2.81k
      xmlFree(defaultValue);
6078
12.0k
        if (tree != NULL)
6079
1.77k
      xmlFreeEnumeration(tree);
6080
12.0k
        break;
6081
12.0k
    }
6082
82.8k
      }
6083
109k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6084
104k
    (ctxt->sax->attributeDecl != NULL))
6085
104k
    ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName,
6086
104k
                          type, def, defaultValue, tree);
6087
4.90k
      else if (tree != NULL)
6088
2.06k
    xmlFreeEnumeration(tree);
6089
6090
109k
      if ((ctxt->sax2) && (defaultValue != NULL) &&
6091
56.8k
          (def != XML_ATTRIBUTE_IMPLIED) &&
6092
56.8k
    (def != XML_ATTRIBUTE_REQUIRED)) {
6093
56.8k
    xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue);
6094
56.8k
      }
6095
109k
      if (ctxt->sax2) {
6096
87.4k
    xmlAddSpecialAttr(ctxt, elemName, attrName, type);
6097
87.4k
      }
6098
109k
      if (defaultValue != NULL)
6099
67.1k
          xmlFree(defaultValue);
6100
109k
      GROW;
6101
109k
  }
6102
88.1k
  if (RAW == '>') {
6103
54.9k
#ifdef LIBXML_VALID_ENABLED
6104
54.9k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6105
132
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6106
132
                                 "Attribute list declaration doesn't start and"
6107
132
                                 " stop in the same entity\n",
6108
132
                                 NULL, NULL);
6109
132
      }
6110
54.9k
#endif
6111
54.9k
      NEXT;
6112
54.9k
  }
6113
88.1k
    }
6114
90.8k
}
6115
6116
/**
6117
 * Handle PEs and check that we don't pop the entity that started
6118
 * a balanced group.
6119
 *
6120
 * @param ctxt  parser context
6121
 * @param openInputNr  input nr of the entity with opening '('
6122
 */
6123
static void
6124
3.85M
xmlSkipBlankCharsPEBalanced(xmlParserCtxt *ctxt, int openInputNr) {
6125
3.85M
    SKIP_BLANKS;
6126
3.85M
    GROW;
6127
6128
3.85M
    (void) openInputNr;
6129
6130
3.85M
    if (!PARSER_EXTERNAL(ctxt) && !PARSER_IN_PE(ctxt))
6131
3.68M
        return;
6132
6133
183k
    while (!PARSER_STOPPED(ctxt)) {
6134
183k
        if (ctxt->input->cur >= ctxt->input->end) {
6135
6.60k
#ifdef LIBXML_VALID_ENABLED
6136
6.60k
            if ((ctxt->validate) && (ctxt->inputNr <= openInputNr)) {
6137
2.22k
                xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6138
2.22k
                                 "Element content declaration doesn't start "
6139
2.22k
                                 "and stop in the same entity\n",
6140
2.22k
                                 NULL, NULL);
6141
2.22k
            }
6142
6.60k
#endif
6143
6.60k
            if (PARSER_IN_PE(ctxt))
6144
6.46k
                xmlPopPE(ctxt);
6145
144
            else
6146
144
                break;
6147
177k
        } else if (RAW == '%') {
6148
8.29k
            xmlParsePERefInternal(ctxt, 0);
6149
168k
        } else {
6150
168k
            break;
6151
168k
        }
6152
6153
14.7k
        SKIP_BLANKS;
6154
14.7k
        GROW;
6155
14.7k
    }
6156
169k
}
6157
6158
/**
6159
 * Parse the declaration for a Mixed Element content
6160
 * The leading '(' and spaces have been skipped in #xmlParseElementContentDecl
6161
 *
6162
 * @deprecated Internal function, don't use.
6163
 *
6164
 *     [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' |
6165
 *                    '(' S? '#PCDATA' S? ')'
6166
 *
6167
 * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49])
6168
 *
6169
 * [ VC: No Duplicate Types ]
6170
 * The same name must not appear more than once in a single
6171
 * mixed-content declaration.
6172
 *
6173
 * @param ctxt  an XML parser context
6174
 * @param openInputNr  the input used for the current entity, needed for
6175
 * boundary checks
6176
 * @returns the list of the xmlElementContent describing the element choices
6177
 */
6178
xmlElementContent *
6179
10.4k
xmlParseElementMixedContentDecl(xmlParserCtxt *ctxt, int openInputNr) {
6180
10.4k
    xmlElementContentPtr ret = NULL, cur = NULL, n;
6181
10.4k
    const xmlChar *elem = NULL;
6182
6183
10.4k
    GROW;
6184
10.4k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6185
10.4k
  SKIP(7);
6186
10.4k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6187
10.4k
  if (RAW == ')') {
6188
5.19k
#ifdef LIBXML_VALID_ENABLED
6189
5.19k
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6190
3
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6191
3
                                 "Element content declaration doesn't start "
6192
3
                                 "and stop in the same entity\n",
6193
3
                                 NULL, NULL);
6194
3
      }
6195
5.19k
#endif
6196
5.19k
      NEXT;
6197
5.19k
      ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6198
5.19k
      if (ret == NULL)
6199
8
                goto mem_error;
6200
5.19k
      if (RAW == '*') {
6201
360
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6202
360
    NEXT;
6203
360
      }
6204
5.19k
      return(ret);
6205
5.19k
  }
6206
5.29k
  if ((RAW == '(') || (RAW == '|')) {
6207
5.01k
      ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6208
5.01k
      if (ret == NULL)
6209
8
                goto mem_error;
6210
5.01k
  }
6211
38.2k
  while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) {
6212
33.2k
      NEXT;
6213
33.2k
            n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6214
33.2k
            if (n == NULL)
6215
12
                goto mem_error;
6216
33.2k
      if (elem == NULL) {
6217
4.98k
    n->c1 = cur;
6218
4.98k
    if (cur != NULL)
6219
4.98k
        cur->parent = n;
6220
4.98k
    ret = cur = n;
6221
28.2k
      } else {
6222
28.2k
          cur->c2 = n;
6223
28.2k
    n->parent = cur;
6224
28.2k
    n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6225
28.2k
                if (n->c1 == NULL)
6226
11
                    goto mem_error;
6227
28.2k
    n->c1->parent = n;
6228
28.2k
    cur = n;
6229
28.2k
      }
6230
33.2k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6231
33.2k
      elem = xmlParseName(ctxt);
6232
33.2k
      if (elem == NULL) {
6233
252
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6234
252
      "xmlParseElementMixedContentDecl : Name expected\n");
6235
252
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6236
252
    return(NULL);
6237
252
      }
6238
32.9k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6239
32.9k
  }
6240
5.01k
  if ((RAW == ')') && (NXT(1) == '*')) {
6241
4.45k
      if (elem != NULL) {
6242
4.45k
    cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem,
6243
4.45k
                                   XML_ELEMENT_CONTENT_ELEMENT);
6244
4.45k
    if (cur->c2 == NULL)
6245
12
                    goto mem_error;
6246
4.43k
    cur->c2->parent = cur;
6247
4.43k
            }
6248
4.43k
            if (ret != NULL)
6249
4.43k
                ret->ocur = XML_ELEMENT_CONTENT_MULT;
6250
4.43k
#ifdef LIBXML_VALID_ENABLED
6251
4.43k
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6252
3
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6253
3
                                 "Element content declaration doesn't start "
6254
3
                                 "and stop in the same entity\n",
6255
3
                                 NULL, NULL);
6256
3
      }
6257
4.43k
#endif
6258
4.43k
      SKIP(2);
6259
4.43k
  } else {
6260
559
      xmlFreeDocElementContent(ctxt->myDoc, ret);
6261
559
      xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL);
6262
559
      return(NULL);
6263
559
  }
6264
6265
5.01k
    } else {
6266
0
  xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL);
6267
0
    }
6268
4.43k
    return(ret);
6269
6270
51
mem_error:
6271
51
    xmlErrMemory(ctxt);
6272
51
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6273
51
    return(NULL);
6274
10.4k
}
6275
6276
/**
6277
 * Parse the declaration for a Mixed Element content
6278
 * The leading '(' and spaces have been skipped in #xmlParseElementContentDecl
6279
 *
6280
 *     [47] children ::= (choice | seq) ('?' | '*' | '+')?
6281
 *
6282
 *     [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')?
6283
 *
6284
 *     [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')'
6285
 *
6286
 *     [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
6287
 *
6288
 * [ VC: Proper Group/PE Nesting ] applies to [49] and [50]
6289
 * TODO Parameter-entity replacement text must be properly nested
6290
 *  with parenthesized groups. That is to say, if either of the
6291
 *  opening or closing parentheses in a choice, seq, or Mixed
6292
 *  construct is contained in the replacement text for a parameter
6293
 *  entity, both must be contained in the same replacement text. For
6294
 *  interoperability, if a parameter-entity reference appears in a
6295
 *  choice, seq, or Mixed construct, its replacement text should not
6296
 *  be empty, and neither the first nor last non-blank character of
6297
 *  the replacement text should be a connector (| or ,).
6298
 *
6299
 * @param ctxt  an XML parser context
6300
 * @param openInputNr  the input used for the current entity, needed for
6301
 * boundary checks
6302
 * @param depth  the level of recursion
6303
 * @returns the tree of xmlElementContent describing the element
6304
 *          hierarchy.
6305
 */
6306
static xmlElementContentPtr
6307
xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int openInputNr,
6308
125k
                                       int depth) {
6309
125k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
6310
125k
    xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL;
6311
125k
    const xmlChar *elem;
6312
125k
    xmlChar type = 0;
6313
6314
125k
    if (depth > maxDepth) {
6315
9
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
6316
9
                "xmlParseElementChildrenContentDecl : depth %d too deep, "
6317
9
                "use XML_PARSE_HUGE\n", depth);
6318
9
  return(NULL);
6319
9
    }
6320
125k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6321
125k
    if (RAW == '(') {
6322
91.9k
        int newInputNr = ctxt->inputNr;
6323
6324
        /* Recurse on first child */
6325
91.9k
  NEXT;
6326
91.9k
        cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6327
91.9k
                                                           depth + 1);
6328
91.9k
        if (cur == NULL)
6329
82.7k
            return(NULL);
6330
91.9k
    } else {
6331
33.9k
  elem = xmlParseName(ctxt);
6332
33.9k
  if (elem == NULL) {
6333
840
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6334
840
      return(NULL);
6335
840
  }
6336
33.0k
        cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6337
33.0k
  if (cur == NULL) {
6338
62
      xmlErrMemory(ctxt);
6339
62
      return(NULL);
6340
62
  }
6341
33.0k
  GROW;
6342
33.0k
  if (RAW == '?') {
6343
2.88k
      cur->ocur = XML_ELEMENT_CONTENT_OPT;
6344
2.88k
      NEXT;
6345
30.1k
  } else if (RAW == '*') {
6346
2.28k
      cur->ocur = XML_ELEMENT_CONTENT_MULT;
6347
2.28k
      NEXT;
6348
27.8k
  } else if (RAW == '+') {
6349
2.26k
      cur->ocur = XML_ELEMENT_CONTENT_PLUS;
6350
2.26k
      NEXT;
6351
25.5k
  } else {
6352
25.5k
      cur->ocur = XML_ELEMENT_CONTENT_ONCE;
6353
25.5k
  }
6354
33.0k
  GROW;
6355
33.0k
    }
6356
1.83M
    while (!PARSER_STOPPED(ctxt)) {
6357
1.82M
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6358
1.82M
        if (RAW == ')')
6359
35.4k
            break;
6360
        /*
6361
   * Each loop we parse one separator and one element.
6362
   */
6363
1.79M
        if (RAW == ',') {
6364
1.54M
      if (type == 0) type = CUR;
6365
6366
      /*
6367
       * Detect "Name | Name , Name" error
6368
       */
6369
1.54M
      else if (type != CUR) {
6370
11
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6371
11
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6372
11
                      type);
6373
11
    if ((last != NULL) && (last != ret))
6374
11
        xmlFreeDocElementContent(ctxt->myDoc, last);
6375
11
    if (ret != NULL)
6376
11
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6377
11
    return(NULL);
6378
11
      }
6379
1.54M
      NEXT;
6380
6381
1.54M
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ);
6382
1.54M
      if (op == NULL) {
6383
25
                xmlErrMemory(ctxt);
6384
25
    if ((last != NULL) && (last != ret))
6385
9
        xmlFreeDocElementContent(ctxt->myDoc, last);
6386
25
          xmlFreeDocElementContent(ctxt->myDoc, ret);
6387
25
    return(NULL);
6388
25
      }
6389
1.54M
      if (last == NULL) {
6390
9.67k
    op->c1 = ret;
6391
9.67k
    if (ret != NULL)
6392
9.67k
        ret->parent = op;
6393
9.67k
    ret = cur = op;
6394
1.54M
      } else {
6395
1.54M
          cur->c2 = op;
6396
1.54M
    if (op != NULL)
6397
1.54M
        op->parent = cur;
6398
1.54M
    op->c1 = last;
6399
1.54M
    if (last != NULL)
6400
1.54M
        last->parent = op;
6401
1.54M
    cur =op;
6402
1.54M
    last = NULL;
6403
1.54M
      }
6404
1.54M
  } else if (RAW == '|') {
6405
241k
      if (type == 0) type = CUR;
6406
6407
      /*
6408
       * Detect "Name , Name | Name" error
6409
       */
6410
225k
      else if (type != CUR) {
6411
10
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6412
10
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6413
10
          type);
6414
10
    if ((last != NULL) && (last != ret))
6415
10
        xmlFreeDocElementContent(ctxt->myDoc, last);
6416
10
    if (ret != NULL)
6417
10
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6418
10
    return(NULL);
6419
10
      }
6420
241k
      NEXT;
6421
6422
241k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6423
241k
      if (op == NULL) {
6424
18
                xmlErrMemory(ctxt);
6425
18
    if ((last != NULL) && (last != ret))
6426
11
        xmlFreeDocElementContent(ctxt->myDoc, last);
6427
18
    if (ret != NULL)
6428
18
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6429
18
    return(NULL);
6430
18
      }
6431
241k
      if (last == NULL) {
6432
16.1k
    op->c1 = ret;
6433
16.1k
    if (ret != NULL)
6434
16.1k
        ret->parent = op;
6435
16.1k
    ret = cur = op;
6436
225k
      } else {
6437
225k
          cur->c2 = op;
6438
225k
    if (op != NULL)
6439
225k
        op->parent = cur;
6440
225k
    op->c1 = last;
6441
225k
    if (last != NULL)
6442
225k
        last->parent = op;
6443
225k
    cur =op;
6444
225k
    last = NULL;
6445
225k
      }
6446
241k
  } else {
6447
2.14k
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL);
6448
2.14k
      if ((last != NULL) && (last != ret))
6449
1.22k
          xmlFreeDocElementContent(ctxt->myDoc, last);
6450
2.14k
      if (ret != NULL)
6451
2.14k
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6452
2.14k
      return(NULL);
6453
2.14k
  }
6454
1.79M
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6455
1.79M
        if (RAW == '(') {
6456
15.8k
            int newInputNr = ctxt->inputNr;
6457
6458
      /* Recurse on second child */
6459
15.8k
      NEXT;
6460
15.8k
      last = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6461
15.8k
                                                          depth + 1);
6462
15.8k
            if (last == NULL) {
6463
1.82k
    if (ret != NULL)
6464
1.82k
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6465
1.82k
    return(NULL);
6466
1.82k
            }
6467
1.77M
  } else {
6468
1.77M
      elem = xmlParseName(ctxt);
6469
1.77M
      if (elem == NULL) {
6470
1.12k
    xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6471
1.12k
    if (ret != NULL)
6472
1.12k
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6473
1.12k
    return(NULL);
6474
1.12k
      }
6475
1.77M
      last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6476
1.77M
      if (last == NULL) {
6477
48
                xmlErrMemory(ctxt);
6478
48
    if (ret != NULL)
6479
48
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6480
48
    return(NULL);
6481
48
      }
6482
1.77M
      if (RAW == '?') {
6483
90.5k
    last->ocur = XML_ELEMENT_CONTENT_OPT;
6484
90.5k
    NEXT;
6485
1.68M
      } else if (RAW == '*') {
6486
8.86k
    last->ocur = XML_ELEMENT_CONTENT_MULT;
6487
8.86k
    NEXT;
6488
1.67M
      } else if (RAW == '+') {
6489
18.8k
    last->ocur = XML_ELEMENT_CONTENT_PLUS;
6490
18.8k
    NEXT;
6491
1.65M
      } else {
6492
1.65M
    last->ocur = XML_ELEMENT_CONTENT_ONCE;
6493
1.65M
      }
6494
1.77M
  }
6495
1.79M
    }
6496
36.9k
    if ((cur != NULL) && (last != NULL)) {
6497
21.5k
        cur->c2 = last;
6498
21.5k
  if (last != NULL)
6499
21.5k
      last->parent = cur;
6500
21.5k
    }
6501
36.9k
#ifdef LIBXML_VALID_ENABLED
6502
36.9k
    if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6503
10
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6504
10
                         "Element content declaration doesn't start "
6505
10
                         "and stop in the same entity\n",
6506
10
                         NULL, NULL);
6507
10
    }
6508
36.9k
#endif
6509
36.9k
    NEXT;
6510
36.9k
    if (RAW == '?') {
6511
7.99k
  if (ret != NULL) {
6512
7.99k
      if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) ||
6513
7.91k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6514
171
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6515
7.82k
      else
6516
7.82k
          ret->ocur = XML_ELEMENT_CONTENT_OPT;
6517
7.99k
  }
6518
7.99k
  NEXT;
6519
28.9k
    } else if (RAW == '*') {
6520
4.91k
  if (ret != NULL) {
6521
4.91k
      ret->ocur = XML_ELEMENT_CONTENT_MULT;
6522
4.91k
      cur = ret;
6523
      /*
6524
       * Some normalization:
6525
       * (a | b* | c?)* == (a | b | c)*
6526
       */
6527
20.0k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6528
15.1k
    if ((cur->c1 != NULL) &&
6529
15.1k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6530
13.7k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT)))
6531
1.90k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6532
15.1k
    if ((cur->c2 != NULL) &&
6533
15.1k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6534
15.0k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT)))
6535
355
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6536
15.1k
    cur = cur->c2;
6537
15.1k
      }
6538
4.91k
  }
6539
4.91k
  NEXT;
6540
24.0k
    } else if (RAW == '+') {
6541
6.21k
  if (ret != NULL) {
6542
6.21k
      int found = 0;
6543
6544
6.21k
      if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) ||
6545
5.58k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6546
921
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6547
5.29k
      else
6548
5.29k
          ret->ocur = XML_ELEMENT_CONTENT_PLUS;
6549
      /*
6550
       * Some normalization:
6551
       * (a | b*)+ == (a | b)*
6552
       * (a | b?)+ == (a | b)*
6553
       */
6554
10.5k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6555
4.35k
    if ((cur->c1 != NULL) &&
6556
4.35k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6557
3.66k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) {
6558
1.22k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6559
1.22k
        found = 1;
6560
1.22k
    }
6561
4.35k
    if ((cur->c2 != NULL) &&
6562
4.35k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6563
4.27k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) {
6564
983
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6565
983
        found = 1;
6566
983
    }
6567
4.35k
    cur = cur->c2;
6568
4.35k
      }
6569
6.21k
      if (found)
6570
1.34k
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6571
6.21k
  }
6572
6.21k
  NEXT;
6573
6.21k
    }
6574
36.9k
    return(ret);
6575
42.1k
}
6576
6577
/**
6578
 * Parse the declaration for a Mixed Element content
6579
 * The leading '(' and spaces have been skipped in #xmlParseElementContentDecl
6580
 *
6581
 * @deprecated Internal function, don't use.
6582
 *
6583
 *     [47] children ::= (choice | seq) ('?' | '*' | '+')?
6584
 *
6585
 *     [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')?
6586
 *
6587
 *     [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')'
6588
 *
6589
 *     [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
6590
 *
6591
 * [ VC: Proper Group/PE Nesting ] applies to [49] and [50]
6592
 * TODO Parameter-entity replacement text must be properly nested
6593
 *  with parenthesized groups. That is to say, if either of the
6594
 *  opening or closing parentheses in a choice, seq, or Mixed
6595
 *  construct is contained in the replacement text for a parameter
6596
 *  entity, both must be contained in the same replacement text. For
6597
 *  interoperability, if a parameter-entity reference appears in a
6598
 *  choice, seq, or Mixed construct, its replacement text should not
6599
 *  be empty, and neither the first nor last non-blank character of
6600
 *  the replacement text should be a connector (| or ,).
6601
 *
6602
 * @param ctxt  an XML parser context
6603
 * @param inputchk  the input used for the current entity, needed for boundary checks
6604
 * @returns the tree of xmlElementContent describing the element
6605
 *          hierarchy.
6606
 */
6607
xmlElementContent *
6608
0
xmlParseElementChildrenContentDecl(xmlParserCtxt *ctxt, int inputchk) {
6609
    /* stub left for API/ABI compat */
6610
0
    return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1));
6611
0
}
6612
6613
/**
6614
 * Parse the declaration for an Element content either Mixed or Children,
6615
 * the cases EMPTY and ANY are handled directly in #xmlParseElementDecl
6616
 *
6617
 * @deprecated Internal function, don't use.
6618
 *
6619
 *     [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children
6620
 *
6621
 * @param ctxt  an XML parser context
6622
 * @param name  the name of the element being defined.
6623
 * @param result  the Element Content pointer will be stored here if any
6624
 * @returns an xmlElementTypeVal value or -1 on error
6625
 */
6626
6627
int
6628
xmlParseElementContentDecl(xmlParserCtxt *ctxt, const xmlChar *name,
6629
28.5k
                           xmlElementContent **result) {
6630
6631
28.5k
    xmlElementContentPtr tree = NULL;
6632
28.5k
    int openInputNr = ctxt->inputNr;
6633
28.5k
    int res;
6634
6635
28.5k
    *result = NULL;
6636
6637
28.5k
    if (RAW != '(') {
6638
0
  xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6639
0
    "xmlParseElementContentDecl : %s '(' expected\n", name);
6640
0
  return(-1);
6641
0
    }
6642
28.5k
    NEXT;
6643
28.5k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6644
28.5k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6645
10.4k
        tree = xmlParseElementMixedContentDecl(ctxt, openInputNr);
6646
10.4k
  res = XML_ELEMENT_TYPE_MIXED;
6647
18.1k
    } else {
6648
18.1k
        tree = xmlParseElementChildrenContentDeclPriv(ctxt, openInputNr, 1);
6649
18.1k
  res = XML_ELEMENT_TYPE_ELEMENT;
6650
18.1k
    }
6651
28.5k
    if (tree == NULL)
6652
5.15k
        return(-1);
6653
23.4k
    SKIP_BLANKS_PE;
6654
23.4k
    *result = tree;
6655
23.4k
    return(res);
6656
28.5k
}
6657
6658
/**
6659
 * Parse an element declaration. Always consumes '<!'.
6660
 *
6661
 * @deprecated Internal function, don't use.
6662
 *
6663
 *     [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>'
6664
 *
6665
 * [ VC: Unique Element Type Declaration ]
6666
 * No element type may be declared more than once
6667
 *
6668
 * @param ctxt  an XML parser context
6669
 * @returns the type of the element, or -1 in case of error
6670
 */
6671
int
6672
37.2k
xmlParseElementDecl(xmlParserCtxt *ctxt) {
6673
37.2k
    const xmlChar *name;
6674
37.2k
    int ret = -1;
6675
37.2k
    xmlElementContentPtr content  = NULL;
6676
6677
37.2k
    if ((CUR != '<') || (NXT(1) != '!'))
6678
0
        return(ret);
6679
37.2k
    SKIP(2);
6680
6681
    /* GROW; done in the caller */
6682
37.2k
    if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) {
6683
37.1k
#ifdef LIBXML_VALID_ENABLED
6684
37.1k
  int oldInputNr = ctxt->inputNr;
6685
37.1k
#endif
6686
6687
37.1k
  SKIP(7);
6688
37.1k
  if (SKIP_BLANKS_PE == 0) {
6689
335
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6690
335
               "Space required after 'ELEMENT'\n");
6691
335
      return(-1);
6692
335
  }
6693
36.8k
        name = xmlParseName(ctxt);
6694
36.8k
  if (name == NULL) {
6695
588
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6696
588
         "xmlParseElementDecl: no name for Element\n");
6697
588
      return(-1);
6698
588
  }
6699
36.2k
  if (SKIP_BLANKS_PE == 0) {
6700
6.57k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6701
6.57k
         "Space required after the element name\n");
6702
6.57k
  }
6703
36.2k
  if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) {
6704
4.63k
      SKIP(5);
6705
      /*
6706
       * Element must always be empty.
6707
       */
6708
4.63k
      ret = XML_ELEMENT_TYPE_EMPTY;
6709
31.6k
  } else if ((RAW == 'A') && (NXT(1) == 'N') &&
6710
1.48k
             (NXT(2) == 'Y')) {
6711
1.47k
      SKIP(3);
6712
      /*
6713
       * Element is a generic container.
6714
       */
6715
1.47k
      ret = XML_ELEMENT_TYPE_ANY;
6716
30.1k
  } else if (RAW == '(') {
6717
28.5k
      ret = xmlParseElementContentDecl(ctxt, name, &content);
6718
28.5k
            if (ret <= 0)
6719
5.15k
                return(-1);
6720
28.5k
  } else {
6721
      /*
6722
       * [ WFC: PEs in Internal Subset ] error handling.
6723
       */
6724
1.55k
            xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6725
1.55k
                  "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n");
6726
1.55k
      return(-1);
6727
1.55k
  }
6728
6729
29.5k
  SKIP_BLANKS_PE;
6730
6731
29.5k
  if (RAW != '>') {
6732
2.53k
      xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
6733
2.53k
      if (content != NULL) {
6734
2.11k
    xmlFreeDocElementContent(ctxt->myDoc, content);
6735
2.11k
      }
6736
27.0k
  } else {
6737
27.0k
#ifdef LIBXML_VALID_ENABLED
6738
27.0k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6739
4
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6740
4
                                 "Element declaration doesn't start and stop in"
6741
4
                                 " the same entity\n",
6742
4
                                 NULL, NULL);
6743
4
      }
6744
27.0k
#endif
6745
6746
27.0k
      NEXT;
6747
27.0k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6748
25.5k
    (ctxt->sax->elementDecl != NULL)) {
6749
25.5k
    if (content != NULL)
6750
20.1k
        content->parent = NULL;
6751
25.5k
          ctxt->sax->elementDecl(ctxt->userData, name, ret,
6752
25.5k
                           content);
6753
25.5k
    if ((content != NULL) && (content->parent == NULL)) {
6754
        /*
6755
         * this is a trick: if xmlAddElementDecl is called,
6756
         * instead of copying the full tree it is plugged directly
6757
         * if called from the parser. Avoid duplicating the
6758
         * interfaces or change the API/ABI
6759
         */
6760
3.48k
        xmlFreeDocElementContent(ctxt->myDoc, content);
6761
3.48k
    }
6762
25.5k
      } else if (content != NULL) {
6763
1.17k
    xmlFreeDocElementContent(ctxt->myDoc, content);
6764
1.17k
      }
6765
27.0k
  }
6766
29.5k
    }
6767
29.6k
    return(ret);
6768
37.2k
}
6769
6770
/**
6771
 * Parse a conditional section. Always consumes '<!['.
6772
 *
6773
 *     [61] conditionalSect ::= includeSect | ignoreSect
6774
 *     [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>'
6775
 *     [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>'
6776
 *     [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>'
6777
 *                                 Ignore)*
6778
 *     [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*)
6779
 * @param ctxt  an XML parser context
6780
 */
6781
6782
static void
6783
6.45k
xmlParseConditionalSections(xmlParserCtxtPtr ctxt) {
6784
6.45k
    size_t depth = 0;
6785
6.45k
    int isFreshPE = 0;
6786
6.45k
    int oldInputNr = ctxt->inputNr;
6787
6.45k
    int declInputNr = ctxt->inputNr;
6788
6789
14.9k
    while (!PARSER_STOPPED(ctxt)) {
6790
14.9k
        if (ctxt->input->cur >= ctxt->input->end) {
6791
1.37k
            if (ctxt->inputNr <= oldInputNr) {
6792
882
                xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6793
882
                return;
6794
882
            }
6795
6796
495
            xmlPopPE(ctxt);
6797
495
            declInputNr = ctxt->inputNr;
6798
13.5k
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6799
7.80k
            SKIP(3);
6800
7.80k
            SKIP_BLANKS_PE;
6801
6802
7.80k
            isFreshPE = 0;
6803
6804
7.80k
            if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) {
6805
4.63k
                SKIP(7);
6806
4.63k
                SKIP_BLANKS_PE;
6807
4.63k
                if (RAW != '[') {
6808
368
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6809
368
                    return;
6810
368
                }
6811
4.26k
#ifdef LIBXML_VALID_ENABLED
6812
4.26k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6813
94
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6814
94
                                     "All markup of the conditional section is"
6815
94
                                     " not in the same entity\n",
6816
94
                                     NULL, NULL);
6817
94
                }
6818
4.26k
#endif
6819
4.26k
                NEXT;
6820
6821
4.26k
                depth++;
6822
4.26k
            } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) {
6823
2.72k
                size_t ignoreDepth = 0;
6824
6825
2.72k
                SKIP(6);
6826
2.72k
                SKIP_BLANKS_PE;
6827
2.72k
                if (RAW != '[') {
6828
233
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6829
233
                    return;
6830
233
                }
6831
2.48k
#ifdef LIBXML_VALID_ENABLED
6832
2.48k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6833
38
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6834
38
                                     "All markup of the conditional section is"
6835
38
                                     " not in the same entity\n",
6836
38
                                     NULL, NULL);
6837
38
                }
6838
2.48k
#endif
6839
2.48k
                NEXT;
6840
6841
303k
                while (PARSER_STOPPED(ctxt) == 0) {
6842
303k
                    if (RAW == 0) {
6843
1.25k
                        xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL);
6844
1.25k
                        return;
6845
1.25k
                    }
6846
302k
                    if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6847
2.24k
                        SKIP(3);
6848
2.24k
                        ignoreDepth++;
6849
                        /* Check for integer overflow */
6850
2.24k
                        if (ignoreDepth == 0) {
6851
0
                            xmlErrMemory(ctxt);
6852
0
                            return;
6853
0
                        }
6854
300k
                    } else if ((RAW == ']') && (NXT(1) == ']') &&
6855
3.53k
                               (NXT(2) == '>')) {
6856
1.99k
                        SKIP(3);
6857
1.99k
                        if (ignoreDepth == 0)
6858
1.21k
                            break;
6859
780
                        ignoreDepth--;
6860
298k
                    } else {
6861
298k
                        NEXT;
6862
298k
                    }
6863
302k
                }
6864
6865
1.23k
#ifdef LIBXML_VALID_ENABLED
6866
1.23k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6867
34
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6868
34
                                     "All markup of the conditional section is"
6869
34
                                     " not in the same entity\n",
6870
34
                                     NULL, NULL);
6871
34
                }
6872
1.23k
#endif
6873
1.23k
            } else {
6874
445
                xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL);
6875
445
                return;
6876
445
            }
6877
7.80k
        } else if ((depth > 0) &&
6878
5.76k
                   (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) {
6879
2.14k
            if (isFreshPE) {
6880
6
                xmlFatalErrMsg(ctxt, XML_ERR_CONDSEC_INVALID,
6881
6
                               "Parameter entity must match "
6882
6
                               "extSubsetDecl\n");
6883
6
                return;
6884
6
            }
6885
6886
2.14k
            depth--;
6887
2.14k
#ifdef LIBXML_VALID_ENABLED
6888
2.14k
            if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6889
10
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6890
10
                                 "All markup of the conditional section is not"
6891
10
                                 " in the same entity\n",
6892
10
                                 NULL, NULL);
6893
10
            }
6894
2.14k
#endif
6895
2.14k
            SKIP(3);
6896
3.62k
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
6897
1.01k
            isFreshPE = 0;
6898
1.01k
            xmlParseMarkupDecl(ctxt);
6899
2.60k
        } else if (RAW == '%') {
6900
2.50k
            xmlParsePERefInternal(ctxt, 1);
6901
2.50k
            if (ctxt->inputNr > declInputNr) {
6902
504
                isFreshPE = 1;
6903
504
                declInputNr = ctxt->inputNr;
6904
504
            }
6905
2.50k
        } else {
6906
100
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6907
100
            return;
6908
100
        }
6909
6910
11.6k
        if (depth == 0)
6911
3.13k
            break;
6912
6913
8.52k
        SKIP_BLANKS;
6914
8.52k
        SHRINK;
6915
8.52k
        GROW;
6916
8.52k
    }
6917
6.45k
}
6918
6919
/**
6920
 * Parse markup declarations. Always consumes '<!' or '<?'.
6921
 *
6922
 * @deprecated Internal function, don't use.
6923
 *
6924
 *     [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl |
6925
 *                         NotationDecl | PI | Comment
6926
 *
6927
 * [ VC: Proper Declaration/PE Nesting ]
6928
 * Parameter-entity replacement text must be properly nested with
6929
 * markup declarations. That is to say, if either the first character
6930
 * or the last character of a markup declaration (markupdecl above) is
6931
 * contained in the replacement text for a parameter-entity reference,
6932
 * both must be contained in the same replacement text.
6933
 *
6934
 * [ WFC: PEs in Internal Subset ]
6935
 * In the internal DTD subset, parameter-entity references can occur
6936
 * only where markup declarations can occur, not within markup declarations.
6937
 * (This does not apply to references that occur in external parameter
6938
 * entities or to the external subset.)
6939
 *
6940
 * @param ctxt  an XML parser context
6941
 */
6942
void
6943
573k
xmlParseMarkupDecl(xmlParserCtxt *ctxt) {
6944
573k
    GROW;
6945
573k
    if (CUR == '<') {
6946
573k
        if (NXT(1) == '!') {
6947
565k
      switch (NXT(2)) {
6948
132k
          case 'E':
6949
132k
        if (NXT(3) == 'L')
6950
37.2k
      xmlParseElementDecl(ctxt);
6951
95.2k
        else if (NXT(3) == 'N')
6952
95.1k
      xmlParseEntityDecl(ctxt);
6953
75
                    else
6954
75
                        SKIP(2);
6955
132k
        break;
6956
90.8k
          case 'A':
6957
90.8k
        xmlParseAttributeListDecl(ctxt);
6958
90.8k
        break;
6959
8.10k
          case 'N':
6960
8.10k
        xmlParseNotationDecl(ctxt);
6961
8.10k
        break;
6962
319k
          case '-':
6963
319k
        xmlParseComment(ctxt);
6964
319k
        break;
6965
14.1k
    default:
6966
14.1k
                    xmlFatalErr(ctxt,
6967
14.1k
                                ctxt->inSubset == 2 ?
6968
5.38k
                                    XML_ERR_EXT_SUBSET_NOT_FINISHED :
6969
14.1k
                                    XML_ERR_INT_SUBSET_NOT_FINISHED,
6970
14.1k
                                NULL);
6971
14.1k
                    SKIP(2);
6972
14.1k
        break;
6973
565k
      }
6974
565k
  } else if (NXT(1) == '?') {
6975
8.05k
      xmlParsePI(ctxt);
6976
8.05k
  }
6977
573k
    }
6978
573k
}
6979
6980
/**
6981
 * Parse an XML declaration header for external entities
6982
 *
6983
 * @deprecated Internal function, don't use.
6984
 *
6985
 *     [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>'
6986
 * @param ctxt  an XML parser context
6987
 */
6988
6989
void
6990
38.7k
xmlParseTextDecl(xmlParserCtxt *ctxt) {
6991
38.7k
    xmlChar *version;
6992
6993
    /*
6994
     * We know that '<?xml' is here.
6995
     */
6996
38.7k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
6997
38.7k
  SKIP(5);
6998
38.7k
    } else {
6999
10
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL);
7000
10
  return;
7001
10
    }
7002
7003
38.7k
    if (SKIP_BLANKS == 0) {
7004
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7005
0
           "Space needed after '<?xml'\n");
7006
0
    }
7007
7008
    /*
7009
     * We may have the VersionInfo here.
7010
     */
7011
38.7k
    version = xmlParseVersionInfo(ctxt);
7012
38.7k
    if (version == NULL) {
7013
20.0k
  version = xmlCharStrdup(XML_DEFAULT_VERSION);
7014
20.0k
        if (version == NULL) {
7015
15
            xmlErrMemory(ctxt);
7016
15
            return;
7017
15
        }
7018
20.0k
    } else {
7019
18.6k
  if (SKIP_BLANKS == 0) {
7020
1.34k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7021
1.34k
               "Space needed here\n");
7022
1.34k
  }
7023
18.6k
    }
7024
38.7k
    ctxt->input->version = version;
7025
7026
    /*
7027
     * We must have the encoding declaration
7028
     */
7029
38.7k
    xmlParseEncodingDecl(ctxt);
7030
7031
38.7k
    SKIP_BLANKS;
7032
38.7k
    if ((RAW == '?') && (NXT(1) == '>')) {
7033
3.41k
        SKIP(2);
7034
35.3k
    } else if (RAW == '>') {
7035
        /* Deprecated old WD ... */
7036
645
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7037
645
  NEXT;
7038
34.6k
    } else {
7039
34.6k
        int c;
7040
7041
34.6k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7042
839M
        while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) {
7043
839M
            NEXT;
7044
839M
            if (c == '>')
7045
12.2k
                break;
7046
839M
        }
7047
34.6k
    }
7048
38.7k
}
7049
7050
/**
7051
 * Parse Markup declarations from an external subset
7052
 *
7053
 * @deprecated Internal function, don't use.
7054
 *
7055
 *     [30] extSubset ::= textDecl? extSubsetDecl
7056
 *
7057
 *     [31] extSubsetDecl ::= (markupdecl | conditionalSect |
7058
 *                             PEReference | S) *
7059
 * @param ctxt  an XML parser context
7060
 * @param publicId  the public identifier
7061
 * @param systemId  the system identifier (URL)
7062
 */
7063
void
7064
xmlParseExternalSubset(xmlParserCtxt *ctxt, const xmlChar *publicId,
7065
3.94k
                       const xmlChar *systemId) {
7066
3.94k
    int oldInputNr;
7067
7068
3.94k
    xmlCtxtInitializeLate(ctxt);
7069
7070
3.94k
    xmlDetectEncoding(ctxt);
7071
7072
3.94k
    if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) {
7073
340
  xmlParseTextDecl(ctxt);
7074
340
    }
7075
3.94k
    if (ctxt->myDoc == NULL) {
7076
0
        ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0");
7077
0
  if (ctxt->myDoc == NULL) {
7078
0
      xmlErrMemory(ctxt);
7079
0
      return;
7080
0
  }
7081
0
  ctxt->myDoc->properties = XML_DOC_INTERNAL;
7082
0
    }
7083
3.94k
    if ((ctxt->myDoc->intSubset == NULL) &&
7084
723
        (xmlCreateIntSubset(ctxt->myDoc, NULL, publicId, systemId) == NULL)) {
7085
7
        xmlErrMemory(ctxt);
7086
7
    }
7087
7088
3.94k
    ctxt->inSubset = 2;
7089
3.94k
    oldInputNr = ctxt->inputNr;
7090
7091
3.94k
    SKIP_BLANKS;
7092
200k
    while (!PARSER_STOPPED(ctxt)) {
7093
200k
        if (ctxt->input->cur >= ctxt->input->end) {
7094
2.83k
            if (ctxt->inputNr <= oldInputNr) {
7095
1.54k
                xmlParserCheckEOF(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED);
7096
1.54k
                break;
7097
1.54k
            }
7098
7099
1.28k
            xmlPopPE(ctxt);
7100
197k
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
7101
1.58k
            xmlParseConditionalSections(ctxt);
7102
196k
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
7103
191k
            xmlParseMarkupDecl(ctxt);
7104
191k
        } else if (RAW == '%') {
7105
2.44k
            xmlParsePERefInternal(ctxt, 1);
7106
2.44k
        } else {
7107
1.99k
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
7108
7109
2.33k
            while (ctxt->inputNr > oldInputNr)
7110
344
                xmlPopPE(ctxt);
7111
1.99k
            break;
7112
1.99k
        }
7113
197k
        SKIP_BLANKS;
7114
197k
        SHRINK;
7115
197k
        GROW;
7116
197k
    }
7117
3.94k
}
7118
7119
/**
7120
 * Parse and handle entity references in content, depending on the SAX
7121
 * interface, this may end-up in a call to character() if this is a
7122
 * CharRef, a predefined entity, if there is no reference() callback.
7123
 * or if the parser was asked to switch to that mode.
7124
 *
7125
 * @deprecated Internal function, don't use.
7126
 *
7127
 * Always consumes '&'.
7128
 *
7129
 *     [67] Reference ::= EntityRef | CharRef
7130
 * @param ctxt  an XML parser context
7131
 */
7132
void
7133
198k
xmlParseReference(xmlParserCtxt *ctxt) {
7134
198k
    xmlEntityPtr ent = NULL;
7135
198k
    const xmlChar *name;
7136
198k
    xmlChar *val;
7137
7138
198k
    if (RAW != '&')
7139
0
        return;
7140
7141
    /*
7142
     * Simple case of a CharRef
7143
     */
7144
198k
    if (NXT(1) == '#') {
7145
50.6k
  int i = 0;
7146
50.6k
  xmlChar out[16];
7147
50.6k
  int value = xmlParseCharRef(ctxt);
7148
7149
50.6k
  if (value == 0)
7150
21.2k
      return;
7151
7152
        /*
7153
         * Just encode the value in UTF-8
7154
         */
7155
29.3k
        COPY_BUF(out, i, value);
7156
29.3k
        out[i] = 0;
7157
29.3k
        if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7158
29.3k
            (!ctxt->disableSAX))
7159
13.7k
            ctxt->sax->characters(ctxt->userData, out, i);
7160
29.3k
  return;
7161
50.6k
    }
7162
7163
    /*
7164
     * We are seeing an entity reference
7165
     */
7166
148k
    name = xmlParseEntityRefInternal(ctxt);
7167
148k
    if (name == NULL)
7168
33.4k
        return;
7169
114k
    ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0);
7170
114k
    if (ent == NULL) {
7171
        /*
7172
         * Create a reference for undeclared entities.
7173
         */
7174
55.3k
        if ((ctxt->replaceEntities == 0) &&
7175
43.7k
            (ctxt->sax != NULL) &&
7176
43.7k
            (ctxt->disableSAX == 0) &&
7177
41.3k
            (ctxt->sax->reference != NULL)) {
7178
41.3k
            ctxt->sax->reference(ctxt->userData, name);
7179
41.3k
        }
7180
55.3k
        return;
7181
55.3k
    }
7182
59.3k
    if (!ctxt->wellFormed)
7183
20.7k
  return;
7184
7185
    /* special case of predefined entities */
7186
38.5k
    if ((ent->name == NULL) ||
7187
38.5k
        (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
7188
708
  val = ent->content;
7189
708
  if (val == NULL) return;
7190
  /*
7191
   * inline the entity.
7192
   */
7193
708
  if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7194
708
      (!ctxt->disableSAX))
7195
708
      ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val));
7196
708
  return;
7197
708
    }
7198
7199
    /*
7200
     * Some users try to parse entities on their own and used to set
7201
     * the renamed "checked" member. Fix the flags to cover this
7202
     * case.
7203
     */
7204
37.8k
    if (((ent->flags & XML_ENT_PARSED) == 0) && (ent->children != NULL))
7205
0
        ent->flags |= XML_ENT_PARSED;
7206
7207
    /*
7208
     * The first reference to the entity trigger a parsing phase
7209
     * where the ent->children is filled with the result from
7210
     * the parsing.
7211
     * Note: external parsed entities will not be loaded, it is not
7212
     * required for a non-validating parser, unless the parsing option
7213
     * of validating, or substituting entities were given. Doing so is
7214
     * far more secure as the parser will only process data coming from
7215
     * the document entity by default.
7216
     *
7217
     * FIXME: This doesn't work correctly since entities can be
7218
     * expanded with different namespace declarations in scope.
7219
     * For example:
7220
     *
7221
     * <!DOCTYPE doc [
7222
     *   <!ENTITY ent "<ns:elem/>">
7223
     * ]>
7224
     * <doc>
7225
     *   <decl1 xmlns:ns="urn:ns1">
7226
     *     &ent;
7227
     *   </decl1>
7228
     *   <decl2 xmlns:ns="urn:ns2">
7229
     *     &ent;
7230
     *   </decl2>
7231
     * </doc>
7232
     *
7233
     * Proposed fix:
7234
     *
7235
     * - Ignore current namespace declarations when parsing the
7236
     *   entity. If a prefix can't be resolved, don't report an error
7237
     *   but mark it as unresolved.
7238
     * - Try to resolve these prefixes when expanding the entity.
7239
     *   This will require a specialized version of xmlStaticCopyNode
7240
     *   which can also make use of the namespace hash table to avoid
7241
     *   quadratic behavior.
7242
     *
7243
     * Alternatively, we could simply reparse the entity on each
7244
     * expansion like we already do with custom SAX callbacks.
7245
     * External entity content should be cached in this case.
7246
     */
7247
37.8k
    if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) ||
7248
13.9k
        (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
7249
13.6k
         ((ctxt->replaceEntities) ||
7250
37.0k
          (ctxt->validate)))) {
7251
37.0k
        if ((ent->flags & XML_ENT_PARSED) == 0) {
7252
7.12k
            xmlCtxtParseEntity(ctxt, ent);
7253
29.9k
        } else if (ent->children == NULL) {
7254
            /*
7255
             * Probably running in SAX mode and the callbacks don't
7256
             * build the entity content. Parse the entity again.
7257
             *
7258
             * This will also be triggered in normal tree builder mode
7259
             * if an entity happens to be empty, causing unnecessary
7260
             * reloads. It's hard to come up with a reliable check in
7261
             * which mode we're running.
7262
             */
7263
3.15k
            xmlCtxtParseEntity(ctxt, ent);
7264
3.15k
        }
7265
37.0k
    }
7266
7267
    /*
7268
     * We also check for amplification if entities aren't substituted.
7269
     * They might be expanded later.
7270
     */
7271
37.8k
    if (xmlParserEntityCheck(ctxt, ent->expandedSize))
7272
136
        return;
7273
7274
37.7k
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
7275
1.80k
        return;
7276
7277
35.9k
    if (ctxt->replaceEntities == 0) {
7278
  /*
7279
   * Create a reference
7280
   */
7281
5.21k
        if (ctxt->sax->reference != NULL)
7282
5.21k
      ctxt->sax->reference(ctxt->userData, ent->name);
7283
30.7k
    } else if ((ent->children != NULL) && (ctxt->node != NULL)) {
7284
27.6k
        xmlNodePtr copy, cur;
7285
7286
        /*
7287
         * Seems we are generating the DOM content, copy the tree
7288
   */
7289
27.6k
        cur = ent->children;
7290
7291
        /*
7292
         * Handle first text node with SAX to coalesce text efficiently
7293
         */
7294
27.6k
        if ((cur->type == XML_TEXT_NODE) ||
7295
18.4k
            (cur->type == XML_CDATA_SECTION_NODE)) {
7296
9.26k
            int len = xmlStrlen(cur->content);
7297
7298
9.26k
            if ((cur->type == XML_TEXT_NODE) ||
7299
9.15k
                (ctxt->options & XML_PARSE_NOCDATA)) {
7300
9.15k
                if (ctxt->sax->characters != NULL)
7301
9.15k
                    ctxt->sax->characters(ctxt->userData, cur->content, len);
7302
9.15k
            } else {
7303
108
                if (ctxt->sax->cdataBlock != NULL)
7304
108
                    ctxt->sax->cdataBlock(ctxt->userData, cur->content, len);
7305
108
            }
7306
7307
9.26k
            cur = cur->next;
7308
9.26k
        }
7309
7310
462k
        while (cur != NULL) {
7311
439k
            xmlNodePtr last;
7312
7313
            /*
7314
             * Handle last text node with SAX to coalesce text efficiently
7315
             */
7316
439k
            if ((cur->next == NULL) &&
7317
22.7k
                ((cur->type == XML_TEXT_NODE) ||
7318
18.8k
                 (cur->type == XML_CDATA_SECTION_NODE))) {
7319
4.09k
                int len = xmlStrlen(cur->content);
7320
7321
4.09k
                if ((cur->type == XML_TEXT_NODE) ||
7322
3.86k
                    (ctxt->options & XML_PARSE_NOCDATA)) {
7323
3.86k
                    if (ctxt->sax->characters != NULL)
7324
3.86k
                        ctxt->sax->characters(ctxt->userData, cur->content,
7325
3.86k
                                              len);
7326
3.86k
                } else {
7327
225
                    if (ctxt->sax->cdataBlock != NULL)
7328
225
                        ctxt->sax->cdataBlock(ctxt->userData, cur->content,
7329
225
                                              len);
7330
225
                }
7331
7332
4.09k
                break;
7333
4.09k
            }
7334
7335
            /*
7336
             * Reset coalesce buffer stats only for non-text nodes.
7337
             */
7338
435k
            ctxt->nodemem = 0;
7339
435k
            ctxt->nodelen = 0;
7340
7341
435k
            copy = xmlDocCopyNode(cur, ctxt->myDoc, 1);
7342
7343
435k
            if (copy == NULL) {
7344
313
                xmlErrMemory(ctxt);
7345
313
                break;
7346
313
            }
7347
7348
435k
            if (ctxt->parseMode == XML_PARSE_READER) {
7349
                /* Needed for reader */
7350
0
                copy->extra = cur->extra;
7351
                /* Maybe needed for reader */
7352
0
                copy->_private = cur->_private;
7353
0
            }
7354
7355
435k
            copy->parent = ctxt->node;
7356
435k
            last = ctxt->node->last;
7357
435k
            if (last == NULL) {
7358
359
                ctxt->node->children = copy;
7359
434k
            } else {
7360
434k
                last->next = copy;
7361
434k
                copy->prev = last;
7362
434k
            }
7363
435k
            ctxt->node->last = copy;
7364
7365
435k
            cur = cur->next;
7366
435k
        }
7367
27.6k
    }
7368
35.9k
}
7369
7370
static void
7371
288k
xmlHandleUndeclaredEntity(xmlParserCtxtPtr ctxt, const xmlChar *name) {
7372
    /*
7373
     * [ WFC: Entity Declared ]
7374
     * In a document without any DTD, a document with only an
7375
     * internal DTD subset which contains no parameter entity
7376
     * references, or a document with "standalone='yes'", the
7377
     * Name given in the entity reference must match that in an
7378
     * entity declaration, except that well-formed documents
7379
     * need not declare any of the following entities: amp, lt,
7380
     * gt, apos, quot.
7381
     * The declaration of a parameter entity must precede any
7382
     * reference to it.
7383
     * Similarly, the declaration of a general entity must
7384
     * precede any reference to it which appears in a default
7385
     * value in an attribute-list declaration. Note that if
7386
     * entities are declared in the external subset or in
7387
     * external parameter entities, a non-validating processor
7388
     * is not obligated to read and process their declarations;
7389
     * for such documents, the rule that an entity must be
7390
     * declared is a well-formedness constraint only if
7391
     * standalone='yes'.
7392
     */
7393
288k
    if ((ctxt->standalone == 1) ||
7394
287k
        ((ctxt->hasExternalSubset == 0) &&
7395
282k
         (ctxt->hasPErefs == 0))) {
7396
263k
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY,
7397
263k
                          "Entity '%s' not defined\n", name);
7398
263k
#ifdef LIBXML_VALID_ENABLED
7399
263k
    } else if (ctxt->validate) {
7400
        /*
7401
         * [ VC: Entity Declared ]
7402
         * In a document with an external subset or external
7403
         * parameter entities with "standalone='no'", ...
7404
         * ... The declaration of a parameter entity must
7405
         * precede any reference to it...
7406
         */
7407
15.0k
        xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY,
7408
15.0k
                         "Entity '%s' not defined\n", name, NULL);
7409
15.0k
#endif
7410
15.0k
    } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) ||
7411
2.39k
               ((ctxt->replaceEntities) &&
7412
7.89k
                ((ctxt->options & XML_PARSE_NO_XXE) == 0))) {
7413
        /*
7414
         * Also raise a non-fatal error
7415
         *
7416
         * - if the external subset is loaded and all entity declarations
7417
         *   should be available, or
7418
         * - entity substition was requested without restricting
7419
         *   external entity access.
7420
         */
7421
7.89k
        xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY,
7422
7.89k
                     "Entity '%s' not defined\n", name);
7423
7.89k
    } else {
7424
1.52k
        xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7425
1.52k
                      "Entity '%s' not defined\n", name, NULL);
7426
1.52k
    }
7427
7428
288k
    ctxt->valid = 0;
7429
288k
}
7430
7431
static xmlEntityPtr
7432
2.39M
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) {
7433
2.39M
    xmlEntityPtr ent = NULL;
7434
7435
    /*
7436
     * Predefined entities override any extra definition
7437
     */
7438
2.39M
    if ((ctxt->options & XML_PARSE_OLDSAX) == 0) {
7439
1.54M
        ent = xmlGetPredefinedEntity(name);
7440
1.54M
        if (ent != NULL)
7441
644k
            return(ent);
7442
1.54M
    }
7443
7444
    /*
7445
     * Ask first SAX for entity resolution, otherwise try the
7446
     * entities which may have stored in the parser context.
7447
     */
7448
1.74M
    if (ctxt->sax != NULL) {
7449
1.74M
  if (ctxt->sax->getEntity != NULL)
7450
1.74M
      ent = ctxt->sax->getEntity(ctxt->userData, name);
7451
1.74M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7452
3.58k
      (ctxt->options & XML_PARSE_OLDSAX))
7453
452
      ent = xmlGetPredefinedEntity(name);
7454
1.74M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7455
3.58k
      (ctxt->userData==ctxt)) {
7456
3.58k
      ent = xmlSAX2GetEntity(ctxt, name);
7457
3.58k
  }
7458
1.74M
    }
7459
7460
1.74M
    if (ent == NULL) {
7461
270k
        xmlHandleUndeclaredEntity(ctxt, name);
7462
270k
    }
7463
7464
    /*
7465
     * [ WFC: Parsed Entity ]
7466
     * An entity reference must not contain the name of an
7467
     * unparsed entity
7468
     */
7469
1.47M
    else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) {
7470
389
  xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY,
7471
389
     "Entity reference to unparsed entity %s\n", name);
7472
389
        ent = NULL;
7473
389
    }
7474
7475
    /*
7476
     * [ WFC: No External Entity References ]
7477
     * Attribute values cannot contain direct or indirect
7478
     * entity references to external entities.
7479
     */
7480
1.47M
    else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) {
7481
23.9k
        if (inAttr) {
7482
1.40k
            xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL,
7483
1.40k
                 "Attribute references external entity '%s'\n", name);
7484
1.40k
            ent = NULL;
7485
1.40k
        }
7486
23.9k
    }
7487
7488
1.74M
    return(ent);
7489
2.39M
}
7490
7491
/**
7492
 * Parse an entity reference. Always consumes '&'.
7493
 *
7494
 *     [68] EntityRef ::= '&' Name ';'
7495
 *
7496
 * @param ctxt  an XML parser context
7497
 * @returns the name, or NULL in case of error.
7498
 */
7499
static const xmlChar *
7500
541k
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) {
7501
541k
    const xmlChar *name;
7502
7503
541k
    GROW;
7504
7505
541k
    if (RAW != '&')
7506
0
        return(NULL);
7507
541k
    NEXT;
7508
541k
    name = xmlParseName(ctxt);
7509
541k
    if (name == NULL) {
7510
55.3k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7511
55.3k
           "xmlParseEntityRef: no name\n");
7512
55.3k
        return(NULL);
7513
55.3k
    }
7514
486k
    if (RAW != ';') {
7515
40.9k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7516
40.9k
  return(NULL);
7517
40.9k
    }
7518
445k
    NEXT;
7519
7520
445k
    return(name);
7521
486k
}
7522
7523
/**
7524
 * @deprecated Internal function, don't use.
7525
 *
7526
 * @param ctxt  an XML parser context
7527
 * @returns the xmlEntity if found, or NULL otherwise.
7528
 */
7529
xmlEntity *
7530
0
xmlParseEntityRef(xmlParserCtxt *ctxt) {
7531
0
    const xmlChar *name;
7532
7533
0
    if (ctxt == NULL)
7534
0
        return(NULL);
7535
7536
0
    name = xmlParseEntityRefInternal(ctxt);
7537
0
    if (name == NULL)
7538
0
        return(NULL);
7539
7540
0
    return(xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 0));
7541
0
}
7542
7543
/**
7544
 * Parse ENTITY references declarations, but this version parses it from
7545
 * a string value.
7546
 *
7547
 *     [68] EntityRef ::= '&' Name ';'
7548
 *
7549
 * [ WFC: Entity Declared ]
7550
 * In a document without any DTD, a document with only an internal DTD
7551
 * subset which contains no parameter entity references, or a document
7552
 * with "standalone='yes'", the Name given in the entity reference
7553
 * must match that in an entity declaration, except that well-formed
7554
 * documents need not declare any of the following entities: amp, lt,
7555
 * gt, apos, quot.  The declaration of a parameter entity must precede
7556
 * any reference to it.  Similarly, the declaration of a general entity
7557
 * must precede any reference to it which appears in a default value in an
7558
 * attribute-list declaration. Note that if entities are declared in the
7559
 * external subset or in external parameter entities, a non-validating
7560
 * processor is not obligated to read and process their declarations;
7561
 * for such documents, the rule that an entity must be declared is a
7562
 * well-formedness constraint only if standalone='yes'.
7563
 *
7564
 * [ WFC: Parsed Entity ]
7565
 * An entity reference must not contain the name of an unparsed entity
7566
 *
7567
 * @param ctxt  an XML parser context
7568
 * @param str  a pointer to an index in the string
7569
 * @returns the xmlEntity if found, or NULL otherwise. The str pointer
7570
 * is updated to the current location in the string.
7571
 */
7572
static xmlChar *
7573
1.94M
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) {
7574
1.94M
    xmlChar *name;
7575
1.94M
    const xmlChar *ptr;
7576
1.94M
    xmlChar cur;
7577
7578
1.94M
    if ((str == NULL) || (*str == NULL))
7579
0
        return(NULL);
7580
1.94M
    ptr = *str;
7581
1.94M
    cur = *ptr;
7582
1.94M
    if (cur != '&')
7583
0
  return(NULL);
7584
7585
1.94M
    ptr++;
7586
1.94M
    name = xmlParseStringName(ctxt, &ptr);
7587
1.94M
    if (name == NULL) {
7588
130
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7589
130
           "xmlParseStringEntityRef: no name\n");
7590
130
  *str = ptr;
7591
130
  return(NULL);
7592
130
    }
7593
1.94M
    if (*ptr != ';') {
7594
44
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7595
44
        xmlFree(name);
7596
44
  *str = ptr;
7597
44
  return(NULL);
7598
44
    }
7599
1.94M
    ptr++;
7600
7601
1.94M
    *str = ptr;
7602
1.94M
    return(name);
7603
1.94M
}
7604
7605
/**
7606
 * Parse a parameter entity reference. Always consumes '%'.
7607
 *
7608
 * The entity content is handled directly by pushing it's content as
7609
 * a new input stream.
7610
 *
7611
 *     [69] PEReference ::= '%' Name ';'
7612
 *
7613
 * [ WFC: No Recursion ]
7614
 * A parsed entity must not contain a recursive
7615
 * reference to itself, either directly or indirectly.
7616
 *
7617
 * [ WFC: Entity Declared ]
7618
 * In a document without any DTD, a document with only an internal DTD
7619
 * subset which contains no parameter entity references, or a document
7620
 * with "standalone='yes'", ...  ... The declaration of a parameter
7621
 * entity must precede any reference to it...
7622
 *
7623
 * [ VC: Entity Declared ]
7624
 * In a document with an external subset or external parameter entities
7625
 * with "standalone='no'", ...  ... The declaration of a parameter entity
7626
 * must precede any reference to it...
7627
 *
7628
 * [ WFC: In DTD ]
7629
 * Parameter-entity references may only appear in the DTD.
7630
 * NOTE: misleading but this is handled.
7631
 *
7632
 * @param ctxt  an XML parser context
7633
 * @param markupDecl  whether the PERef starts a markup declaration
7634
 */
7635
static void
7636
156k
xmlParsePERefInternal(xmlParserCtxt *ctxt, int markupDecl) {
7637
156k
    const xmlChar *name;
7638
156k
    xmlEntityPtr entity = NULL;
7639
156k
    xmlParserInputPtr input;
7640
7641
156k
    if (RAW != '%')
7642
0
        return;
7643
156k
    NEXT;
7644
156k
    name = xmlParseName(ctxt);
7645
156k
    if (name == NULL) {
7646
14.7k
  xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n");
7647
14.7k
  return;
7648
14.7k
    }
7649
141k
    if (RAW != ';') {
7650
9.92k
  xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL);
7651
9.92k
        return;
7652
9.92k
    }
7653
7654
131k
    NEXT;
7655
7656
    /* Must be set before xmlHandleUndeclaredEntity */
7657
131k
    ctxt->hasPErefs = 1;
7658
7659
    /*
7660
     * Request the entity from SAX
7661
     */
7662
131k
    if ((ctxt->sax != NULL) &&
7663
131k
  (ctxt->sax->getParameterEntity != NULL))
7664
131k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7665
7666
131k
    if (entity == NULL) {
7667
14.9k
        xmlHandleUndeclaredEntity(ctxt, name);
7668
116k
    } else {
7669
  /*
7670
   * Internal checking in case the entity quest barfed
7671
   */
7672
116k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7673
76.9k
      (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) {
7674
0
      xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7675
0
      "Internal: %%%s; is not a parameter entity\n",
7676
0
        name, NULL);
7677
116k
  } else {
7678
116k
      if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) &&
7679
76.9k
                ((ctxt->options & XML_PARSE_NO_XXE) ||
7680
76.7k
     (((ctxt->loadsubset & ~XML_SKIP_IDS) == 0) &&
7681
26.4k
      (ctxt->replaceEntities == 0) &&
7682
3.18k
      (ctxt->validate == 0))))
7683
1.36k
    return;
7684
7685
115k
            if (entity->flags & XML_ENT_EXPANDING) {
7686
35
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
7687
35
                return;
7688
35
            }
7689
7690
115k
      input = xmlNewEntityInputStream(ctxt, entity);
7691
115k
      if (xmlCtxtPushInput(ctxt, input) < 0) {
7692
3.08k
                xmlFreeInputStream(input);
7693
3.08k
    return;
7694
3.08k
            }
7695
7696
112k
            entity->flags |= XML_ENT_EXPANDING;
7697
7698
112k
            if (markupDecl)
7699
93.7k
                input->flags |= XML_INPUT_MARKUP_DECL;
7700
7701
112k
            GROW;
7702
7703
112k
      if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
7704
72.5k
                xmlDetectEncoding(ctxt);
7705
7706
72.5k
                if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
7707
14.2k
                    (IS_BLANK_CH(NXT(5)))) {
7708
13.5k
                    xmlParseTextDecl(ctxt);
7709
13.5k
                }
7710
72.5k
            }
7711
112k
  }
7712
116k
    }
7713
131k
}
7714
7715
/**
7716
 * Parse a parameter entity reference.
7717
 *
7718
 * @deprecated Internal function, don't use.
7719
 *
7720
 * @param ctxt  an XML parser context
7721
 */
7722
void
7723
0
xmlParsePEReference(xmlParserCtxt *ctxt) {
7724
0
    xmlParsePERefInternal(ctxt, 0);
7725
0
}
7726
7727
/**
7728
 * Load the content of an entity.
7729
 *
7730
 * @param ctxt  an XML parser context
7731
 * @param entity  an unloaded system entity
7732
 * @returns 0 in case of success and -1 in case of failure
7733
 */
7734
static int
7735
35.0k
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) {
7736
35.0k
    xmlParserInputPtr oldinput, input = NULL;
7737
35.0k
    xmlParserInputPtr *oldinputTab;
7738
35.0k
    xmlChar *oldencoding;
7739
35.0k
    xmlChar *content = NULL;
7740
35.0k
    xmlResourceType rtype;
7741
35.0k
    size_t length, i;
7742
35.0k
    int oldinputNr, oldinputMax;
7743
35.0k
    int ret = -1;
7744
35.0k
    int res;
7745
7746
35.0k
    if ((ctxt == NULL) || (entity == NULL) ||
7747
35.0k
        ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) &&
7748
0
   (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) ||
7749
35.0k
  (entity->content != NULL)) {
7750
0
  xmlFatalErr(ctxt, XML_ERR_ARGUMENT,
7751
0
              "xmlLoadEntityContent parameter error");
7752
0
        return(-1);
7753
0
    }
7754
7755
35.0k
    if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY)
7756
35.0k
        rtype = XML_RESOURCE_PARAMETER_ENTITY;
7757
0
    else
7758
0
        rtype = XML_RESOURCE_GENERAL_ENTITY;
7759
7760
35.0k
    input = xmlLoadResource(ctxt, (char *) entity->URI,
7761
35.0k
                            (char *) entity->ExternalID, rtype);
7762
35.0k
    if (input == NULL)
7763
1.25k
        return(-1);
7764
7765
33.8k
    oldinput = ctxt->input;
7766
33.8k
    oldinputNr = ctxt->inputNr;
7767
33.8k
    oldinputMax = ctxt->inputMax;
7768
33.8k
    oldinputTab = ctxt->inputTab;
7769
33.8k
    oldencoding = ctxt->encoding;
7770
7771
33.8k
    ctxt->input = NULL;
7772
33.8k
    ctxt->inputNr = 0;
7773
33.8k
    ctxt->inputMax = 1;
7774
33.8k
    ctxt->encoding = NULL;
7775
33.8k
    ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr));
7776
33.8k
    if (ctxt->inputTab == NULL) {
7777
11
        xmlErrMemory(ctxt);
7778
11
        xmlFreeInputStream(input);
7779
11
        goto error;
7780
11
    }
7781
7782
33.8k
    xmlBufResetInput(input->buf->buffer, input);
7783
7784
33.8k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
7785
21
        xmlFreeInputStream(input);
7786
21
        goto error;
7787
21
    }
7788
7789
33.7k
    xmlDetectEncoding(ctxt);
7790
7791
    /*
7792
     * Parse a possible text declaration first
7793
     */
7794
33.7k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
7795
22.8k
  xmlParseTextDecl(ctxt);
7796
        /*
7797
         * An XML-1.0 document can't reference an entity not XML-1.0
7798
         */
7799
22.8k
        if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
7800
22.4k
            (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
7801
14.2k
            xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
7802
14.2k
                           "Version mismatch between document and entity\n");
7803
14.2k
        }
7804
22.8k
    }
7805
7806
33.7k
    length = input->cur - input->base;
7807
33.7k
    xmlBufShrink(input->buf->buffer, length);
7808
33.7k
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7809
7810
45.8k
    while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0)
7811
12.0k
        ;
7812
7813
33.7k
    xmlBufResetInput(input->buf->buffer, input);
7814
7815
33.7k
    if (res < 0) {
7816
11.6k
        xmlCtxtErrIO(ctxt, input->buf->error, NULL);
7817
11.6k
        goto error;
7818
11.6k
    }
7819
7820
22.1k
    length = xmlBufUse(input->buf->buffer);
7821
22.1k
    if (length > INT_MAX) {
7822
0
        xmlErrMemory(ctxt);
7823
0
        goto error;
7824
0
    }
7825
7826
22.1k
    content = xmlStrndup(xmlBufContent(input->buf->buffer), length);
7827
22.1k
    if (content == NULL) {
7828
22
        xmlErrMemory(ctxt);
7829
22
        goto error;
7830
22
    }
7831
7832
18.4M
    for (i = 0; i < length; ) {
7833
18.4M
        int clen = length - i;
7834
18.4M
        int c = xmlGetUTF8Char(content + i, &clen);
7835
7836
18.4M
        if ((c < 0) || (!IS_CHAR(c))) {
7837
21.9k
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
7838
21.9k
                              "xmlLoadEntityContent: invalid char value %d\n",
7839
21.9k
                              content[i]);
7840
21.9k
            goto error;
7841
21.9k
        }
7842
18.4M
        i += clen;
7843
18.4M
    }
7844
7845
202
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7846
202
    entity->content = content;
7847
202
    entity->length = length;
7848
202
    content = NULL;
7849
202
    ret = 0;
7850
7851
33.8k
error:
7852
67.6k
    while (ctxt->inputNr > 0)
7853
33.7k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
7854
33.8k
    xmlFree(ctxt->inputTab);
7855
33.8k
    xmlFree(ctxt->encoding);
7856
7857
33.8k
    ctxt->input = oldinput;
7858
33.8k
    ctxt->inputNr = oldinputNr;
7859
33.8k
    ctxt->inputMax = oldinputMax;
7860
33.8k
    ctxt->inputTab = oldinputTab;
7861
33.8k
    ctxt->encoding = oldencoding;
7862
7863
33.8k
    xmlFree(content);
7864
7865
33.8k
    return(ret);
7866
202
}
7867
7868
/**
7869
 * Parse PEReference declarations
7870
 *
7871
 *     [69] PEReference ::= '%' Name ';'
7872
 *
7873
 * [ WFC: No Recursion ]
7874
 * A parsed entity must not contain a recursive
7875
 * reference to itself, either directly or indirectly.
7876
 *
7877
 * [ WFC: Entity Declared ]
7878
 * In a document without any DTD, a document with only an internal DTD
7879
 * subset which contains no parameter entity references, or a document
7880
 * with "standalone='yes'", ...  ... The declaration of a parameter
7881
 * entity must precede any reference to it...
7882
 *
7883
 * [ VC: Entity Declared ]
7884
 * In a document with an external subset or external parameter entities
7885
 * with "standalone='no'", ...  ... The declaration of a parameter entity
7886
 * must precede any reference to it...
7887
 *
7888
 * [ WFC: In DTD ]
7889
 * Parameter-entity references may only appear in the DTD.
7890
 * NOTE: misleading but this is handled.
7891
 *
7892
 * @param ctxt  an XML parser context
7893
 * @param str  a pointer to an index in the string
7894
 * @returns the string of the entity content.
7895
 *         str is updated to the current value of the index
7896
 */
7897
static xmlEntityPtr
7898
229k
xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) {
7899
229k
    const xmlChar *ptr;
7900
229k
    xmlChar cur;
7901
229k
    xmlChar *name;
7902
229k
    xmlEntityPtr entity = NULL;
7903
7904
229k
    if ((str == NULL) || (*str == NULL)) return(NULL);
7905
229k
    ptr = *str;
7906
229k
    cur = *ptr;
7907
229k
    if (cur != '%')
7908
0
        return(NULL);
7909
229k
    ptr++;
7910
229k
    name = xmlParseStringName(ctxt, &ptr);
7911
229k
    if (name == NULL) {
7912
10.2k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7913
10.2k
           "xmlParseStringPEReference: no name\n");
7914
10.2k
  *str = ptr;
7915
10.2k
  return(NULL);
7916
10.2k
    }
7917
219k
    cur = *ptr;
7918
219k
    if (cur != ';') {
7919
3.97k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7920
3.97k
  xmlFree(name);
7921
3.97k
  *str = ptr;
7922
3.97k
  return(NULL);
7923
3.97k
    }
7924
215k
    ptr++;
7925
7926
    /* Must be set before xmlHandleUndeclaredEntity */
7927
215k
    ctxt->hasPErefs = 1;
7928
7929
    /*
7930
     * Request the entity from SAX
7931
     */
7932
215k
    if ((ctxt->sax != NULL) &&
7933
215k
  (ctxt->sax->getParameterEntity != NULL))
7934
215k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7935
7936
215k
    if (entity == NULL) {
7937
3.16k
        xmlHandleUndeclaredEntity(ctxt, name);
7938
211k
    } else {
7939
  /*
7940
   * Internal checking in case the entity quest barfed
7941
   */
7942
211k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7943
38.8k
      (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) {
7944
0
      xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7945
0
        "%%%s; is not a parameter entity\n",
7946
0
        name, NULL);
7947
0
  }
7948
211k
    }
7949
7950
215k
    xmlFree(name);
7951
215k
    *str = ptr;
7952
215k
    return(entity);
7953
219k
}
7954
7955
/**
7956
 * Parse a DOCTYPE declaration
7957
 *
7958
 * @deprecated Internal function, don't use.
7959
 *
7960
 *     [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S?
7961
 *                          ('[' (markupdecl | PEReference | S)* ']' S?)? '>'
7962
 *
7963
 * [ VC: Root Element Type ]
7964
 * The Name in the document type declaration must match the element
7965
 * type of the root element.
7966
 *
7967
 * @param ctxt  an XML parser context
7968
 */
7969
7970
void
7971
52.6k
xmlParseDocTypeDecl(xmlParserCtxt *ctxt) {
7972
52.6k
    const xmlChar *name = NULL;
7973
52.6k
    xmlChar *publicId = NULL;
7974
52.6k
    xmlChar *URI = NULL;
7975
7976
    /*
7977
     * We know that '<!DOCTYPE' has been detected.
7978
     */
7979
52.6k
    SKIP(9);
7980
7981
52.6k
    if (SKIP_BLANKS == 0) {
7982
19.2k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7983
19.2k
                       "Space required after 'DOCTYPE'\n");
7984
19.2k
    }
7985
7986
    /*
7987
     * Parse the DOCTYPE name.
7988
     */
7989
52.6k
    name = xmlParseName(ctxt);
7990
52.6k
    if (name == NULL) {
7991
17.9k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7992
17.9k
           "xmlParseDocTypeDecl : no DOCTYPE name !\n");
7993
17.9k
    }
7994
52.6k
    ctxt->intSubName = name;
7995
7996
52.6k
    SKIP_BLANKS;
7997
7998
    /*
7999
     * Check for public and system identifier (URI)
8000
     */
8001
52.6k
    URI = xmlParseExternalID(ctxt, &publicId, 1);
8002
8003
52.6k
    if ((URI != NULL) || (publicId != NULL)) {
8004
10.1k
        ctxt->hasExternalSubset = 1;
8005
10.1k
    }
8006
52.6k
    ctxt->extSubURI = URI;
8007
52.6k
    ctxt->extSubSystem = publicId;
8008
8009
52.6k
    SKIP_BLANKS;
8010
8011
    /*
8012
     * Create and update the internal subset.
8013
     */
8014
52.6k
    if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) &&
8015
52.6k
  (!ctxt->disableSAX))
8016
49.3k
  ctxt->sax->internalSubset(ctxt->userData, name, publicId, URI);
8017
8018
52.6k
    if ((RAW != '[') && (RAW != '>')) {
8019
3.56k
  xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
8020
3.56k
    }
8021
52.6k
}
8022
8023
/**
8024
 * Parse the internal subset declaration
8025
 *
8026
 *     [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>'
8027
 * @param ctxt  an XML parser context
8028
 */
8029
8030
static void
8031
42.0k
xmlParseInternalSubset(xmlParserCtxtPtr ctxt) {
8032
    /*
8033
     * Is there any DTD definition ?
8034
     */
8035
42.0k
    if (RAW == '[') {
8036
42.0k
        int oldInputNr = ctxt->inputNr;
8037
8038
42.0k
        NEXT;
8039
  /*
8040
   * Parse the succession of Markup declarations and
8041
   * PEReferences.
8042
   * Subsequence (markupdecl | PEReference | S)*
8043
   */
8044
42.0k
  SKIP_BLANKS;
8045
640k
        while (1) {
8046
640k
            if (PARSER_STOPPED(ctxt)) {
8047
4.31k
                return;
8048
636k
            } else if (ctxt->input->cur >= ctxt->input->end) {
8049
94.4k
                if (ctxt->inputNr <= oldInputNr) {
8050
5.71k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8051
5.71k
                    return;
8052
5.71k
                }
8053
88.7k
                xmlPopPE(ctxt);
8054
541k
            } else if ((RAW == ']') && (ctxt->inputNr <= oldInputNr)) {
8055
17.6k
                NEXT;
8056
17.6k
                SKIP_BLANKS;
8057
17.6k
                break;
8058
524k
            } else if ((PARSER_EXTERNAL(ctxt)) &&
8059
321k
                       (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
8060
                /*
8061
                 * Conditional sections are allowed in external entities
8062
                 * included by PE References in the internal subset.
8063
                 */
8064
4.87k
                xmlParseConditionalSections(ctxt);
8065
519k
            } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
8066
380k
                xmlParseMarkupDecl(ctxt);
8067
380k
            } else if (RAW == '%') {
8068
124k
                xmlParsePERefInternal(ctxt, 1);
8069
124k
            } else {
8070
14.3k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8071
8072
15.3k
                while (ctxt->inputNr > oldInputNr)
8073
972
                    xmlPopPE(ctxt);
8074
14.3k
                return;
8075
14.3k
            }
8076
598k
            SKIP_BLANKS;
8077
598k
            SHRINK;
8078
598k
            GROW;
8079
598k
        }
8080
42.0k
    }
8081
8082
    /*
8083
     * We should be at the end of the DOCTYPE declaration.
8084
     */
8085
17.6k
    if (RAW != '>') {
8086
1.17k
        xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
8087
1.17k
        return;
8088
1.17k
    }
8089
16.4k
    NEXT;
8090
16.4k
}
8091
8092
#ifdef LIBXML_SAX1_ENABLED
8093
/**
8094
 * Parse an attribute
8095
 *
8096
 * @deprecated Internal function, don't use.
8097
 *
8098
 *     [41] Attribute ::= Name Eq AttValue
8099
 *
8100
 * [ WFC: No External Entity References ]
8101
 * Attribute values cannot contain direct or indirect entity references
8102
 * to external entities.
8103
 *
8104
 * [ WFC: No < in Attribute Values ]
8105
 * The replacement text of any entity referred to directly or indirectly in
8106
 * an attribute value (other than "&lt;") must not contain a <.
8107
 *
8108
 * [ VC: Attribute Value Type ]
8109
 * The attribute must have been declared; the value must be of the type
8110
 * declared for it.
8111
 *
8112
 *     [25] Eq ::= S? '=' S?
8113
 *
8114
 * With namespace:
8115
 *
8116
 *     [NS 11] Attribute ::= QName Eq AttValue
8117
 *
8118
 * Also the case QName == xmlns:??? is handled independently as a namespace
8119
 * definition.
8120
 *
8121
 * @param ctxt  an XML parser context
8122
 * @param value  a xmlChar ** used to store the value of the attribute
8123
 * @returns the attribute name, and the value in *value.
8124
 */
8125
8126
const xmlChar *
8127
139k
xmlParseAttribute(xmlParserCtxt *ctxt, xmlChar **value) {
8128
139k
    const xmlChar *name;
8129
139k
    xmlChar *val;
8130
8131
139k
    *value = NULL;
8132
139k
    GROW;
8133
139k
    name = xmlParseName(ctxt);
8134
139k
    if (name == NULL) {
8135
47.8k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8136
47.8k
                 "error parsing attribute name\n");
8137
47.8k
        return(NULL);
8138
47.8k
    }
8139
8140
    /*
8141
     * read the value
8142
     */
8143
91.4k
    SKIP_BLANKS;
8144
91.4k
    if (RAW == '=') {
8145
64.5k
        NEXT;
8146
64.5k
  SKIP_BLANKS;
8147
64.5k
  val = xmlParseAttValue(ctxt);
8148
64.5k
    } else {
8149
26.8k
  xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8150
26.8k
         "Specification mandates value for attribute %s\n", name);
8151
26.8k
  return(name);
8152
26.8k
    }
8153
8154
    /*
8155
     * Check that xml:lang conforms to the specification
8156
     * No more registered as an error, just generate a warning now
8157
     * since this was deprecated in XML second edition
8158
     */
8159
64.5k
    if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) {
8160
6.25k
  if (!xmlCheckLanguageID(val)) {
8161
4.95k
      xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8162
4.95k
              "Malformed value for xml:lang : %s\n",
8163
4.95k
        val, NULL);
8164
4.95k
  }
8165
6.25k
    }
8166
8167
    /*
8168
     * Check that xml:space conforms to the specification
8169
     */
8170
64.5k
    if (xmlStrEqual(name, BAD_CAST "xml:space")) {
8171
707
  if (xmlStrEqual(val, BAD_CAST "default"))
8172
260
      *(ctxt->space) = 0;
8173
447
  else if (xmlStrEqual(val, BAD_CAST "preserve"))
8174
211
      *(ctxt->space) = 1;
8175
236
  else {
8176
236
    xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8177
236
"Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8178
236
                                 val, NULL);
8179
236
  }
8180
707
    }
8181
8182
64.5k
    *value = val;
8183
64.5k
    return(name);
8184
91.4k
}
8185
8186
/**
8187
 * Parse a start tag. Always consumes '<'.
8188
 *
8189
 * @deprecated Internal function, don't use.
8190
 *
8191
 *     [40] STag ::= '<' Name (S Attribute)* S? '>'
8192
 *
8193
 * [ WFC: Unique Att Spec ]
8194
 * No attribute name may appear more than once in the same start-tag or
8195
 * empty-element tag.
8196
 *
8197
 *     [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
8198
 *
8199
 * [ WFC: Unique Att Spec ]
8200
 * No attribute name may appear more than once in the same start-tag or
8201
 * empty-element tag.
8202
 *
8203
 * With namespace:
8204
 *
8205
 *     [NS 8] STag ::= '<' QName (S Attribute)* S? '>'
8206
 *
8207
 *     [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>'
8208
 *
8209
 * @param ctxt  an XML parser context
8210
 * @returns the element name parsed
8211
 */
8212
8213
const xmlChar *
8214
264k
xmlParseStartTag(xmlParserCtxt *ctxt) {
8215
264k
    const xmlChar *name;
8216
264k
    const xmlChar *attname;
8217
264k
    xmlChar *attvalue;
8218
264k
    const xmlChar **atts = ctxt->atts;
8219
264k
    int nbatts = 0;
8220
264k
    int maxatts = ctxt->maxatts;
8221
264k
    int i;
8222
8223
264k
    if (RAW != '<') return(NULL);
8224
264k
    NEXT1;
8225
8226
264k
    name = xmlParseName(ctxt);
8227
264k
    if (name == NULL) {
8228
27.5k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8229
27.5k
       "xmlParseStartTag: invalid element name\n");
8230
27.5k
        return(NULL);
8231
27.5k
    }
8232
8233
    /*
8234
     * Now parse the attributes, it ends up with the ending
8235
     *
8236
     * (S Attribute)* S?
8237
     */
8238
237k
    SKIP_BLANKS;
8239
237k
    GROW;
8240
8241
303k
    while (((RAW != '>') &&
8242
162k
     ((RAW != '/') || (NXT(1) != '>')) &&
8243
155k
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8244
139k
  attname = xmlParseAttribute(ctxt, &attvalue);
8245
139k
        if (attname == NULL)
8246
47.8k
      break;
8247
91.4k
        if (attvalue != NULL) {
8248
      /*
8249
       * [ WFC: Unique Att Spec ]
8250
       * No attribute name may appear more than once in the same
8251
       * start-tag or empty-element tag.
8252
       */
8253
346k
      for (i = 0; i < nbatts;i += 2) {
8254
290k
          if (xmlStrEqual(atts[i], attname)) {
8255
3.80k
        xmlErrAttributeDup(ctxt, NULL, attname);
8256
3.80k
        goto failed;
8257
3.80k
    }
8258
290k
      }
8259
      /*
8260
       * Add the pair to atts
8261
       */
8262
55.3k
      if (nbatts + 4 > maxatts) {
8263
9.56k
          const xmlChar **n;
8264
9.56k
                int newSize;
8265
8266
9.56k
                newSize = xmlGrowCapacity(maxatts, sizeof(n[0]) * 2,
8267
9.56k
                                          11, XML_MAX_ATTRS);
8268
9.56k
                if (newSize < 0) {
8269
0
        xmlErrMemory(ctxt);
8270
0
        goto failed;
8271
0
    }
8272
9.56k
#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
8273
9.56k
                if (newSize < 2)
8274
7.92k
                    newSize = 2;
8275
9.56k
#endif
8276
9.56k
          n = xmlRealloc(atts, newSize * sizeof(n[0]) * 2);
8277
9.56k
    if (n == NULL) {
8278
18
        xmlErrMemory(ctxt);
8279
18
        goto failed;
8280
18
    }
8281
9.54k
    atts = n;
8282
9.54k
                maxatts = newSize * 2;
8283
9.54k
    ctxt->atts = atts;
8284
9.54k
    ctxt->maxatts = maxatts;
8285
9.54k
      }
8286
8287
55.3k
      atts[nbatts++] = attname;
8288
55.3k
      atts[nbatts++] = attvalue;
8289
55.3k
      atts[nbatts] = NULL;
8290
55.3k
      atts[nbatts + 1] = NULL;
8291
8292
55.3k
            attvalue = NULL;
8293
55.3k
  }
8294
8295
91.4k
failed:
8296
8297
91.4k
        if (attvalue != NULL)
8298
3.82k
            xmlFree(attvalue);
8299
8300
91.4k
  GROW
8301
91.4k
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
8302
24.8k
      break;
8303
66.6k
  if (SKIP_BLANKS == 0) {
8304
49.0k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
8305
49.0k
         "attributes construct error\n");
8306
49.0k
  }
8307
66.6k
  SHRINK;
8308
66.6k
        GROW;
8309
66.6k
    }
8310
8311
    /*
8312
     * SAX: Start of Element !
8313
     */
8314
237k
    if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) &&
8315
237k
  (!ctxt->disableSAX)) {
8316
221k
  if (nbatts > 0)
8317
39.3k
      ctxt->sax->startElement(ctxt->userData, name, atts);
8318
182k
  else
8319
182k
      ctxt->sax->startElement(ctxt->userData, name, NULL);
8320
221k
    }
8321
8322
237k
    if (atts != NULL) {
8323
        /* Free only the content strings */
8324
141k
        for (i = 1;i < nbatts;i+=2)
8325
55.3k
      if (atts[i] != NULL)
8326
55.3k
         xmlFree((xmlChar *) atts[i]);
8327
86.1k
    }
8328
237k
    return(name);
8329
237k
}
8330
8331
/**
8332
 * Parse an end tag. Always consumes '</'.
8333
 *
8334
 *     [42] ETag ::= '</' Name S? '>'
8335
 *
8336
 * With namespace
8337
 *
8338
 *     [NS 9] ETag ::= '</' QName S? '>'
8339
 * @param ctxt  an XML parser context
8340
 * @param line  line of the start tag
8341
 */
8342
8343
static void
8344
18.1k
xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) {
8345
18.1k
    const xmlChar *name;
8346
8347
18.1k
    GROW;
8348
18.1k
    if ((RAW != '<') || (NXT(1) != '/')) {
8349
1.00k
  xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED,
8350
1.00k
           "xmlParseEndTag: '</' not found\n");
8351
1.00k
  return;
8352
1.00k
    }
8353
17.1k
    SKIP(2);
8354
8355
17.1k
    name = xmlParseNameAndCompare(ctxt,ctxt->name);
8356
8357
    /*
8358
     * We should definitely be at the ending "S? '>'" part
8359
     */
8360
17.1k
    GROW;
8361
17.1k
    SKIP_BLANKS;
8362
17.1k
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
8363
3.45k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
8364
3.45k
    } else
8365
13.6k
  NEXT1;
8366
8367
    /*
8368
     * [ WFC: Element Type Match ]
8369
     * The Name in an element's end-tag must match the element type in the
8370
     * start-tag.
8371
     *
8372
     */
8373
17.1k
    if (name != (xmlChar*)1) {
8374
3.76k
        if (name == NULL) name = BAD_CAST "unparsable";
8375
3.76k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
8376
3.76k
         "Opening and ending tag mismatch: %s line %d and %s\n",
8377
3.76k
                    ctxt->name, line, name);
8378
3.76k
    }
8379
8380
    /*
8381
     * SAX: End of Tag
8382
     */
8383
17.1k
    if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
8384
17.1k
  (!ctxt->disableSAX))
8385
16.0k
        ctxt->sax->endElement(ctxt->userData, ctxt->name);
8386
8387
17.1k
    namePop(ctxt);
8388
17.1k
    spacePop(ctxt);
8389
17.1k
}
8390
8391
/**
8392
 * Parse an end of tag
8393
 *
8394
 * @deprecated Internal function, don't use.
8395
 *
8396
 *     [42] ETag ::= '</' Name S? '>'
8397
 *
8398
 * With namespace
8399
 *
8400
 *     [NS 9] ETag ::= '</' QName S? '>'
8401
 * @param ctxt  an XML parser context
8402
 */
8403
8404
void
8405
0
xmlParseEndTag(xmlParserCtxt *ctxt) {
8406
0
    xmlParseEndTag1(ctxt, 0);
8407
0
}
8408
#endif /* LIBXML_SAX1_ENABLED */
8409
8410
/************************************************************************
8411
 *                  *
8412
 *          SAX 2 specific operations       *
8413
 *                  *
8414
 ************************************************************************/
8415
8416
/**
8417
 * Parse an XML Namespace QName
8418
 *
8419
 *     [6]  QName  ::= (Prefix ':')? LocalPart
8420
 *     [7]  Prefix  ::= NCName
8421
 *     [8]  LocalPart  ::= NCName
8422
 *
8423
 * @param ctxt  an XML parser context
8424
 * @param prefix  pointer to store the prefix part
8425
 * @returns the Name parsed or NULL
8426
 */
8427
8428
static xmlHashedString
8429
1.32M
xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) {
8430
1.32M
    xmlHashedString l, p;
8431
1.32M
    int start, isNCName = 0;
8432
8433
1.32M
    l.name = NULL;
8434
1.32M
    p.name = NULL;
8435
8436
1.32M
    GROW;
8437
1.32M
    start = CUR_PTR - BASE_PTR;
8438
8439
1.32M
    l = xmlParseNCName(ctxt);
8440
1.32M
    if (l.name != NULL) {
8441
925k
        isNCName = 1;
8442
925k
        if (CUR == ':') {
8443
110k
            NEXT;
8444
110k
            p = l;
8445
110k
            l = xmlParseNCName(ctxt);
8446
110k
        }
8447
925k
    }
8448
1.32M
    if ((l.name == NULL) || (CUR == ':')) {
8449
408k
        xmlChar *tmp;
8450
8451
408k
        l.name = NULL;
8452
408k
        p.name = NULL;
8453
408k
        if ((isNCName == 0) && (CUR != ':'))
8454
391k
            return(l);
8455
16.1k
        tmp = xmlParseNmtoken(ctxt);
8456
16.1k
        if (tmp != NULL)
8457
10.4k
            xmlFree(tmp);
8458
16.1k
        l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start,
8459
16.1k
                                CUR_PTR - (BASE_PTR + start));
8460
16.1k
        if (l.name == NULL) {
8461
13
            xmlErrMemory(ctxt);
8462
13
            return(l);
8463
13
        }
8464
16.1k
        xmlNsErr(ctxt, XML_NS_ERR_QNAME,
8465
16.1k
                 "Failed to parse QName '%s'\n", l.name, NULL, NULL);
8466
16.1k
    }
8467
8468
932k
    *prefix = p;
8469
932k
    return(l);
8470
1.32M
}
8471
8472
/**
8473
 * Parse an XML Namespace QName
8474
 *
8475
 *     [6]  QName  ::= (Prefix ':')? LocalPart
8476
 *     [7]  Prefix  ::= NCName
8477
 *     [8]  LocalPart  ::= NCName
8478
 *
8479
 * @param ctxt  an XML parser context
8480
 * @param prefix  pointer to store the prefix part
8481
 * @returns the Name parsed or NULL
8482
 */
8483
8484
static const xmlChar *
8485
4.21k
xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) {
8486
4.21k
    xmlHashedString n, p;
8487
8488
4.21k
    n = xmlParseQNameHashed(ctxt, &p);
8489
4.21k
    if (n.name == NULL)
8490
2.19k
        return(NULL);
8491
2.02k
    *prefix = p.name;
8492
2.02k
    return(n.name);
8493
4.21k
}
8494
8495
/**
8496
 * Parse an XML name and compares for match
8497
 * (specialized for endtag parsing)
8498
 *
8499
 * @param ctxt  an XML parser context
8500
 * @param name  the localname
8501
 * @param prefix  the prefix, if any.
8502
 * @returns NULL for an illegal name, (xmlChar*) 1 for success
8503
 * and the name for mismatch
8504
 */
8505
8506
static const xmlChar *
8507
xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name,
8508
8.11k
                        xmlChar const *prefix) {
8509
8.11k
    const xmlChar *cmp;
8510
8.11k
    const xmlChar *in;
8511
8.11k
    const xmlChar *ret;
8512
8.11k
    const xmlChar *prefix2;
8513
8514
8.11k
    if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name));
8515
8516
8.11k
    GROW;
8517
8.11k
    in = ctxt->input->cur;
8518
8519
8.11k
    cmp = prefix;
8520
17.1k
    while (*in != 0 && *in == *cmp) {
8521
9.06k
  ++in;
8522
9.06k
  ++cmp;
8523
9.06k
    }
8524
8.11k
    if ((*cmp == 0) && (*in == ':')) {
8525
4.90k
        in++;
8526
4.90k
  cmp = name;
8527
10.9k
  while (*in != 0 && *in == *cmp) {
8528
6.09k
      ++in;
8529
6.09k
      ++cmp;
8530
6.09k
  }
8531
4.90k
  if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
8532
      /* success */
8533
3.89k
            ctxt->input->col += in - ctxt->input->cur;
8534
3.89k
      ctxt->input->cur = in;
8535
3.89k
      return((const xmlChar*) 1);
8536
3.89k
  }
8537
4.90k
    }
8538
    /*
8539
     * all strings coms from the dictionary, equality can be done directly
8540
     */
8541
4.21k
    ret = xmlParseQName (ctxt, &prefix2);
8542
4.21k
    if (ret == NULL)
8543
2.19k
        return(NULL);
8544
2.02k
    if ((ret == name) && (prefix == prefix2))
8545
596
  return((const xmlChar*) 1);
8546
1.42k
    return ret;
8547
2.02k
}
8548
8549
/**
8550
 * Parse an attribute in the new SAX2 framework.
8551
 *
8552
 * @param ctxt  an XML parser context
8553
 * @param pref  the element prefix
8554
 * @param elem  the element name
8555
 * @param hprefix  resulting attribute prefix
8556
 * @param value  resulting value of the attribute
8557
 * @param len  resulting length of the attribute
8558
 * @param alloc  resulting indicator if the attribute was allocated
8559
 * @returns the attribute name, and the value in *value, .
8560
 */
8561
8562
static xmlHashedString
8563
xmlParseAttribute2(xmlParserCtxtPtr ctxt,
8564
                   const xmlChar * pref, const xmlChar * elem,
8565
                   xmlHashedString * hprefix, xmlChar ** value,
8566
                   int *len, int *alloc)
8567
499k
{
8568
499k
    xmlHashedString hname;
8569
499k
    const xmlChar *prefix, *name;
8570
499k
    xmlChar *val = NULL, *internal_val = NULL;
8571
499k
    int special = 0;
8572
499k
    int isNamespace;
8573
499k
    int flags;
8574
8575
499k
    *value = NULL;
8576
499k
    GROW;
8577
499k
    hname = xmlParseQNameHashed(ctxt, hprefix);
8578
499k
    if (hname.name == NULL) {
8579
311k
        xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8580
311k
                       "error parsing attribute name\n");
8581
311k
        return(hname);
8582
311k
    }
8583
188k
    name = hname.name;
8584
188k
    prefix = hprefix->name;
8585
8586
    /*
8587
     * get the type if needed
8588
     */
8589
188k
    if (ctxt->attsSpecial != NULL) {
8590
69.6k
        special = XML_PTR_TO_INT(xmlHashQLookup2(ctxt->attsSpecial, pref, elem,
8591
69.6k
                                              prefix, name));
8592
69.6k
    }
8593
8594
    /*
8595
     * read the value
8596
     */
8597
188k
    SKIP_BLANKS;
8598
188k
    if (RAW != '=') {
8599
23.6k
        xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8600
23.6k
                          "Specification mandates value for attribute %s\n",
8601
23.6k
                          name);
8602
23.6k
        goto error;
8603
23.6k
    }
8604
8605
8606
164k
    NEXT;
8607
164k
    SKIP_BLANKS;
8608
164k
    flags = 0;
8609
164k
    isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) ||
8610
133k
                   (prefix == ctxt->str_xmlns));
8611
164k
    val = xmlParseAttValueInternal(ctxt, len, &flags, special,
8612
164k
                                   isNamespace);
8613
164k
    if (val == NULL)
8614
6.70k
        goto error;
8615
8616
157k
    *alloc = (flags & XML_ATTVAL_ALLOC) != 0;
8617
8618
157k
#ifdef LIBXML_VALID_ENABLED
8619
157k
    if ((ctxt->validate) &&
8620
67.2k
        (ctxt->standalone == 1) &&
8621
614
        (special & XML_SPECIAL_EXTERNAL) &&
8622
414
        (flags & XML_ATTVAL_NORM_CHANGE)) {
8623
194
        xmlValidityError(ctxt, XML_DTD_NOT_STANDALONE,
8624
194
                         "standalone: normalization of attribute %s on %s "
8625
194
                         "by external subset declaration\n",
8626
194
                         name, elem);
8627
194
    }
8628
157k
#endif
8629
8630
157k
    if (prefix == ctxt->str_xml) {
8631
        /*
8632
         * Check that xml:lang conforms to the specification
8633
         * No more registered as an error, just generate a warning now
8634
         * since this was deprecated in XML second edition
8635
         */
8636
9.27k
        if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) {
8637
4.36k
            internal_val = xmlStrndup(val, *len);
8638
4.36k
            if (internal_val == NULL)
8639
19
                goto mem_error;
8640
4.34k
            if (!xmlCheckLanguageID(internal_val)) {
8641
3.29k
                xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8642
3.29k
                              "Malformed value for xml:lang : %s\n",
8643
3.29k
                              internal_val, NULL);
8644
3.29k
            }
8645
4.34k
        }
8646
8647
        /*
8648
         * Check that xml:space conforms to the specification
8649
         */
8650
9.25k
        if (xmlStrEqual(name, BAD_CAST "space")) {
8651
1.09k
            internal_val = xmlStrndup(val, *len);
8652
1.09k
            if (internal_val == NULL)
8653
8
                goto mem_error;
8654
1.08k
            if (xmlStrEqual(internal_val, BAD_CAST "default"))
8655
570
                *(ctxt->space) = 0;
8656
513
            else if (xmlStrEqual(internal_val, BAD_CAST "preserve"))
8657
225
                *(ctxt->space) = 1;
8658
288
            else {
8659
288
                xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8660
288
                              "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8661
288
                              internal_val, NULL);
8662
288
            }
8663
1.08k
        }
8664
9.24k
        if (internal_val) {
8665
5.42k
            xmlFree(internal_val);
8666
5.42k
        }
8667
9.24k
    }
8668
8669
157k
    *value = val;
8670
157k
    return (hname);
8671
8672
27
mem_error:
8673
27
    xmlErrMemory(ctxt);
8674
30.4k
error:
8675
30.4k
    if ((val != NULL) && (*alloc != 0))
8676
9
        xmlFree(val);
8677
30.4k
    return(hname);
8678
27
}
8679
8680
/**
8681
 * Inserts a new attribute into the hash table.
8682
 *
8683
 * @param ctxt  parser context
8684
 * @param size  size of the hash table
8685
 * @param name  attribute name
8686
 * @param uri  namespace uri
8687
 * @param hashValue  combined hash value of name and uri
8688
 * @param aindex  attribute index (this is a multiple of 5)
8689
 * @returns INT_MAX if no existing attribute was found, the attribute
8690
 * index if an attribute was found, -1 if a memory allocation failed.
8691
 */
8692
static int
8693
xmlAttrHashInsert(xmlParserCtxtPtr ctxt, unsigned size, const xmlChar *name,
8694
319k
                  const xmlChar *uri, unsigned hashValue, int aindex) {
8695
319k
    xmlAttrHashBucket *table = ctxt->attrHash;
8696
319k
    xmlAttrHashBucket *bucket;
8697
319k
    unsigned hindex;
8698
8699
319k
    hindex = hashValue & (size - 1);
8700
319k
    bucket = &table[hindex];
8701
8702
379k
    while (bucket->index >= 0) {
8703
90.4k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8704
8705
90.4k
        if (name == atts[0]) {
8706
31.3k
            int nsIndex = XML_PTR_TO_INT(atts[2]);
8707
8708
31.3k
            if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) :
8709
31.3k
                (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) :
8710
5.76k
                (uri == ctxt->nsTab[nsIndex * 2 + 1]))
8711
30.1k
                return(bucket->index);
8712
31.3k
        }
8713
8714
60.3k
        hindex++;
8715
60.3k
        bucket++;
8716
60.3k
        if (hindex >= size) {
8717
1.86k
            hindex = 0;
8718
1.86k
            bucket = table;
8719
1.86k
        }
8720
60.3k
    }
8721
8722
288k
    bucket->index = aindex;
8723
8724
288k
    return(INT_MAX);
8725
319k
}
8726
8727
static int
8728
xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size,
8729
                       const xmlChar *name, const xmlChar *prefix,
8730
4.23k
                       unsigned hashValue, int aindex) {
8731
4.23k
    xmlAttrHashBucket *table = ctxt->attrHash;
8732
4.23k
    xmlAttrHashBucket *bucket;
8733
4.23k
    unsigned hindex;
8734
8735
4.23k
    hindex = hashValue & (size - 1);
8736
4.23k
    bucket = &table[hindex];
8737
8738
6.36k
    while (bucket->index >= 0) {
8739
3.46k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8740
8741
3.46k
        if ((name == atts[0]) && (prefix == atts[1]))
8742
1.33k
            return(bucket->index);
8743
8744
2.13k
        hindex++;
8745
2.13k
        bucket++;
8746
2.13k
        if (hindex >= size) {
8747
213
            hindex = 0;
8748
213
            bucket = table;
8749
213
        }
8750
2.13k
    }
8751
8752
2.90k
    bucket->index = aindex;
8753
8754
2.90k
    return(INT_MAX);
8755
4.23k
}
8756
/**
8757
 * Parse a start tag. Always consumes '<'.
8758
 *
8759
 * This routine is called when running SAX2 parsing
8760
 *
8761
 *     [40] STag ::= '<' Name (S Attribute)* S? '>'
8762
 *
8763
 * [ WFC: Unique Att Spec ]
8764
 * No attribute name may appear more than once in the same start-tag or
8765
 * empty-element tag.
8766
 *
8767
 *     [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
8768
 *
8769
 * [ WFC: Unique Att Spec ]
8770
 * No attribute name may appear more than once in the same start-tag or
8771
 * empty-element tag.
8772
 *
8773
 * With namespace:
8774
 *
8775
 *     [NS 8] STag ::= '<' QName (S Attribute)* S? '>'
8776
 *
8777
 *     [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>'
8778
 *
8779
 * @param ctxt  an XML parser context
8780
 * @param pref  resulting namespace prefix
8781
 * @param URI  resulting namespace URI
8782
 * @param nbNsPtr  resulting number of namespace declarations
8783
 * @returns the element name parsed
8784
 */
8785
8786
static const xmlChar *
8787
xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref,
8788
820k
                  const xmlChar **URI, int *nbNsPtr) {
8789
820k
    xmlHashedString hlocalname;
8790
820k
    xmlHashedString hprefix;
8791
820k
    xmlHashedString hattname;
8792
820k
    xmlHashedString haprefix;
8793
820k
    const xmlChar *localname;
8794
820k
    const xmlChar *prefix;
8795
820k
    const xmlChar *attname;
8796
820k
    const xmlChar *aprefix;
8797
820k
    const xmlChar *uri;
8798
820k
    xmlChar *attvalue = NULL;
8799
820k
    const xmlChar **atts = ctxt->atts;
8800
820k
    unsigned attrHashSize = 0;
8801
820k
    int maxatts = ctxt->maxatts;
8802
820k
    int nratts, nbatts, nbdef;
8803
820k
    int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts;
8804
820k
    int alloc = 0;
8805
820k
    int numNsErr = 0;
8806
820k
    int numDupErr = 0;
8807
8808
820k
    if (RAW != '<') return(NULL);
8809
820k
    NEXT1;
8810
8811
820k
    nbatts = 0;
8812
820k
    nratts = 0;
8813
820k
    nbdef = 0;
8814
820k
    nbNs = 0;
8815
820k
    nbTotalDef = 0;
8816
820k
    attval = 0;
8817
8818
820k
    if (xmlParserNsStartElement(ctxt->nsdb) < 0) {
8819
0
        xmlErrMemory(ctxt);
8820
0
        return(NULL);
8821
0
    }
8822
8823
820k
    hlocalname = xmlParseQNameHashed(ctxt, &hprefix);
8824
820k
    if (hlocalname.name == NULL) {
8825
78.5k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8826
78.5k
           "StartTag: invalid element name\n");
8827
78.5k
        return(NULL);
8828
78.5k
    }
8829
741k
    localname = hlocalname.name;
8830
741k
    prefix = hprefix.name;
8831
8832
    /*
8833
     * Now parse the attributes, it ends up with the ending
8834
     *
8835
     * (S Attribute)* S?
8836
     */
8837
741k
    SKIP_BLANKS;
8838
741k
    GROW;
8839
8840
    /*
8841
     * The ctxt->atts array will be ultimately passed to the SAX callback
8842
     * containing five xmlChar pointers for each attribute:
8843
     *
8844
     * [0] attribute name
8845
     * [1] attribute prefix
8846
     * [2] namespace URI
8847
     * [3] attribute value
8848
     * [4] end of attribute value
8849
     *
8850
     * To save memory, we reuse this array temporarily and store integers
8851
     * in these pointer variables.
8852
     *
8853
     * [0] attribute name
8854
     * [1] attribute prefix
8855
     * [2] hash value of attribute prefix, and later namespace index
8856
     * [3] for non-allocated values: ptrdiff_t offset into input buffer
8857
     * [4] for non-allocated values: ptrdiff_t offset into input buffer
8858
     *
8859
     * The ctxt->attallocs array contains an additional unsigned int for
8860
     * each attribute, containing the hash value of the attribute name
8861
     * and the alloc flag in bit 31.
8862
     */
8863
8864
827k
    while (((RAW != '>') &&
8865
585k
     ((RAW != '/') || (NXT(1) != '>')) &&
8866
517k
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8867
499k
  int len = -1;
8868
8869
499k
  hattname = xmlParseAttribute2(ctxt, prefix, localname,
8870
499k
                                          &haprefix, &attvalue, &len,
8871
499k
                                          &alloc);
8872
499k
        if (hattname.name == NULL)
8873
311k
      break;
8874
188k
        if (attvalue == NULL)
8875
30.4k
            goto next_attr;
8876
157k
        attname = hattname.name;
8877
157k
        aprefix = haprefix.name;
8878
157k
  if (len < 0) len = xmlStrlen(attvalue);
8879
8880
157k
        if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
8881
30.5k
            xmlHashedString huri;
8882
30.5k
            xmlURIPtr parsedUri;
8883
8884
30.5k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8885
30.5k
            uri = huri.name;
8886
30.5k
            if (uri == NULL) {
8887
10
                xmlErrMemory(ctxt);
8888
10
                goto next_attr;
8889
10
            }
8890
30.5k
            if (*uri != 0) {
8891
19.6k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8892
109
                    xmlErrMemory(ctxt);
8893
109
                    goto next_attr;
8894
109
                }
8895
19.4k
                if (parsedUri == NULL) {
8896
11.5k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8897
11.5k
                             "xmlns: '%s' is not a valid URI\n",
8898
11.5k
                                       uri, NULL, NULL);
8899
11.5k
                } else {
8900
7.98k
                    if (parsedUri->scheme == NULL) {
8901
3.15k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8902
3.15k
                                  "xmlns: URI %s is not absolute\n",
8903
3.15k
                                  uri, NULL, NULL);
8904
3.15k
                    }
8905
7.98k
                    xmlFreeURI(parsedUri);
8906
7.98k
                }
8907
19.4k
                if (uri == ctxt->str_xml_ns) {
8908
197
                    if (attname != ctxt->str_xml) {
8909
197
                        xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8910
197
                     "xml namespace URI cannot be the default namespace\n",
8911
197
                                 NULL, NULL, NULL);
8912
197
                    }
8913
197
                    goto next_attr;
8914
197
                }
8915
19.2k
                if ((len == 29) &&
8916
417
                    (xmlStrEqual(uri,
8917
417
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8918
201
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8919
201
                         "reuse of the xmlns namespace name is forbidden\n",
8920
201
                             NULL, NULL, NULL);
8921
201
                    goto next_attr;
8922
201
                }
8923
19.2k
            }
8924
8925
30.0k
            if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0)
8926
20.0k
                nbNs++;
8927
127k
        } else if (aprefix == ctxt->str_xmlns) {
8928
20.0k
            xmlHashedString huri;
8929
20.0k
            xmlURIPtr parsedUri;
8930
8931
20.0k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8932
20.0k
            uri = huri.name;
8933
20.0k
            if (uri == NULL) {
8934
10
                xmlErrMemory(ctxt);
8935
10
                goto next_attr;
8936
10
            }
8937
8938
20.0k
            if (attname == ctxt->str_xml) {
8939
437
                if (uri != ctxt->str_xml_ns) {
8940
227
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8941
227
                             "xml namespace prefix mapped to wrong URI\n",
8942
227
                             NULL, NULL, NULL);
8943
227
                }
8944
                /*
8945
                 * Do not keep a namespace definition node
8946
                 */
8947
437
                goto next_attr;
8948
437
            }
8949
19.6k
            if (uri == ctxt->str_xml_ns) {
8950
203
                if (attname != ctxt->str_xml) {
8951
203
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8952
203
                             "xml namespace URI mapped to wrong prefix\n",
8953
203
                             NULL, NULL, NULL);
8954
203
                }
8955
203
                goto next_attr;
8956
203
            }
8957
19.3k
            if (attname == ctxt->str_xmlns) {
8958
259
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8959
259
                         "redefinition of the xmlns prefix is forbidden\n",
8960
259
                         NULL, NULL, NULL);
8961
259
                goto next_attr;
8962
259
            }
8963
19.1k
            if ((len == 29) &&
8964
779
                (xmlStrEqual(uri,
8965
779
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8966
354
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8967
354
                         "reuse of the xmlns namespace name is forbidden\n",
8968
354
                         NULL, NULL, NULL);
8969
354
                goto next_attr;
8970
354
            }
8971
18.7k
            if ((uri == NULL) || (uri[0] == 0)) {
8972
390
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8973
390
                         "xmlns:%s: Empty XML namespace is not allowed\n",
8974
390
                              attname, NULL, NULL);
8975
390
                goto next_attr;
8976
18.3k
            } else {
8977
18.3k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8978
28
                    xmlErrMemory(ctxt);
8979
28
                    goto next_attr;
8980
28
                }
8981
18.3k
                if (parsedUri == NULL) {
8982
6.79k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8983
6.79k
                         "xmlns:%s: '%s' is not a valid URI\n",
8984
6.79k
                                       attname, uri, NULL);
8985
11.5k
                } else {
8986
11.5k
                    if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) {
8987
1.90k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8988
1.90k
                                  "xmlns:%s: URI %s is not absolute\n",
8989
1.90k
                                  attname, uri, NULL);
8990
1.90k
                    }
8991
11.5k
                    xmlFreeURI(parsedUri);
8992
11.5k
                }
8993
18.3k
            }
8994
8995
18.3k
            if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0)
8996
16.1k
                nbNs++;
8997
107k
        } else {
8998
            /*
8999
             * Populate attributes array, see above for repurposing
9000
             * of xmlChar pointers.
9001
             */
9002
107k
            if ((atts == NULL) || (nbatts + 5 > maxatts)) {
9003
11.4k
                int res = xmlCtxtGrowAttrs(ctxt);
9004
9005
11.4k
                maxatts = ctxt->maxatts;
9006
11.4k
                atts = ctxt->atts;
9007
9008
11.4k
                if (res < 0)
9009
71
                    goto next_attr;
9010
11.4k
            }
9011
107k
            ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) |
9012
107k
                                        ((unsigned) alloc << 31);
9013
107k
            atts[nbatts++] = attname;
9014
107k
            atts[nbatts++] = aprefix;
9015
107k
            atts[nbatts++] = XML_INT_TO_PTR(haprefix.hashValue);
9016
107k
            if (alloc) {
9017
18.4k
                atts[nbatts++] = attvalue;
9018
18.4k
                attvalue += len;
9019
18.4k
                atts[nbatts++] = attvalue;
9020
88.8k
            } else {
9021
                /*
9022
                 * attvalue points into the input buffer which can be
9023
                 * reallocated. Store differences to input->base instead.
9024
                 * The pointers will be reconstructed later.
9025
                 */
9026
88.8k
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
9027
88.8k
                attvalue += len;
9028
88.8k
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
9029
88.8k
            }
9030
            /*
9031
             * tag if some deallocation is needed
9032
             */
9033
107k
            if (alloc != 0) attval = 1;
9034
107k
            attvalue = NULL; /* moved into atts */
9035
107k
        }
9036
9037
188k
next_attr:
9038
188k
        if ((attvalue != NULL) && (alloc != 0)) {
9039
11.0k
            xmlFree(attvalue);
9040
11.0k
            attvalue = NULL;
9041
11.0k
        }
9042
9043
188k
  GROW
9044
188k
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
9045
51.6k
      break;
9046
136k
  if (SKIP_BLANKS == 0) {
9047
50.8k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
9048
50.8k
         "attributes construct error\n");
9049
50.8k
      break;
9050
50.8k
  }
9051
85.8k
        GROW;
9052
85.8k
    }
9053
9054
    /*
9055
     * Namespaces from default attributes
9056
     */
9057
741k
    if (ctxt->attsDefault != NULL) {
9058
473k
        xmlDefAttrsPtr defaults;
9059
9060
473k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9061
473k
  if (defaults != NULL) {
9062
1.40M
      for (i = 0; i < defaults->nbAttrs; i++) {
9063
989k
                xmlDefAttr *attr = &defaults->attrs[i];
9064
9065
989k
          attname = attr->name.name;
9066
989k
    aprefix = attr->prefix.name;
9067
9068
989k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
9069
9.33k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9070
9071
9.33k
                    if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0)
9072
8.29k
                        nbNs++;
9073
980k
    } else if (aprefix == ctxt->str_xmlns) {
9074
637k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9075
9076
637k
                    if (xmlParserNsPush(ctxt, &attr->name, &attr->value,
9077
637k
                                      NULL, 1) > 0)
9078
627k
                        nbNs++;
9079
637k
    } else {
9080
342k
                    if (nratts + nbTotalDef >= XML_MAX_ATTRS) {
9081
0
                        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
9082
0
                                    "Maximum number of attributes exceeded");
9083
0
                        break;
9084
0
                    }
9085
342k
                    nbTotalDef += 1;
9086
342k
                }
9087
989k
      }
9088
414k
  }
9089
473k
    }
9090
9091
    /*
9092
     * Resolve attribute namespaces
9093
     */
9094
849k
    for (i = 0; i < nbatts; i += 5) {
9095
107k
        attname = atts[i];
9096
107k
        aprefix = atts[i+1];
9097
9098
        /*
9099
  * The default namespace does not apply to attribute names.
9100
  */
9101
107k
  if (aprefix == NULL) {
9102
59.4k
            nsIndex = NS_INDEX_EMPTY;
9103
59.4k
        } else if (aprefix == ctxt->str_xml) {
9104
9.23k
            nsIndex = NS_INDEX_XML;
9105
38.5k
        } else {
9106
38.5k
            haprefix.name = aprefix;
9107
38.5k
            haprefix.hashValue = (size_t) atts[i+2];
9108
38.5k
            nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL);
9109
9110
38.5k
      if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) {
9111
30.6k
                xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9112
30.6k
        "Namespace prefix %s for %s on %s is not defined\n",
9113
30.6k
        aprefix, attname, localname);
9114
30.6k
                nsIndex = NS_INDEX_EMPTY;
9115
30.6k
            }
9116
38.5k
        }
9117
9118
107k
        atts[i+2] = XML_INT_TO_PTR(nsIndex);
9119
107k
    }
9120
9121
    /*
9122
     * Maximum number of attributes including default attributes.
9123
     */
9124
741k
    maxAtts = nratts + nbTotalDef;
9125
9126
    /*
9127
     * Verify that attribute names are unique.
9128
     */
9129
741k
    if (maxAtts > 1) {
9130
34.7k
        attrHashSize = 4;
9131
58.6k
        while (attrHashSize / 2 < (unsigned) maxAtts)
9132
23.9k
            attrHashSize *= 2;
9133
9134
34.7k
        if (attrHashSize > ctxt->attrHashMax) {
9135
2.74k
            xmlAttrHashBucket *tmp;
9136
9137
2.74k
            tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0]));
9138
2.74k
            if (tmp == NULL) {
9139
12
                xmlErrMemory(ctxt);
9140
12
                goto done;
9141
12
            }
9142
9143
2.72k
            ctxt->attrHash = tmp;
9144
2.72k
            ctxt->attrHashMax = attrHashSize;
9145
2.72k
        }
9146
9147
34.7k
        memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0]));
9148
9149
107k
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9150
72.2k
            const xmlChar *nsuri;
9151
72.2k
            unsigned hashValue, nameHashValue, uriHashValue;
9152
72.2k
            int res;
9153
9154
72.2k
            attname = atts[i];
9155
72.2k
            aprefix = atts[i+1];
9156
72.2k
            nsIndex = XML_PTR_TO_INT(atts[i+2]);
9157
            /* Hash values always have bit 31 set, see dict.c */
9158
72.2k
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9159
9160
72.2k
            if (nsIndex == NS_INDEX_EMPTY) {
9161
                /*
9162
                 * Prefix with empty namespace means an undeclared
9163
                 * prefix which was already reported above.
9164
                 */
9165
63.6k
                if (aprefix != NULL)
9166
28.6k
                    continue;
9167
34.9k
                nsuri = NULL;
9168
34.9k
                uriHashValue = URI_HASH_EMPTY;
9169
34.9k
            } else if (nsIndex == NS_INDEX_XML) {
9170
1.99k
                nsuri = ctxt->str_xml_ns;
9171
1.99k
                uriHashValue = URI_HASH_XML;
9172
6.64k
            } else {
9173
6.64k
                nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9174
6.64k
                uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9175
6.64k
            }
9176
9177
43.5k
            hashValue = xmlDictCombineHash(nameHashValue, uriHashValue);
9178
43.5k
            res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9179
43.5k
                                    hashValue, i);
9180
43.5k
            if (res < 0)
9181
0
                continue;
9182
9183
            /*
9184
             * [ WFC: Unique Att Spec ]
9185
             * No attribute name may appear more than once in the same
9186
             * start-tag or empty-element tag.
9187
             * As extended by the Namespace in XML REC.
9188
             */
9189
43.5k
            if (res < INT_MAX) {
9190
26.9k
                if (aprefix == atts[res+1]) {
9191
24.7k
                    xmlErrAttributeDup(ctxt, aprefix, attname);
9192
24.7k
                    numDupErr += 1;
9193
24.7k
                } else {
9194
2.23k
                    xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9195
2.23k
                             "Namespaced Attribute %s in '%s' redefined\n",
9196
2.23k
                             attname, nsuri, NULL);
9197
2.23k
                    numNsErr += 1;
9198
2.23k
                }
9199
26.9k
            }
9200
43.5k
        }
9201
34.7k
    }
9202
9203
    /*
9204
     * Default attributes
9205
     */
9206
741k
    if (ctxt->attsDefault != NULL) {
9207
473k
        xmlDefAttrsPtr defaults;
9208
9209
473k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9210
473k
  if (defaults != NULL) {
9211
1.40M
      for (i = 0; i < defaults->nbAttrs; i++) {
9212
989k
                xmlDefAttr *attr = &defaults->attrs[i];
9213
989k
                const xmlChar *nsuri = NULL;
9214
989k
                unsigned hashValue, uriHashValue = 0;
9215
989k
                int res;
9216
9217
989k
          attname = attr->name.name;
9218
989k
    aprefix = attr->prefix.name;
9219
9220
989k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL))
9221
9.33k
                    continue;
9222
980k
    if (aprefix == ctxt->str_xmlns)
9223
637k
                    continue;
9224
9225
342k
                if (aprefix == NULL) {
9226
294k
                    nsIndex = NS_INDEX_EMPTY;
9227
294k
                    nsuri = NULL;
9228
294k
                    uriHashValue = URI_HASH_EMPTY;
9229
294k
                } else if (aprefix == ctxt->str_xml) {
9230
11.0k
                    nsIndex = NS_INDEX_XML;
9231
11.0k
                    nsuri = ctxt->str_xml_ns;
9232
11.0k
                    uriHashValue = URI_HASH_XML;
9233
36.4k
                } else {
9234
36.4k
                    nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL);
9235
36.4k
                    if ((nsIndex == INT_MAX) ||
9236
28.1k
                        (nsIndex < ctxt->nsdb->minNsIndex)) {
9237
28.1k
                        xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9238
28.1k
                                 "Namespace prefix %s for %s on %s is not "
9239
28.1k
                                 "defined\n",
9240
28.1k
                                 aprefix, attname, localname);
9241
28.1k
                        nsIndex = NS_INDEX_EMPTY;
9242
28.1k
                        nsuri = NULL;
9243
28.1k
                        uriHashValue = URI_HASH_EMPTY;
9244
28.1k
                    } else {
9245
8.37k
                        nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9246
8.37k
                        uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9247
8.37k
                    }
9248
36.4k
                }
9249
9250
                /*
9251
                 * Check whether the attribute exists
9252
                 */
9253
342k
                if (maxAtts > 1) {
9254
275k
                    hashValue = xmlDictCombineHash(attr->name.hashValue,
9255
275k
                                                   uriHashValue);
9256
275k
                    res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9257
275k
                                            hashValue, nbatts);
9258
275k
                    if (res < 0)
9259
0
                        continue;
9260
275k
                    if (res < INT_MAX) {
9261
3.15k
                        if (aprefix == atts[res+1])
9262
2.57k
                            continue;
9263
576
                        xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9264
576
                                 "Namespaced Attribute %s in '%s' redefined\n",
9265
576
                                 attname, nsuri, NULL);
9266
576
                    }
9267
275k
                }
9268
9269
339k
                xmlParserEntityCheck(ctxt, attr->expandedSize);
9270
9271
339k
                if ((atts == NULL) || (nbatts + 5 > maxatts)) {
9272
4.25k
                    res = xmlCtxtGrowAttrs(ctxt);
9273
9274
4.25k
                    maxatts = ctxt->maxatts;
9275
4.25k
                    atts = ctxt->atts;
9276
9277
4.25k
                    if (res < 0) {
9278
23
                        localname = NULL;
9279
23
                        goto done;
9280
23
                    }
9281
4.25k
                }
9282
9283
339k
                atts[nbatts++] = attname;
9284
339k
                atts[nbatts++] = aprefix;
9285
339k
                atts[nbatts++] = XML_INT_TO_PTR(nsIndex);
9286
339k
                atts[nbatts++] = attr->value.name;
9287
339k
                atts[nbatts++] = attr->valueEnd;
9288
9289
339k
#ifdef LIBXML_VALID_ENABLED
9290
                /*
9291
                 * This should be moved to valid.c, but we don't keep track
9292
                 * whether an attribute was defaulted.
9293
                 */
9294
339k
                if ((ctxt->validate) &&
9295
217k
                    (ctxt->standalone == 1) &&
9296
452
                    (attr->external != 0)) {
9297
256
                    xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED,
9298
256
                            "standalone: attribute %s on %s defaulted "
9299
256
                            "from external subset\n",
9300
256
                            attname, localname);
9301
256
                }
9302
339k
#endif
9303
339k
                nbdef++;
9304
339k
      }
9305
414k
  }
9306
473k
    }
9307
9308
    /*
9309
     * Using a single hash table for nsUri/localName pairs cannot
9310
     * detect duplicate QNames reliably. The following example will
9311
     * only result in two namespace errors.
9312
     *
9313
     * <doc xmlns:a="a" xmlns:b="a">
9314
     *   <elem a:a="" b:a="" b:a=""/>
9315
     * </doc>
9316
     *
9317
     * If we saw more than one namespace error but no duplicate QNames
9318
     * were found, we have to scan for duplicate QNames.
9319
     */
9320
741k
    if ((numDupErr == 0) && (numNsErr > 1)) {
9321
884
        memset(ctxt->attrHash, -1,
9322
884
               attrHashSize * sizeof(ctxt->attrHash[0]));
9323
9324
5.84k
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9325
4.95k
            unsigned hashValue, nameHashValue, prefixHashValue;
9326
4.95k
            int res;
9327
9328
4.95k
            aprefix = atts[i+1];
9329
4.95k
            if (aprefix == NULL)
9330
720
                continue;
9331
9332
4.23k
            attname = atts[i];
9333
            /* Hash values always have bit 31 set, see dict.c */
9334
4.23k
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9335
4.23k
            prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix);
9336
9337
4.23k
            hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue);
9338
4.23k
            res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname,
9339
4.23k
                                         aprefix, hashValue, i);
9340
4.23k
            if (res < INT_MAX)
9341
1.33k
                xmlErrAttributeDup(ctxt, aprefix, attname);
9342
4.23k
        }
9343
884
    }
9344
9345
    /*
9346
     * Reconstruct attribute pointers
9347
     */
9348
1.18M
    for (i = 0, j = 0; i < nbatts; i += 5, j++) {
9349
        /* namespace URI */
9350
447k
        nsIndex = XML_PTR_TO_INT(atts[i+2]);
9351
447k
        if (nsIndex == INT_MAX)
9352
411k
            atts[i+2] = NULL;
9353
36.1k
        else if (nsIndex == INT_MAX - 1)
9354
20.1k
            atts[i+2] = ctxt->str_xml_ns;
9355
16.0k
        else
9356
16.0k
            atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1];
9357
9358
447k
        if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) {
9359
88.7k
            atts[i+3] = BASE_PTR + XML_PTR_TO_INT(atts[i+3]);  /* value */
9360
88.7k
            atts[i+4] = BASE_PTR + XML_PTR_TO_INT(atts[i+4]);  /* valuend */
9361
88.7k
        }
9362
447k
    }
9363
9364
741k
    uri = xmlParserNsLookupUri(ctxt, &hprefix);
9365
741k
    if ((prefix != NULL) && (uri == NULL)) {
9366
21.1k
  xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9367
21.1k
           "Namespace prefix %s on %s is not defined\n",
9368
21.1k
     prefix, localname, NULL);
9369
21.1k
    }
9370
741k
    *pref = prefix;
9371
741k
    *URI = uri;
9372
9373
    /*
9374
     * SAX callback
9375
     */
9376
741k
    if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) &&
9377
741k
  (!ctxt->disableSAX)) {
9378
424k
  if (nbNs > 0)
9379
111k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9380
111k
                          nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs),
9381
111k
        nbatts / 5, nbdef, atts);
9382
312k
  else
9383
312k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9384
312k
                          0, NULL, nbatts / 5, nbdef, atts);
9385
424k
    }
9386
9387
741k
done:
9388
    /*
9389
     * Free allocated attribute values
9390
     */
9391
741k
    if (attval != 0) {
9392
66.0k
  for (i = 0, j = 0; j < nratts; i += 5, j++)
9393
48.6k
      if (ctxt->attallocs[j] & 0x80000000)
9394
18.4k
          xmlFree((xmlChar *) atts[i+3]);
9395
17.4k
    }
9396
9397
741k
    *nbNsPtr = nbNs;
9398
741k
    return(localname);
9399
741k
}
9400
9401
/**
9402
 * Parse an end tag. Always consumes '</'.
9403
 *
9404
 *     [42] ETag ::= '</' Name S? '>'
9405
 *
9406
 * With namespace
9407
 *
9408
 *     [NS 9] ETag ::= '</' QName S? '>'
9409
 * @param ctxt  an XML parser context
9410
 * @param tag  the corresponding start tag
9411
 */
9412
9413
static void
9414
44.2k
xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) {
9415
44.2k
    const xmlChar *name;
9416
9417
44.2k
    GROW;
9418
44.2k
    if ((RAW != '<') || (NXT(1) != '/')) {
9419
1.90k
  xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL);
9420
1.90k
  return;
9421
1.90k
    }
9422
42.3k
    SKIP(2);
9423
9424
42.3k
    if (tag->prefix == NULL)
9425
34.1k
        name = xmlParseNameAndCompare(ctxt, ctxt->name);
9426
8.11k
    else
9427
8.11k
        name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix);
9428
9429
    /*
9430
     * We should definitely be at the ending "S? '>'" part
9431
     */
9432
42.3k
    GROW;
9433
42.3k
    SKIP_BLANKS;
9434
42.3k
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
9435
10.2k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
9436
10.2k
    } else
9437
32.0k
  NEXT1;
9438
9439
    /*
9440
     * [ WFC: Element Type Match ]
9441
     * The Name in an element's end-tag must match the element type in the
9442
     * start-tag.
9443
     *
9444
     */
9445
42.3k
    if (name != (xmlChar*)1) {
9446
11.6k
        if (name == NULL) name = BAD_CAST "unparsable";
9447
11.6k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
9448
11.6k
         "Opening and ending tag mismatch: %s line %d and %s\n",
9449
11.6k
                    ctxt->name, tag->line, name);
9450
11.6k
    }
9451
9452
    /*
9453
     * SAX: End of Tag
9454
     */
9455
42.3k
    if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9456
42.3k
  (!ctxt->disableSAX))
9457
33.2k
  ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix,
9458
33.2k
                                tag->URI);
9459
9460
42.3k
    spacePop(ctxt);
9461
42.3k
    if (tag->nsNr != 0)
9462
3.21k
  xmlParserNsPop(ctxt, tag->nsNr);
9463
42.3k
}
9464
9465
/**
9466
 * Parse escaped pure raw content. Always consumes '<!['.
9467
 *
9468
 * @deprecated Internal function, don't use.
9469
 *
9470
 *     [18] CDSect ::= CDStart CData CDEnd
9471
 *
9472
 *     [19] CDStart ::= '<![CDATA['
9473
 *
9474
 *     [20] Data ::= (Char* - (Char* ']]>' Char*))
9475
 *
9476
 *     [21] CDEnd ::= ']]>'
9477
 * @param ctxt  an XML parser context
9478
 */
9479
void
9480
46.3k
xmlParseCDSect(xmlParserCtxt *ctxt) {
9481
46.3k
    xmlChar *buf = NULL;
9482
46.3k
    int len = 0;
9483
46.3k
    int size = XML_PARSER_BUFFER_SIZE;
9484
46.3k
    int r, rl;
9485
46.3k
    int s, sl;
9486
46.3k
    int cur, l;
9487
46.3k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9488
14.5k
                    XML_MAX_HUGE_LENGTH :
9489
46.3k
                    XML_MAX_TEXT_LENGTH;
9490
9491
46.3k
    if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '['))
9492
0
        return;
9493
46.3k
    SKIP(3);
9494
9495
46.3k
    if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '['))
9496
0
        return;
9497
46.3k
    SKIP(6);
9498
9499
46.3k
    r = xmlCurrentCharRecover(ctxt, &rl);
9500
46.3k
    if (!IS_CHAR(r)) {
9501
623
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9502
623
        goto out;
9503
623
    }
9504
45.7k
    NEXTL(rl);
9505
45.7k
    s = xmlCurrentCharRecover(ctxt, &sl);
9506
45.7k
    if (!IS_CHAR(s)) {
9507
684
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9508
684
        goto out;
9509
684
    }
9510
45.0k
    NEXTL(sl);
9511
45.0k
    cur = xmlCurrentCharRecover(ctxt, &l);
9512
45.0k
    buf = xmlMalloc(size);
9513
45.0k
    if (buf == NULL) {
9514
24
  xmlErrMemory(ctxt);
9515
24
        goto out;
9516
24
    }
9517
1.97M
    while (IS_CHAR(cur) &&
9518
1.94M
           ((r != ']') || (s != ']') || (cur != '>'))) {
9519
1.92M
  if (len + 5 >= size) {
9520
5.43k
      xmlChar *tmp;
9521
5.43k
            int newSize;
9522
9523
5.43k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9524
5.43k
            if (newSize < 0) {
9525
0
                xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9526
0
                               "CData section too big found\n");
9527
0
                goto out;
9528
0
            }
9529
5.43k
      tmp = xmlRealloc(buf, newSize);
9530
5.43k
      if (tmp == NULL) {
9531
16
    xmlErrMemory(ctxt);
9532
16
                goto out;
9533
16
      }
9534
5.42k
      buf = tmp;
9535
5.42k
      size = newSize;
9536
5.42k
  }
9537
1.92M
  COPY_BUF(buf, len, r);
9538
1.92M
  r = s;
9539
1.92M
  rl = sl;
9540
1.92M
  s = cur;
9541
1.92M
  sl = l;
9542
1.92M
  NEXTL(l);
9543
1.92M
  cur = xmlCurrentCharRecover(ctxt, &l);
9544
1.92M
    }
9545
45.0k
    buf[len] = 0;
9546
45.0k
    if (cur != '>') {
9547
24.2k
  xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9548
24.2k
                       "CData section not finished\n%.50s\n", buf);
9549
24.2k
        goto out;
9550
24.2k
    }
9551
20.7k
    NEXTL(l);
9552
9553
    /*
9554
     * OK the buffer is to be consumed as cdata.
9555
     */
9556
20.7k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) {
9557
19.1k
        if ((ctxt->sax->cdataBlock != NULL) &&
9558
19.1k
            ((ctxt->options & XML_PARSE_NOCDATA) == 0)) {
9559
12.5k
            ctxt->sax->cdataBlock(ctxt->userData, buf, len);
9560
12.5k
        } else if (ctxt->sax->characters != NULL) {
9561
6.64k
            ctxt->sax->characters(ctxt->userData, buf, len);
9562
6.64k
        }
9563
19.1k
    }
9564
9565
46.3k
out:
9566
46.3k
    xmlFree(buf);
9567
46.3k
}
9568
9569
/**
9570
 * Parse a content sequence. Stops at EOF or '</'. Leaves checking of
9571
 * unexpected EOF to the caller.
9572
 *
9573
 * @param ctxt  an XML parser context
9574
 */
9575
9576
static void
9577
31.7k
xmlParseContentInternal(xmlParserCtxtPtr ctxt) {
9578
31.7k
    int oldNameNr = ctxt->nameNr;
9579
31.7k
    int oldSpaceNr = ctxt->spaceNr;
9580
31.7k
    int oldNodeNr = ctxt->nodeNr;
9581
9582
31.7k
    GROW;
9583
16.0M
    while ((ctxt->input->cur < ctxt->input->end) &&
9584
15.9M
     (PARSER_STOPPED(ctxt) == 0)) {
9585
15.9M
  const xmlChar *cur = ctxt->input->cur;
9586
9587
  /*
9588
   * First case : a Processing Instruction.
9589
   */
9590
15.9M
  if ((*cur == '<') && (cur[1] == '?')) {
9591
22.3k
      xmlParsePI(ctxt);
9592
22.3k
  }
9593
9594
  /*
9595
   * Second case : a CDSection
9596
   */
9597
  /* 2.6.0 test was *cur not RAW */
9598
15.9M
  else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) {
9599
41.6k
      xmlParseCDSect(ctxt);
9600
41.6k
  }
9601
9602
  /*
9603
   * Third case :  a comment
9604
   */
9605
15.9M
  else if ((*cur == '<') && (NXT(1) == '!') &&
9606
132k
     (NXT(2) == '-') && (NXT(3) == '-')) {
9607
86.9k
      xmlParseComment(ctxt);
9608
86.9k
  }
9609
9610
  /*
9611
   * Fourth case :  a sub-element.
9612
   */
9613
15.8M
  else if (*cur == '<') {
9614
970k
            if (NXT(1) == '/') {
9615
49.5k
                if (ctxt->nameNr <= oldNameNr)
9616
1.30k
                    break;
9617
48.2k
          xmlParseElementEnd(ctxt);
9618
920k
            } else {
9619
920k
          xmlParseElementStart(ctxt);
9620
920k
            }
9621
970k
  }
9622
9623
  /*
9624
   * Fifth case : a reference. If if has not been resolved,
9625
   *    parsing returns it's Name, create the node
9626
   */
9627
9628
14.8M
  else if (*cur == '&') {
9629
180k
      xmlParseReference(ctxt);
9630
180k
  }
9631
9632
  /*
9633
   * Last case, text. Note that References are handled directly.
9634
   */
9635
14.6M
  else {
9636
14.6M
      xmlParseCharDataInternal(ctxt, 0);
9637
14.6M
  }
9638
9639
15.9M
  SHRINK;
9640
15.9M
  GROW;
9641
15.9M
    }
9642
9643
31.7k
    if ((ctxt->nameNr > oldNameNr) &&
9644
13.4k
        (ctxt->input->cur >= ctxt->input->end) &&
9645
12.3k
        (ctxt->wellFormed)) {
9646
515
        const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9647
515
        int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9648
515
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9649
515
                "Premature end of data in tag %s line %d\n",
9650
515
                name, line, NULL);
9651
515
    }
9652
9653
    /*
9654
     * Clean up in error case
9655
     */
9656
9657
282k
    while (ctxt->nodeNr > oldNodeNr)
9658
250k
        nodePop(ctxt);
9659
9660
337k
    while (ctxt->nameNr > oldNameNr) {
9661
306k
        xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1];
9662
9663
306k
        if (tag->nsNr != 0)
9664
62.4k
            xmlParserNsPop(ctxt, tag->nsNr);
9665
9666
306k
        namePop(ctxt);
9667
306k
    }
9668
9669
338k
    while (ctxt->spaceNr > oldSpaceNr)
9670
306k
        spacePop(ctxt);
9671
31.7k
}
9672
9673
/**
9674
 * Parse XML element content. This is useful if you're only interested
9675
 * in custom SAX callbacks. If you want a node list, use
9676
 * #xmlCtxtParseContent.
9677
 *
9678
 * @param ctxt  an XML parser context
9679
 */
9680
void
9681
0
xmlParseContent(xmlParserCtxt *ctxt) {
9682
0
    if ((ctxt == NULL) || (ctxt->input == NULL))
9683
0
        return;
9684
9685
0
    xmlCtxtInitializeLate(ctxt);
9686
9687
0
    xmlParseContentInternal(ctxt);
9688
9689
0
    xmlParserCheckEOF(ctxt, XML_ERR_NOT_WELL_BALANCED);
9690
0
}
9691
9692
/**
9693
 * Parse an XML element
9694
 *
9695
 * @deprecated Internal function, don't use.
9696
 *
9697
 *     [39] element ::= EmptyElemTag | STag content ETag
9698
 *
9699
 * [ WFC: Element Type Match ]
9700
 * The Name in an element's end-tag must match the element type in the
9701
 * start-tag.
9702
 *
9703
 * @param ctxt  an XML parser context
9704
 */
9705
9706
void
9707
37.6k
xmlParseElement(xmlParserCtxt *ctxt) {
9708
37.6k
    if (xmlParseElementStart(ctxt) != 0)
9709
14.8k
        return;
9710
9711
22.8k
    xmlParseContentInternal(ctxt);
9712
9713
22.8k
    if (ctxt->input->cur >= ctxt->input->end) {
9714
18.3k
        if (ctxt->wellFormed) {
9715
1.07k
            const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9716
1.07k
            int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9717
1.07k
            xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9718
1.07k
                    "Premature end of data in tag %s line %d\n",
9719
1.07k
                    name, line, NULL);
9720
1.07k
        }
9721
18.3k
        return;
9722
18.3k
    }
9723
9724
4.48k
    xmlParseElementEnd(ctxt);
9725
4.48k
}
9726
9727
/**
9728
 * Parse the start of an XML element. Returns -1 in case of error, 0 if an
9729
 * opening tag was parsed, 1 if an empty element was parsed.
9730
 *
9731
 * Always consumes '<'.
9732
 *
9733
 * @param ctxt  an XML parser context
9734
 */
9735
static int
9736
958k
xmlParseElementStart(xmlParserCtxtPtr ctxt) {
9737
958k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
9738
958k
    const xmlChar *name;
9739
958k
    const xmlChar *prefix = NULL;
9740
958k
    const xmlChar *URI = NULL;
9741
958k
    xmlParserNodeInfo node_info;
9742
958k
    int line;
9743
958k
    xmlNodePtr cur;
9744
958k
    int nbNs = 0;
9745
9746
958k
    if (ctxt->nameNr > maxDepth) {
9747
17
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
9748
17
                "Excessive depth in document: %d use XML_PARSE_HUGE option\n",
9749
17
                ctxt->nameNr);
9750
17
  return(-1);
9751
17
    }
9752
9753
    /* Capture start position */
9754
958k
    if (ctxt->record_info) {
9755
0
        node_info.begin_pos = ctxt->input->consumed +
9756
0
                          (CUR_PTR - ctxt->input->base);
9757
0
  node_info.begin_line = ctxt->input->line;
9758
0
    }
9759
9760
958k
    if (ctxt->spaceNr == 0)
9761
37.6k
  spacePush(ctxt, -1);
9762
920k
    else if (*ctxt->space == -2)
9763
116k
  spacePush(ctxt, -1);
9764
804k
    else
9765
804k
  spacePush(ctxt, *ctxt->space);
9766
9767
958k
    line = ctxt->input->line;
9768
958k
#ifdef LIBXML_SAX1_ENABLED
9769
958k
    if (ctxt->sax2)
9770
740k
#endif /* LIBXML_SAX1_ENABLED */
9771
740k
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
9772
218k
#ifdef LIBXML_SAX1_ENABLED
9773
218k
    else
9774
218k
  name = xmlParseStartTag(ctxt);
9775
958k
#endif /* LIBXML_SAX1_ENABLED */
9776
958k
    if (name == NULL) {
9777
102k
  spacePop(ctxt);
9778
102k
        return(-1);
9779
102k
    }
9780
855k
    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
9781
855k
    cur = ctxt->node;
9782
9783
855k
#ifdef LIBXML_VALID_ENABLED
9784
    /*
9785
     * [ VC: Root Element Type ]
9786
     * The Name in the document type declaration must match the element
9787
     * type of the root element.
9788
     */
9789
855k
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
9790
34.9k
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
9791
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
9792
855k
#endif /* LIBXML_VALID_ENABLED */
9793
9794
    /*
9795
     * Check for an Empty Element.
9796
     */
9797
855k
    if ((RAW == '/') && (NXT(1) == '>')) {
9798
62.2k
        SKIP(2);
9799
62.2k
  if (ctxt->sax2) {
9800
55.3k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9801
55.3k
    (!ctxt->disableSAX))
9802
52.9k
    ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI);
9803
55.3k
#ifdef LIBXML_SAX1_ENABLED
9804
55.3k
  } else {
9805
6.83k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
9806
6.83k
    (!ctxt->disableSAX))
9807
6.17k
    ctxt->sax->endElement(ctxt->userData, name);
9808
6.83k
#endif /* LIBXML_SAX1_ENABLED */
9809
6.83k
  }
9810
62.2k
  namePop(ctxt);
9811
62.2k
  spacePop(ctxt);
9812
62.2k
  if (nbNs > 0)
9813
29.2k
      xmlParserNsPop(ctxt, nbNs);
9814
62.2k
  if (cur != NULL && ctxt->record_info) {
9815
0
            node_info.node = cur;
9816
0
            node_info.end_pos = ctxt->input->consumed +
9817
0
                                (CUR_PTR - ctxt->input->base);
9818
0
            node_info.end_line = ctxt->input->line;
9819
0
            xmlParserAddNodeInfo(ctxt, &node_info);
9820
0
  }
9821
62.2k
  return(1);
9822
62.2k
    }
9823
793k
    if (RAW == '>') {
9824
377k
        NEXT1;
9825
377k
        if (cur != NULL && ctxt->record_info) {
9826
0
            node_info.node = cur;
9827
0
            node_info.end_pos = 0;
9828
0
            node_info.end_line = 0;
9829
0
            xmlParserAddNodeInfo(ctxt, &node_info);
9830
0
        }
9831
416k
    } else {
9832
416k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED,
9833
416k
         "Couldn't find end of Start Tag %s line %d\n",
9834
416k
                    name, line, NULL);
9835
9836
  /*
9837
   * end of parsing of this node.
9838
   */
9839
416k
  nodePop(ctxt);
9840
416k
  namePop(ctxt);
9841
416k
  spacePop(ctxt);
9842
416k
  if (nbNs > 0)
9843
252k
      xmlParserNsPop(ctxt, nbNs);
9844
416k
  return(-1);
9845
416k
    }
9846
9847
377k
    return(0);
9848
793k
}
9849
9850
/**
9851
 * Parse the end of an XML element. Always consumes '</'.
9852
 *
9853
 * @param ctxt  an XML parser context
9854
 */
9855
static void
9856
52.6k
xmlParseElementEnd(xmlParserCtxtPtr ctxt) {
9857
52.6k
    xmlNodePtr cur = ctxt->node;
9858
9859
52.6k
    if (ctxt->nameNr <= 0) {
9860
130
        if ((RAW == '<') && (NXT(1) == '/'))
9861
40
            SKIP(2);
9862
130
        return;
9863
130
    }
9864
9865
    /*
9866
     * parse the end of tag: '</' should be here.
9867
     */
9868
52.5k
    if (ctxt->sax2) {
9869
37.1k
  xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
9870
37.1k
  namePop(ctxt);
9871
37.1k
    }
9872
15.3k
#ifdef LIBXML_SAX1_ENABLED
9873
15.3k
    else
9874
15.3k
  xmlParseEndTag1(ctxt, 0);
9875
52.5k
#endif /* LIBXML_SAX1_ENABLED */
9876
9877
    /*
9878
     * Capture end position
9879
     */
9880
52.5k
    if (cur != NULL && ctxt->record_info) {
9881
0
        xmlParserNodeInfoPtr node_info;
9882
9883
0
        node_info = (xmlParserNodeInfoPtr) xmlParserFindNodeInfo(ctxt, cur);
9884
0
        if (node_info != NULL) {
9885
0
            node_info->end_pos = ctxt->input->consumed +
9886
0
                                 (CUR_PTR - ctxt->input->base);
9887
0
            node_info->end_line = ctxt->input->line;
9888
0
        }
9889
0
    }
9890
52.5k
}
9891
9892
/**
9893
 * Parse the XML version value.
9894
 *
9895
 * @deprecated Internal function, don't use.
9896
 *
9897
 *     [26] VersionNum ::= '1.' [0-9]+
9898
 *
9899
 * In practice allow [0-9].[0-9]+ at that level
9900
 *
9901
 * @param ctxt  an XML parser context
9902
 * @returns the string giving the XML version number, or NULL
9903
 */
9904
xmlChar *
9905
26.1k
xmlParseVersionNum(xmlParserCtxt *ctxt) {
9906
26.1k
    xmlChar *buf = NULL;
9907
26.1k
    int len = 0;
9908
26.1k
    int size = 10;
9909
26.1k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9910
11.6k
                    XML_MAX_TEXT_LENGTH :
9911
26.1k
                    XML_MAX_NAME_LENGTH;
9912
26.1k
    xmlChar cur;
9913
9914
26.1k
    buf = xmlMalloc(size);
9915
26.1k
    if (buf == NULL) {
9916
105
  xmlErrMemory(ctxt);
9917
105
  return(NULL);
9918
105
    }
9919
26.0k
    cur = CUR;
9920
26.0k
    if (!((cur >= '0') && (cur <= '9'))) {
9921
2.22k
  xmlFree(buf);
9922
2.22k
  return(NULL);
9923
2.22k
    }
9924
23.8k
    buf[len++] = cur;
9925
23.8k
    NEXT;
9926
23.8k
    cur=CUR;
9927
23.8k
    if (cur != '.') {
9928
384
  xmlFree(buf);
9929
384
  return(NULL);
9930
384
    }
9931
23.4k
    buf[len++] = cur;
9932
23.4k
    NEXT;
9933
23.4k
    cur=CUR;
9934
33.7M
    while ((cur >= '0') && (cur <= '9')) {
9935
33.7M
  if (len + 1 >= size) {
9936
18.8k
      xmlChar *tmp;
9937
18.8k
            int newSize;
9938
9939
18.8k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9940
18.8k
            if (newSize < 0) {
9941
275
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "VersionNum");
9942
275
                xmlFree(buf);
9943
275
                return(NULL);
9944
275
            }
9945
18.5k
      tmp = xmlRealloc(buf, newSize);
9946
18.5k
      if (tmp == NULL) {
9947
10
    xmlErrMemory(ctxt);
9948
10
          xmlFree(buf);
9949
10
    return(NULL);
9950
10
      }
9951
18.5k
      buf = tmp;
9952
18.5k
            size = newSize;
9953
18.5k
  }
9954
33.7M
  buf[len++] = cur;
9955
33.7M
  NEXT;
9956
33.7M
  cur=CUR;
9957
33.7M
    }
9958
23.1k
    buf[len] = 0;
9959
23.1k
    return(buf);
9960
23.4k
}
9961
9962
/**
9963
 * Parse the XML version.
9964
 *
9965
 * @deprecated Internal function, don't use.
9966
 *
9967
 *     [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ")
9968
 *
9969
 *     [25] Eq ::= S? '=' S?
9970
 *
9971
 * @param ctxt  an XML parser context
9972
 * @returns the version string, e.g. "1.0"
9973
 */
9974
9975
xmlChar *
9976
47.1k
xmlParseVersionInfo(xmlParserCtxt *ctxt) {
9977
47.1k
    xmlChar *version = NULL;
9978
9979
47.1k
    if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) {
9980
27.8k
  SKIP(7);
9981
27.8k
  SKIP_BLANKS;
9982
27.8k
  if (RAW != '=') {
9983
640
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
9984
640
      return(NULL);
9985
640
        }
9986
27.2k
  NEXT;
9987
27.2k
  SKIP_BLANKS;
9988
27.2k
  if (RAW == '"') {
9989
19.9k
      NEXT;
9990
19.9k
      version = xmlParseVersionNum(ctxt);
9991
19.9k
      if (RAW != '"') {
9992
1.68k
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9993
1.68k
      } else
9994
18.2k
          NEXT;
9995
19.9k
  } else if (RAW == '\''){
9996
6.18k
      NEXT;
9997
6.18k
      version = xmlParseVersionNum(ctxt);
9998
6.18k
      if (RAW != '\'') {
9999
1.98k
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10000
1.98k
      } else
10001
4.19k
          NEXT;
10002
6.18k
  } else {
10003
1.10k
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10004
1.10k
  }
10005
27.2k
    }
10006
46.5k
    return(version);
10007
47.1k
}
10008
10009
/**
10010
 * Parse the XML encoding name
10011
 *
10012
 * @deprecated Internal function, don't use.
10013
 *
10014
 *     [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')*
10015
 *
10016
 * @param ctxt  an XML parser context
10017
 * @returns the encoding name value or NULL
10018
 */
10019
xmlChar *
10020
24.8k
xmlParseEncName(xmlParserCtxt *ctxt) {
10021
24.8k
    xmlChar *buf = NULL;
10022
24.8k
    int len = 0;
10023
24.8k
    int size = 10;
10024
24.8k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
10025
11.5k
                    XML_MAX_TEXT_LENGTH :
10026
24.8k
                    XML_MAX_NAME_LENGTH;
10027
24.8k
    xmlChar cur;
10028
10029
24.8k
    cur = CUR;
10030
24.8k
    if (((cur >= 'a') && (cur <= 'z')) ||
10031
24.2k
        ((cur >= 'A') && (cur <= 'Z'))) {
10032
24.2k
  buf = xmlMalloc(size);
10033
24.2k
  if (buf == NULL) {
10034
66
      xmlErrMemory(ctxt);
10035
66
      return(NULL);
10036
66
  }
10037
10038
24.1k
  buf[len++] = cur;
10039
24.1k
  NEXT;
10040
24.1k
  cur = CUR;
10041
25.9M
  while (((cur >= 'a') && (cur <= 'z')) ||
10042
130k
         ((cur >= 'A') && (cur <= 'Z')) ||
10043
78.6k
         ((cur >= '0') && (cur <= '9')) ||
10044
38.7k
         (cur == '.') || (cur == '_') ||
10045
25.9M
         (cur == '-')) {
10046
25.9M
      if (len + 1 >= size) {
10047
19.5k
          xmlChar *tmp;
10048
19.5k
                int newSize;
10049
10050
19.5k
                newSize = xmlGrowCapacity(size, 1, 1, maxLength);
10051
19.5k
                if (newSize < 0) {
10052
303
                    xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName");
10053
303
                    xmlFree(buf);
10054
303
                    return(NULL);
10055
303
                }
10056
19.2k
    tmp = xmlRealloc(buf, newSize);
10057
19.2k
    if (tmp == NULL) {
10058
17
        xmlErrMemory(ctxt);
10059
17
        xmlFree(buf);
10060
17
        return(NULL);
10061
17
    }
10062
19.2k
    buf = tmp;
10063
19.2k
                size = newSize;
10064
19.2k
      }
10065
25.9M
      buf[len++] = cur;
10066
25.9M
      NEXT;
10067
25.9M
      cur = CUR;
10068
25.9M
        }
10069
23.8k
  buf[len] = 0;
10070
23.8k
    } else {
10071
635
  xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL);
10072
635
    }
10073
24.4k
    return(buf);
10074
24.8k
}
10075
10076
/**
10077
 * Parse the XML encoding declaration
10078
 *
10079
 * @deprecated Internal function, don't use.
10080
 *
10081
 *     [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | 
10082
 *                           "'" EncName "'")
10083
 *
10084
 * this setups the conversion filters.
10085
 *
10086
 * @param ctxt  an XML parser context
10087
 * @returns the encoding value or NULL
10088
 */
10089
10090
const xmlChar *
10091
46.3k
xmlParseEncodingDecl(xmlParserCtxt *ctxt) {
10092
46.3k
    xmlChar *encoding = NULL;
10093
10094
46.3k
    SKIP_BLANKS;
10095
46.3k
    if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0)
10096
19.8k
        return(NULL);
10097
10098
26.4k
    SKIP(8);
10099
26.4k
    SKIP_BLANKS;
10100
26.4k
    if (RAW != '=') {
10101
1.04k
        xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10102
1.04k
        return(NULL);
10103
1.04k
    }
10104
25.3k
    NEXT;
10105
25.3k
    SKIP_BLANKS;
10106
25.3k
    if (RAW == '"') {
10107
18.6k
        NEXT;
10108
18.6k
        encoding = xmlParseEncName(ctxt);
10109
18.6k
        if (RAW != '"') {
10110
1.08k
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10111
1.08k
            xmlFree(encoding);
10112
1.08k
            return(NULL);
10113
1.08k
        } else
10114
17.5k
            NEXT;
10115
18.6k
    } else if (RAW == '\''){
10116
6.24k
        NEXT;
10117
6.24k
        encoding = xmlParseEncName(ctxt);
10118
6.24k
        if (RAW != '\'') {
10119
466
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10120
466
            xmlFree(encoding);
10121
466
            return(NULL);
10122
466
        } else
10123
5.78k
            NEXT;
10124
6.24k
    } else {
10125
520
        xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10126
520
    }
10127
10128
23.8k
    if (encoding == NULL)
10129
654
        return(NULL);
10130
10131
23.1k
    xmlSetDeclaredEncoding(ctxt, encoding);
10132
10133
23.1k
    return(ctxt->encoding);
10134
23.8k
}
10135
10136
/**
10137
 * Parse the XML standalone declaration
10138
 *
10139
 * @deprecated Internal function, don't use.
10140
 *
10141
 *     [32] SDDecl ::= S 'standalone' Eq
10142
 *                     (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"'))
10143
 *
10144
 * [ VC: Standalone Document Declaration ]
10145
 * TODO The standalone document declaration must have the value "no"
10146
 * if any external markup declarations contain declarations of:
10147
 *  - attributes with default values, if elements to which these
10148
 *    attributes apply appear in the document without specifications
10149
 *    of values for these attributes, or
10150
 *  - entities (other than amp, lt, gt, apos, quot), if references
10151
 *    to those entities appear in the document, or
10152
 *  - attributes with values subject to normalization, where the
10153
 *    attribute appears in the document with a value which will change
10154
 *    as a result of normalization, or
10155
 *  - element types with element content, if white space occurs directly
10156
 *    within any instance of those types.
10157
 *
10158
 * @param ctxt  an XML parser context
10159
 * @returns
10160
 *   1 if standalone="yes"
10161
 *   0 if standalone="no"
10162
 *  -2 if standalone attribute is missing or invalid
10163
 *    (A standalone value of -2 means that the XML declaration was found,
10164
 *     but no value was specified for the standalone attribute).
10165
 */
10166
10167
int
10168
4.23k
xmlParseSDDecl(xmlParserCtxt *ctxt) {
10169
4.23k
    int standalone = -2;
10170
10171
4.23k
    SKIP_BLANKS;
10172
4.23k
    if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) {
10173
907
  SKIP(10);
10174
907
        SKIP_BLANKS;
10175
907
  if (RAW != '=') {
10176
15
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10177
15
      return(standalone);
10178
15
        }
10179
892
  NEXT;
10180
892
  SKIP_BLANKS;
10181
892
        if (RAW == '\''){
10182
124
      NEXT;
10183
124
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10184
69
          standalone = 0;
10185
69
                SKIP(2);
10186
69
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10187
24
                 (NXT(2) == 's')) {
10188
12
          standalone = 1;
10189
12
    SKIP(3);
10190
43
            } else {
10191
43
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10192
43
      }
10193
124
      if (RAW != '\'') {
10194
61
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10195
61
      } else
10196
63
          NEXT;
10197
768
  } else if (RAW == '"'){
10198
762
      NEXT;
10199
762
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10200
12
          standalone = 0;
10201
12
    SKIP(2);
10202
750
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10203
732
                 (NXT(2) == 's')) {
10204
726
          standalone = 1;
10205
726
                SKIP(3);
10206
726
            } else {
10207
24
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10208
24
      }
10209
762
      if (RAW != '"') {
10210
83
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10211
83
      } else
10212
679
          NEXT;
10213
762
  } else {
10214
6
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10215
6
        }
10216
892
    }
10217
4.21k
    return(standalone);
10218
4.23k
}
10219
10220
/**
10221
 * Parse an XML declaration header
10222
 *
10223
 * @deprecated Internal function, don't use.
10224
 *
10225
 *     [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>'
10226
 * @param ctxt  an XML parser context
10227
 */
10228
10229
void
10230
8.43k
xmlParseXMLDecl(xmlParserCtxt *ctxt) {
10231
8.43k
    xmlChar *version;
10232
10233
    /*
10234
     * This value for standalone indicates that the document has an
10235
     * XML declaration but it does not have a standalone attribute.
10236
     * It will be overwritten later if a standalone attribute is found.
10237
     */
10238
10239
8.43k
    ctxt->standalone = -2;
10240
10241
    /*
10242
     * We know that '<?xml' is here.
10243
     */
10244
8.43k
    SKIP(5);
10245
10246
8.43k
    if (!IS_BLANK_CH(RAW)) {
10247
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
10248
0
                 "Blank needed after '<?xml'\n");
10249
0
    }
10250
8.43k
    SKIP_BLANKS;
10251
10252
    /*
10253
     * We must have the VersionInfo here.
10254
     */
10255
8.43k
    version = xmlParseVersionInfo(ctxt);
10256
8.43k
    if (version == NULL) {
10257
3.99k
  xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL);
10258
4.44k
    } else {
10259
4.44k
  if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) {
10260
      /*
10261
       * Changed here for XML-1.0 5th edition
10262
       */
10263
1.73k
      if (ctxt->options & XML_PARSE_OLD10) {
10264
227
    xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10265
227
                "Unsupported version '%s'\n",
10266
227
                version);
10267
1.51k
      } else {
10268
1.51k
          if ((version[0] == '1') && ((version[1] == '.'))) {
10269
585
        xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION,
10270
585
                      "Unsupported version '%s'\n",
10271
585
          version, NULL);
10272
927
    } else {
10273
927
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10274
927
              "Unsupported version '%s'\n",
10275
927
              version);
10276
927
    }
10277
1.51k
      }
10278
1.73k
  }
10279
4.44k
  if (ctxt->version != NULL)
10280
0
      xmlFree(ctxt->version);
10281
4.44k
  ctxt->version = version;
10282
4.44k
    }
10283
10284
    /*
10285
     * We may have the encoding declaration
10286
     */
10287
8.43k
    if (!IS_BLANK_CH(RAW)) {
10288
4.88k
        if ((RAW == '?') && (NXT(1) == '>')) {
10289
822
      SKIP(2);
10290
822
      return;
10291
822
  }
10292
4.06k
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10293
4.06k
    }
10294
7.61k
    xmlParseEncodingDecl(ctxt);
10295
10296
    /*
10297
     * We may have the standalone status.
10298
     */
10299
7.61k
    if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) {
10300
4.05k
        if ((RAW == '?') && (NXT(1) == '>')) {
10301
3.37k
      SKIP(2);
10302
3.37k
      return;
10303
3.37k
  }
10304
681
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10305
681
    }
10306
10307
    /*
10308
     * We can grow the input buffer freely at that point
10309
     */
10310
4.23k
    GROW;
10311
10312
4.23k
    SKIP_BLANKS;
10313
4.23k
    ctxt->standalone = xmlParseSDDecl(ctxt);
10314
10315
4.23k
    SKIP_BLANKS;
10316
4.23k
    if ((RAW == '?') && (NXT(1) == '>')) {
10317
749
        SKIP(2);
10318
3.48k
    } else if (RAW == '>') {
10319
        /* Deprecated old WD ... */
10320
403
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10321
403
  NEXT;
10322
3.08k
    } else {
10323
3.08k
        int c;
10324
10325
3.08k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10326
4.12M
        while ((PARSER_STOPPED(ctxt) == 0) &&
10327
4.12M
               ((c = CUR) != 0)) {
10328
4.12M
            NEXT;
10329
4.12M
            if (c == '>')
10330
1.49k
                break;
10331
4.12M
        }
10332
3.08k
    }
10333
4.23k
}
10334
10335
/**
10336
 * @since 2.14.0
10337
 *
10338
 * @param ctxt  parser context
10339
 * @returns the version from the XML declaration.
10340
 */
10341
const xmlChar *
10342
0
xmlCtxtGetVersion(xmlParserCtxt *ctxt) {
10343
0
    if (ctxt == NULL)
10344
0
        return(NULL);
10345
10346
0
    return(ctxt->version);
10347
0
}
10348
10349
/**
10350
 * @since 2.14.0
10351
 *
10352
 * @param ctxt  parser context
10353
 * @returns the value from the standalone document declaration.
10354
 */
10355
int
10356
0
xmlCtxtGetStandalone(xmlParserCtxt *ctxt) {
10357
0
    if (ctxt == NULL)
10358
0
        return(0);
10359
10360
0
    return(ctxt->standalone);
10361
0
}
10362
10363
/**
10364
 * Parse an XML Misc* optional field.
10365
 *
10366
 * @deprecated Internal function, don't use.
10367
 *
10368
 *     [27] Misc ::= Comment | PI |  S
10369
 * @param ctxt  an XML parser context
10370
 */
10371
10372
void
10373
131k
xmlParseMisc(xmlParserCtxt *ctxt) {
10374
244k
    while (PARSER_STOPPED(ctxt) == 0) {
10375
229k
        SKIP_BLANKS;
10376
229k
        GROW;
10377
229k
        if ((RAW == '<') && (NXT(1) == '?')) {
10378
15.1k
      xmlParsePI(ctxt);
10379
213k
        } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) {
10380
98.4k
      xmlParseComment(ctxt);
10381
115k
        } else {
10382
115k
            break;
10383
115k
        }
10384
229k
    }
10385
131k
}
10386
10387
static void
10388
75.6k
xmlFinishDocument(xmlParserCtxtPtr ctxt) {
10389
75.6k
    xmlDocPtr doc;
10390
10391
    /*
10392
     * SAX: end of the document processing.
10393
     */
10394
75.6k
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10395
75.6k
        ctxt->sax->endDocument(ctxt->userData);
10396
10397
    /*
10398
     * Remove locally kept entity definitions if the tree was not built
10399
     */
10400
75.6k
    doc = ctxt->myDoc;
10401
75.6k
    if ((doc != NULL) &&
10402
73.5k
        (xmlStrEqual(doc->version, SAX_COMPAT_MODE))) {
10403
520
        xmlFreeDoc(doc);
10404
520
        ctxt->myDoc = NULL;
10405
520
    }
10406
75.6k
}
10407
10408
/**
10409
 * Parse an XML document and invoke the SAX handlers. This is useful
10410
 * if you're only interested in custom SAX callbacks. If you want a
10411
 * document tree, use #xmlCtxtParseDocument.
10412
 *
10413
 * @param ctxt  an XML parser context
10414
 * @returns 0, -1 in case of error.
10415
 */
10416
10417
int
10418
58.7k
xmlParseDocument(xmlParserCtxt *ctxt) {
10419
58.7k
    if ((ctxt == NULL) || (ctxt->input == NULL))
10420
0
        return(-1);
10421
10422
58.7k
    GROW;
10423
10424
    /*
10425
     * SAX: detecting the level.
10426
     */
10427
58.7k
    xmlCtxtInitializeLate(ctxt);
10428
10429
58.7k
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10430
58.7k
        ctxt->sax->setDocumentLocator(ctxt->userData,
10431
58.7k
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10432
58.7k
    }
10433
10434
58.7k
    xmlDetectEncoding(ctxt);
10435
10436
58.7k
    if (CUR == 0) {
10437
580
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10438
580
  return(-1);
10439
580
    }
10440
10441
58.1k
    GROW;
10442
58.1k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
10443
10444
  /*
10445
   * Note that we will switch encoding on the fly.
10446
   */
10447
5.66k
  xmlParseXMLDecl(ctxt);
10448
5.66k
  SKIP_BLANKS;
10449
52.5k
    } else {
10450
52.5k
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10451
52.5k
        if (ctxt->version == NULL) {
10452
30
            xmlErrMemory(ctxt);
10453
30
            return(-1);
10454
30
        }
10455
52.5k
    }
10456
58.1k
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10457
56.0k
        ctxt->sax->startDocument(ctxt->userData);
10458
58.1k
    if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) &&
10459
55.8k
        (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) {
10460
0
  ctxt->myDoc->compression = ctxt->input->buf->compressed;
10461
0
    }
10462
10463
    /*
10464
     * The Misc part of the Prolog
10465
     */
10466
58.1k
    xmlParseMisc(ctxt);
10467
10468
    /*
10469
     * Then possibly doc type declaration(s) and more Misc
10470
     * (doctypedecl Misc*)?
10471
     */
10472
58.1k
    GROW;
10473
58.1k
    if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) {
10474
10475
35.2k
  ctxt->inSubset = 1;
10476
35.2k
  xmlParseDocTypeDecl(ctxt);
10477
35.2k
  if (RAW == '[') {
10478
28.7k
      xmlParseInternalSubset(ctxt);
10479
28.7k
  } else if (RAW == '>') {
10480
4.18k
            NEXT;
10481
4.18k
        }
10482
10483
  /*
10484
   * Create and update the external subset.
10485
   */
10486
35.2k
  ctxt->inSubset = 2;
10487
35.2k
  if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) &&
10488
35.2k
      (!ctxt->disableSAX))
10489
28.9k
      ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
10490
28.9k
                                ctxt->extSubSystem, ctxt->extSubURI);
10491
35.2k
  ctxt->inSubset = 0;
10492
10493
35.2k
        xmlCleanSpecialAttr(ctxt);
10494
10495
35.2k
  xmlParseMisc(ctxt);
10496
35.2k
    }
10497
10498
    /*
10499
     * Time to start parsing the tree itself
10500
     */
10501
58.1k
    GROW;
10502
58.1k
    if (RAW != '<') {
10503
20.5k
        if (ctxt->wellFormed)
10504
2.78k
            xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10505
2.78k
                           "Start tag expected, '<' not found\n");
10506
37.6k
    } else {
10507
37.6k
  xmlParseElement(ctxt);
10508
10509
  /*
10510
   * The Misc part at the end
10511
   */
10512
37.6k
  xmlParseMisc(ctxt);
10513
10514
37.6k
        xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
10515
37.6k
    }
10516
10517
58.1k
    ctxt->instate = XML_PARSER_EOF;
10518
58.1k
    xmlFinishDocument(ctxt);
10519
10520
58.1k
    if (! ctxt->wellFormed) {
10521
57.4k
  ctxt->valid = 0;
10522
57.4k
  return(-1);
10523
57.4k
    }
10524
10525
695
    return(0);
10526
58.1k
}
10527
10528
/**
10529
 * Parse a general parsed entity
10530
 * An external general parsed entity is well-formed if it matches the
10531
 * production labeled extParsedEnt.
10532
 *
10533
 * @deprecated Internal function, don't use.
10534
 *
10535
 *     [78] extParsedEnt ::= TextDecl? content
10536
 *
10537
 * @param ctxt  an XML parser context
10538
 * @returns 0, -1 in case of error. the parser context is augmented
10539
 *                as a result of the parsing.
10540
 */
10541
10542
int
10543
0
xmlParseExtParsedEnt(xmlParserCtxt *ctxt) {
10544
0
    if ((ctxt == NULL) || (ctxt->input == NULL))
10545
0
        return(-1);
10546
10547
0
    xmlCtxtInitializeLate(ctxt);
10548
10549
0
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10550
0
        ctxt->sax->setDocumentLocator(ctxt->userData,
10551
0
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10552
0
    }
10553
10554
0
    xmlDetectEncoding(ctxt);
10555
10556
0
    if (CUR == 0) {
10557
0
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10558
0
    }
10559
10560
    /*
10561
     * Check for the XMLDecl in the Prolog.
10562
     */
10563
0
    GROW;
10564
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
10565
10566
  /*
10567
   * Note that we will switch encoding on the fly.
10568
   */
10569
0
  xmlParseXMLDecl(ctxt);
10570
0
  SKIP_BLANKS;
10571
0
    } else {
10572
0
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10573
0
    }
10574
0
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10575
0
        ctxt->sax->startDocument(ctxt->userData);
10576
10577
    /*
10578
     * Doing validity checking on chunk doesn't make sense
10579
     */
10580
0
    ctxt->options &= ~XML_PARSE_DTDVALID;
10581
0
    ctxt->validate = 0;
10582
0
    ctxt->depth = 0;
10583
10584
0
    xmlParseContentInternal(ctxt);
10585
10586
0
    if (ctxt->input->cur < ctxt->input->end)
10587
0
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
10588
10589
    /*
10590
     * SAX: end of the document processing.
10591
     */
10592
0
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10593
0
        ctxt->sax->endDocument(ctxt->userData);
10594
10595
0
    if (! ctxt->wellFormed) return(-1);
10596
0
    return(0);
10597
0
}
10598
10599
#ifdef LIBXML_PUSH_ENABLED
10600
/************************************************************************
10601
 *                  *
10602
 *    Progressive parsing interfaces        *
10603
 *                  *
10604
 ************************************************************************/
10605
10606
/**
10607
 * Check whether the input buffer contains a character.
10608
 *
10609
 * @param ctxt  an XML parser context
10610
 * @param c  character
10611
 */
10612
static int
10613
61.7k
xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) {
10614
61.7k
    const xmlChar *cur;
10615
10616
61.7k
    if (ctxt->checkIndex == 0) {
10617
23.7k
        cur = ctxt->input->cur + 1;
10618
38.0k
    } else {
10619
38.0k
        cur = ctxt->input->cur + ctxt->checkIndex;
10620
38.0k
    }
10621
10622
61.7k
    if (memchr(cur, c, ctxt->input->end - cur) == NULL) {
10623
38.7k
        size_t index = ctxt->input->end - ctxt->input->cur;
10624
10625
38.7k
        if (index > LONG_MAX) {
10626
0
            ctxt->checkIndex = 0;
10627
0
            return(1);
10628
0
        }
10629
38.7k
        ctxt->checkIndex = index;
10630
38.7k
        return(0);
10631
38.7k
    } else {
10632
23.0k
        ctxt->checkIndex = 0;
10633
23.0k
        return(1);
10634
23.0k
    }
10635
61.7k
}
10636
10637
/**
10638
 * Check whether the input buffer contains a string.
10639
 *
10640
 * @param ctxt  an XML parser context
10641
 * @param startDelta  delta to apply at the start
10642
 * @param str  string
10643
 * @param strLen  length of string
10644
 */
10645
static const xmlChar *
10646
xmlParseLookupString(xmlParserCtxtPtr ctxt, size_t startDelta,
10647
379k
                     const char *str, size_t strLen) {
10648
379k
    const xmlChar *cur, *term;
10649
10650
379k
    if (ctxt->checkIndex == 0) {
10651
107k
        cur = ctxt->input->cur + startDelta;
10652
272k
    } else {
10653
272k
        cur = ctxt->input->cur + ctxt->checkIndex;
10654
272k
    }
10655
10656
379k
    term = BAD_CAST strstr((const char *) cur, str);
10657
379k
    if (term == NULL) {
10658
274k
        const xmlChar *end = ctxt->input->end;
10659
274k
        size_t index;
10660
10661
        /* Rescan (strLen - 1) characters. */
10662
274k
        if ((size_t) (end - cur) < strLen)
10663
2.39k
            end = cur;
10664
272k
        else
10665
272k
            end -= strLen - 1;
10666
274k
        index = end - ctxt->input->cur;
10667
274k
        if (index > LONG_MAX) {
10668
0
            ctxt->checkIndex = 0;
10669
0
            return(ctxt->input->end - strLen);
10670
0
        }
10671
274k
        ctxt->checkIndex = index;
10672
274k
    } else {
10673
105k
        ctxt->checkIndex = 0;
10674
105k
    }
10675
10676
379k
    return(term);
10677
379k
}
10678
10679
/**
10680
 * Check whether the input buffer contains terminated char data.
10681
 *
10682
 * @param ctxt  an XML parser context
10683
 */
10684
static int
10685
64.7k
xmlParseLookupCharData(xmlParserCtxtPtr ctxt) {
10686
64.7k
    const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex;
10687
64.7k
    const xmlChar *end = ctxt->input->end;
10688
64.7k
    size_t index;
10689
10690
4.28M
    while (cur < end) {
10691
4.26M
        if ((*cur == '<') || (*cur == '&')) {
10692
42.8k
            ctxt->checkIndex = 0;
10693
42.8k
            return(1);
10694
42.8k
        }
10695
4.21M
        cur++;
10696
4.21M
    }
10697
10698
21.9k
    index = cur - ctxt->input->cur;
10699
21.9k
    if (index > LONG_MAX) {
10700
0
        ctxt->checkIndex = 0;
10701
0
        return(1);
10702
0
    }
10703
21.9k
    ctxt->checkIndex = index;
10704
21.9k
    return(0);
10705
21.9k
}
10706
10707
/**
10708
 * Check whether there's enough data in the input buffer to finish parsing
10709
 * a start tag. This has to take quotes into account.
10710
 *
10711
 * @param ctxt  an XML parser context
10712
 */
10713
static int
10714
1.52M
xmlParseLookupGt(xmlParserCtxtPtr ctxt) {
10715
1.52M
    const xmlChar *cur;
10716
1.52M
    const xmlChar *end = ctxt->input->end;
10717
1.52M
    int state = ctxt->endCheckState;
10718
1.52M
    size_t index;
10719
10720
1.52M
    if (ctxt->checkIndex == 0)
10721
113k
        cur = ctxt->input->cur + 1;
10722
1.41M
    else
10723
1.41M
        cur = ctxt->input->cur + ctxt->checkIndex;
10724
10725
333M
    while (cur < end) {
10726
331M
        if (state) {
10727
292M
            if (*cur == state)
10728
73.2k
                state = 0;
10729
292M
        } else if (*cur == '\'' || *cur == '"') {
10730
75.8k
            state = *cur;
10731
39.7M
        } else if (*cur == '>') {
10732
102k
            ctxt->checkIndex = 0;
10733
102k
            ctxt->endCheckState = 0;
10734
102k
            return(1);
10735
102k
        }
10736
331M
        cur++;
10737
331M
    }
10738
10739
1.42M
    index = cur - ctxt->input->cur;
10740
1.42M
    if (index > LONG_MAX) {
10741
0
        ctxt->checkIndex = 0;
10742
0
        ctxt->endCheckState = 0;
10743
0
        return(1);
10744
0
    }
10745
1.42M
    ctxt->checkIndex = index;
10746
1.42M
    ctxt->endCheckState = state;
10747
1.42M
    return(0);
10748
1.42M
}
10749
10750
/**
10751
 * Check whether there's enough data in the input buffer to finish parsing
10752
 * the internal subset.
10753
 *
10754
 * @param ctxt  an XML parser context
10755
 */
10756
static int
10757
494k
xmlParseLookupInternalSubset(xmlParserCtxtPtr ctxt) {
10758
    /*
10759
     * Sorry, but progressive parsing of the internal subset is not
10760
     * supported. We first check that the full content of the internal
10761
     * subset is available and parsing is launched only at that point.
10762
     * Internal subset ends with "']' S? '>'" in an unescaped section and
10763
     * not in a ']]>' sequence which are conditional sections.
10764
     */
10765
494k
    const xmlChar *cur, *start;
10766
494k
    const xmlChar *end = ctxt->input->end;
10767
494k
    int state = ctxt->endCheckState;
10768
494k
    size_t index;
10769
10770
494k
    if (ctxt->checkIndex == 0) {
10771
10.8k
        cur = ctxt->input->cur + 1;
10772
483k
    } else {
10773
483k
        cur = ctxt->input->cur + ctxt->checkIndex;
10774
483k
    }
10775
494k
    start = cur;
10776
10777
209M
    while (cur < end) {
10778
208M
        if (state == '-') {
10779
1.42M
            if ((*cur == '-') &&
10780
1.82k
                (cur[1] == '-') &&
10781
1.15k
                (cur[2] == '>')) {
10782
411
                state = 0;
10783
411
                cur += 3;
10784
411
                start = cur;
10785
411
                continue;
10786
411
            }
10787
1.42M
        }
10788
207M
        else if (state == ']') {
10789
13.1k
            if (*cur == '>') {
10790
5.43k
                ctxt->checkIndex = 0;
10791
5.43k
                ctxt->endCheckState = 0;
10792
5.43k
                return(1);
10793
5.43k
            }
10794
7.67k
            if (IS_BLANK_CH(*cur)) {
10795
1.10k
                state = ' ';
10796
6.57k
            } else if (*cur != ']') {
10797
563
                state = 0;
10798
563
                start = cur;
10799
563
                continue;
10800
563
            }
10801
7.67k
        }
10802
207M
        else if (state == ' ') {
10803
4.72k
            if (*cur == '>') {
10804
109
                ctxt->checkIndex = 0;
10805
109
                ctxt->endCheckState = 0;
10806
109
                return(1);
10807
109
            }
10808
4.61k
            if (!IS_BLANK_CH(*cur)) {
10809
989
                state = 0;
10810
989
                start = cur;
10811
989
                continue;
10812
989
            }
10813
4.61k
        }
10814
207M
        else if (state != 0) {
10815
193M
            if (*cur == state) {
10816
37.1k
                state = 0;
10817
37.1k
                start = cur + 1;
10818
37.1k
            }
10819
193M
        }
10820
13.4M
        else if (*cur == '<') {
10821
64.0k
            if ((cur[1] == '!') &&
10822
22.3k
                (cur[2] == '-') &&
10823
660
                (cur[3] == '-')) {
10824
458
                state = '-';
10825
458
                cur += 4;
10826
                /* Don't treat <!--> as comment */
10827
458
                start = cur;
10828
458
                continue;
10829
458
            }
10830
64.0k
        }
10831
13.3M
        else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) {
10832
44.8k
            state = *cur;
10833
44.8k
        }
10834
10835
208M
        cur++;
10836
208M
    }
10837
10838
    /*
10839
     * Rescan the three last characters to detect "<!--" and "-->"
10840
     * split across chunks.
10841
     */
10842
488k
    if ((state == 0) || (state == '-')) {
10843
31.0k
        if (cur - start < 3)
10844
994
            cur = start;
10845
30.0k
        else
10846
30.0k
            cur -= 3;
10847
31.0k
    }
10848
488k
    index = cur - ctxt->input->cur;
10849
488k
    if (index > LONG_MAX) {
10850
0
        ctxt->checkIndex = 0;
10851
0
        ctxt->endCheckState = 0;
10852
0
        return(1);
10853
0
    }
10854
488k
    ctxt->checkIndex = index;
10855
488k
    ctxt->endCheckState = state;
10856
488k
    return(0);
10857
488k
}
10858
10859
/**
10860
 * Try to progress on parsing
10861
 *
10862
 * @param ctxt  an XML parser context
10863
 * @param terminate  last chunk indicator
10864
 * @returns zero if no parsing was possible
10865
 */
10866
static int
10867
2.68M
xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) {
10868
2.68M
    int ret = 0;
10869
2.68M
    size_t avail;
10870
2.68M
    xmlChar cur, next;
10871
10872
2.68M
    if (ctxt->input == NULL)
10873
0
        return(0);
10874
10875
2.68M
    if ((ctxt->input != NULL) &&
10876
2.68M
        (ctxt->input->cur - ctxt->input->base > 4096)) {
10877
2.08k
        xmlParserShrink(ctxt);
10878
2.08k
    }
10879
10880
5.38M
    while (ctxt->disableSAX == 0) {
10881
5.37M
        avail = ctxt->input->end - ctxt->input->cur;
10882
5.37M
        if (avail < 1)
10883
20.8k
      goto done;
10884
5.35M
        switch (ctxt->instate) {
10885
386k
            case XML_PARSER_EOF:
10886
          /*
10887
     * Document parsing is done !
10888
     */
10889
386k
          goto done;
10890
30.0k
            case XML_PARSER_START:
10891
                /*
10892
                 * Very first chars read from the document flow.
10893
                 */
10894
30.0k
                if ((!terminate) && (avail < 4))
10895
743
                    goto done;
10896
10897
                /*
10898
                 * We need more bytes to detect EBCDIC code pages.
10899
                 * See xmlDetectEBCDIC.
10900
                 */
10901
29.3k
                if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) &&
10902
694
                    (!terminate) && (avail < 200))
10903
348
                    goto done;
10904
10905
29.0k
                xmlDetectEncoding(ctxt);
10906
29.0k
                ctxt->instate = XML_PARSER_XML_DECL;
10907
29.0k
    break;
10908
10909
112k
            case XML_PARSER_XML_DECL:
10910
112k
    if ((!terminate) && (avail < 2))
10911
22
        goto done;
10912
112k
    cur = ctxt->input->cur[0];
10913
112k
    next = ctxt->input->cur[1];
10914
112k
          if ((cur == '<') && (next == '?')) {
10915
        /* PI or XML decl */
10916
88.1k
        if ((!terminate) &&
10917
86.4k
                        (!xmlParseLookupString(ctxt, 2, "?>", 2)))
10918
83.8k
      goto done;
10919
4.27k
        if ((ctxt->input->cur[2] == 'x') &&
10920
3.63k
      (ctxt->input->cur[3] == 'm') &&
10921
3.37k
      (ctxt->input->cur[4] == 'l') &&
10922
2.86k
      (IS_BLANK_CH(ctxt->input->cur[5]))) {
10923
2.76k
      ret += 5;
10924
2.76k
      xmlParseXMLDecl(ctxt);
10925
2.76k
        } else {
10926
1.50k
      ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10927
1.50k
                        if (ctxt->version == NULL) {
10928
5
                            xmlErrMemory(ctxt);
10929
5
                            break;
10930
5
                        }
10931
1.50k
        }
10932
24.6k
    } else {
10933
24.6k
        ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10934
24.6k
        if (ctxt->version == NULL) {
10935
44
            xmlErrMemory(ctxt);
10936
44
      break;
10937
44
        }
10938
24.6k
    }
10939
28.8k
                if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10940
28.8k
                    ctxt->sax->setDocumentLocator(ctxt->userData,
10941
28.8k
                            (xmlSAXLocator *) &xmlDefaultSAXLocator);
10942
28.8k
                }
10943
28.8k
                if ((ctxt->sax) && (ctxt->sax->startDocument) &&
10944
28.8k
                    (!ctxt->disableSAX))
10945
28.1k
                    ctxt->sax->startDocument(ctxt->userData);
10946
28.8k
                ctxt->instate = XML_PARSER_MISC;
10947
28.8k
    break;
10948
1.15M
            case XML_PARSER_START_TAG: {
10949
1.15M
          const xmlChar *name;
10950
1.15M
    const xmlChar *prefix = NULL;
10951
1.15M
    const xmlChar *URI = NULL;
10952
1.15M
                int line = ctxt->input->line;
10953
1.15M
    int nbNs = 0;
10954
10955
1.15M
    if ((!terminate) && (avail < 2))
10956
116
        goto done;
10957
1.15M
    cur = ctxt->input->cur[0];
10958
1.15M
          if (cur != '<') {
10959
2.73k
        xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10960
2.73k
                                   "Start tag expected, '<' not found");
10961
2.73k
                    ctxt->instate = XML_PARSER_EOF;
10962
2.73k
                    xmlFinishDocument(ctxt);
10963
2.73k
        goto done;
10964
2.73k
    }
10965
1.15M
    if ((!terminate) && (!xmlParseLookupGt(ctxt)))
10966
1.02M
                    goto done;
10967
126k
    if (ctxt->spaceNr == 0)
10968
0
        spacePush(ctxt, -1);
10969
126k
    else if (*ctxt->space == -2)
10970
12.9k
        spacePush(ctxt, -1);
10971
113k
    else
10972
113k
        spacePush(ctxt, *ctxt->space);
10973
126k
#ifdef LIBXML_SAX1_ENABLED
10974
126k
    if (ctxt->sax2)
10975
80.0k
#endif /* LIBXML_SAX1_ENABLED */
10976
80.0k
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
10977
46.5k
#ifdef LIBXML_SAX1_ENABLED
10978
46.5k
    else
10979
46.5k
        name = xmlParseStartTag(ctxt);
10980
126k
#endif /* LIBXML_SAX1_ENABLED */
10981
126k
    if (name == NULL) {
10982
3.39k
        spacePop(ctxt);
10983
3.39k
                    ctxt->instate = XML_PARSER_EOF;
10984
3.39k
                    xmlFinishDocument(ctxt);
10985
3.39k
        goto done;
10986
3.39k
    }
10987
123k
#ifdef LIBXML_VALID_ENABLED
10988
    /*
10989
     * [ VC: Root Element Type ]
10990
     * The Name in the document type declaration must match
10991
     * the element type of the root element.
10992
     */
10993
123k
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
10994
33.7k
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
10995
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
10996
123k
#endif /* LIBXML_VALID_ENABLED */
10997
10998
    /*
10999
     * Check for an Empty Element.
11000
     */
11001
123k
    if ((RAW == '/') && (NXT(1) == '>')) {
11002
18.6k
        SKIP(2);
11003
11004
18.6k
        if (ctxt->sax2) {
11005
16.6k
      if ((ctxt->sax != NULL) &&
11006
16.6k
          (ctxt->sax->endElementNs != NULL) &&
11007
16.6k
          (!ctxt->disableSAX))
11008
16.6k
          ctxt->sax->endElementNs(ctxt->userData, name,
11009
16.6k
                                  prefix, URI);
11010
16.6k
      if (nbNs > 0)
11011
9.54k
          xmlParserNsPop(ctxt, nbNs);
11012
16.6k
#ifdef LIBXML_SAX1_ENABLED
11013
16.6k
        } else {
11014
1.97k
      if ((ctxt->sax != NULL) &&
11015
1.97k
          (ctxt->sax->endElement != NULL) &&
11016
1.97k
          (!ctxt->disableSAX))
11017
1.96k
          ctxt->sax->endElement(ctxt->userData, name);
11018
1.97k
#endif /* LIBXML_SAX1_ENABLED */
11019
1.97k
        }
11020
18.6k
        spacePop(ctxt);
11021
104k
    } else if (RAW == '>') {
11022
77.0k
        NEXT;
11023
77.0k
                    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
11024
77.0k
    } else {
11025
27.5k
        xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED,
11026
27.5k
           "Couldn't find end of Start Tag %s\n",
11027
27.5k
           name);
11028
27.5k
        nodePop(ctxt);
11029
27.5k
        spacePop(ctxt);
11030
27.5k
                    if (nbNs > 0)
11031
4.16k
                        xmlParserNsPop(ctxt, nbNs);
11032
27.5k
    }
11033
11034
123k
                if (ctxt->nameNr == 0)
11035
5.94k
                    ctxt->instate = XML_PARSER_EPILOG;
11036
117k
                else
11037
117k
                    ctxt->instate = XML_PARSER_CONTENT;
11038
123k
                break;
11039
126k
      }
11040
2.58M
            case XML_PARSER_CONTENT: {
11041
2.58M
    cur = ctxt->input->cur[0];
11042
11043
2.58M
    if (cur == '<') {
11044
313k
                    if ((!terminate) && (avail < 2))
11045
1.35k
                        goto done;
11046
312k
        next = ctxt->input->cur[1];
11047
11048
312k
                    if (next == '/') {
11049
9.76k
                        ctxt->instate = XML_PARSER_END_TAG;
11050
9.76k
                        break;
11051
302k
                    } else if (next == '?') {
11052
64.5k
                        if ((!terminate) &&
11053
63.5k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11054
55.0k
                            goto done;
11055
9.51k
                        xmlParsePI(ctxt);
11056
9.51k
                        ctxt->instate = XML_PARSER_CONTENT;
11057
9.51k
                        break;
11058
238k
                    } else if (next == '!') {
11059
129k
                        if ((!terminate) && (avail < 3))
11060
374
                            goto done;
11061
129k
                        next = ctxt->input->cur[2];
11062
11063
129k
                        if (next == '-') {
11064
64.7k
                            if ((!terminate) && (avail < 4))
11065
348
                                goto done;
11066
64.3k
                            if (ctxt->input->cur[3] == '-') {
11067
64.3k
                                if ((!terminate) &&
11068
63.8k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11069
28.0k
                                    goto done;
11070
36.3k
                                xmlParseComment(ctxt);
11071
36.3k
                                ctxt->instate = XML_PARSER_CONTENT;
11072
36.3k
                                break;
11073
64.3k
                            }
11074
64.6k
                        } else if (next == '[') {
11075
64.0k
                            if ((!terminate) && (avail < 9))
11076
242
                                goto done;
11077
63.8k
                            if ((ctxt->input->cur[2] == '[') &&
11078
63.8k
                                (ctxt->input->cur[3] == 'C') &&
11079
63.8k
                                (ctxt->input->cur[4] == 'D') &&
11080
63.8k
                                (ctxt->input->cur[5] == 'A') &&
11081
63.7k
                                (ctxt->input->cur[6] == 'T') &&
11082
63.7k
                                (ctxt->input->cur[7] == 'A') &&
11083
63.7k
                                (ctxt->input->cur[8] == '[')) {
11084
63.7k
                                if ((!terminate) &&
11085
62.6k
                                    (!xmlParseLookupString(ctxt, 9, "]]>", 3)))
11086
58.9k
                                    goto done;
11087
4.77k
                                ctxt->instate = XML_PARSER_CDATA_SECTION;
11088
4.77k
                                xmlParseCDSect(ctxt);
11089
4.77k
                                ctxt->instate = XML_PARSER_CONTENT;
11090
4.77k
                                break;
11091
63.7k
                            }
11092
63.8k
                        }
11093
129k
                    }
11094
2.26M
    } else if (cur == '&') {
11095
31.8k
        if ((!terminate) && (!xmlParseLookupChar(ctxt, ';')))
11096
13.2k
      goto done;
11097
18.6k
        xmlParseReference(ctxt);
11098
18.6k
                    break;
11099
2.23M
    } else {
11100
        /* TODO Avoid the extra copy, handle directly !!! */
11101
        /*
11102
         * Goal of the following test is:
11103
         *  - minimize calls to the SAX 'character' callback
11104
         *    when they are mergeable
11105
         *  - handle an problem for isBlank when we only parse
11106
         *    a sequence of blank chars and the next one is
11107
         *    not available to check against '<' presence.
11108
         *  - tries to homogenize the differences in SAX
11109
         *    callbacks between the push and pull versions
11110
         *    of the parser.
11111
         */
11112
2.23M
        if (avail < XML_PARSER_BIG_BUFFER_SIZE) {
11113
79.2k
      if ((!terminate) && (!xmlParseLookupCharData(ctxt)))
11114
21.9k
          goto done;
11115
79.2k
                    }
11116
2.21M
                    ctxt->checkIndex = 0;
11117
2.21M
        xmlParseCharDataInternal(ctxt, !terminate);
11118
2.21M
                    break;
11119
2.23M
    }
11120
11121
109k
                ctxt->instate = XML_PARSER_START_TAG;
11122
109k
    break;
11123
2.58M
      }
11124
35.3k
            case XML_PARSER_END_TAG:
11125
35.3k
    if ((!terminate) && (!xmlParseLookupChar(ctxt, '>')))
11126
25.5k
        goto done;
11127
9.76k
    if (ctxt->sax2) {
11128
7.00k
              xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
11129
7.00k
        nameNsPop(ctxt);
11130
7.00k
    }
11131
2.75k
#ifdef LIBXML_SAX1_ENABLED
11132
2.75k
      else
11133
2.75k
        xmlParseEndTag1(ctxt, 0);
11134
9.76k
#endif /* LIBXML_SAX1_ENABLED */
11135
9.76k
    if (ctxt->nameNr == 0) {
11136
520
        ctxt->instate = XML_PARSER_EPILOG;
11137
9.24k
    } else {
11138
9.24k
        ctxt->instate = XML_PARSER_CONTENT;
11139
9.24k
    }
11140
9.76k
    break;
11141
529k
            case XML_PARSER_MISC:
11142
545k
            case XML_PARSER_PROLOG:
11143
547k
            case XML_PARSER_EPILOG:
11144
547k
    SKIP_BLANKS;
11145
547k
                avail = ctxt->input->end - ctxt->input->cur;
11146
547k
    if (avail < 1)
11147
426
        goto done;
11148
547k
    if (ctxt->input->cur[0] == '<') {
11149
543k
                    if ((!terminate) && (avail < 2))
11150
503
                        goto done;
11151
543k
                    next = ctxt->input->cur[1];
11152
543k
                    if (next == '?') {
11153
29.6k
                        if ((!terminate) &&
11154
28.2k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11155
22.1k
                            goto done;
11156
7.46k
                        xmlParsePI(ctxt);
11157
7.46k
                        break;
11158
513k
                    } else if (next == '!') {
11159
496k
                        if ((!terminate) && (avail < 3))
11160
494
                            goto done;
11161
11162
495k
                        if (ctxt->input->cur[2] == '-') {
11163
76.1k
                            if ((!terminate) && (avail < 4))
11164
476
                                goto done;
11165
75.6k
                            if (ctxt->input->cur[3] == '-') {
11166
75.6k
                                if ((!terminate) &&
11167
74.8k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11168
26.4k
                                    goto done;
11169
49.1k
                                xmlParseComment(ctxt);
11170
49.1k
                                break;
11171
75.6k
                            }
11172
419k
                        } else if (ctxt->instate == XML_PARSER_MISC) {
11173
419k
                            if ((!terminate) && (avail < 9))
11174
35
                                goto done;
11175
419k
                            if ((ctxt->input->cur[2] == 'D') &&
11176
419k
                                (ctxt->input->cur[3] == 'O') &&
11177
419k
                                (ctxt->input->cur[4] == 'C') &&
11178
419k
                                (ctxt->input->cur[5] == 'T') &&
11179
419k
                                (ctxt->input->cur[6] == 'Y') &&
11180
419k
                                (ctxt->input->cur[7] == 'P') &&
11181
419k
                                (ctxt->input->cur[8] == 'E')) {
11182
419k
                                if ((!terminate) && (!xmlParseLookupGt(ctxt)))
11183
401k
                                    goto done;
11184
17.3k
                                ctxt->inSubset = 1;
11185
17.3k
                                xmlParseDocTypeDecl(ctxt);
11186
17.3k
                                if (RAW == '[') {
11187
14.0k
                                    ctxt->instate = XML_PARSER_DTD;
11188
14.0k
                                } else {
11189
3.26k
                                    if (RAW == '>')
11190
2.09k
                                        NEXT;
11191
                                    /*
11192
                                     * Create and update the external subset.
11193
                                     */
11194
3.26k
                                    ctxt->inSubset = 2;
11195
3.26k
                                    if ((ctxt->sax != NULL) &&
11196
3.26k
                                        (!ctxt->disableSAX) &&
11197
3.10k
                                        (ctxt->sax->externalSubset != NULL))
11198
3.10k
                                        ctxt->sax->externalSubset(
11199
3.10k
                                                ctxt->userData,
11200
3.10k
                                                ctxt->intSubName,
11201
3.10k
                                                ctxt->extSubSystem,
11202
3.10k
                                                ctxt->extSubURI);
11203
3.26k
                                    ctxt->inSubset = 0;
11204
3.26k
                                    xmlCleanSpecialAttr(ctxt);
11205
3.26k
                                    ctxt->instate = XML_PARSER_PROLOG;
11206
3.26k
                                }
11207
17.3k
                                break;
11208
419k
                            }
11209
419k
                        }
11210
495k
                    }
11211
543k
                }
11212
11213
21.1k
                if (ctxt->instate == XML_PARSER_EPILOG) {
11214
948
                    if (ctxt->errNo == XML_ERR_OK)
11215
11
                        xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL);
11216
948
        ctxt->instate = XML_PARSER_EOF;
11217
948
                    xmlFinishDocument(ctxt);
11218
20.2k
                } else {
11219
20.2k
        ctxt->instate = XML_PARSER_START_TAG;
11220
20.2k
    }
11221
21.1k
    break;
11222
501k
            case XML_PARSER_DTD: {
11223
501k
                if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt)))
11224
488k
                    goto done;
11225
13.3k
    xmlParseInternalSubset(ctxt);
11226
13.3k
    ctxt->inSubset = 2;
11227
13.3k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
11228
11.3k
        (ctxt->sax->externalSubset != NULL))
11229
11.3k
        ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
11230
11.3k
          ctxt->extSubSystem, ctxt->extSubURI);
11231
13.3k
    ctxt->inSubset = 0;
11232
13.3k
    xmlCleanSpecialAttr(ctxt);
11233
13.3k
    ctxt->instate = XML_PARSER_PROLOG;
11234
13.3k
                break;
11235
501k
      }
11236
0
            default:
11237
0
                xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
11238
0
      "PP: internal error\n");
11239
0
    ctxt->instate = XML_PARSER_EOF;
11240
0
    break;
11241
5.35M
  }
11242
5.35M
    }
11243
2.68M
done:
11244
2.68M
    return(ret);
11245
2.68M
}
11246
11247
/**
11248
 * Parse a chunk of memory in push parser mode.
11249
 *
11250
 * Assumes that the parser context was initialized with
11251
 * #xmlCreatePushParserCtxt.
11252
 *
11253
 * The last chunk, which will often be empty, must be marked with
11254
 * the `terminate` flag. With the default SAX callbacks, the resulting
11255
 * document will be available in ctxt->myDoc. This pointer will not
11256
 * be freed when calling #xmlFreeParserCtxt and must be freed by the
11257
 * caller. If the document isn't well-formed, it will still be returned
11258
 * in ctxt->myDoc.
11259
 *
11260
 * As an exception, #xmlCtxtResetPush will free the document in
11261
 * ctxt->myDoc. So ctxt->myDoc should be set to NULL after extracting
11262
 * the document.
11263
 *
11264
 * Since 2.14.0, #xmlCtxtGetDocument can be used to retrieve the
11265
 * result document.
11266
 *
11267
 * @param ctxt  an XML parser context
11268
 * @param chunk  chunk of memory
11269
 * @param size  size of chunk in bytes
11270
 * @param terminate  last chunk indicator
11271
 * @returns an xmlParserErrors code (0 on success).
11272
 */
11273
int
11274
xmlParseChunk(xmlParserCtxt *ctxt, const char *chunk, int size,
11275
3.34M
              int terminate) {
11276
3.34M
    size_t curBase;
11277
3.34M
    size_t maxLength;
11278
3.34M
    size_t pos;
11279
3.34M
    int end_in_lf = 0;
11280
3.34M
    int res;
11281
11282
3.34M
    if ((ctxt == NULL) || (size < 0))
11283
0
        return(XML_ERR_ARGUMENT);
11284
3.34M
    if ((chunk == NULL) && (size > 0))
11285
0
        return(XML_ERR_ARGUMENT);
11286
3.34M
    if ((ctxt->input == NULL) || (ctxt->input->buf == NULL))
11287
0
        return(XML_ERR_ARGUMENT);
11288
3.34M
    if (ctxt->disableSAX != 0)
11289
663k
        return(ctxt->errNo);
11290
11291
2.68M
    ctxt->input->flags |= XML_INPUT_PROGRESSIVE;
11292
2.68M
    if (ctxt->instate == XML_PARSER_START)
11293
30.4k
        xmlCtxtInitializeLate(ctxt);
11294
2.68M
    if ((size > 0) && (chunk != NULL) && (!terminate) &&
11295
2.65M
        (chunk[size - 1] == '\r')) {
11296
4.04k
  end_in_lf = 1;
11297
4.04k
  size--;
11298
4.04k
    }
11299
11300
    /*
11301
     * Also push an empty chunk to make sure that the raw buffer
11302
     * will be flushed if there is an encoder.
11303
     */
11304
2.68M
    pos = ctxt->input->cur - ctxt->input->base;
11305
2.68M
    res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk);
11306
2.68M
    xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11307
2.68M
    if (res < 0) {
11308
325
        xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11309
325
        return(ctxt->errNo);
11310
325
    }
11311
11312
2.68M
    xmlParseTryOrFinish(ctxt, terminate);
11313
11314
2.68M
    curBase = ctxt->input->cur - ctxt->input->base;
11315
2.68M
    maxLength = (ctxt->options & XML_PARSE_HUGE) ?
11316
1.03M
                XML_MAX_HUGE_LENGTH :
11317
2.68M
                XML_MAX_LOOKUP_LIMIT;
11318
2.68M
    if (curBase > maxLength) {
11319
0
        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
11320
0
                    "Buffer size limit exceeded, try XML_PARSE_HUGE\n");
11321
0
    }
11322
11323
2.68M
    if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0))
11324
12.6k
        return(ctxt->errNo);
11325
11326
2.66M
    if (end_in_lf == 1) {
11327
4.01k
  pos = ctxt->input->cur - ctxt->input->base;
11328
4.01k
  res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r");
11329
4.01k
  xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11330
4.01k
        if (res < 0) {
11331
8
            xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11332
8
            return(ctxt->errNo);
11333
8
        }
11334
4.01k
    }
11335
2.66M
    if (terminate) {
11336
  /*
11337
   * Check for termination
11338
   */
11339
16.6k
        if ((ctxt->instate != XML_PARSER_EOF) &&
11340
10.6k
            (ctxt->instate != XML_PARSER_EPILOG)) {
11341
8.28k
            if (ctxt->nameNr > 0) {
11342
4.62k
                const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
11343
4.62k
                int line = ctxt->pushTab[ctxt->nameNr - 1].line;
11344
4.62k
                xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
11345
4.62k
                        "Premature end of data in tag %s line %d\n",
11346
4.62k
                        name, line, NULL);
11347
4.62k
            } else if (ctxt->instate == XML_PARSER_START) {
11348
184
                xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
11349
3.48k
            } else {
11350
3.48k
                xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
11351
3.48k
                               "Start tag expected, '<' not found\n");
11352
3.48k
            }
11353
8.40k
        } else {
11354
8.40k
            xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
11355
8.40k
        }
11356
16.6k
  if (ctxt->instate != XML_PARSER_EOF) {
11357
10.3k
            ctxt->instate = XML_PARSER_EOF;
11358
10.3k
            xmlFinishDocument(ctxt);
11359
10.3k
  }
11360
16.6k
    }
11361
2.66M
    if (ctxt->wellFormed == 0)
11362
1.60M
  return((xmlParserErrors) ctxt->errNo);
11363
1.06M
    else
11364
1.06M
        return(0);
11365
2.66M
}
11366
11367
/************************************************************************
11368
 *                  *
11369
 *    I/O front end functions to the parser     *
11370
 *                  *
11371
 ************************************************************************/
11372
11373
/**
11374
 * Create a parser context for using the XML parser in push mode.
11375
 * See #xmlParseChunk.
11376
 *
11377
 * Passing an initial chunk is useless and deprecated.
11378
 *
11379
 * The push parser doesn't support recovery mode or the
11380
 * XML_PARSE_NOBLANKS option.
11381
 *
11382
 * `filename` is used as base URI to fetch external entities and for
11383
 * error reports.
11384
 *
11385
 * @param sax  a SAX handler (optional)
11386
 * @param user_data  user data for SAX callbacks (optional)
11387
 * @param chunk  initial chunk (optional, deprecated)
11388
 * @param size  size of initial chunk in bytes
11389
 * @param filename  file name or URI (optional)
11390
 * @returns the new parser context or NULL if a memory allocation
11391
 * failed.
11392
 */
11393
11394
xmlParserCtxt *
11395
xmlCreatePushParserCtxt(xmlSAXHandler *sax, void *user_data,
11396
29.4k
                        const char *chunk, int size, const char *filename) {
11397
29.4k
    xmlParserCtxtPtr ctxt;
11398
29.4k
    xmlParserInputPtr input;
11399
11400
29.4k
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11401
29.4k
    if (ctxt == NULL)
11402
30
  return(NULL);
11403
11404
29.4k
    ctxt->options &= ~XML_PARSE_NODICT;
11405
29.4k
    ctxt->dictNames = 1;
11406
11407
29.4k
    input = xmlNewPushInput(filename, chunk, size);
11408
29.4k
    if (input == NULL) {
11409
13
  xmlFreeParserCtxt(ctxt);
11410
13
  return(NULL);
11411
13
    }
11412
29.3k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11413
8
        xmlFreeInputStream(input);
11414
8
        xmlFreeParserCtxt(ctxt);
11415
8
        return(NULL);
11416
8
    }
11417
11418
29.3k
    return(ctxt);
11419
29.3k
}
11420
#endif /* LIBXML_PUSH_ENABLED */
11421
11422
/**
11423
 * Blocks further parser processing
11424
 *
11425
 * @param ctxt  an XML parser context
11426
 */
11427
void
11428
0
xmlStopParser(xmlParserCtxt *ctxt) {
11429
0
    if (ctxt == NULL)
11430
0
        return;
11431
11432
    /* This stops the parser */
11433
0
    ctxt->disableSAX = 2;
11434
11435
    /*
11436
     * xmlStopParser is often called from error handlers,
11437
     * so we can't raise an error here to avoid infinite
11438
     * loops. Just make sure that an error condition is
11439
     * reported.
11440
     */
11441
0
    if (ctxt->errNo == XML_ERR_OK) {
11442
0
        ctxt->errNo = XML_ERR_USER_STOP;
11443
0
        ctxt->lastError.code = XML_ERR_USER_STOP;
11444
0
        ctxt->wellFormed = 0;
11445
0
    }
11446
0
}
11447
11448
/**
11449
 * Create a parser context for using the XML parser with an existing
11450
 * I/O stream
11451
 *
11452
 * @param sax  a SAX handler (optional)
11453
 * @param user_data  user data for SAX callbacks (optional)
11454
 * @param ioread  an I/O read function
11455
 * @param ioclose  an I/O close function (optional)
11456
 * @param ioctx  an I/O handler
11457
 * @param enc  the charset encoding if known (deprecated)
11458
 * @returns the new parser context or NULL
11459
 */
11460
xmlParserCtxt *
11461
xmlCreateIOParserCtxt(xmlSAXHandler *sax, void *user_data,
11462
                      xmlInputReadCallback ioread,
11463
                      xmlInputCloseCallback ioclose,
11464
0
                      void *ioctx, xmlCharEncoding enc) {
11465
0
    xmlParserCtxtPtr ctxt;
11466
0
    xmlParserInputPtr input;
11467
0
    const char *encoding;
11468
11469
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11470
0
    if (ctxt == NULL)
11471
0
  return(NULL);
11472
11473
0
    encoding = xmlGetCharEncodingName(enc);
11474
0
    input = xmlCtxtNewInputFromIO(ctxt, NULL, ioread, ioclose, ioctx,
11475
0
                                  encoding, 0);
11476
0
    if (input == NULL) {
11477
0
  xmlFreeParserCtxt(ctxt);
11478
0
        return (NULL);
11479
0
    }
11480
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11481
0
        xmlFreeInputStream(input);
11482
0
        xmlFreeParserCtxt(ctxt);
11483
0
        return(NULL);
11484
0
    }
11485
11486
0
    return(ctxt);
11487
0
}
11488
11489
#ifdef LIBXML_VALID_ENABLED
11490
/************************************************************************
11491
 *                  *
11492
 *    Front ends when parsing a DTD       *
11493
 *                  *
11494
 ************************************************************************/
11495
11496
/**
11497
 * Parse a DTD.
11498
 *
11499
 * Option XML_PARSE_DTDLOAD should be enabled in the parser context
11500
 * to make external entities work.
11501
 *
11502
 * @since 2.14.0
11503
 *
11504
 * @param ctxt  a parser context
11505
 * @param input  a parser input
11506
 * @param publicId  public ID of the DTD (optional)
11507
 * @param systemId  system ID of the DTD (optional)
11508
 * @returns the resulting xmlDtd or NULL in case of error.
11509
 * `input` will be freed by the function in any case.
11510
 */
11511
xmlDtd *
11512
xmlCtxtParseDtd(xmlParserCtxt *ctxt, xmlParserInput *input,
11513
737
                const xmlChar *publicId, const xmlChar *systemId) {
11514
737
    xmlDtdPtr ret = NULL;
11515
11516
737
    if ((ctxt == NULL) || (input == NULL)) {
11517
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
11518
0
        xmlFreeInputStream(input);
11519
0
        return(NULL);
11520
0
    }
11521
11522
737
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11523
4
        xmlFreeInputStream(input);
11524
4
        return(NULL);
11525
4
    }
11526
11527
733
    if (publicId == NULL)
11528
607
        publicId = BAD_CAST "none";
11529
733
    if (systemId == NULL)
11530
0
        systemId = BAD_CAST "none";
11531
11532
733
    ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0");
11533
733
    if (ctxt->myDoc == NULL) {
11534
1
        xmlErrMemory(ctxt);
11535
1
        goto error;
11536
1
    }
11537
732
    ctxt->myDoc->properties = XML_DOC_INTERNAL;
11538
732
    ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none",
11539
732
                                       publicId, systemId);
11540
732
    if (ctxt->myDoc->extSubset == NULL) {
11541
9
        xmlErrMemory(ctxt);
11542
9
        xmlFreeDoc(ctxt->myDoc);
11543
9
        goto error;
11544
9
    }
11545
11546
723
    xmlParseExternalSubset(ctxt, publicId, systemId);
11547
11548
723
    if (ctxt->wellFormed) {
11549
94
        ret = ctxt->myDoc->extSubset;
11550
94
        ctxt->myDoc->extSubset = NULL;
11551
94
        if (ret != NULL) {
11552
94
            xmlNodePtr tmp;
11553
11554
94
            ret->doc = NULL;
11555
94
            tmp = ret->children;
11556
4.27k
            while (tmp != NULL) {
11557
4.17k
                tmp->doc = NULL;
11558
4.17k
                tmp = tmp->next;
11559
4.17k
            }
11560
94
        }
11561
629
    } else {
11562
629
        ret = NULL;
11563
629
    }
11564
723
    xmlFreeDoc(ctxt->myDoc);
11565
723
    ctxt->myDoc = NULL;
11566
11567
733
error:
11568
733
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
11569
11570
733
    return(ret);
11571
723
}
11572
11573
/**
11574
 * Load and parse a DTD
11575
 *
11576
 * @deprecated Use #xmlCtxtParseDtd.
11577
 *
11578
 * @param sax  the SAX handler block or NULL
11579
 * @param input  an Input Buffer
11580
 * @param enc  the charset encoding if known
11581
 * @returns the resulting xmlDtd or NULL in case of error.
11582
 * `input` will be freed by the function in any case.
11583
 */
11584
11585
xmlDtd *
11586
xmlIOParseDTD(xmlSAXHandler *sax, xmlParserInputBuffer *input,
11587
0
        xmlCharEncoding enc) {
11588
0
    xmlDtdPtr ret = NULL;
11589
0
    xmlParserCtxtPtr ctxt;
11590
0
    xmlParserInputPtr pinput = NULL;
11591
11592
0
    if (input == NULL)
11593
0
  return(NULL);
11594
11595
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
11596
0
    if (ctxt == NULL) {
11597
0
        xmlFreeParserInputBuffer(input);
11598
0
  return(NULL);
11599
0
    }
11600
0
    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
11601
11602
    /*
11603
     * generate a parser input from the I/O handler
11604
     */
11605
11606
0
    pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE);
11607
0
    if (pinput == NULL) {
11608
0
  xmlFreeParserCtxt(ctxt);
11609
0
  return(NULL);
11610
0
    }
11611
11612
0
    if (enc != XML_CHAR_ENCODING_NONE) {
11613
0
        xmlSwitchEncoding(ctxt, enc);
11614
0
    }
11615
11616
0
    ret = xmlCtxtParseDtd(ctxt, pinput, NULL, NULL);
11617
11618
0
    xmlFreeParserCtxt(ctxt);
11619
0
    return(ret);
11620
0
}
11621
11622
/**
11623
 * Load and parse an external subset.
11624
 *
11625
 * @deprecated Use #xmlCtxtParseDtd.
11626
 *
11627
 * @param sax  the SAX handler block
11628
 * @param publicId  public identifier of the DTD (optional)
11629
 * @param systemId  system identifier (URL) of the DTD
11630
 * @returns the resulting xmlDtd or NULL in case of error.
11631
 */
11632
11633
xmlDtd *
11634
xmlSAXParseDTD(xmlSAXHandler *sax, const xmlChar *publicId,
11635
0
               const xmlChar *systemId) {
11636
0
    xmlDtdPtr ret = NULL;
11637
0
    xmlParserCtxtPtr ctxt;
11638
0
    xmlParserInputPtr input = NULL;
11639
0
    xmlChar* systemIdCanonic;
11640
11641
0
    if ((publicId == NULL) && (systemId == NULL)) return(NULL);
11642
11643
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
11644
0
    if (ctxt == NULL) {
11645
0
  return(NULL);
11646
0
    }
11647
0
    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
11648
11649
    /*
11650
     * Canonicalise the system ID
11651
     */
11652
0
    systemIdCanonic = xmlCanonicPath(systemId);
11653
0
    if ((systemId != NULL) && (systemIdCanonic == NULL)) {
11654
0
  xmlFreeParserCtxt(ctxt);
11655
0
  return(NULL);
11656
0
    }
11657
11658
    /*
11659
     * Ask the Entity resolver to load the damn thing
11660
     */
11661
11662
0
    if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL))
11663
0
  input = ctxt->sax->resolveEntity(ctxt->userData, publicId,
11664
0
                                   systemIdCanonic);
11665
0
    if (input == NULL) {
11666
0
  xmlFreeParserCtxt(ctxt);
11667
0
  if (systemIdCanonic != NULL)
11668
0
      xmlFree(systemIdCanonic);
11669
0
  return(NULL);
11670
0
    }
11671
11672
0
    if (input->filename == NULL)
11673
0
  input->filename = (char *) systemIdCanonic;
11674
0
    else
11675
0
  xmlFree(systemIdCanonic);
11676
11677
0
    ret = xmlCtxtParseDtd(ctxt, input, publicId, systemId);
11678
11679
0
    xmlFreeParserCtxt(ctxt);
11680
0
    return(ret);
11681
0
}
11682
11683
11684
/**
11685
 * Load and parse an external subset.
11686
 *
11687
 * @param publicId  public identifier of the DTD (optional)
11688
 * @param systemId  system identifier (URL) of the DTD
11689
 * @returns the resulting xmlDtd or NULL in case of error.
11690
 */
11691
11692
xmlDtd *
11693
0
xmlParseDTD(const xmlChar *publicId, const xmlChar *systemId) {
11694
0
    return(xmlSAXParseDTD(NULL, publicId, systemId));
11695
0
}
11696
#endif /* LIBXML_VALID_ENABLED */
11697
11698
/************************************************************************
11699
 *                  *
11700
 *    Front ends when parsing an Entity     *
11701
 *                  *
11702
 ************************************************************************/
11703
11704
static xmlNodePtr
11705
xmlCtxtParseContentInternal(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
11706
8.96k
                            int hasTextDecl, int buildTree) {
11707
8.96k
    xmlNodePtr root = NULL;
11708
8.96k
    xmlNodePtr list = NULL;
11709
8.96k
    xmlChar *rootName = BAD_CAST "#root";
11710
8.96k
    int result;
11711
11712
8.96k
    if (buildTree) {
11713
8.96k
        root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL);
11714
8.96k
        if (root == NULL) {
11715
23
            xmlErrMemory(ctxt);
11716
23
            goto error;
11717
23
        }
11718
8.96k
    }
11719
11720
8.94k
    if (xmlCtxtPushInput(ctxt, input) < 0)
11721
22
        goto error;
11722
11723
8.92k
    nameNsPush(ctxt, rootName, NULL, NULL, 0, 0);
11724
8.92k
    spacePush(ctxt, -1);
11725
11726
8.92k
    if (buildTree)
11727
8.92k
        nodePush(ctxt, root);
11728
11729
8.92k
    if (hasTextDecl) {
11730
5.95k
        xmlDetectEncoding(ctxt);
11731
11732
        /*
11733
         * Parse a possible text declaration first
11734
         */
11735
5.95k
        if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
11736
2.08k
            (IS_BLANK_CH(NXT(5)))) {
11737
2.06k
            xmlParseTextDecl(ctxt);
11738
            /*
11739
             * An XML-1.0 document can't reference an entity not XML-1.0
11740
             */
11741
2.06k
            if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
11742
2.00k
                (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
11743
9
                xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
11744
9
                               "Version mismatch between document and "
11745
9
                               "entity\n");
11746
9
            }
11747
2.06k
        }
11748
5.95k
    }
11749
11750
8.92k
    xmlParseContentInternal(ctxt);
11751
11752
8.92k
    if (ctxt->input->cur < ctxt->input->end)
11753
819
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
11754
11755
8.92k
    if ((ctxt->wellFormed) ||
11756
7.25k
        ((ctxt->recovery) && (!xmlCtxtIsCatastrophicError(ctxt)))) {
11757
7.25k
        if (root != NULL) {
11758
7.25k
            xmlNodePtr cur;
11759
11760
            /*
11761
             * Unlink newly created node list.
11762
             */
11763
7.25k
            list = root->children;
11764
7.25k
            root->children = NULL;
11765
7.25k
            root->last = NULL;
11766
90.1k
            for (cur = list; cur != NULL; cur = cur->next)
11767
82.8k
                cur->parent = NULL;
11768
7.25k
        }
11769
7.25k
    }
11770
11771
    /*
11772
     * Read the rest of the stream in case of errors. We want
11773
     * to account for the whole entity size.
11774
     */
11775
10.2k
    do {
11776
10.2k
        ctxt->input->cur = ctxt->input->end;
11777
10.2k
        xmlParserShrink(ctxt);
11778
10.2k
        result = xmlParserGrow(ctxt);
11779
10.2k
    } while (result > 0);
11780
11781
8.92k
    if (buildTree)
11782
8.92k
        nodePop(ctxt);
11783
11784
8.92k
    namePop(ctxt);
11785
8.92k
    spacePop(ctxt);
11786
11787
8.92k
    xmlCtxtPopInput(ctxt);
11788
11789
8.96k
error:
11790
8.96k
    xmlFreeNode(root);
11791
11792
8.96k
    return(list);
11793
8.92k
}
11794
11795
static void
11796
10.2k
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
11797
10.2k
    xmlParserInputPtr input;
11798
10.2k
    xmlNodePtr list;
11799
10.2k
    unsigned long consumed;
11800
10.2k
    int isExternal;
11801
10.2k
    int buildTree;
11802
10.2k
    int oldMinNsIndex;
11803
10.2k
    int oldNodelen, oldNodemem;
11804
11805
10.2k
    isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY);
11806
10.2k
    buildTree = (ctxt->node != NULL);
11807
11808
    /*
11809
     * Recursion check
11810
     */
11811
10.2k
    if (ent->flags & XML_ENT_EXPANDING) {
11812
12
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
11813
12
        goto error;
11814
12
    }
11815
11816
    /*
11817
     * Load entity
11818
     */
11819
10.2k
    input = xmlNewEntityInputStream(ctxt, ent);
11820
10.2k
    if (input == NULL)
11821
1.29k
        goto error;
11822
11823
    /*
11824
     * When building a tree, we need to limit the scope of namespace
11825
     * declarations, so that entities don't reference xmlNs structs
11826
     * from the parent of a reference.
11827
     */
11828
8.96k
    oldMinNsIndex = ctxt->nsdb->minNsIndex;
11829
8.96k
    if (buildTree)
11830
8.96k
        ctxt->nsdb->minNsIndex = ctxt->nsNr;
11831
11832
8.96k
    oldNodelen = ctxt->nodelen;
11833
8.96k
    oldNodemem = ctxt->nodemem;
11834
8.96k
    ctxt->nodelen = 0;
11835
8.96k
    ctxt->nodemem = 0;
11836
11837
    /*
11838
     * Parse content
11839
     *
11840
     * This initiates a recursive call chain:
11841
     *
11842
     * - xmlCtxtParseContentInternal
11843
     * - xmlParseContentInternal
11844
     * - xmlParseReference
11845
     * - xmlCtxtParseEntity
11846
     *
11847
     * The nesting depth is limited by the maximum number of inputs,
11848
     * see xmlCtxtPushInput.
11849
     *
11850
     * It's possible to make this non-recursive (minNsIndex must be
11851
     * stored in the input struct) at the expense of code readability.
11852
     */
11853
11854
8.96k
    ent->flags |= XML_ENT_EXPANDING;
11855
11856
8.96k
    list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree);
11857
11858
8.96k
    ent->flags &= ~XML_ENT_EXPANDING;
11859
11860
8.96k
    ctxt->nsdb->minNsIndex = oldMinNsIndex;
11861
8.96k
    ctxt->nodelen = oldNodelen;
11862
8.96k
    ctxt->nodemem = oldNodemem;
11863
11864
    /*
11865
     * Entity size accounting
11866
     */
11867
8.96k
    consumed = input->consumed;
11868
8.96k
    xmlSaturatedAddSizeT(&consumed, input->end - input->base);
11869
11870
8.96k
    if ((ent->flags & XML_ENT_CHECKED) == 0)
11871
6.87k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
11872
11873
8.96k
    if ((ent->flags & XML_ENT_PARSED) == 0) {
11874
6.87k
        if (isExternal)
11875
4.31k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
11876
11877
6.87k
        ent->children = list;
11878
11879
89.7k
        while (list != NULL) {
11880
82.8k
            list->parent = (xmlNodePtr) ent;
11881
11882
            /*
11883
             * Downstream code like the nginx xslt module can set
11884
             * ctxt->myDoc->extSubset to a separate DTD, so the entity
11885
             * might have a different or a NULL document.
11886
             */
11887
82.8k
            if (list->doc != ent->doc)
11888
0
                xmlSetTreeDoc(list, ent->doc);
11889
11890
82.8k
            if (list->next == NULL)
11891
4.92k
                ent->last = list;
11892
82.8k
            list = list->next;
11893
82.8k
        }
11894
6.87k
    } else {
11895
2.09k
        xmlFreeNodeList(list);
11896
2.09k
    }
11897
11898
8.96k
    xmlFreeInputStream(input);
11899
11900
10.2k
error:
11901
10.2k
    ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED;
11902
10.2k
}
11903
11904
/**
11905
 * Parse an external general entity within an existing parsing context
11906
 * An external general parsed entity is well-formed if it matches the
11907
 * production labeled extParsedEnt.
11908
 *
11909
 *     [78] extParsedEnt ::= TextDecl? content
11910
 *
11911
 * @param ctxt  the existing parsing context
11912
 * @param URL  the URL for the entity to load
11913
 * @param ID  the System ID for the entity to load
11914
 * @param listOut  the return value for the set of parsed nodes
11915
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11916
 *    the parser error code otherwise
11917
 */
11918
11919
int
11920
xmlParseCtxtExternalEntity(xmlParserCtxt *ctxt, const xmlChar *URL,
11921
0
                           const xmlChar *ID, xmlNode **listOut) {
11922
0
    xmlParserInputPtr input;
11923
0
    xmlNodePtr list;
11924
11925
0
    if (listOut != NULL)
11926
0
        *listOut = NULL;
11927
11928
0
    if (ctxt == NULL)
11929
0
        return(XML_ERR_ARGUMENT);
11930
11931
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
11932
0
                            XML_RESOURCE_GENERAL_ENTITY);
11933
0
    if (input == NULL)
11934
0
        return(ctxt->errNo);
11935
11936
0
    xmlCtxtInitializeLate(ctxt);
11937
11938
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 1, 1);
11939
0
    if (listOut != NULL)
11940
0
        *listOut = list;
11941
0
    else
11942
0
        xmlFreeNodeList(list);
11943
11944
0
    xmlFreeInputStream(input);
11945
0
    return(ctxt->errNo);
11946
0
}
11947
11948
#ifdef LIBXML_SAX1_ENABLED
11949
/**
11950
 * Parse an external general entity
11951
 * An external general parsed entity is well-formed if it matches the
11952
 * production labeled extParsedEnt.
11953
 *
11954
 * This function uses deprecated global variables to set parser options
11955
 * which default to XML_PARSE_NODICT.
11956
 *
11957
 * @deprecated Use #xmlParseCtxtExternalEntity.
11958
 *
11959
 *     [78] extParsedEnt ::= TextDecl? content
11960
 *
11961
 * @param doc  the document the chunk pertains to
11962
 * @param sax  the SAX handler block (possibly NULL)
11963
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
11964
 * @param depth  Used for loop detection, use 0
11965
 * @param URL  the URL for the entity to load
11966
 * @param ID  the System ID for the entity to load
11967
 * @param list  the return value for the set of parsed nodes
11968
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11969
 *    the parser error code otherwise
11970
 */
11971
11972
int
11973
xmlParseExternalEntity(xmlDoc *doc, xmlSAXHandler *sax, void *user_data,
11974
0
    int depth, const xmlChar *URL, const xmlChar *ID, xmlNode **list) {
11975
0
    xmlParserCtxtPtr ctxt;
11976
0
    int ret;
11977
11978
0
    if (list != NULL)
11979
0
        *list = NULL;
11980
11981
0
    if (doc == NULL)
11982
0
        return(XML_ERR_ARGUMENT);
11983
11984
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11985
0
    if (ctxt == NULL)
11986
0
        return(XML_ERR_NO_MEMORY);
11987
11988
0
    ctxt->depth = depth;
11989
0
    ctxt->myDoc = doc;
11990
0
    ret = xmlParseCtxtExternalEntity(ctxt, URL, ID, list);
11991
11992
0
    xmlFreeParserCtxt(ctxt);
11993
0
    return(ret);
11994
0
}
11995
11996
/**
11997
 * Parse a well-balanced chunk of an XML document
11998
 * called by the parser
11999
 * The allowed sequence for the Well Balanced Chunk is the one defined by
12000
 * the content production in the XML grammar:
12001
 *
12002
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12003
 *                       Comment)*
12004
 *
12005
 * This function uses deprecated global variables to set parser options
12006
 * which default to XML_PARSE_NODICT.
12007
 *
12008
 * @param doc  the document the chunk pertains to (must not be NULL)
12009
 * @param sax  the SAX handler block (possibly NULL)
12010
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
12011
 * @param depth  Used for loop detection, use 0
12012
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
12013
 * @param lst  the return value for the set of parsed nodes
12014
 * @returns 0 if the chunk is well balanced, -1 in case of args problem and
12015
 *    the parser error code otherwise
12016
 */
12017
12018
int
12019
xmlParseBalancedChunkMemory(xmlDoc *doc, xmlSAXHandler *sax,
12020
0
     void *user_data, int depth, const xmlChar *string, xmlNode **lst) {
12021
0
    return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data,
12022
0
                                                depth, string, lst, 0 );
12023
0
}
12024
#endif /* LIBXML_SAX1_ENABLED */
12025
12026
/**
12027
 * Parse a well-balanced chunk of XML matching the 'content' production.
12028
 *
12029
 * Namespaces in scope of `node` and entities of `node`'s document are
12030
 * recognized. When validating, the DTD of `node`'s document is used.
12031
 *
12032
 * Always consumes `input` even in error case.
12033
 *
12034
 * @since 2.14.0
12035
 *
12036
 * @param ctxt  parser context
12037
 * @param input  parser input
12038
 * @param node  target node or document
12039
 * @param hasTextDecl  whether to parse text declaration
12040
 * @returns a node list or NULL in case of error.
12041
 */
12042
xmlNode *
12043
xmlCtxtParseContent(xmlParserCtxt *ctxt, xmlParserInput *input,
12044
0
                    xmlNode *node, int hasTextDecl) {
12045
0
    xmlDocPtr doc;
12046
0
    xmlNodePtr cur, list = NULL;
12047
0
    int nsnr = 0;
12048
0
    xmlDictPtr oldDict;
12049
0
    int oldOptions, oldDictNames, oldLoadSubset;
12050
12051
0
    if ((ctxt == NULL) || (input == NULL) || (node == NULL)) {
12052
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12053
0
        goto exit;
12054
0
    }
12055
12056
0
    doc = node->doc;
12057
0
    if (doc == NULL) {
12058
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12059
0
        goto exit;
12060
0
    }
12061
12062
0
    switch (node->type) {
12063
0
        case XML_ELEMENT_NODE:
12064
0
        case XML_DOCUMENT_NODE:
12065
0
        case XML_HTML_DOCUMENT_NODE:
12066
0
            break;
12067
12068
0
        case XML_ATTRIBUTE_NODE:
12069
0
        case XML_TEXT_NODE:
12070
0
        case XML_CDATA_SECTION_NODE:
12071
0
        case XML_ENTITY_REF_NODE:
12072
0
        case XML_PI_NODE:
12073
0
        case XML_COMMENT_NODE:
12074
0
            for (cur = node->parent; cur != NULL; cur = cur->parent) {
12075
0
                if ((cur->type == XML_ELEMENT_NODE) ||
12076
0
                    (cur->type == XML_DOCUMENT_NODE) ||
12077
0
                    (cur->type == XML_HTML_DOCUMENT_NODE)) {
12078
0
                    node = cur;
12079
0
                    break;
12080
0
                }
12081
0
            }
12082
0
            break;
12083
12084
0
        default:
12085
0
            xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12086
0
            goto exit;
12087
0
    }
12088
12089
0
    xmlCtxtReset(ctxt);
12090
12091
0
    oldDict = ctxt->dict;
12092
0
    oldOptions = ctxt->options;
12093
0
    oldDictNames = ctxt->dictNames;
12094
0
    oldLoadSubset = ctxt->loadsubset;
12095
12096
    /*
12097
     * Use input doc's dict if present, else assure XML_PARSE_NODICT is set.
12098
     */
12099
0
    if (doc->dict != NULL) {
12100
0
        ctxt->dict = doc->dict;
12101
0
    } else {
12102
0
        ctxt->options |= XML_PARSE_NODICT;
12103
0
        ctxt->dictNames = 0;
12104
0
    }
12105
12106
    /*
12107
     * Disable IDs
12108
     */
12109
0
    ctxt->loadsubset |= XML_SKIP_IDS;
12110
0
    ctxt->options |= XML_PARSE_SKIP_IDS;
12111
12112
0
    ctxt->myDoc = doc;
12113
12114
0
#ifdef LIBXML_HTML_ENABLED
12115
0
    if (ctxt->html) {
12116
        /*
12117
         * When parsing in context, it makes no sense to add implied
12118
         * elements like html/body/etc...
12119
         */
12120
0
        ctxt->options |= HTML_PARSE_NOIMPLIED;
12121
12122
0
        list = htmlCtxtParseContentInternal(ctxt, input);
12123
0
    } else
12124
0
#endif
12125
0
    {
12126
0
        xmlCtxtInitializeLate(ctxt);
12127
12128
        /*
12129
         * initialize the SAX2 namespaces stack
12130
         */
12131
0
        cur = node;
12132
0
        while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) {
12133
0
            xmlNsPtr ns = cur->nsDef;
12134
0
            xmlHashedString hprefix, huri;
12135
12136
0
            while (ns != NULL) {
12137
0
                hprefix = xmlDictLookupHashed(ctxt->dict, ns->prefix, -1);
12138
0
                huri = xmlDictLookupHashed(ctxt->dict, ns->href, -1);
12139
0
                if (xmlParserNsPush(ctxt, &hprefix, &huri, ns, 1) > 0)
12140
0
                    nsnr++;
12141
0
                ns = ns->next;
12142
0
            }
12143
0
            cur = cur->parent;
12144
0
        }
12145
12146
0
        list = xmlCtxtParseContentInternal(ctxt, input, hasTextDecl, 1);
12147
12148
0
        if (nsnr > 0)
12149
0
            xmlParserNsPop(ctxt, nsnr);
12150
0
    }
12151
12152
0
    ctxt->dict = oldDict;
12153
0
    ctxt->options = oldOptions;
12154
0
    ctxt->dictNames = oldDictNames;
12155
0
    ctxt->loadsubset = oldLoadSubset;
12156
0
    ctxt->myDoc = NULL;
12157
0
    ctxt->node = NULL;
12158
12159
0
exit:
12160
0
    xmlFreeInputStream(input);
12161
0
    return(list);
12162
0
}
12163
12164
/**
12165
 * Parse a well-balanced chunk of an XML document
12166
 * within the context (DTD, namespaces, etc ...) of the given node.
12167
 *
12168
 * The allowed sequence for the data is a Well Balanced Chunk defined by
12169
 * the content production in the XML grammar:
12170
 *
12171
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12172
 *                       Comment)*
12173
 *
12174
 * This function assumes the encoding of `node`'s document which is
12175
 * typically not what you want. A better alternative is
12176
 * #xmlCtxtParseContent.
12177
 *
12178
 * @param node  the context node
12179
 * @param data  the input string
12180
 * @param datalen  the input string length in bytes
12181
 * @param options  a combination of xmlParserOption
12182
 * @param listOut  the return value for the set of parsed nodes
12183
 * @returns XML_ERR_OK if the chunk is well balanced, and the parser
12184
 * error code otherwise
12185
 */
12186
xmlParserErrors
12187
xmlParseInNodeContext(xmlNode *node, const char *data, int datalen,
12188
0
                      int options, xmlNode **listOut) {
12189
0
    xmlParserCtxtPtr ctxt;
12190
0
    xmlParserInputPtr input;
12191
0
    xmlDocPtr doc;
12192
0
    xmlNodePtr list;
12193
0
    xmlParserErrors ret;
12194
12195
0
    if (listOut == NULL)
12196
0
        return(XML_ERR_INTERNAL_ERROR);
12197
0
    *listOut = NULL;
12198
12199
0
    if ((node == NULL) || (data == NULL) || (datalen < 0))
12200
0
        return(XML_ERR_INTERNAL_ERROR);
12201
12202
0
    doc = node->doc;
12203
0
    if (doc == NULL)
12204
0
        return(XML_ERR_INTERNAL_ERROR);
12205
12206
0
#ifdef LIBXML_HTML_ENABLED
12207
0
    if (doc->type == XML_HTML_DOCUMENT_NODE) {
12208
0
        ctxt = htmlNewParserCtxt();
12209
0
    }
12210
0
    else
12211
0
#endif
12212
0
        ctxt = xmlNewParserCtxt();
12213
12214
0
    if (ctxt == NULL)
12215
0
        return(XML_ERR_NO_MEMORY);
12216
12217
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, data, datalen,
12218
0
                                      (const char *) doc->encoding,
12219
0
                                      XML_INPUT_BUF_STATIC);
12220
0
    if (input == NULL) {
12221
0
        xmlFreeParserCtxt(ctxt);
12222
0
        return(XML_ERR_NO_MEMORY);
12223
0
    }
12224
12225
0
    xmlCtxtUseOptions(ctxt, options);
12226
12227
0
    list = xmlCtxtParseContent(ctxt, input, node, /* hasTextDecl */ 0);
12228
12229
0
    if (list == NULL) {
12230
0
        ret = ctxt->errNo;
12231
0
        if (ret == XML_ERR_ARGUMENT)
12232
0
            ret = XML_ERR_INTERNAL_ERROR;
12233
0
    } else {
12234
0
        ret = XML_ERR_OK;
12235
0
        *listOut = list;
12236
0
    }
12237
12238
0
    xmlFreeParserCtxt(ctxt);
12239
12240
0
    return(ret);
12241
0
}
12242
12243
#ifdef LIBXML_SAX1_ENABLED
12244
/**
12245
 * Parse a well-balanced chunk of an XML document
12246
 *
12247
 * The allowed sequence for the Well Balanced Chunk is the one defined by
12248
 * the content production in the XML grammar:
12249
 *
12250
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12251
 *                       Comment)*
12252
 *
12253
 * In case recover is set to 1, the nodelist will not be empty even if
12254
 * the parsed chunk is not well balanced, assuming the parsing succeeded to
12255
 * some extent.
12256
 *
12257
 * This function uses deprecated global variables to set parser options
12258
 * which default to XML_PARSE_NODICT.
12259
 *
12260
 * @param doc  the document the chunk pertains to (must not be NULL)
12261
 * @param sax  the SAX handler block (possibly NULL)
12262
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
12263
 * @param depth  Used for loop detection, use 0
12264
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
12265
 * @param listOut  the return value for the set of parsed nodes
12266
 * @param recover  return nodes even if the data is broken (use 0)
12267
 * @returns 0 if the chunk is well balanced, or thehe parser error code
12268
 * otherwise.
12269
 */
12270
int
12271
xmlParseBalancedChunkMemoryRecover(xmlDoc *doc, xmlSAXHandler *sax,
12272
     void *user_data, int depth, const xmlChar *string, xmlNode **listOut,
12273
0
     int recover) {
12274
0
    xmlParserCtxtPtr ctxt;
12275
0
    xmlParserInputPtr input;
12276
0
    xmlNodePtr list;
12277
0
    int ret;
12278
12279
0
    if (listOut != NULL)
12280
0
        *listOut = NULL;
12281
12282
0
    if (string == NULL)
12283
0
        return(XML_ERR_ARGUMENT);
12284
12285
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
12286
0
    if (ctxt == NULL)
12287
0
        return(XML_ERR_NO_MEMORY);
12288
12289
0
    xmlCtxtInitializeLate(ctxt);
12290
12291
0
    ctxt->depth = depth;
12292
0
    ctxt->myDoc = doc;
12293
0
    if (recover) {
12294
0
        ctxt->options |= XML_PARSE_RECOVER;
12295
0
        ctxt->recovery = 1;
12296
0
    }
12297
12298
0
    input = xmlNewStringInputStream(ctxt, string);
12299
0
    if (input == NULL) {
12300
0
        ret = ctxt->errNo;
12301
0
        goto error;
12302
0
    }
12303
12304
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 0, 1);
12305
0
    if (listOut != NULL)
12306
0
        *listOut = list;
12307
0
    else
12308
0
        xmlFreeNodeList(list);
12309
12310
0
    if (!ctxt->wellFormed)
12311
0
        ret = ctxt->errNo;
12312
0
    else
12313
0
        ret = XML_ERR_OK;
12314
12315
0
error:
12316
0
    xmlFreeInputStream(input);
12317
0
    xmlFreeParserCtxt(ctxt);
12318
0
    return(ret);
12319
0
}
12320
12321
/**
12322
 * Parse an XML external entity out of context and build a tree.
12323
 * It use the given SAX function block to handle the parsing callback.
12324
 * If sax is NULL, fallback to the default DOM tree building routines.
12325
 *
12326
 * @deprecated Don't use.
12327
 *
12328
 *     [78] extParsedEnt ::= TextDecl? content
12329
 *
12330
 * This correspond to a "Well Balanced" chunk
12331
 *
12332
 * This function uses deprecated global variables to set parser options
12333
 * which default to XML_PARSE_NODICT.
12334
 *
12335
 * @param sax  the SAX handler block
12336
 * @param filename  the filename
12337
 * @returns the resulting document tree
12338
 */
12339
12340
xmlDoc *
12341
0
xmlSAXParseEntity(xmlSAXHandler *sax, const char *filename) {
12342
0
    xmlDocPtr ret;
12343
0
    xmlParserCtxtPtr ctxt;
12344
12345
0
    ctxt = xmlCreateFileParserCtxt(filename);
12346
0
    if (ctxt == NULL) {
12347
0
  return(NULL);
12348
0
    }
12349
0
    if (sax != NULL) {
12350
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12351
0
            *ctxt->sax = *sax;
12352
0
        } else {
12353
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12354
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12355
0
        }
12356
0
        ctxt->userData = NULL;
12357
0
    }
12358
12359
0
    xmlParseExtParsedEnt(ctxt);
12360
12361
0
    if (ctxt->wellFormed) {
12362
0
  ret = ctxt->myDoc;
12363
0
    } else {
12364
0
        ret = NULL;
12365
0
        xmlFreeDoc(ctxt->myDoc);
12366
0
    }
12367
12368
0
    xmlFreeParserCtxt(ctxt);
12369
12370
0
    return(ret);
12371
0
}
12372
12373
/**
12374
 * Parse an XML external entity out of context and build a tree.
12375
 *
12376
 *     [78] extParsedEnt ::= TextDecl? content
12377
 *
12378
 * This correspond to a "Well Balanced" chunk
12379
 *
12380
 * This function uses deprecated global variables to set parser options
12381
 * which default to XML_PARSE_NODICT.
12382
 *
12383
 * @deprecated Don't use.
12384
 *
12385
 * @param filename  the filename
12386
 * @returns the resulting document tree
12387
 */
12388
12389
xmlDoc *
12390
0
xmlParseEntity(const char *filename) {
12391
0
    return(xmlSAXParseEntity(NULL, filename));
12392
0
}
12393
#endif /* LIBXML_SAX1_ENABLED */
12394
12395
/**
12396
 * Create a parser context for an external entity
12397
 * Automatic support for ZLIB/Compress compressed document is provided
12398
 * by default if found at compile-time.
12399
 *
12400
 * @deprecated Don't use.
12401
 *
12402
 * @param URL  the entity URL
12403
 * @param ID  the entity PUBLIC ID
12404
 * @param base  a possible base for the target URI
12405
 * @returns the new parser context or NULL
12406
 */
12407
xmlParserCtxt *
12408
xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID,
12409
0
                    const xmlChar *base) {
12410
0
    xmlParserCtxtPtr ctxt;
12411
0
    xmlParserInputPtr input;
12412
0
    xmlChar *uri = NULL;
12413
12414
0
    ctxt = xmlNewParserCtxt();
12415
0
    if (ctxt == NULL)
12416
0
  return(NULL);
12417
12418
0
    if (base != NULL) {
12419
0
        if (xmlBuildURISafe(URL, base, &uri) < 0)
12420
0
            goto error;
12421
0
        if (uri != NULL)
12422
0
            URL = uri;
12423
0
    }
12424
12425
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
12426
0
                            XML_RESOURCE_UNKNOWN);
12427
0
    if (input == NULL)
12428
0
        goto error;
12429
12430
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12431
0
        xmlFreeInputStream(input);
12432
0
        goto error;
12433
0
    }
12434
12435
0
    xmlFree(uri);
12436
0
    return(ctxt);
12437
12438
0
error:
12439
0
    xmlFree(uri);
12440
0
    xmlFreeParserCtxt(ctxt);
12441
0
    return(NULL);
12442
0
}
12443
12444
/************************************************************************
12445
 *                  *
12446
 *    Front ends when parsing from a file     *
12447
 *                  *
12448
 ************************************************************************/
12449
12450
/**
12451
 * Create a parser context for a file or URL content.
12452
 * Automatic support for ZLIB/Compress compressed document is provided
12453
 * by default if found at compile-time and for file accesses
12454
 *
12455
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12456
 *
12457
 * @param filename  the filename or URL
12458
 * @param options  a combination of xmlParserOption
12459
 * @returns the new parser context or NULL
12460
 */
12461
xmlParserCtxt *
12462
xmlCreateURLParserCtxt(const char *filename, int options)
12463
0
{
12464
0
    xmlParserCtxtPtr ctxt;
12465
0
    xmlParserInputPtr input;
12466
12467
0
    ctxt = xmlNewParserCtxt();
12468
0
    if (ctxt == NULL)
12469
0
  return(NULL);
12470
12471
0
    xmlCtxtUseOptions(ctxt, options);
12472
12473
0
    input = xmlLoadResource(ctxt, filename, NULL, XML_RESOURCE_MAIN_DOCUMENT);
12474
0
    if (input == NULL) {
12475
0
  xmlFreeParserCtxt(ctxt);
12476
0
  return(NULL);
12477
0
    }
12478
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12479
0
        xmlFreeInputStream(input);
12480
0
        xmlFreeParserCtxt(ctxt);
12481
0
        return(NULL);
12482
0
    }
12483
12484
0
    return(ctxt);
12485
0
}
12486
12487
/**
12488
 * Create a parser context for a file content.
12489
 * Automatic support for ZLIB/Compress compressed document is provided
12490
 * by default if found at compile-time.
12491
 *
12492
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12493
 *
12494
 * @param filename  the filename
12495
 * @returns the new parser context or NULL
12496
 */
12497
xmlParserCtxt *
12498
xmlCreateFileParserCtxt(const char *filename)
12499
0
{
12500
0
    return(xmlCreateURLParserCtxt(filename, 0));
12501
0
}
12502
12503
#ifdef LIBXML_SAX1_ENABLED
12504
/**
12505
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12506
 * compressed document is provided by default if found at compile-time.
12507
 * It use the given SAX function block to handle the parsing callback.
12508
 * If sax is NULL, fallback to the default DOM tree building routines.
12509
 *
12510
 * This function uses deprecated global variables to set parser options
12511
 * which default to XML_PARSE_NODICT.
12512
 *
12513
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12514
 *
12515
 * User data (void *) is stored within the parser context in the
12516
 * context's _private member, so it is available nearly everywhere in libxml
12517
 *
12518
 * @param sax  the SAX handler block
12519
 * @param filename  the filename
12520
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12521
 *             documents
12522
 * @param data  the userdata
12523
 * @returns the resulting document tree
12524
 */
12525
12526
xmlDoc *
12527
xmlSAXParseFileWithData(xmlSAXHandler *sax, const char *filename,
12528
0
                        int recovery, void *data) {
12529
0
    xmlDocPtr ret = NULL;
12530
0
    xmlParserCtxtPtr ctxt;
12531
0
    xmlParserInputPtr input;
12532
12533
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12534
0
    if (ctxt == NULL)
12535
0
  return(NULL);
12536
12537
0
    if (data != NULL)
12538
0
  ctxt->_private = data;
12539
12540
0
    if (recovery) {
12541
0
        ctxt->options |= XML_PARSE_RECOVER;
12542
0
        ctxt->recovery = 1;
12543
0
    }
12544
12545
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
12546
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, NULL, 0);
12547
0
    else
12548
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, NULL, 0);
12549
12550
0
    if (input != NULL)
12551
0
        ret = xmlCtxtParseDocument(ctxt, input);
12552
12553
0
    xmlFreeParserCtxt(ctxt);
12554
0
    return(ret);
12555
0
}
12556
12557
/**
12558
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12559
 * compressed document is provided by default if found at compile-time.
12560
 * It use the given SAX function block to handle the parsing callback.
12561
 * If sax is NULL, fallback to the default DOM tree building routines.
12562
 *
12563
 * This function uses deprecated global variables to set parser options
12564
 * which default to XML_PARSE_NODICT.
12565
 *
12566
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12567
 *
12568
 * @param sax  the SAX handler block
12569
 * @param filename  the filename
12570
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12571
 *             documents
12572
 * @returns the resulting document tree
12573
 */
12574
12575
xmlDoc *
12576
xmlSAXParseFile(xmlSAXHandler *sax, const char *filename,
12577
0
                          int recovery) {
12578
0
    return(xmlSAXParseFileWithData(sax,filename,recovery,NULL));
12579
0
}
12580
12581
/**
12582
 * Parse an XML in-memory document and build a tree.
12583
 * In the case the document is not Well Formed, a attempt to build a
12584
 * tree is tried anyway
12585
 *
12586
 * This function uses deprecated global variables to set parser options
12587
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12588
 *
12589
 * @deprecated Use #xmlReadDoc with XML_PARSE_RECOVER.
12590
 *
12591
 * @param cur  a pointer to an array of xmlChar
12592
 * @returns the resulting document tree or NULL in case of failure
12593
 */
12594
12595
xmlDoc *
12596
0
xmlRecoverDoc(const xmlChar *cur) {
12597
0
    return(xmlSAXParseDoc(NULL, cur, 1));
12598
0
}
12599
12600
/**
12601
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12602
 * compressed document is provided by default if found at compile-time.
12603
 *
12604
 * This function uses deprecated global variables to set parser options
12605
 * which default to XML_PARSE_NODICT.
12606
 *
12607
 * @deprecated Use #xmlReadFile.
12608
 *
12609
 * @param filename  the filename
12610
 * @returns the resulting document tree if the file was wellformed,
12611
 * NULL otherwise.
12612
 */
12613
12614
xmlDoc *
12615
0
xmlParseFile(const char *filename) {
12616
0
    return(xmlSAXParseFile(NULL, filename, 0));
12617
0
}
12618
12619
/**
12620
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12621
 * compressed document is provided by default if found at compile-time.
12622
 * In the case the document is not Well Formed, it attempts to build
12623
 * a tree anyway
12624
 *
12625
 * This function uses deprecated global variables to set parser options
12626
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12627
 *
12628
 * @deprecated Use #xmlReadFile with XML_PARSE_RECOVER.
12629
 *
12630
 * @param filename  the filename
12631
 * @returns the resulting document tree or NULL in case of failure
12632
 */
12633
12634
xmlDoc *
12635
0
xmlRecoverFile(const char *filename) {
12636
0
    return(xmlSAXParseFile(NULL, filename, 1));
12637
0
}
12638
12639
12640
/**
12641
 * Setup the parser context to parse a new buffer; Clears any prior
12642
 * contents from the parser context. The buffer parameter must not be
12643
 * NULL, but the filename parameter can be
12644
 *
12645
 * @deprecated Don't use.
12646
 *
12647
 * @param ctxt  an XML parser context
12648
 * @param buffer  a xmlChar * buffer
12649
 * @param filename  a file name
12650
 */
12651
void
12652
xmlSetupParserForBuffer(xmlParserCtxt *ctxt, const xmlChar* buffer,
12653
                             const char* filename)
12654
0
{
12655
0
    xmlParserInputPtr input;
12656
12657
0
    if ((ctxt == NULL) || (buffer == NULL))
12658
0
        return;
12659
12660
0
    xmlCtxtReset(ctxt);
12661
12662
0
    input = xmlCtxtNewInputFromString(ctxt, filename, (const char *) buffer,
12663
0
                                      NULL, 0);
12664
0
    if (input == NULL)
12665
0
        return;
12666
0
    if (xmlCtxtPushInput(ctxt, input) < 0)
12667
0
        xmlFreeInputStream(input);
12668
0
}
12669
12670
/**
12671
 * Parse an XML file and call the given SAX handler routines.
12672
 * Automatic support for ZLIB/Compress compressed document is provided
12673
 *
12674
 * This function uses deprecated global variables to set parser options
12675
 * which default to XML_PARSE_NODICT.
12676
 *
12677
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12678
 *
12679
 * @param sax  a SAX handler
12680
 * @param user_data  The user data returned on SAX callbacks
12681
 * @param filename  a file name
12682
 * @returns 0 in case of success or a error number otherwise
12683
 */
12684
int
12685
xmlSAXUserParseFile(xmlSAXHandler *sax, void *user_data,
12686
0
                    const char *filename) {
12687
0
    int ret = 0;
12688
0
    xmlParserCtxtPtr ctxt;
12689
12690
0
    ctxt = xmlCreateFileParserCtxt(filename);
12691
0
    if (ctxt == NULL) return -1;
12692
0
    if (sax != NULL) {
12693
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12694
0
            *ctxt->sax = *sax;
12695
0
        } else {
12696
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12697
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12698
0
        }
12699
0
  ctxt->userData = user_data;
12700
0
    }
12701
12702
0
    xmlParseDocument(ctxt);
12703
12704
0
    if (ctxt->wellFormed)
12705
0
  ret = 0;
12706
0
    else {
12707
0
        if (ctxt->errNo != 0)
12708
0
      ret = ctxt->errNo;
12709
0
  else
12710
0
      ret = -1;
12711
0
    }
12712
0
    if (ctxt->myDoc != NULL) {
12713
0
        xmlFreeDoc(ctxt->myDoc);
12714
0
  ctxt->myDoc = NULL;
12715
0
    }
12716
0
    xmlFreeParserCtxt(ctxt);
12717
12718
0
    return ret;
12719
0
}
12720
#endif /* LIBXML_SAX1_ENABLED */
12721
12722
/************************************************************************
12723
 *                  *
12724
 *    Front ends when parsing from memory     *
12725
 *                  *
12726
 ************************************************************************/
12727
12728
/**
12729
 * Create a parser context for an XML in-memory document. The input buffer
12730
 * must not contain a terminating null byte.
12731
 *
12732
 * @param buffer  a pointer to a char array
12733
 * @param size  the size of the array
12734
 * @returns the new parser context or NULL
12735
 */
12736
xmlParserCtxt *
12737
0
xmlCreateMemoryParserCtxt(const char *buffer, int size) {
12738
0
    xmlParserCtxtPtr ctxt;
12739
0
    xmlParserInputPtr input;
12740
12741
0
    if (size < 0)
12742
0
  return(NULL);
12743
12744
0
    ctxt = xmlNewParserCtxt();
12745
0
    if (ctxt == NULL)
12746
0
  return(NULL);
12747
12748
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, 0);
12749
0
    if (input == NULL) {
12750
0
  xmlFreeParserCtxt(ctxt);
12751
0
  return(NULL);
12752
0
    }
12753
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12754
0
        xmlFreeInputStream(input);
12755
0
        xmlFreeParserCtxt(ctxt);
12756
0
        return(NULL);
12757
0
    }
12758
12759
0
    return(ctxt);
12760
0
}
12761
12762
#ifdef LIBXML_SAX1_ENABLED
12763
/**
12764
 * Parse an XML in-memory block and use the given SAX function block
12765
 * to handle the parsing callback. If sax is NULL, fallback to the default
12766
 * DOM tree building routines.
12767
 *
12768
 * This function uses deprecated global variables to set parser options
12769
 * which default to XML_PARSE_NODICT.
12770
 *
12771
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12772
 *
12773
 * User data (void *) is stored within the parser context in the
12774
 * context's _private member, so it is available nearly everywhere in libxml
12775
 *
12776
 * @param sax  the SAX handler block
12777
 * @param buffer  an pointer to a char array
12778
 * @param size  the size of the array
12779
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12780
 *             documents
12781
 * @param data  the userdata
12782
 * @returns the resulting document tree
12783
 */
12784
12785
xmlDoc *
12786
xmlSAXParseMemoryWithData(xmlSAXHandler *sax, const char *buffer,
12787
0
                          int size, int recovery, void *data) {
12788
0
    xmlDocPtr ret = NULL;
12789
0
    xmlParserCtxtPtr ctxt;
12790
0
    xmlParserInputPtr input;
12791
12792
0
    if (size < 0)
12793
0
        return(NULL);
12794
12795
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12796
0
    if (ctxt == NULL)
12797
0
        return(NULL);
12798
12799
0
    if (data != NULL)
12800
0
  ctxt->_private=data;
12801
12802
0
    if (recovery) {
12803
0
        ctxt->options |= XML_PARSE_RECOVER;
12804
0
        ctxt->recovery = 1;
12805
0
    }
12806
12807
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL,
12808
0
                                      XML_INPUT_BUF_STATIC);
12809
12810
0
    if (input != NULL)
12811
0
        ret = xmlCtxtParseDocument(ctxt, input);
12812
12813
0
    xmlFreeParserCtxt(ctxt);
12814
0
    return(ret);
12815
0
}
12816
12817
/**
12818
 * Parse an XML in-memory block and use the given SAX function block
12819
 * to handle the parsing callback. If sax is NULL, fallback to the default
12820
 * DOM tree building routines.
12821
 *
12822
 * This function uses deprecated global variables to set parser options
12823
 * which default to XML_PARSE_NODICT.
12824
 *
12825
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12826
 *
12827
 * @param sax  the SAX handler block
12828
 * @param buffer  an pointer to a char array
12829
 * @param size  the size of the array
12830
 * @param recovery  work in recovery mode, i.e. tries to read not Well Formed
12831
 *             documents
12832
 * @returns the resulting document tree
12833
 */
12834
xmlDoc *
12835
xmlSAXParseMemory(xmlSAXHandler *sax, const char *buffer,
12836
0
            int size, int recovery) {
12837
0
    return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL);
12838
0
}
12839
12840
/**
12841
 * Parse an XML in-memory block and build a tree.
12842
 *
12843
 * This function uses deprecated global variables to set parser options
12844
 * which default to XML_PARSE_NODICT.
12845
 *
12846
 * @deprecated Use #xmlReadMemory.
12847
 *
12848
 * @param buffer  an pointer to a char array
12849
 * @param size  the size of the array
12850
 * @returns the resulting document tree
12851
 */
12852
12853
0
xmlDoc *xmlParseMemory(const char *buffer, int size) {
12854
0
   return(xmlSAXParseMemory(NULL, buffer, size, 0));
12855
0
}
12856
12857
/**
12858
 * Parse an XML in-memory block and build a tree.
12859
 * In the case the document is not Well Formed, an attempt to
12860
 * build a tree is tried anyway
12861
 *
12862
 * This function uses deprecated global variables to set parser options
12863
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12864
 *
12865
 * @deprecated Use #xmlReadMemory with XML_PARSE_RECOVER.
12866
 *
12867
 * @param buffer  an pointer to a char array
12868
 * @param size  the size of the array
12869
 * @returns the resulting document tree or NULL in case of error
12870
 */
12871
12872
0
xmlDoc *xmlRecoverMemory(const char *buffer, int size) {
12873
0
   return(xmlSAXParseMemory(NULL, buffer, size, 1));
12874
0
}
12875
12876
/**
12877
 * Parse an XML in-memory buffer and call the given SAX handler routines.
12878
 *
12879
 * This function uses deprecated global variables to set parser options
12880
 * which default to XML_PARSE_NODICT.
12881
 *
12882
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12883
 *
12884
 * @param sax  a SAX handler
12885
 * @param user_data  The user data returned on SAX callbacks
12886
 * @param buffer  an in-memory XML document input
12887
 * @param size  the length of the XML document in bytes
12888
 * @returns 0 in case of success or a error number otherwise
12889
 */
12890
int xmlSAXUserParseMemory(xmlSAXHandler *sax, void *user_data,
12891
0
        const char *buffer, int size) {
12892
0
    int ret = 0;
12893
0
    xmlParserCtxtPtr ctxt;
12894
12895
0
    ctxt = xmlCreateMemoryParserCtxt(buffer, size);
12896
0
    if (ctxt == NULL) return -1;
12897
0
    if (sax != NULL) {
12898
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12899
0
            *ctxt->sax = *sax;
12900
0
        } else {
12901
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12902
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12903
0
        }
12904
0
  ctxt->userData = user_data;
12905
0
    }
12906
12907
0
    xmlParseDocument(ctxt);
12908
12909
0
    if (ctxt->wellFormed)
12910
0
  ret = 0;
12911
0
    else {
12912
0
        if (ctxt->errNo != 0)
12913
0
      ret = ctxt->errNo;
12914
0
  else
12915
0
      ret = -1;
12916
0
    }
12917
0
    if (ctxt->myDoc != NULL) {
12918
0
        xmlFreeDoc(ctxt->myDoc);
12919
0
  ctxt->myDoc = NULL;
12920
0
    }
12921
0
    xmlFreeParserCtxt(ctxt);
12922
12923
0
    return ret;
12924
0
}
12925
#endif /* LIBXML_SAX1_ENABLED */
12926
12927
/**
12928
 * Creates a parser context for an XML in-memory document.
12929
 *
12930
 * @param str  a pointer to an array of xmlChar
12931
 * @returns the new parser context or NULL
12932
 */
12933
xmlParserCtxt *
12934
0
xmlCreateDocParserCtxt(const xmlChar *str) {
12935
0
    xmlParserCtxtPtr ctxt;
12936
0
    xmlParserInputPtr input;
12937
12938
0
    ctxt = xmlNewParserCtxt();
12939
0
    if (ctxt == NULL)
12940
0
  return(NULL);
12941
12942
0
    input = xmlCtxtNewInputFromString(ctxt, NULL, (const char *) str, NULL, 0);
12943
0
    if (input == NULL) {
12944
0
  xmlFreeParserCtxt(ctxt);
12945
0
  return(NULL);
12946
0
    }
12947
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12948
0
        xmlFreeInputStream(input);
12949
0
        xmlFreeParserCtxt(ctxt);
12950
0
        return(NULL);
12951
0
    }
12952
12953
0
    return(ctxt);
12954
0
}
12955
12956
#ifdef LIBXML_SAX1_ENABLED
12957
/**
12958
 * Parse an XML in-memory document and build a tree.
12959
 * It use the given SAX function block to handle the parsing callback.
12960
 * If sax is NULL, fallback to the default DOM tree building routines.
12961
 *
12962
 * This function uses deprecated global variables to set parser options
12963
 * which default to XML_PARSE_NODICT.
12964
 *
12965
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadDoc.
12966
 *
12967
 * @param sax  the SAX handler block
12968
 * @param cur  a pointer to an array of xmlChar
12969
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12970
 *             documents
12971
 * @returns the resulting document tree
12972
 */
12973
12974
xmlDoc *
12975
0
xmlSAXParseDoc(xmlSAXHandler *sax, const xmlChar *cur, int recovery) {
12976
0
    xmlDocPtr ret;
12977
0
    xmlParserCtxtPtr ctxt;
12978
0
    xmlSAXHandlerPtr oldsax = NULL;
12979
12980
0
    if (cur == NULL) return(NULL);
12981
12982
12983
0
    ctxt = xmlCreateDocParserCtxt(cur);
12984
0
    if (ctxt == NULL) return(NULL);
12985
0
    if (sax != NULL) {
12986
0
        oldsax = ctxt->sax;
12987
0
        ctxt->sax = sax;
12988
0
        ctxt->userData = NULL;
12989
0
    }
12990
12991
0
    xmlParseDocument(ctxt);
12992
0
    if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc;
12993
0
    else {
12994
0
       ret = NULL;
12995
0
       xmlFreeDoc(ctxt->myDoc);
12996
0
       ctxt->myDoc = NULL;
12997
0
    }
12998
0
    if (sax != NULL)
12999
0
  ctxt->sax = oldsax;
13000
0
    xmlFreeParserCtxt(ctxt);
13001
13002
0
    return(ret);
13003
0
}
13004
13005
/**
13006
 * Parse an XML in-memory document and build a tree.
13007
 *
13008
 * This function uses deprecated global variables to set parser options
13009
 * which default to XML_PARSE_NODICT.
13010
 *
13011
 * @deprecated Use #xmlReadDoc.
13012
 *
13013
 * @param cur  a pointer to an array of xmlChar
13014
 * @returns the resulting document tree
13015
 */
13016
13017
xmlDoc *
13018
0
xmlParseDoc(const xmlChar *cur) {
13019
0
    return(xmlSAXParseDoc(NULL, cur, 0));
13020
0
}
13021
#endif /* LIBXML_SAX1_ENABLED */
13022
13023
/************************************************************************
13024
 *                  *
13025
 *  New set (2.6.0) of simpler and more flexible APIs   *
13026
 *                  *
13027
 ************************************************************************/
13028
13029
/**
13030
 * Reset a parser context
13031
 *
13032
 * @param ctxt  an XML parser context
13033
 */
13034
void
13035
xmlCtxtReset(xmlParserCtxt *ctxt)
13036
75.7k
{
13037
75.7k
    xmlParserInputPtr input;
13038
13039
75.7k
    if (ctxt == NULL)
13040
0
        return;
13041
13042
75.7k
    while ((input = xmlCtxtPopInput(ctxt)) != NULL) { /* Non consuming */
13043
0
        xmlFreeInputStream(input);
13044
0
    }
13045
75.7k
    ctxt->inputNr = 0;
13046
75.7k
    ctxt->input = NULL;
13047
13048
75.7k
    ctxt->spaceNr = 0;
13049
75.7k
    if (ctxt->spaceTab != NULL) {
13050
75.7k
  ctxt->spaceTab[0] = -1;
13051
75.7k
  ctxt->space = &ctxt->spaceTab[0];
13052
75.7k
    } else {
13053
0
        ctxt->space = NULL;
13054
0
    }
13055
13056
13057
75.7k
    ctxt->nodeNr = 0;
13058
75.7k
    ctxt->node = NULL;
13059
13060
75.7k
    ctxt->nameNr = 0;
13061
75.7k
    ctxt->name = NULL;
13062
13063
75.7k
    ctxt->nsNr = 0;
13064
75.7k
    xmlParserNsReset(ctxt->nsdb);
13065
13066
75.7k
    if (ctxt->version != NULL) {
13067
16.3k
        xmlFree(ctxt->version);
13068
16.3k
        ctxt->version = NULL;
13069
16.3k
    }
13070
75.7k
    if (ctxt->encoding != NULL) {
13071
877
        xmlFree(ctxt->encoding);
13072
877
        ctxt->encoding = NULL;
13073
877
    }
13074
75.7k
    if (ctxt->extSubURI != NULL) {
13075
2.44k
        xmlFree(ctxt->extSubURI);
13076
2.44k
        ctxt->extSubURI = NULL;
13077
2.44k
    }
13078
75.7k
    if (ctxt->extSubSystem != NULL) {
13079
328
        xmlFree(ctxt->extSubSystem);
13080
328
        ctxt->extSubSystem = NULL;
13081
328
    }
13082
75.7k
    if (ctxt->directory != NULL) {
13083
16.9k
        xmlFree(ctxt->directory);
13084
16.9k
        ctxt->directory = NULL;
13085
16.9k
    }
13086
13087
75.7k
    if (ctxt->myDoc != NULL)
13088
0
        xmlFreeDoc(ctxt->myDoc);
13089
75.7k
    ctxt->myDoc = NULL;
13090
13091
75.7k
    ctxt->standalone = -1;
13092
75.7k
    ctxt->hasExternalSubset = 0;
13093
75.7k
    ctxt->hasPErefs = 0;
13094
75.7k
    ctxt->html = ctxt->html ? 1 : 0;
13095
75.7k
    ctxt->instate = XML_PARSER_START;
13096
13097
75.7k
    ctxt->wellFormed = 1;
13098
75.7k
    ctxt->nsWellFormed = 1;
13099
75.7k
    ctxt->disableSAX = 0;
13100
75.7k
    ctxt->valid = 1;
13101
75.7k
    ctxt->record_info = 0;
13102
75.7k
    ctxt->checkIndex = 0;
13103
75.7k
    ctxt->endCheckState = 0;
13104
75.7k
    ctxt->inSubset = 0;
13105
75.7k
    ctxt->errNo = XML_ERR_OK;
13106
75.7k
    ctxt->depth = 0;
13107
75.7k
    ctxt->catalogs = NULL;
13108
75.7k
    ctxt->sizeentities = 0;
13109
75.7k
    ctxt->sizeentcopy = 0;
13110
75.7k
    xmlInitNodeInfoSeq(&ctxt->node_seq);
13111
13112
75.7k
    if (ctxt->attsDefault != NULL) {
13113
1.64k
        xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator);
13114
1.64k
        ctxt->attsDefault = NULL;
13115
1.64k
    }
13116
75.7k
    if (ctxt->attsSpecial != NULL) {
13117
2.11k
        xmlHashFree(ctxt->attsSpecial, NULL);
13118
2.11k
        ctxt->attsSpecial = NULL;
13119
2.11k
    }
13120
13121
75.7k
#ifdef LIBXML_CATALOG_ENABLED
13122
75.7k
    if (ctxt->catalogs != NULL)
13123
0
  xmlCatalogFreeLocal(ctxt->catalogs);
13124
75.7k
#endif
13125
75.7k
    ctxt->nbErrors = 0;
13126
75.7k
    ctxt->nbWarnings = 0;
13127
75.7k
    if (ctxt->lastError.code != XML_ERR_OK)
13128
16.6k
        xmlResetError(&ctxt->lastError);
13129
75.7k
}
13130
13131
/**
13132
 * Reset a push parser context
13133
 *
13134
 * @param ctxt  an XML parser context
13135
 * @param chunk  a pointer to an array of chars
13136
 * @param size  number of chars in the array
13137
 * @param filename  an optional file name or URI
13138
 * @param encoding  the document encoding, or NULL
13139
 * @returns 0 in case of success and 1 in case of error
13140
 */
13141
int
13142
xmlCtxtResetPush(xmlParserCtxt *ctxt, const char *chunk,
13143
                 int size, const char *filename, const char *encoding)
13144
0
{
13145
0
    xmlParserInputPtr input;
13146
13147
0
    if (ctxt == NULL)
13148
0
        return(1);
13149
13150
0
    xmlCtxtReset(ctxt);
13151
13152
0
    input = xmlNewPushInput(filename, chunk, size);
13153
0
    if (input == NULL)
13154
0
        return(1);
13155
13156
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13157
0
        xmlFreeInputStream(input);
13158
0
        return(1);
13159
0
    }
13160
13161
0
    if (encoding != NULL)
13162
0
        xmlSwitchEncodingName(ctxt, encoding);
13163
13164
0
    return(0);
13165
0
}
13166
13167
static int
13168
xmlCtxtSetOptionsInternal(xmlParserCtxtPtr ctxt, int options, int keepMask)
13169
88.2k
{
13170
88.2k
    int allMask;
13171
13172
88.2k
    if (ctxt == NULL)
13173
0
        return(-1);
13174
13175
    /*
13176
     * XInclude options aren't handled by the parser.
13177
     *
13178
     * XML_PARSE_XINCLUDE
13179
     * XML_PARSE_NOXINCNODE
13180
     * XML_PARSE_NOBASEFIX
13181
     */
13182
88.2k
    allMask = XML_PARSE_RECOVER |
13183
88.2k
              XML_PARSE_NOENT |
13184
88.2k
              XML_PARSE_DTDLOAD |
13185
88.2k
              XML_PARSE_DTDATTR |
13186
88.2k
              XML_PARSE_DTDVALID |
13187
88.2k
              XML_PARSE_NOERROR |
13188
88.2k
              XML_PARSE_NOWARNING |
13189
88.2k
              XML_PARSE_PEDANTIC |
13190
88.2k
              XML_PARSE_NOBLANKS |
13191
88.2k
#ifdef LIBXML_SAX1_ENABLED
13192
88.2k
              XML_PARSE_SAX1 |
13193
88.2k
#endif
13194
88.2k
              XML_PARSE_NONET |
13195
88.2k
              XML_PARSE_NODICT |
13196
88.2k
              XML_PARSE_NSCLEAN |
13197
88.2k
              XML_PARSE_NOCDATA |
13198
88.2k
              XML_PARSE_COMPACT |
13199
88.2k
              XML_PARSE_OLD10 |
13200
88.2k
              XML_PARSE_HUGE |
13201
88.2k
              XML_PARSE_OLDSAX |
13202
88.2k
              XML_PARSE_IGNORE_ENC |
13203
88.2k
              XML_PARSE_BIG_LINES |
13204
88.2k
              XML_PARSE_NO_XXE |
13205
88.2k
              XML_PARSE_UNZIP |
13206
88.2k
              XML_PARSE_NO_SYS_CATALOG |
13207
88.2k
              XML_PARSE_CATALOG_PI;
13208
13209
88.2k
    ctxt->options = (ctxt->options & keepMask) | (options & allMask);
13210
13211
    /*
13212
     * For some options, struct members are historically the source
13213
     * of truth. The values are initalized from global variables and
13214
     * old code could also modify them directly. Several older API
13215
     * functions that don't take an options argument rely on these
13216
     * deprecated mechanisms.
13217
     *
13218
     * Once public access to struct members and the globals are
13219
     * disabled, we can use the options bitmask as source of
13220
     * truth, making all these struct members obsolete.
13221
     *
13222
     * The XML_DETECT_IDS flags is misnamed. It simply enables
13223
     * loading of the external subset.
13224
     */
13225
88.2k
    ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0;
13226
88.2k
    ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0;
13227
88.2k
    ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0;
13228
88.2k
    ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0;
13229
88.2k
    ctxt->loadsubset |= (options & XML_PARSE_SKIP_IDS) ? XML_SKIP_IDS : 0;
13230
88.2k
    ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0;
13231
88.2k
    ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0;
13232
88.2k
    ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1;
13233
88.2k
    ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1;
13234
13235
88.2k
    return(options & ~allMask);
13236
88.2k
}
13237
13238
/**
13239
 * Applies the options to the parser context. Unset options are
13240
 * cleared.
13241
 *
13242
 * @since 2.13.0
13243
 *
13244
 * With older versions, you can use #xmlCtxtUseOptions.
13245
 *
13246
 * @param ctxt  an XML parser context
13247
 * @param options  a bitmask of xmlParserOption values
13248
 * @returns 0 in case of success, the set of unknown or unimplemented options
13249
 *         in case of error.
13250
 */
13251
int
13252
xmlCtxtSetOptions(xmlParserCtxt *ctxt, int options)
13253
0
{
13254
0
#ifdef LIBXML_HTML_ENABLED
13255
0
    if ((ctxt != NULL) && (ctxt->html))
13256
0
        return(htmlCtxtSetOptions(ctxt, options));
13257
0
#endif
13258
13259
0
    return(xmlCtxtSetOptionsInternal(ctxt, options, 0));
13260
0
}
13261
13262
/**
13263
 * Get the current options of the parser context.
13264
 *
13265
 * @since 2.14.0
13266
 *
13267
 * @param ctxt  an XML parser context
13268
 * @returns the current options set in the parser context, or -1 if ctxt is NULL.
13269
 */
13270
int
13271
xmlCtxtGetOptions(xmlParserCtxt *ctxt)
13272
0
{
13273
0
    if (ctxt == NULL)
13274
0
        return(-1);
13275
13276
0
    return(ctxt->options);
13277
0
}
13278
13279
/**
13280
 * Applies the options to the parser context. The following options
13281
 * are never cleared and can only be enabled:
13282
 *
13283
 * - XML_PARSE_NOERROR
13284
 * - XML_PARSE_NOWARNING
13285
 * - XML_PARSE_NONET
13286
 * - XML_PARSE_NSCLEAN
13287
 * - XML_PARSE_NOCDATA
13288
 * - XML_PARSE_COMPACT
13289
 * - XML_PARSE_OLD10
13290
 * - XML_PARSE_HUGE
13291
 * - XML_PARSE_OLDSAX
13292
 * - XML_PARSE_IGNORE_ENC
13293
 * - XML_PARSE_BIG_LINES
13294
 *
13295
 * @deprecated Use #xmlCtxtSetOptions.
13296
 *
13297
 * @param ctxt  an XML parser context
13298
 * @param options  a combination of xmlParserOption
13299
 * @returns 0 in case of success, the set of unknown or unimplemented options
13300
 *         in case of error.
13301
 */
13302
int
13303
xmlCtxtUseOptions(xmlParserCtxt *ctxt, int options)
13304
88.2k
{
13305
88.2k
    int keepMask;
13306
13307
88.2k
#ifdef LIBXML_HTML_ENABLED
13308
88.2k
    if ((ctxt != NULL) && (ctxt->html))
13309
0
        return(htmlCtxtUseOptions(ctxt, options));
13310
88.2k
#endif
13311
13312
    /*
13313
     * For historic reasons, some options can only be enabled.
13314
     */
13315
88.2k
    keepMask = XML_PARSE_NOERROR |
13316
88.2k
               XML_PARSE_NOWARNING |
13317
88.2k
               XML_PARSE_NONET |
13318
88.2k
               XML_PARSE_NSCLEAN |
13319
88.2k
               XML_PARSE_NOCDATA |
13320
88.2k
               XML_PARSE_COMPACT |
13321
88.2k
               XML_PARSE_OLD10 |
13322
88.2k
               XML_PARSE_HUGE |
13323
88.2k
               XML_PARSE_OLDSAX |
13324
88.2k
               XML_PARSE_IGNORE_ENC |
13325
88.2k
               XML_PARSE_BIG_LINES;
13326
13327
88.2k
    return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask));
13328
88.2k
}
13329
13330
/**
13331
 * To protect against exponential entity expansion ("billion laughs"), the
13332
 * size of serialized output is (roughly) limited to the input size
13333
 * multiplied by this factor. The default value is 5.
13334
 *
13335
 * When working with documents making heavy use of entity expansion, it can
13336
 * be necessary to increase the value. For security reasons, this should only
13337
 * be considered when processing trusted input.
13338
 *
13339
 * @param ctxt  an XML parser context
13340
 * @param maxAmpl  maximum amplification factor
13341
 */
13342
void
13343
xmlCtxtSetMaxAmplification(xmlParserCtxt *ctxt, unsigned maxAmpl)
13344
0
{
13345
0
    if (ctxt == NULL)
13346
0
        return;
13347
0
    if (maxAmpl == 0)
13348
0
        return;
13349
0
    ctxt->maxAmpl = maxAmpl;
13350
0
}
13351
13352
/**
13353
 * Parse an XML document and return the resulting document tree.
13354
 * Takes ownership of the input object.
13355
 *
13356
 * @since 2.13.0
13357
 *
13358
 * @param ctxt  an XML parser context
13359
 * @param input  parser input
13360
 * @returns the resulting document tree or NULL
13361
 */
13362
xmlDoc *
13363
xmlCtxtParseDocument(xmlParserCtxt *ctxt, xmlParserInput *input)
13364
58.7k
{
13365
58.7k
    xmlDocPtr ret = NULL;
13366
13367
58.7k
    if ((ctxt == NULL) || (input == NULL)) {
13368
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
13369
0
        xmlFreeInputStream(input);
13370
0
        return(NULL);
13371
0
    }
13372
13373
    /* assert(ctxt->inputNr == 0); */
13374
58.7k
    while (ctxt->inputNr > 0)
13375
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13376
13377
58.7k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13378
6
        xmlFreeInputStream(input);
13379
6
        return(NULL);
13380
6
    }
13381
13382
58.7k
    xmlParseDocument(ctxt);
13383
13384
58.7k
    ret = xmlCtxtGetDocument(ctxt);
13385
13386
    /* assert(ctxt->inputNr == 1); */
13387
118k
    while (ctxt->inputNr > 0)
13388
59.7k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13389
13390
58.7k
    return(ret);
13391
58.7k
}
13392
13393
/**
13394
 * Convenience function to parse an XML document from a
13395
 * zero-terminated string.
13396
 *
13397
 * See #xmlCtxtReadDoc for details.
13398
 *
13399
 * @param cur  a pointer to a zero terminated string
13400
 * @param URL  base URL (optional)
13401
 * @param encoding  the document encoding (optional)
13402
 * @param options  a combination of xmlParserOption
13403
 * @returns the resulting document tree
13404
 */
13405
xmlDoc *
13406
xmlReadDoc(const xmlChar *cur, const char *URL, const char *encoding,
13407
           int options)
13408
0
{
13409
0
    xmlParserCtxtPtr ctxt;
13410
0
    xmlParserInputPtr input;
13411
0
    xmlDocPtr doc = NULL;
13412
13413
0
    ctxt = xmlNewParserCtxt();
13414
0
    if (ctxt == NULL)
13415
0
        return(NULL);
13416
13417
0
    xmlCtxtUseOptions(ctxt, options);
13418
13419
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) cur, encoding,
13420
0
                                      XML_INPUT_BUF_STATIC);
13421
13422
0
    if (input != NULL)
13423
0
        doc = xmlCtxtParseDocument(ctxt, input);
13424
13425
0
    xmlFreeParserCtxt(ctxt);
13426
0
    return(doc);
13427
0
}
13428
13429
/**
13430
 * Convenience function to parse an XML file from the filesystem
13431
 * or a global, user-defined resource loader.
13432
 *
13433
 * If a "-" filename is passed, the function will read from stdin.
13434
 * This feature is potentially insecure and might be removed from
13435
 * later versions.
13436
 *
13437
 * See #xmlCtxtReadFile for details.
13438
 *
13439
 * @param filename  a file or URL
13440
 * @param encoding  the document encoding (optional)
13441
 * @param options  a combination of xmlParserOption
13442
 * @returns the resulting document tree
13443
 */
13444
xmlDoc *
13445
xmlReadFile(const char *filename, const char *encoding, int options)
13446
0
{
13447
0
    xmlParserCtxtPtr ctxt;
13448
0
    xmlParserInputPtr input;
13449
0
    xmlDocPtr doc = NULL;
13450
13451
0
    ctxt = xmlNewParserCtxt();
13452
0
    if (ctxt == NULL)
13453
0
        return(NULL);
13454
13455
0
    xmlCtxtUseOptions(ctxt, options);
13456
13457
    /*
13458
     * Backward compatibility for users of command line utilities like
13459
     * xmlstarlet expecting "-" to mean stdin. This is dangerous and
13460
     * should be removed at some point.
13461
     */
13462
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
13463
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO,
13464
0
                                      encoding, 0);
13465
0
    else
13466
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13467
13468
0
    if (input != NULL)
13469
0
        doc = xmlCtxtParseDocument(ctxt, input);
13470
13471
0
    xmlFreeParserCtxt(ctxt);
13472
0
    return(doc);
13473
0
}
13474
13475
/**
13476
 * Parse an XML in-memory document and build a tree. The input buffer must
13477
 * not contain a terminating null byte.
13478
 *
13479
 * See #xmlCtxtReadMemory for details.
13480
 *
13481
 * @param buffer  a pointer to a char array
13482
 * @param size  the size of the array
13483
 * @param url  base URL (optional)
13484
 * @param encoding  the document encoding (optional)
13485
 * @param options  a combination of xmlParserOption
13486
 * @returns the resulting document tree
13487
 */
13488
xmlDoc *
13489
xmlReadMemory(const char *buffer, int size, const char *url,
13490
              const char *encoding, int options)
13491
0
{
13492
0
    xmlParserCtxtPtr ctxt;
13493
0
    xmlParserInputPtr input;
13494
0
    xmlDocPtr doc = NULL;
13495
13496
0
    if (size < 0)
13497
0
  return(NULL);
13498
13499
0
    ctxt = xmlNewParserCtxt();
13500
0
    if (ctxt == NULL)
13501
0
        return(NULL);
13502
13503
0
    xmlCtxtUseOptions(ctxt, options);
13504
13505
0
    input = xmlCtxtNewInputFromMemory(ctxt, url, buffer, size, encoding,
13506
0
                                      XML_INPUT_BUF_STATIC);
13507
13508
0
    if (input != NULL)
13509
0
        doc = xmlCtxtParseDocument(ctxt, input);
13510
13511
0
    xmlFreeParserCtxt(ctxt);
13512
0
    return(doc);
13513
0
}
13514
13515
/**
13516
 * Parse an XML from a file descriptor and build a tree.
13517
 *
13518
 * See #xmlCtxtReadFd for details.
13519
 *
13520
 * NOTE that the file descriptor will not be closed when the
13521
 * context is freed or reset.
13522
 *
13523
 * @param fd  an open file descriptor
13524
 * @param URL  base URL (optional)
13525
 * @param encoding  the document encoding (optional)
13526
 * @param options  a combination of xmlParserOption
13527
 * @returns the resulting document tree
13528
 */
13529
xmlDoc *
13530
xmlReadFd(int fd, const char *URL, const char *encoding, int options)
13531
0
{
13532
0
    xmlParserCtxtPtr ctxt;
13533
0
    xmlParserInputPtr input;
13534
0
    xmlDocPtr doc = NULL;
13535
13536
0
    ctxt = xmlNewParserCtxt();
13537
0
    if (ctxt == NULL)
13538
0
        return(NULL);
13539
13540
0
    xmlCtxtUseOptions(ctxt, options);
13541
13542
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
13543
13544
0
    if (input != NULL)
13545
0
        doc = xmlCtxtParseDocument(ctxt, input);
13546
13547
0
    xmlFreeParserCtxt(ctxt);
13548
0
    return(doc);
13549
0
}
13550
13551
/**
13552
 * Parse an XML document from I/O functions and context and build a tree.
13553
 *
13554
 * See #xmlCtxtReadIO for details.
13555
 *
13556
 * @param ioread  an I/O read function
13557
 * @param ioclose  an I/O close function (optional)
13558
 * @param ioctx  an I/O handler
13559
 * @param URL  base URL (optional)
13560
 * @param encoding  the document encoding (optional)
13561
 * @param options  a combination of xmlParserOption
13562
 * @returns the resulting document tree
13563
 */
13564
xmlDoc *
13565
xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose,
13566
          void *ioctx, const char *URL, const char *encoding, int options)
13567
0
{
13568
0
    xmlParserCtxtPtr ctxt;
13569
0
    xmlParserInputPtr input;
13570
0
    xmlDocPtr doc = NULL;
13571
13572
0
    ctxt = xmlNewParserCtxt();
13573
0
    if (ctxt == NULL)
13574
0
        return(NULL);
13575
13576
0
    xmlCtxtUseOptions(ctxt, options);
13577
13578
0
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
13579
0
                                  encoding, 0);
13580
13581
0
    if (input != NULL)
13582
0
        doc = xmlCtxtParseDocument(ctxt, input);
13583
13584
0
    xmlFreeParserCtxt(ctxt);
13585
0
    return(doc);
13586
0
}
13587
13588
/**
13589
 * Parse an XML in-memory document and build a tree.
13590
 *
13591
 * `URL` is used as base to resolve external entities and for error
13592
 * reporting.
13593
 *
13594
 * @param ctxt  an XML parser context
13595
 * @param str  a pointer to a zero terminated string
13596
 * @param URL  base URL (optional)
13597
 * @param encoding  the document encoding (optional)
13598
 * @param options  a combination of xmlParserOption
13599
 * @returns the resulting document tree
13600
 */
13601
xmlDoc *
13602
xmlCtxtReadDoc(xmlParserCtxt *ctxt, const xmlChar *str,
13603
               const char *URL, const char *encoding, int options)
13604
0
{
13605
0
    xmlParserInputPtr input;
13606
13607
0
    if (ctxt == NULL)
13608
0
        return(NULL);
13609
13610
0
    xmlCtxtReset(ctxt);
13611
0
    xmlCtxtUseOptions(ctxt, options);
13612
13613
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) str, encoding,
13614
0
                                      XML_INPUT_BUF_STATIC);
13615
0
    if (input == NULL)
13616
0
        return(NULL);
13617
13618
0
    return(xmlCtxtParseDocument(ctxt, input));
13619
0
}
13620
13621
/**
13622
 * Parse an XML file from the filesystem or a global, user-defined
13623
 * resource loader.
13624
 *
13625
 * @param ctxt  an XML parser context
13626
 * @param filename  a file or URL
13627
 * @param encoding  the document encoding (optional)
13628
 * @param options  a combination of xmlParserOption
13629
 * @returns the resulting document tree
13630
 */
13631
xmlDoc *
13632
xmlCtxtReadFile(xmlParserCtxt *ctxt, const char *filename,
13633
                const char *encoding, int options)
13634
0
{
13635
0
    xmlParserInputPtr input;
13636
13637
0
    if (ctxt == NULL)
13638
0
        return(NULL);
13639
13640
0
    xmlCtxtReset(ctxt);
13641
0
    xmlCtxtUseOptions(ctxt, options);
13642
13643
0
    input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13644
0
    if (input == NULL)
13645
0
        return(NULL);
13646
13647
0
    return(xmlCtxtParseDocument(ctxt, input));
13648
0
}
13649
13650
/**
13651
 * Parse an XML in-memory document and build a tree. The input buffer must
13652
 * not contain a terminating null byte.
13653
 *
13654
 * `URL` is used as base to resolve external entities and for error
13655
 * reporting.
13656
 *
13657
 * @param ctxt  an XML parser context
13658
 * @param buffer  a pointer to a char array
13659
 * @param size  the size of the array
13660
 * @param URL  base URL (optional)
13661
 * @param encoding  the document encoding (optional)
13662
 * @param options  a combination of xmlParserOption
13663
 * @returns the resulting document tree
13664
 */
13665
xmlDoc *
13666
xmlCtxtReadMemory(xmlParserCtxt *ctxt, const char *buffer, int size,
13667
                  const char *URL, const char *encoding, int options)
13668
58.8k
{
13669
58.8k
    xmlParserInputPtr input;
13670
13671
58.8k
    if ((ctxt == NULL) || (size < 0))
13672
0
        return(NULL);
13673
13674
58.8k
    xmlCtxtReset(ctxt);
13675
58.8k
    xmlCtxtUseOptions(ctxt, options);
13676
13677
58.8k
    input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding,
13678
58.8k
                                      XML_INPUT_BUF_STATIC);
13679
58.8k
    if (input == NULL)
13680
42
        return(NULL);
13681
13682
58.7k
    return(xmlCtxtParseDocument(ctxt, input));
13683
58.8k
}
13684
13685
/**
13686
 * Parse an XML document from a file descriptor and build a tree.
13687
 *
13688
 * NOTE that the file descriptor will not be closed when the
13689
 * context is freed or reset.
13690
 *
13691
 * `URL` is used as base to resolve external entities and for error
13692
 * reporting.
13693
 *
13694
 * @param ctxt  an XML parser context
13695
 * @param fd  an open file descriptor
13696
 * @param URL  base URL (optional)
13697
 * @param encoding  the document encoding (optional)
13698
 * @param options  a combination of xmlParserOption
13699
 * @returns the resulting document tree
13700
 */
13701
xmlDoc *
13702
xmlCtxtReadFd(xmlParserCtxt *ctxt, int fd,
13703
              const char *URL, const char *encoding, int options)
13704
0
{
13705
0
    xmlParserInputPtr input;
13706
13707
0
    if (ctxt == NULL)
13708
0
        return(NULL);
13709
13710
0
    xmlCtxtReset(ctxt);
13711
0
    xmlCtxtUseOptions(ctxt, options);
13712
13713
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
13714
0
    if (input == NULL)
13715
0
        return(NULL);
13716
13717
0
    return(xmlCtxtParseDocument(ctxt, input));
13718
0
}
13719
13720
/**
13721
 * Parse an XML document from I/O functions and source and build a tree.
13722
 * This reuses the existing `ctxt` parser context
13723
 *
13724
 * `URL` is used as base to resolve external entities and for error
13725
 * reporting.
13726
 *
13727
 * @param ctxt  an XML parser context
13728
 * @param ioread  an I/O read function
13729
 * @param ioclose  an I/O close function
13730
 * @param ioctx  an I/O handler
13731
 * @param URL  the base URL to use for the document
13732
 * @param encoding  the document encoding, or NULL
13733
 * @param options  a combination of xmlParserOption
13734
 * @returns the resulting document tree
13735
 */
13736
xmlDoc *
13737
xmlCtxtReadIO(xmlParserCtxt *ctxt, xmlInputReadCallback ioread,
13738
              xmlInputCloseCallback ioclose, void *ioctx,
13739
        const char *URL,
13740
              const char *encoding, int options)
13741
0
{
13742
0
    xmlParserInputPtr input;
13743
13744
0
    if (ctxt == NULL)
13745
0
        return(NULL);
13746
13747
0
    xmlCtxtReset(ctxt);
13748
0
    xmlCtxtUseOptions(ctxt, options);
13749
13750
0
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
13751
0
                                  encoding, 0);
13752
0
    if (input == NULL)
13753
0
        return(NULL);
13754
13755
0
    return(xmlCtxtParseDocument(ctxt, input));
13756
0
}
13757