Coverage Report

Created: 2025-12-04 06:50

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
405k
#define NS_INDEX_EMPTY  INT_MAX
81
74.2k
#define NS_INDEX_XML    (INT_MAX - 1)
82
178k
#define URI_HASH_EMPTY  0xD943A04E
83
7.72k
#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
178k
#define XML_MAX_ATTRS 100000000 /* 100 million */
94
95
367k
#define XML_SPECIAL_EXTERNAL    (1 << 20)
96
333k
#define XML_SPECIAL_TYPE_MASK   (XML_SPECIAL_EXTERNAL - 1)
97
98
370k
#define XML_ATTVAL_ALLOC        (1 << 0)
99
620k
#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
1.03M
#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
1.04M
#define XML_ENT_FIXED_COST 20
170
171
28.0M
#define XML_PARSER_BIG_BUFFER_SIZE 300
172
884k
#define XML_PARSER_BUFFER_SIZE 100
173
97.8k
#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
418
xmlErrMemory(xmlParserCtxtPtr ctxt) {
225
418
    xmlCtxtErrMemory(ctxt);
226
418
}
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
19.3k
{
239
19.3k
    if (prefix == NULL)
240
10.6k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
241
10.6k
                   XML_ERR_FATAL, localname, NULL, NULL, 0,
242
10.6k
                   "Attribute %s redefined\n", localname);
243
8.70k
    else
244
8.70k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
245
8.70k
                   XML_ERR_FATAL, prefix, localname, NULL, 0,
246
8.70k
                   "Attribute %s:%s redefined\n", prefix, localname);
247
19.3k
}
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
23.1M
{
260
23.1M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
261
23.1M
               NULL, NULL, NULL, 0, "%s", msg);
262
23.1M
}
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
21.6k
{
277
21.6k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING,
278
21.6k
               str1, str2, NULL, 0, msg, str1, str2);
279
21.6k
}
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
0
{
295
0
    ctxt->valid = 0;
296
297
0
    xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR,
298
0
               str1, str2, NULL, 0, msg, str1, str2);
299
0
}
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
2.16M
{
314
2.16M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
315
2.16M
               NULL, NULL, NULL, val, msg, val);
316
2.16M
}
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
887k
{
333
887k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
334
887k
               str1, str2, NULL, val, msg, str1, val, str2);
335
887k
}
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
641k
{
349
641k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
350
641k
               val, NULL, NULL, 0, msg, val);
351
641k
}
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
106k
{
365
106k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR,
366
106k
               val, NULL, NULL, 0, msg, val);
367
106k
}
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
142k
{
385
142k
    ctxt->nsWellFormed = 0;
386
387
142k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR,
388
142k
               info1, info2, info3, 0, msg, info1, info2, info3);
389
142k
}
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
13.4k
{
407
13.4k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING,
408
13.4k
               info1, info2, info3, 0, msg, info1, info2, info3);
409
13.4k
}
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
1.10M
{
436
1.10M
    unsigned long consumed;
437
1.10M
    unsigned long *expandedSize;
438
1.10M
    xmlParserInputPtr input = ctxt->input;
439
1.10M
    xmlEntityPtr entity = input->entity;
440
441
1.10M
    if ((entity) && (entity->flags & XML_ENT_CHECKED))
442
64.4k
        return(0);
443
444
    /*
445
     * Compute total consumed bytes so far, including input streams of
446
     * external entities.
447
     */
448
1.03M
    consumed = input->consumed;
449
1.03M
    xmlSaturatedAddSizeT(&consumed, input->cur - input->base);
450
1.03M
    xmlSaturatedAdd(&consumed, ctxt->sizeentities);
451
452
1.03M
    if (entity)
453
22.1k
        expandedSize = &entity->expandedSize;
454
1.01M
    else
455
1.01M
        expandedSize = &ctxt->sizeentcopy;
456
457
    /*
458
     * Add extra cost and some fixed cost.
459
     */
460
1.03M
    xmlSaturatedAdd(expandedSize, extra);
461
1.03M
    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
1.03M
    if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) &&
469
165k
        ((*expandedSize >= ULONG_MAX) ||
470
165k
         (*expandedSize / ctxt->maxAmpl > consumed))) {
471
459
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
472
459
                       "Maximum entity amplification factor exceeded, see "
473
459
                       "xmlCtxtSetMaxAmplification.\n");
474
459
        return(1);
475
459
    }
476
477
1.03M
    return(0);
478
1.03M
}
479
480
/************************************************************************
481
 *                  *
482
 *    Library wide options          *
483
 *                  *
484
 ************************************************************************/
485
486
/**
487
 * Examines if the library has been compiled with a given feature.
488
 *
489
 * @param feature  the feature to be examined
490
 * @returns zero (0) if the feature does not exist or an unknown
491
 * feature is requested, non-zero otherwise.
492
 */
493
int
494
xmlHasFeature(xmlFeature feature)
495
0
{
496
0
    switch (feature) {
497
0
  case XML_WITH_THREAD:
498
0
#ifdef LIBXML_THREAD_ENABLED
499
0
      return(1);
500
#else
501
      return(0);
502
#endif
503
0
        case XML_WITH_TREE:
504
0
            return(1);
505
0
        case XML_WITH_OUTPUT:
506
0
#ifdef LIBXML_OUTPUT_ENABLED
507
0
            return(1);
508
#else
509
            return(0);
510
#endif
511
0
        case XML_WITH_PUSH:
512
0
#ifdef LIBXML_PUSH_ENABLED
513
0
            return(1);
514
#else
515
            return(0);
516
#endif
517
0
        case XML_WITH_READER:
518
0
#ifdef LIBXML_READER_ENABLED
519
0
            return(1);
520
#else
521
            return(0);
522
#endif
523
0
        case XML_WITH_PATTERN:
524
0
#ifdef LIBXML_PATTERN_ENABLED
525
0
            return(1);
526
#else
527
            return(0);
528
#endif
529
0
        case XML_WITH_WRITER:
530
0
#ifdef LIBXML_WRITER_ENABLED
531
0
            return(1);
532
#else
533
            return(0);
534
#endif
535
0
        case XML_WITH_SAX1:
536
0
#ifdef LIBXML_SAX1_ENABLED
537
0
            return(1);
538
#else
539
            return(0);
540
#endif
541
0
        case XML_WITH_HTTP:
542
0
            return(0);
543
0
        case XML_WITH_VALID:
544
0
#ifdef LIBXML_VALID_ENABLED
545
0
            return(1);
546
#else
547
            return(0);
548
#endif
549
0
        case XML_WITH_HTML:
550
0
#ifdef LIBXML_HTML_ENABLED
551
0
            return(1);
552
#else
553
            return(0);
554
#endif
555
0
        case XML_WITH_LEGACY:
556
0
            return(0);
557
0
        case XML_WITH_C14N:
558
0
#ifdef LIBXML_C14N_ENABLED
559
0
            return(1);
560
#else
561
            return(0);
562
#endif
563
0
        case XML_WITH_CATALOG:
564
0
#ifdef LIBXML_CATALOG_ENABLED
565
0
            return(1);
566
#else
567
            return(0);
568
#endif
569
0
        case XML_WITH_XPATH:
570
0
#ifdef LIBXML_XPATH_ENABLED
571
0
            return(1);
572
#else
573
            return(0);
574
#endif
575
0
        case XML_WITH_XPTR:
576
0
#ifdef LIBXML_XPTR_ENABLED
577
0
            return(1);
578
#else
579
            return(0);
580
#endif
581
0
        case XML_WITH_XINCLUDE:
582
0
#ifdef LIBXML_XINCLUDE_ENABLED
583
0
            return(1);
584
#else
585
            return(0);
586
#endif
587
0
        case XML_WITH_ICONV:
588
0
#ifdef LIBXML_ICONV_ENABLED
589
0
            return(1);
590
#else
591
            return(0);
592
#endif
593
0
        case XML_WITH_ISO8859X:
594
0
#ifdef LIBXML_ISO8859X_ENABLED
595
0
            return(1);
596
#else
597
            return(0);
598
#endif
599
0
        case XML_WITH_UNICODE:
600
0
            return(0);
601
0
        case XML_WITH_REGEXP:
602
0
#ifdef LIBXML_REGEXP_ENABLED
603
0
            return(1);
604
#else
605
            return(0);
606
#endif
607
0
        case XML_WITH_AUTOMATA:
608
0
#ifdef LIBXML_REGEXP_ENABLED
609
0
            return(1);
610
#else
611
            return(0);
612
#endif
613
0
        case XML_WITH_EXPR:
614
0
            return(0);
615
0
        case XML_WITH_RELAXNG:
616
0
#ifdef LIBXML_RELAXNG_ENABLED
617
0
            return(1);
618
#else
619
            return(0);
620
#endif
621
0
        case XML_WITH_SCHEMAS:
622
0
#ifdef LIBXML_SCHEMAS_ENABLED
623
0
            return(1);
624
#else
625
            return(0);
626
#endif
627
0
        case XML_WITH_SCHEMATRON:
628
#ifdef LIBXML_SCHEMATRON_ENABLED
629
            return(1);
630
#else
631
0
            return(0);
632
0
#endif
633
0
        case XML_WITH_MODULES:
634
0
#ifdef LIBXML_MODULES_ENABLED
635
0
            return(1);
636
#else
637
            return(0);
638
#endif
639
0
        case XML_WITH_DEBUG:
640
#ifdef LIBXML_DEBUG_ENABLED
641
            return(1);
642
#else
643
0
            return(0);
644
0
#endif
645
0
        case XML_WITH_DEBUG_MEM:
646
0
            return(0);
647
0
        case XML_WITH_ZLIB:
648
0
#ifdef LIBXML_ZLIB_ENABLED
649
0
            return(1);
650
#else
651
            return(0);
652
#endif
653
0
        case XML_WITH_LZMA:
654
0
            return(0);
655
0
        case XML_WITH_ICU:
656
#ifdef LIBXML_ICU_ENABLED
657
            return(1);
658
#else
659
0
            return(0);
660
0
#endif
661
0
        default:
662
0
      break;
663
0
     }
664
0
     return(0);
665
0
}
666
667
/************************************************************************
668
 *                  *
669
 *      Simple string buffer        *
670
 *                  *
671
 ************************************************************************/
672
673
typedef struct {
674
    xmlChar *mem;
675
    unsigned size;
676
    unsigned cap; /* size < cap */
677
    unsigned max; /* size <= max */
678
    xmlParserErrors code;
679
} xmlSBuf;
680
681
static void
682
407k
xmlSBufInit(xmlSBuf *buf, unsigned max) {
683
407k
    buf->mem = NULL;
684
407k
    buf->size = 0;
685
407k
    buf->cap = 0;
686
407k
    buf->max = max;
687
407k
    buf->code = XML_ERR_OK;
688
407k
}
689
690
static int
691
178k
xmlSBufGrow(xmlSBuf *buf, unsigned len) {
692
178k
    xmlChar *mem;
693
178k
    unsigned cap;
694
695
178k
    if (len >= UINT_MAX / 2 - buf->size) {
696
0
        if (buf->code == XML_ERR_OK)
697
0
            buf->code = XML_ERR_RESOURCE_LIMIT;
698
0
        return(-1);
699
0
    }
700
701
178k
    cap = (buf->size + len) * 2;
702
178k
    if (cap < 240)
703
144k
        cap = 240;
704
705
178k
    mem = xmlRealloc(buf->mem, cap);
706
178k
    if (mem == NULL) {
707
76
        buf->code = XML_ERR_NO_MEMORY;
708
76
        return(-1);
709
76
    }
710
711
177k
    buf->mem = mem;
712
177k
    buf->cap = cap;
713
714
177k
    return(0);
715
178k
}
716
717
static void
718
114M
xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) {
719
114M
    if (buf->max - buf->size < len) {
720
0
        if (buf->code == XML_ERR_OK)
721
0
            buf->code = XML_ERR_RESOURCE_LIMIT;
722
0
        return;
723
0
    }
724
725
114M
    if (buf->cap - buf->size <= len) {
726
171k
        if (xmlSBufGrow(buf, len) < 0)
727
71
            return;
728
171k
    }
729
730
114M
    if (len > 0)
731
114M
        memcpy(buf->mem + buf->size, str, len);
732
114M
    buf->size += len;
733
114M
}
734
735
static void
736
111M
xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) {
737
111M
    xmlSBufAddString(buf, (const xmlChar *) str, len);
738
111M
}
739
740
static void
741
160k
xmlSBufAddChar(xmlSBuf *buf, int c) {
742
160k
    xmlChar *end;
743
744
160k
    if (buf->max - buf->size < 4) {
745
0
        if (buf->code == XML_ERR_OK)
746
0
            buf->code = XML_ERR_RESOURCE_LIMIT;
747
0
        return;
748
0
    }
749
750
160k
    if (buf->cap - buf->size <= 4) {
751
6.39k
        if (xmlSBufGrow(buf, 4) < 0)
752
5
            return;
753
6.39k
    }
754
755
160k
    end = buf->mem + buf->size;
756
757
160k
    if (c < 0x80) {
758
132k
        *end = (xmlChar) c;
759
132k
        buf->size += 1;
760
132k
    } else {
761
28.2k
        buf->size += xmlCopyCharMultiByte(end, c);
762
28.2k
    }
763
160k
}
764
765
static void
766
103M
xmlSBufAddReplChar(xmlSBuf *buf) {
767
103M
    xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3);
768
103M
}
769
770
static void
771
83
xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
772
83
    if (buf->code == XML_ERR_NO_MEMORY)
773
83
        xmlCtxtErrMemory(ctxt);
774
0
    else
775
0
        xmlFatalErr(ctxt, buf->code, errMsg);
776
83
}
777
778
static xmlChar *
779
xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt,
780
148k
              const char *errMsg) {
781
148k
    if (buf->mem == NULL) {
782
12.8k
        buf->mem = xmlMalloc(1);
783
12.8k
        if (buf->mem == NULL) {
784
7
            buf->code = XML_ERR_NO_MEMORY;
785
12.8k
        } else {
786
12.8k
            buf->mem[0] = 0;
787
12.8k
        }
788
135k
    } else {
789
135k
        buf->mem[buf->size] = 0;
790
135k
    }
791
792
148k
    if (buf->code == XML_ERR_OK) {
793
148k
        if (sizeOut != NULL)
794
53.2k
            *sizeOut = buf->size;
795
148k
        return(buf->mem);
796
148k
    }
797
798
50
    xmlSBufReportError(buf, ctxt, errMsg);
799
800
50
    xmlFree(buf->mem);
801
802
50
    if (sizeOut != NULL)
803
16
        *sizeOut = 0;
804
50
    return(NULL);
805
148k
}
806
807
static void
808
255k
xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
809
255k
    if (buf->code != XML_ERR_OK)
810
33
        xmlSBufReportError(buf, ctxt, errMsg);
811
812
255k
    xmlFree(buf->mem);
813
255k
}
814
815
static int
816
xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str,
817
109M
                    const char *errMsg) {
818
109M
    int c = str[0];
819
109M
    int c1 = str[1];
820
821
109M
    if ((c1 & 0xC0) != 0x80)
822
69.5M
        goto encoding_error;
823
824
39.6M
    if (c < 0xE0) {
825
        /* 2-byte sequence */
826
20.2M
        if (c < 0xC2)
827
11.9M
            goto encoding_error;
828
829
8.25M
        return(2);
830
20.2M
    } else {
831
19.4M
        int c2 = str[2];
832
833
19.4M
        if ((c2 & 0xC0) != 0x80)
834
58.1k
            goto encoding_error;
835
836
19.3M
        if (c < 0xF0) {
837
            /* 3-byte sequence */
838
19.1M
            if (c == 0xE0) {
839
                /* overlong */
840
1.96M
                if (c1 < 0xA0)
841
468
                    goto encoding_error;
842
17.2M
            } else if (c == 0xED) {
843
                /* surrogate */
844
290k
                if (c1 >= 0xA0)
845
982
                    goto encoding_error;
846
16.9M
            } else if (c == 0xEF) {
847
                /* U+FFFE and U+FFFF are invalid Chars */
848
12.7M
                if ((c1 == 0xBF) && (c2 >= 0xBE))
849
377
                    xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg);
850
12.7M
            }
851
852
19.1M
            return(3);
853
19.1M
        } else {
854
            /* 4-byte sequence */
855
195k
            if ((str[3] & 0xC0) != 0x80)
856
5.42k
                goto encoding_error;
857
190k
            if (c == 0xF0) {
858
                /* overlong */
859
77.1k
                if (c1 < 0x90)
860
728
                    goto encoding_error;
861
112k
            } else if (c >= 0xF4) {
862
                /* greater than 0x10FFFF */
863
100k
                if ((c > 0xF4) || (c1 >= 0x90))
864
100k
                    goto encoding_error;
865
100k
            }
866
867
88.5k
            return(4);
868
190k
        }
869
19.3M
    }
870
871
81.7M
encoding_error:
872
    /* Only report the first error */
873
81.7M
    if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) {
874
17.5k
        xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL);
875
17.5k
        ctxt->input->flags |= XML_INPUT_ENCODING_ERROR;
876
17.5k
    }
877
878
81.7M
    return(0);
879
39.6M
}
880
881
/************************************************************************
882
 *                  *
883
 *    SAX2 defaulted attributes handling      *
884
 *                  *
885
 ************************************************************************/
886
887
/**
888
 * Final initialization of the parser context before starting to parse.
889
 *
890
 * This accounts for users modifying struct members of parser context
891
 * directly.
892
 *
893
 * @param ctxt  an XML parser context
894
 */
895
static void
896
35.2k
xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) {
897
35.2k
    xmlSAXHandlerPtr sax;
898
899
    /* Avoid unused variable warning if features are disabled. */
900
35.2k
    (void) sax;
901
902
    /*
903
     * Changing the SAX struct directly is still widespread practice
904
     * in internal and external code.
905
     */
906
35.2k
    if (ctxt == NULL) return;
907
35.2k
    sax = ctxt->sax;
908
35.2k
#ifdef LIBXML_SAX1_ENABLED
909
    /*
910
     * Only enable SAX2 if there SAX2 element handlers, except when there
911
     * are no element handlers at all.
912
     */
913
35.2k
    if (((ctxt->options & XML_PARSE_SAX1) == 0) &&
914
35.2k
        (sax) &&
915
35.2k
        (sax->initialized == XML_SAX2_MAGIC) &&
916
35.2k
        ((sax->startElementNs != NULL) ||
917
0
         (sax->endElementNs != NULL) ||
918
0
         ((sax->startElement == NULL) && (sax->endElement == NULL))))
919
35.2k
        ctxt->sax2 = 1;
920
#else
921
    ctxt->sax2 = 1;
922
#endif /* LIBXML_SAX1_ENABLED */
923
924
    /*
925
     * Some users replace the dictionary directly in the context struct.
926
     * We really need an API function to do that cleanly.
927
     */
928
35.2k
    ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3);
929
35.2k
    ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5);
930
35.2k
    ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36);
931
35.2k
    if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) ||
932
35.1k
    (ctxt->str_xml_ns == NULL)) {
933
61
        xmlErrMemory(ctxt);
934
61
    }
935
936
35.2k
    xmlDictSetLimit(ctxt->dict,
937
35.2k
                    (ctxt->options & XML_PARSE_HUGE) ?
938
12.9k
                        0 :
939
35.2k
                        XML_MAX_DICTIONARY_LIMIT);
940
941
35.2k
#ifdef LIBXML_VALID_ENABLED
942
35.2k
    if (ctxt->validate)
943
0
        ctxt->vctxt.flags |= XML_VCTXT_VALIDATE;
944
35.2k
    else
945
35.2k
        ctxt->vctxt.flags &= ~XML_VCTXT_VALIDATE;
946
35.2k
#endif /* LIBXML_VALID_ENABLED */
947
35.2k
}
948
949
typedef struct {
950
    xmlHashedString prefix;
951
    xmlHashedString name;
952
    xmlHashedString value;
953
    const xmlChar *valueEnd;
954
    int external;
955
    int expandedSize;
956
} xmlDefAttr;
957
958
typedef struct _xmlDefAttrs xmlDefAttrs;
959
typedef xmlDefAttrs *xmlDefAttrsPtr;
960
struct _xmlDefAttrs {
961
    int nbAttrs;  /* number of defaulted attributes on that element */
962
    int maxAttrs;       /* the size of the array */
963
#if __STDC_VERSION__ >= 199901L
964
    /* Using a C99 flexible array member avoids UBSan errors. */
965
    xmlDefAttr attrs[] ATTRIBUTE_COUNTED_BY(maxAttrs);
966
#else
967
    xmlDefAttr attrs[1];
968
#endif
969
};
970
971
/**
972
 * Normalize the space in non CDATA attribute values:
973
 * If the attribute type is not CDATA, then the XML processor MUST further
974
 * process the normalized attribute value by discarding any leading and
975
 * trailing space (\#x20) characters, and by replacing sequences of space
976
 * (\#x20) characters by a single space (\#x20) character.
977
 * Note that the size of dst need to be at least src, and if one doesn't need
978
 * to preserve dst (and it doesn't come from a dictionary or read-only) then
979
 * passing src as dst is just fine.
980
 *
981
 * @param src  the source string
982
 * @param dst  the target string
983
 * @returns a pointer to the normalized value (dst) or NULL if no conversion
984
 *         is needed.
985
 */
986
static xmlChar *
987
xmlAttrNormalizeSpace(const xmlChar *src, xmlChar *dst)
988
24.0k
{
989
24.0k
    if ((src == NULL) || (dst == NULL))
990
0
        return(NULL);
991
992
24.5k
    while (*src == 0x20) src++;
993
5.45M
    while (*src != 0) {
994
5.43M
  if (*src == 0x20) {
995
1.46M
      while (*src == 0x20) src++;
996
21.2k
      if (*src != 0)
997
19.1k
    *dst++ = 0x20;
998
5.40M
  } else {
999
5.40M
      *dst++ = *src++;
1000
5.40M
  }
1001
5.43M
    }
1002
24.0k
    *dst = 0;
1003
24.0k
    if (dst == src)
1004
21.4k
       return(NULL);
1005
2.58k
    return(dst);
1006
24.0k
}
1007
1008
/**
1009
 * Add a defaulted attribute for an element
1010
 *
1011
 * @param ctxt  an XML parser context
1012
 * @param fullname  the element fullname
1013
 * @param fullattr  the attribute fullname
1014
 * @param value  the attribute value
1015
 */
1016
static void
1017
xmlAddDefAttrs(xmlParserCtxtPtr ctxt,
1018
               const xmlChar *fullname,
1019
               const xmlChar *fullattr,
1020
24.0k
               const xmlChar *value) {
1021
24.0k
    xmlDefAttrsPtr defaults;
1022
24.0k
    xmlDefAttr *attr;
1023
24.0k
    int len, expandedSize;
1024
24.0k
    xmlHashedString name;
1025
24.0k
    xmlHashedString prefix;
1026
24.0k
    xmlHashedString hvalue;
1027
24.0k
    const xmlChar *localname;
1028
1029
    /*
1030
     * Allows to detect attribute redefinitions
1031
     */
1032
24.0k
    if (ctxt->attsSpecial != NULL) {
1033
22.1k
        if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL)
1034
12.6k
      return;
1035
22.1k
    }
1036
1037
11.4k
    if (ctxt->attsDefault == NULL) {
1038
2.06k
        ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict);
1039
2.06k
  if (ctxt->attsDefault == NULL)
1040
1
      goto mem_error;
1041
2.06k
    }
1042
1043
    /*
1044
     * split the element name into prefix:localname , the string found
1045
     * are within the DTD and then not associated to namespace names.
1046
     */
1047
11.4k
    localname = xmlSplitQName3(fullname, &len);
1048
11.4k
    if (localname == NULL) {
1049
10.8k
        name = xmlDictLookupHashed(ctxt->dict, fullname, -1);
1050
10.8k
  prefix.name = NULL;
1051
10.8k
    } else {
1052
588
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1053
588
  prefix = xmlDictLookupHashed(ctxt->dict, fullname, len);
1054
588
        if (prefix.name == NULL)
1055
1
            goto mem_error;
1056
588
    }
1057
11.4k
    if (name.name == NULL)
1058
1
        goto mem_error;
1059
1060
    /*
1061
     * make sure there is some storage
1062
     */
1063
11.4k
    defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name);
1064
11.4k
    if ((defaults == NULL) ||
1065
8.97k
        (defaults->nbAttrs >= defaults->maxAttrs)) {
1066
3.11k
        xmlDefAttrsPtr temp;
1067
3.11k
        int newSize;
1068
1069
3.11k
        if (defaults == NULL) {
1070
2.42k
            newSize = 4;
1071
2.42k
        } else {
1072
693
            if ((defaults->maxAttrs >= XML_MAX_ATTRS) ||
1073
693
                ((size_t) defaults->maxAttrs >
1074
693
                     SIZE_MAX / 2 / sizeof(temp[0]) - sizeof(*defaults)))
1075
0
                goto mem_error;
1076
1077
693
            if (defaults->maxAttrs > XML_MAX_ATTRS / 2)
1078
0
                newSize = XML_MAX_ATTRS;
1079
693
            else
1080
693
                newSize = defaults->maxAttrs * 2;
1081
693
        }
1082
3.11k
        temp = xmlRealloc(defaults,
1083
3.11k
                          sizeof(*defaults) + newSize * sizeof(xmlDefAttr));
1084
3.11k
  if (temp == NULL)
1085
2
      goto mem_error;
1086
3.11k
        if (defaults == NULL)
1087
2.42k
            temp->nbAttrs = 0;
1088
3.11k
  temp->maxAttrs = newSize;
1089
3.11k
        defaults = temp;
1090
3.11k
  if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name,
1091
3.11k
                          defaults, NULL) < 0) {
1092
0
      xmlFree(defaults);
1093
0
      goto mem_error;
1094
0
  }
1095
3.11k
    }
1096
1097
    /*
1098
     * Split the attribute name into prefix:localname , the string found
1099
     * are within the DTD and hen not associated to namespace names.
1100
     */
1101
11.3k
    localname = xmlSplitQName3(fullattr, &len);
1102
11.3k
    if (localname == NULL) {
1103
10.1k
        name = xmlDictLookupHashed(ctxt->dict, fullattr, -1);
1104
10.1k
  prefix.name = NULL;
1105
10.1k
    } else {
1106
1.26k
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1107
1.26k
  prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len);
1108
1.26k
        if (prefix.name == NULL)
1109
1
            goto mem_error;
1110
1.26k
    }
1111
11.3k
    if (name.name == NULL)
1112
1
        goto mem_error;
1113
1114
    /* intern the string and precompute the end */
1115
11.3k
    len = strlen((const char *) value);
1116
11.3k
    hvalue = xmlDictLookupHashed(ctxt->dict, value, len);
1117
11.3k
    if (hvalue.name == NULL)
1118
2
        goto mem_error;
1119
1120
11.3k
    expandedSize = strlen((const char *) name.name);
1121
11.3k
    if (prefix.name != NULL)
1122
1.26k
        expandedSize += strlen((const char *) prefix.name);
1123
11.3k
    expandedSize += len;
1124
1125
11.3k
    attr = &defaults->attrs[defaults->nbAttrs++];
1126
11.3k
    attr->name = name;
1127
11.3k
    attr->prefix = prefix;
1128
11.3k
    attr->value = hvalue;
1129
11.3k
    attr->valueEnd = hvalue.name + len;
1130
11.3k
    attr->external = PARSER_EXTERNAL(ctxt);
1131
11.3k
    attr->expandedSize = expandedSize;
1132
1133
11.3k
    return;
1134
1135
9
mem_error:
1136
9
    xmlErrMemory(ctxt);
1137
9
}
1138
1139
/**
1140
 * Register this attribute type
1141
 *
1142
 * @param ctxt  an XML parser context
1143
 * @param fullname  the element fullname
1144
 * @param fullattr  the attribute fullname
1145
 * @param type  the attribute type
1146
 */
1147
static void
1148
xmlAddSpecialAttr(xmlParserCtxtPtr ctxt,
1149
      const xmlChar *fullname,
1150
      const xmlChar *fullattr,
1151
      int type)
1152
47.6k
{
1153
47.6k
    if (ctxt->attsSpecial == NULL) {
1154
2.88k
        ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict);
1155
2.88k
  if (ctxt->attsSpecial == NULL)
1156
3
      goto mem_error;
1157
2.88k
    }
1158
1159
47.6k
    if (PARSER_EXTERNAL(ctxt))
1160
33.8k
        type |= XML_SPECIAL_EXTERNAL;
1161
1162
47.6k
    if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr,
1163
47.6k
                    XML_INT_TO_PTR(type)) < 0)
1164
1
        goto mem_error;
1165
47.6k
    return;
1166
1167
47.6k
mem_error:
1168
4
    xmlErrMemory(ctxt);
1169
4
}
1170
1171
/**
1172
 * Removes CDATA attributes from the special attribute table
1173
 */
1174
static void
1175
xmlCleanSpecialAttrCallback(void *payload, void *data,
1176
                            const xmlChar *fullname, const xmlChar *fullattr,
1177
22.3k
                            const xmlChar *unused ATTRIBUTE_UNUSED) {
1178
22.3k
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data;
1179
1180
22.3k
    if (XML_PTR_TO_INT(payload) == XML_ATTRIBUTE_CDATA) {
1181
2.15k
        xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL);
1182
2.15k
    }
1183
22.3k
}
1184
1185
/**
1186
 * Trim the list of attributes defined to remove all those of type
1187
 * CDATA as they are not special. This call should be done when finishing
1188
 * to parse the DTD and before starting to parse the document root.
1189
 *
1190
 * @param ctxt  an XML parser context
1191
 */
1192
static void
1193
xmlCleanSpecialAttr(xmlParserCtxtPtr ctxt)
1194
14.3k
{
1195
14.3k
    if (ctxt->attsSpecial == NULL)
1196
11.4k
        return;
1197
1198
2.88k
    xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt);
1199
1200
2.88k
    if (xmlHashSize(ctxt->attsSpecial) == 0) {
1201
350
        xmlHashFree(ctxt->attsSpecial, NULL);
1202
350
        ctxt->attsSpecial = NULL;
1203
350
    }
1204
2.88k
}
1205
1206
/**
1207
 * Checks that the value conforms to the LanguageID production:
1208
 *
1209
 * @deprecated Internal function, do not use.
1210
 *
1211
 * NOTE: this is somewhat deprecated, those productions were removed from
1212
 * the XML Second edition.
1213
 *
1214
 *     [33] LanguageID ::= Langcode ('-' Subcode)*
1215
 *     [34] Langcode ::= ISO639Code |  IanaCode |  UserCode
1216
 *     [35] ISO639Code ::= ([a-z] | [A-Z]) ([a-z] | [A-Z])
1217
 *     [36] IanaCode ::= ('i' | 'I') '-' ([a-z] | [A-Z])+
1218
 *     [37] UserCode ::= ('x' | 'X') '-' ([a-z] | [A-Z])+
1219
 *     [38] Subcode ::= ([a-z] | [A-Z])+
1220
 *
1221
 * The current REC reference the successors of RFC 1766, currently 5646
1222
 *
1223
 * http://www.rfc-editor.org/rfc/rfc5646.txt
1224
 *
1225
 *     langtag       = language
1226
 *                     ["-" script]
1227
 *                     ["-" region]
1228
 *                     *("-" variant)
1229
 *                     *("-" extension)
1230
 *                     ["-" privateuse]
1231
 *     language      = 2*3ALPHA            ; shortest ISO 639 code
1232
 *                     ["-" extlang]       ; sometimes followed by
1233
 *                                         ; extended language subtags
1234
 *                   / 4ALPHA              ; or reserved for future use
1235
 *                   / 5*8ALPHA            ; or registered language subtag
1236
 *
1237
 *     extlang       = 3ALPHA              ; selected ISO 639 codes
1238
 *                     *2("-" 3ALPHA)      ; permanently reserved
1239
 *
1240
 *     script        = 4ALPHA              ; ISO 15924 code
1241
 *
1242
 *     region        = 2ALPHA              ; ISO 3166-1 code
1243
 *                   / 3DIGIT              ; UN M.49 code
1244
 *
1245
 *     variant       = 5*8alphanum         ; registered variants
1246
 *                   / (DIGIT 3alphanum)
1247
 *
1248
 *     extension     = singleton 1*("-" (2*8alphanum))
1249
 *
1250
 *                                         ; Single alphanumerics
1251
 *                                         ; "x" reserved for private use
1252
 *     singleton     = DIGIT               ; 0 - 9
1253
 *                   / %x41-57             ; A - W
1254
 *                   / %x59-5A             ; Y - Z
1255
 *                   / %x61-77             ; a - w
1256
 *                   / %x79-7A             ; y - z
1257
 *
1258
 * it sounds right to still allow Irregular i-xxx IANA and user codes too
1259
 * The parser below doesn't try to cope with extension or privateuse
1260
 * that could be added but that's not interoperable anyway
1261
 *
1262
 * @param lang  pointer to the string value
1263
 * @returns 1 if correct 0 otherwise
1264
 **/
1265
int
1266
xmlCheckLanguageID(const xmlChar * lang)
1267
24.3k
{
1268
24.3k
    const xmlChar *cur = lang, *nxt;
1269
1270
24.3k
    if (cur == NULL)
1271
0
        return (0);
1272
24.3k
    if (((cur[0] == 'i') && (cur[1] == '-')) ||
1273
23.4k
        ((cur[0] == 'I') && (cur[1] == '-')) ||
1274
23.2k
        ((cur[0] == 'x') && (cur[1] == '-')) ||
1275
23.0k
        ((cur[0] == 'X') && (cur[1] == '-'))) {
1276
        /*
1277
         * Still allow IANA code and user code which were coming
1278
         * from the previous version of the XML-1.0 specification
1279
         * it's deprecated but we should not fail
1280
         */
1281
1.51k
        cur += 2;
1282
2.99k
        while (((cur[0] >= 'A') && (cur[0] <= 'Z')) ||
1283
2.27k
               ((cur[0] >= 'a') && (cur[0] <= 'z')))
1284
1.48k
            cur++;
1285
1.51k
        return(cur[0] == 0);
1286
1.51k
    }
1287
22.8k
    nxt = cur;
1288
608k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1289
569k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1290
585k
           nxt++;
1291
22.8k
    if (nxt - cur >= 4) {
1292
        /*
1293
         * Reserved
1294
         */
1295
1.04k
        if ((nxt - cur > 8) || (nxt[0] != 0))
1296
626
            return(0);
1297
422
        return(1);
1298
1.04k
    }
1299
21.7k
    if (nxt - cur < 2)
1300
892
        return(0);
1301
    /* we got an ISO 639 code */
1302
20.8k
    if (nxt[0] == 0)
1303
478
        return(1);
1304
20.4k
    if (nxt[0] != '-')
1305
385
        return(0);
1306
1307
20.0k
    nxt++;
1308
20.0k
    cur = nxt;
1309
    /* now we can have extlang or script or region or variant */
1310
20.0k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1311
1.22k
        goto region_m49;
1312
1313
311k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1314
231k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1315
292k
           nxt++;
1316
18.8k
    if (nxt - cur == 4)
1317
4.71k
        goto script;
1318
14.0k
    if (nxt - cur == 2)
1319
1.78k
        goto region;
1320
12.3k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1321
3.30k
        goto variant;
1322
9.00k
    if (nxt - cur != 3)
1323
1.06k
        return(0);
1324
    /* we parsed an extlang */
1325
7.93k
    if (nxt[0] == 0)
1326
200
        return(1);
1327
7.73k
    if (nxt[0] != '-')
1328
215
        return(0);
1329
1330
7.52k
    nxt++;
1331
7.52k
    cur = nxt;
1332
    /* now we can have script or region or variant */
1333
7.52k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1334
4.48k
        goto region_m49;
1335
1336
121k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1337
115k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1338
118k
           nxt++;
1339
3.04k
    if (nxt - cur == 2)
1340
901
        goto region;
1341
2.14k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1342
587
        goto variant;
1343
1.55k
    if (nxt - cur != 4)
1344
1.24k
        return(0);
1345
    /* we parsed a script */
1346
5.02k
script:
1347
5.02k
    if (nxt[0] == 0)
1348
294
        return(1);
1349
4.72k
    if (nxt[0] != '-')
1350
409
        return(0);
1351
1352
4.32k
    nxt++;
1353
4.32k
    cur = nxt;
1354
    /* now we can have region or variant */
1355
4.32k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1356
3.09k
        goto region_m49;
1357
1358
5.61k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1359
3.73k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1360
4.39k
           nxt++;
1361
1362
1.22k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1363
222
        goto variant;
1364
1.00k
    if (nxt - cur != 2)
1365
665
        return(0);
1366
    /* we parsed a region */
1367
9.28k
region:
1368
9.28k
    if (nxt[0] == 0)
1369
870
        return(1);
1370
8.41k
    if (nxt[0] != '-')
1371
7.02k
        return(0);
1372
1373
1.38k
    nxt++;
1374
1.38k
    cur = nxt;
1375
    /* now we can just have a variant */
1376
107k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1377
104k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1378
106k
           nxt++;
1379
1380
1.38k
    if ((nxt - cur < 5) || (nxt - cur > 8))
1381
1.18k
        return(0);
1382
1383
    /* we parsed a variant */
1384
4.31k
variant:
1385
4.31k
    if (nxt[0] == 0)
1386
512
        return(1);
1387
3.80k
    if (nxt[0] != '-')
1388
761
        return(0);
1389
    /* extensions and private use subtags not checked */
1390
3.04k
    return (1);
1391
1392
8.79k
region_m49:
1393
8.79k
    if (((nxt[1] >= '0') && (nxt[1] <= '9')) &&
1394
7.96k
        ((nxt[2] >= '0') && (nxt[2] <= '9'))) {
1395
6.25k
        nxt += 3;
1396
6.25k
        goto region;
1397
6.25k
    }
1398
2.54k
    return(0);
1399
8.79k
}
1400
1401
/************************************************************************
1402
 *                  *
1403
 *    Parser stacks related functions and macros    *
1404
 *                  *
1405
 ************************************************************************/
1406
1407
static xmlChar *
1408
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar **str);
1409
1410
/**
1411
 * Create a new namespace database.
1412
 *
1413
 * @returns the new obejct.
1414
 */
1415
xmlParserNsData *
1416
36.2k
xmlParserNsCreate(void) {
1417
36.2k
    xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb));
1418
1419
36.2k
    if (nsdb == NULL)
1420
6
        return(NULL);
1421
36.2k
    memset(nsdb, 0, sizeof(*nsdb));
1422
36.2k
    nsdb->defaultNsIndex = INT_MAX;
1423
1424
36.2k
    return(nsdb);
1425
36.2k
}
1426
1427
/**
1428
 * Free a namespace database.
1429
 *
1430
 * @param nsdb  namespace database
1431
 */
1432
void
1433
36.2k
xmlParserNsFree(xmlParserNsData *nsdb) {
1434
36.2k
    if (nsdb == NULL)
1435
0
        return;
1436
1437
36.2k
    xmlFree(nsdb->extra);
1438
36.2k
    xmlFree(nsdb->hash);
1439
36.2k
    xmlFree(nsdb);
1440
36.2k
}
1441
1442
/**
1443
 * Reset a namespace database.
1444
 *
1445
 * @param nsdb  namespace database
1446
 */
1447
static void
1448
31.4k
xmlParserNsReset(xmlParserNsData *nsdb) {
1449
31.4k
    if (nsdb == NULL)
1450
0
        return;
1451
1452
31.4k
    nsdb->hashElems = 0;
1453
31.4k
    nsdb->elementId = 0;
1454
31.4k
    nsdb->defaultNsIndex = INT_MAX;
1455
1456
31.4k
    if (nsdb->hash)
1457
0
        memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0]));
1458
31.4k
}
1459
1460
/**
1461
 * Signal that a new element has started.
1462
 *
1463
 * @param nsdb  namespace database
1464
 * @returns 0 on success, -1 if the element counter overflowed.
1465
 */
1466
static int
1467
1.36M
xmlParserNsStartElement(xmlParserNsData *nsdb) {
1468
1.36M
    if (nsdb->elementId == UINT_MAX)
1469
0
        return(-1);
1470
1.36M
    nsdb->elementId++;
1471
1472
1.36M
    return(0);
1473
1.36M
}
1474
1475
/**
1476
 * Lookup namespace with given prefix. If `bucketPtr` is non-NULL, it will
1477
 * be set to the matching bucket, or the first empty bucket if no match
1478
 * was found.
1479
 *
1480
 * @param ctxt  parser context
1481
 * @param prefix  namespace prefix
1482
 * @param bucketPtr  optional bucket (return value)
1483
 * @returns the namespace index on success, INT_MAX if no namespace was
1484
 * found.
1485
 */
1486
static int
1487
xmlParserNsLookup(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix,
1488
2.25M
                  xmlParserNsBucket **bucketPtr) {
1489
2.25M
    xmlParserNsBucket *bucket, *tombstone;
1490
2.25M
    unsigned index, hashValue;
1491
1492
2.25M
    if (prefix->name == NULL)
1493
1.61M
        return(ctxt->nsdb->defaultNsIndex);
1494
1495
635k
    if (ctxt->nsdb->hashSize == 0)
1496
32.5k
        return(INT_MAX);
1497
1498
603k
    hashValue = prefix->hashValue;
1499
603k
    index = hashValue & (ctxt->nsdb->hashSize - 1);
1500
603k
    bucket = &ctxt->nsdb->hash[index];
1501
603k
    tombstone = NULL;
1502
1503
720k
    while (bucket->hashValue) {
1504
605k
        if (bucket->index == INT_MAX) {
1505
61.3k
            if (tombstone == NULL)
1506
56.1k
                tombstone = bucket;
1507
544k
        } else if (bucket->hashValue == hashValue) {
1508
488k
            if (ctxt->nsTab[bucket->index * 2] == prefix->name) {
1509
488k
                if (bucketPtr != NULL)
1510
247k
                    *bucketPtr = bucket;
1511
488k
                return(bucket->index);
1512
488k
            }
1513
488k
        }
1514
1515
116k
        index++;
1516
116k
        bucket++;
1517
116k
        if (index == ctxt->nsdb->hashSize) {
1518
9.98k
            index = 0;
1519
9.98k
            bucket = ctxt->nsdb->hash;
1520
9.98k
        }
1521
116k
    }
1522
1523
114k
    if (bucketPtr != NULL)
1524
51.7k
        *bucketPtr = tombstone ? tombstone : bucket;
1525
114k
    return(INT_MAX);
1526
603k
}
1527
1528
/**
1529
 * Lookup namespace URI with given prefix.
1530
 *
1531
 * @param ctxt  parser context
1532
 * @param prefix  namespace prefix
1533
 * @returns the namespace URI on success, NULL if no namespace was found.
1534
 */
1535
static const xmlChar *
1536
1.19M
xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) {
1537
1.19M
    const xmlChar *ret;
1538
1.19M
    int nsIndex;
1539
1540
1.19M
    if (prefix->name == ctxt->str_xml)
1541
553
        return(ctxt->str_xml_ns);
1542
1543
    /*
1544
     * minNsIndex is used when building an entity tree. We must
1545
     * ignore namespaces declared outside the entity.
1546
     */
1547
1.18M
    nsIndex = xmlParserNsLookup(ctxt, prefix, NULL);
1548
1.18M
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1549
603k
        return(NULL);
1550
1551
585k
    ret = ctxt->nsTab[nsIndex * 2 + 1];
1552
585k
    if (ret[0] == 0)
1553
16.0k
        ret = NULL;
1554
585k
    return(ret);
1555
1.18M
}
1556
1557
/**
1558
 * Lookup extra data for the given prefix. This returns data stored
1559
 * with xmlParserNsUdpateSax.
1560
 *
1561
 * @param ctxt  parser context
1562
 * @param prefix  namespace prefix
1563
 * @returns the data on success, NULL if no namespace was found.
1564
 */
1565
void *
1566
594k
xmlParserNsLookupSax(xmlParserCtxt *ctxt, const xmlChar *prefix) {
1567
594k
    xmlHashedString hprefix;
1568
594k
    int nsIndex;
1569
1570
594k
    if (prefix == ctxt->str_xml)
1571
36.3k
        return(NULL);
1572
1573
557k
    hprefix.name = prefix;
1574
557k
    if (prefix != NULL)
1575
64.2k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1576
493k
    else
1577
493k
        hprefix.hashValue = 0;
1578
557k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1579
557k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1580
0
        return(NULL);
1581
1582
557k
    return(ctxt->nsdb->extra[nsIndex].saxData);
1583
557k
}
1584
1585
/**
1586
 * Sets or updates extra data for the given prefix. This value will be
1587
 * returned by xmlParserNsLookupSax as long as the namespace with the
1588
 * given prefix is in scope.
1589
 *
1590
 * @param ctxt  parser context
1591
 * @param prefix  namespace prefix
1592
 * @param saxData  extra data for SAX handler
1593
 * @returns the data on success, NULL if no namespace was found.
1594
 */
1595
int
1596
xmlParserNsUpdateSax(xmlParserCtxt *ctxt, const xmlChar *prefix,
1597
109k
                     void *saxData) {
1598
109k
    xmlHashedString hprefix;
1599
109k
    int nsIndex;
1600
1601
109k
    if (prefix == ctxt->str_xml)
1602
0
        return(-1);
1603
1604
109k
    hprefix.name = prefix;
1605
109k
    if (prefix != NULL)
1606
100k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1607
8.86k
    else
1608
8.86k
        hprefix.hashValue = 0;
1609
109k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1610
109k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1611
0
        return(-1);
1612
1613
109k
    ctxt->nsdb->extra[nsIndex].saxData = saxData;
1614
109k
    return(0);
1615
109k
}
1616
1617
/**
1618
 * Grows the namespace tables.
1619
 *
1620
 * @param ctxt  parser context
1621
 * @returns 0 on success, -1 if a memory allocation failed.
1622
 */
1623
static int
1624
22.0k
xmlParserNsGrow(xmlParserCtxtPtr ctxt) {
1625
22.0k
    const xmlChar **table;
1626
22.0k
    xmlParserNsExtra *extra;
1627
22.0k
    int newSize;
1628
1629
22.0k
    newSize = xmlGrowCapacity(ctxt->nsMax,
1630
22.0k
                              sizeof(table[0]) + sizeof(extra[0]),
1631
22.0k
                              16, XML_MAX_ITEMS);
1632
22.0k
    if (newSize < 0)
1633
0
        goto error;
1634
1635
22.0k
    table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0]));
1636
22.0k
    if (table == NULL)
1637
9
        goto error;
1638
22.0k
    ctxt->nsTab = table;
1639
1640
22.0k
    extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0]));
1641
22.0k
    if (extra == NULL)
1642
14
        goto error;
1643
22.0k
    ctxt->nsdb->extra = extra;
1644
1645
22.0k
    ctxt->nsMax = newSize;
1646
22.0k
    return(0);
1647
1648
23
error:
1649
23
    xmlErrMemory(ctxt);
1650
23
    return(-1);
1651
22.0k
}
1652
1653
/**
1654
 * Push a new namespace on the table.
1655
 *
1656
 * @param ctxt  parser context
1657
 * @param prefix  prefix with hash value
1658
 * @param uri  uri with hash value
1659
 * @param saxData  extra data for SAX handler
1660
 * @param defAttr  whether the namespace comes from a default attribute
1661
 * @returns 1 if the namespace was pushed, 0 if the namespace was ignored,
1662
 * -1 if a memory allocation failed.
1663
 */
1664
static int
1665
xmlParserNsPush(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix,
1666
184k
                const xmlHashedString *uri, void *saxData, int defAttr) {
1667
184k
    xmlParserNsBucket *bucket = NULL;
1668
184k
    xmlParserNsExtra *extra;
1669
184k
    const xmlChar **ns;
1670
184k
    unsigned hashValue, nsIndex, oldIndex;
1671
1672
184k
    if ((prefix != NULL) && (prefix->name == ctxt->str_xml))
1673
210
        return(0);
1674
1675
184k
    if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) {
1676
23
        xmlErrMemory(ctxt);
1677
23
        return(-1);
1678
23
    }
1679
1680
    /*
1681
     * Default namespace and 'xml' namespace
1682
     */
1683
184k
    if ((prefix == NULL) || (prefix->name == NULL)) {
1684
18.6k
        oldIndex = ctxt->nsdb->defaultNsIndex;
1685
1686
18.6k
        if (oldIndex != INT_MAX) {
1687
11.9k
            extra = &ctxt->nsdb->extra[oldIndex];
1688
1689
11.9k
            if (extra->elementId == ctxt->nsdb->elementId) {
1690
1.58k
                if (defAttr == 0)
1691
1.25k
                    xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns");
1692
1.58k
                return(0);
1693
1.58k
            }
1694
1695
10.3k
            if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1696
5.25k
                (uri->name == ctxt->nsTab[oldIndex * 2 + 1]))
1697
3.84k
                return(0);
1698
10.3k
        }
1699
1700
13.2k
        ctxt->nsdb->defaultNsIndex = ctxt->nsNr;
1701
13.2k
        goto populate_entry;
1702
18.6k
    }
1703
1704
    /*
1705
     * Hash table lookup
1706
     */
1707
165k
    oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket);
1708
165k
    if (oldIndex != INT_MAX) {
1709
100k
        extra = &ctxt->nsdb->extra[oldIndex];
1710
1711
        /*
1712
         * Check for duplicate definitions on the same element.
1713
         */
1714
100k
        if (extra->elementId == ctxt->nsdb->elementId) {
1715
2.35k
            if (defAttr == 0)
1716
2.12k
                xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name);
1717
2.35k
            return(0);
1718
2.35k
        }
1719
1720
98.2k
        if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1721
8.20k
            (uri->name == ctxt->nsTab[bucket->index * 2 + 1]))
1722
6.64k
            return(0);
1723
1724
91.5k
        bucket->index = ctxt->nsNr;
1725
91.5k
        goto populate_entry;
1726
98.2k
    }
1727
1728
    /*
1729
     * Insert new bucket
1730
     */
1731
1732
65.1k
    hashValue = prefix->hashValue;
1733
1734
    /*
1735
     * Grow hash table, 50% fill factor
1736
     */
1737
65.1k
    if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) {
1738
14.0k
        xmlParserNsBucket *newHash;
1739
14.0k
        unsigned newSize, i, index;
1740
1741
14.0k
        if (ctxt->nsdb->hashSize > UINT_MAX / 2) {
1742
0
            xmlErrMemory(ctxt);
1743
0
            return(-1);
1744
0
        }
1745
14.0k
        newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16;
1746
14.0k
        newHash = xmlMalloc(newSize * sizeof(newHash[0]));
1747
14.0k
        if (newHash == NULL) {
1748
1
            xmlErrMemory(ctxt);
1749
1
            return(-1);
1750
1
        }
1751
14.0k
        memset(newHash, 0, newSize * sizeof(newHash[0]));
1752
1753
190k
        for (i = 0; i < ctxt->nsdb->hashSize; i++) {
1754
176k
            unsigned hv = ctxt->nsdb->hash[i].hashValue;
1755
176k
            unsigned newIndex;
1756
1757
176k
            if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX))
1758
175k
                continue;
1759
1.17k
            newIndex = hv & (newSize - 1);
1760
1761
1.60k
            while (newHash[newIndex].hashValue != 0) {
1762
433
                newIndex++;
1763
433
                if (newIndex == newSize)
1764
109
                    newIndex = 0;
1765
433
            }
1766
1767
1.17k
            newHash[newIndex] = ctxt->nsdb->hash[i];
1768
1.17k
        }
1769
1770
14.0k
        xmlFree(ctxt->nsdb->hash);
1771
14.0k
        ctxt->nsdb->hash = newHash;
1772
14.0k
        ctxt->nsdb->hashSize = newSize;
1773
1774
        /*
1775
         * Relookup
1776
         */
1777
14.0k
        index = hashValue & (newSize - 1);
1778
1779
14.2k
        while (newHash[index].hashValue != 0) {
1780
216
            index++;
1781
216
            if (index == newSize)
1782
36
                index = 0;
1783
216
        }
1784
1785
14.0k
        bucket = &newHash[index];
1786
14.0k
    }
1787
1788
65.1k
    bucket->hashValue = hashValue;
1789
65.1k
    bucket->index = ctxt->nsNr;
1790
65.1k
    ctxt->nsdb->hashElems++;
1791
65.1k
    oldIndex = INT_MAX;
1792
1793
169k
populate_entry:
1794
169k
    nsIndex = ctxt->nsNr;
1795
1796
169k
    ns = &ctxt->nsTab[nsIndex * 2];
1797
169k
    ns[0] = prefix ? prefix->name : NULL;
1798
169k
    ns[1] = uri->name;
1799
1800
169k
    extra = &ctxt->nsdb->extra[nsIndex];
1801
169k
    extra->saxData = saxData;
1802
169k
    extra->prefixHashValue = prefix ? prefix->hashValue : 0;
1803
169k
    extra->uriHashValue = uri->hashValue;
1804
169k
    extra->elementId = ctxt->nsdb->elementId;
1805
169k
    extra->oldIndex = oldIndex;
1806
1807
169k
    ctxt->nsNr++;
1808
1809
169k
    return(1);
1810
65.1k
}
1811
1812
/**
1813
 * Pops the top `nr` namespaces and restores the hash table.
1814
 *
1815
 * @param ctxt  an XML parser context
1816
 * @param nr  the number to pop
1817
 * @returns the number of namespaces popped.
1818
 */
1819
static int
1820
xmlParserNsPop(xmlParserCtxtPtr ctxt, int nr)
1821
93.1k
{
1822
93.1k
    int i;
1823
1824
    /* assert(nr <= ctxt->nsNr); */
1825
1826
251k
    for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) {
1827
158k
        const xmlChar *prefix = ctxt->nsTab[i * 2];
1828
158k
        xmlParserNsExtra *extra = &ctxt->nsdb->extra[i];
1829
1830
158k
        if (prefix == NULL) {
1831
11.5k
            ctxt->nsdb->defaultNsIndex = extra->oldIndex;
1832
147k
        } else {
1833
147k
            xmlHashedString hprefix;
1834
147k
            xmlParserNsBucket *bucket = NULL;
1835
1836
147k
            hprefix.name = prefix;
1837
147k
            hprefix.hashValue = extra->prefixHashValue;
1838
147k
            xmlParserNsLookup(ctxt, &hprefix, &bucket);
1839
            /* assert(bucket && bucket->hashValue); */
1840
147k
            bucket->index = extra->oldIndex;
1841
147k
        }
1842
158k
    }
1843
1844
93.1k
    ctxt->nsNr -= nr;
1845
93.1k
    return(nr);
1846
93.1k
}
1847
1848
static int
1849
19.9k
xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt) {
1850
19.9k
    const xmlChar **atts;
1851
19.9k
    unsigned *attallocs;
1852
19.9k
    int newSize;
1853
1854
19.9k
    newSize = xmlGrowCapacity(ctxt->maxatts / 5,
1855
19.9k
                              sizeof(atts[0]) * 5 + sizeof(attallocs[0]),
1856
19.9k
                              10, XML_MAX_ATTRS);
1857
19.9k
    if (newSize < 0) {
1858
0
        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
1859
0
                    "Maximum number of attributes exceeded");
1860
0
        return(-1);
1861
0
    }
1862
1863
19.9k
    atts = xmlRealloc(ctxt->atts, newSize * sizeof(atts[0]) * 5);
1864
19.9k
    if (atts == NULL)
1865
10
        goto mem_error;
1866
19.9k
    ctxt->atts = atts;
1867
1868
19.9k
    attallocs = xmlRealloc(ctxt->attallocs,
1869
19.9k
                           newSize * sizeof(attallocs[0]));
1870
19.9k
    if (attallocs == NULL)
1871
7
        goto mem_error;
1872
19.9k
    ctxt->attallocs = attallocs;
1873
1874
19.9k
    ctxt->maxatts = newSize * 5;
1875
1876
19.9k
    return(0);
1877
1878
17
mem_error:
1879
17
    xmlErrMemory(ctxt);
1880
17
    return(-1);
1881
19.9k
}
1882
1883
/**
1884
 * Pushes a new parser input on top of the input stack
1885
 *
1886
 * @param ctxt  an XML parser context
1887
 * @param value  the parser input
1888
 * @returns -1 in case of error, the index in the stack otherwise
1889
 */
1890
int
1891
xmlCtxtPushInput(xmlParserCtxt *ctxt, xmlParserInput *value)
1892
147k
{
1893
147k
    char *directory = NULL;
1894
147k
    int maxDepth;
1895
1896
147k
    if ((ctxt == NULL) || (value == NULL))
1897
14.0k
        return(-1);
1898
1899
133k
    maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
1900
1901
133k
    if (ctxt->inputNr >= ctxt->inputMax) {
1902
7.66k
        xmlParserInputPtr *tmp;
1903
7.66k
        int newSize;
1904
1905
7.66k
        newSize = xmlGrowCapacity(ctxt->inputMax, sizeof(tmp[0]),
1906
7.66k
                                  5, maxDepth);
1907
7.66k
        if (newSize < 0) {
1908
1
            xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
1909
1
                           "Maximum entity nesting depth exceeded");
1910
1
            return(-1);
1911
1
        }
1912
7.66k
        tmp = xmlRealloc(ctxt->inputTab, newSize * sizeof(tmp[0]));
1913
7.66k
        if (tmp == NULL) {
1914
2
            xmlErrMemory(ctxt);
1915
2
            return(-1);
1916
2
        }
1917
7.66k
        ctxt->inputTab = tmp;
1918
7.66k
        ctxt->inputMax = newSize;
1919
7.66k
    }
1920
1921
133k
    if ((ctxt->inputNr == 0) && (value->filename != NULL)) {
1922
48.7k
        directory = xmlParserGetDirectory(value->filename);
1923
48.7k
        if (directory == NULL) {
1924
9
            xmlErrMemory(ctxt);
1925
9
            return(-1);
1926
9
        }
1927
48.7k
    }
1928
1929
133k
    if (ctxt->input_id >= INT_MAX) {
1930
0
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, "Input ID overflow\n");
1931
0
        return(-1);
1932
0
    }
1933
1934
133k
    ctxt->inputTab[ctxt->inputNr] = value;
1935
133k
    ctxt->input = value;
1936
1937
133k
    if (ctxt->inputNr == 0) {
1938
48.6k
        xmlFree(ctxt->directory);
1939
48.6k
        ctxt->directory = directory;
1940
48.6k
    }
1941
1942
    /*
1943
     * The input ID is unused internally, but there are entity
1944
     * loaders in downstream code that detect the main document
1945
     * by checking for "input_id == 1".
1946
     */
1947
133k
    value->id = ctxt->input_id++;
1948
1949
133k
    return(ctxt->inputNr++);
1950
133k
}
1951
1952
/**
1953
 * Pops the top parser input from the input stack
1954
 *
1955
 * @param ctxt  an XML parser context
1956
 * @returns the input just removed
1957
 */
1958
xmlParserInput *
1959
xmlCtxtPopInput(xmlParserCtxt *ctxt)
1960
235k
{
1961
235k
    xmlParserInputPtr ret;
1962
1963
235k
    if (ctxt == NULL)
1964
0
        return(NULL);
1965
235k
    if (ctxt->inputNr <= 0)
1966
104k
        return (NULL);
1967
131k
    ctxt->inputNr--;
1968
131k
    if (ctxt->inputNr > 0)
1969
84.6k
        ctxt->input = ctxt->inputTab[ctxt->inputNr - 1];
1970
46.6k
    else
1971
46.6k
        ctxt->input = NULL;
1972
131k
    ret = ctxt->inputTab[ctxt->inputNr];
1973
131k
    ctxt->inputTab[ctxt->inputNr] = NULL;
1974
131k
    return (ret);
1975
235k
}
1976
1977
/**
1978
 * Pushes a new element node on top of the node stack
1979
 *
1980
 * @deprecated Internal function, do not use.
1981
 *
1982
 * @param ctxt  an XML parser context
1983
 * @param value  the element node
1984
 * @returns -1 in case of error, the index in the stack otherwise
1985
 */
1986
int
1987
nodePush(xmlParserCtxt *ctxt, xmlNode *value)
1988
1.09M
{
1989
1.09M
    if (ctxt == NULL)
1990
0
        return(0);
1991
1992
1.09M
    if (ctxt->nodeNr >= ctxt->nodeMax) {
1993
44.4k
        int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
1994
44.4k
        xmlNodePtr *tmp;
1995
44.4k
        int newSize;
1996
1997
44.4k
        newSize = xmlGrowCapacity(ctxt->nodeMax, sizeof(tmp[0]),
1998
44.4k
                                  10, maxDepth);
1999
44.4k
        if (newSize < 0) {
2000
15
            xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
2001
15
                    "Excessive depth in document: %d,"
2002
15
                    " use XML_PARSE_HUGE option\n",
2003
15
                    ctxt->nodeNr);
2004
15
            return(-1);
2005
15
        }
2006
2007
44.4k
  tmp = xmlRealloc(ctxt->nodeTab, newSize * sizeof(tmp[0]));
2008
44.4k
        if (tmp == NULL) {
2009
9
            xmlErrMemory(ctxt);
2010
9
            return (-1);
2011
9
        }
2012
44.4k
        ctxt->nodeTab = tmp;
2013
44.4k
  ctxt->nodeMax = newSize;
2014
44.4k
    }
2015
2016
1.09M
    ctxt->nodeTab[ctxt->nodeNr] = value;
2017
1.09M
    ctxt->node = value;
2018
1.09M
    return (ctxt->nodeNr++);
2019
1.09M
}
2020
2021
/**
2022
 * Pops the top element node from the node stack
2023
 *
2024
 * @deprecated Internal function, do not use.
2025
 *
2026
 * @param ctxt  an XML parser context
2027
 * @returns the node just removed
2028
 */
2029
xmlNode *
2030
nodePop(xmlParserCtxt *ctxt)
2031
1.13M
{
2032
1.13M
    xmlNodePtr ret;
2033
2034
1.13M
    if (ctxt == NULL) return(NULL);
2035
1.13M
    if (ctxt->nodeNr <= 0)
2036
60.6k
        return (NULL);
2037
1.07M
    ctxt->nodeNr--;
2038
1.07M
    if (ctxt->nodeNr > 0)
2039
1.07M
        ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1];
2040
5.02k
    else
2041
5.02k
        ctxt->node = NULL;
2042
1.07M
    ret = ctxt->nodeTab[ctxt->nodeNr];
2043
1.07M
    ctxt->nodeTab[ctxt->nodeNr] = NULL;
2044
1.07M
    return (ret);
2045
1.13M
}
2046
2047
/**
2048
 * Pushes a new element name/prefix/URL on top of the name stack
2049
 *
2050
 * @param ctxt  an XML parser context
2051
 * @param value  the element name
2052
 * @param prefix  the element prefix
2053
 * @param URI  the element namespace name
2054
 * @param line  the current line number for error messages
2055
 * @param nsNr  the number of namespaces pushed on the namespace table
2056
 * @returns -1 in case of error, the index in the stack otherwise
2057
 */
2058
static int
2059
nameNsPush(xmlParserCtxtPtr ctxt, const xmlChar * value,
2060
           const xmlChar *prefix, const xmlChar *URI, int line, int nsNr)
2061
1.19M
{
2062
1.19M
    xmlStartTag *tag;
2063
2064
1.19M
    if (ctxt->nameNr >= ctxt->nameMax) {
2065
48.2k
        const xmlChar **tmp;
2066
48.2k
        xmlStartTag *tmp2;
2067
48.2k
        int newSize;
2068
2069
48.2k
        newSize = xmlGrowCapacity(ctxt->nameMax,
2070
48.2k
                                  sizeof(tmp[0]) + sizeof(tmp2[0]),
2071
48.2k
                                  10, XML_MAX_ITEMS);
2072
48.2k
        if (newSize < 0)
2073
0
            goto mem_error;
2074
2075
48.2k
        tmp = xmlRealloc(ctxt->nameTab, newSize * sizeof(tmp[0]));
2076
48.2k
        if (tmp == NULL)
2077
11
      goto mem_error;
2078
48.2k
  ctxt->nameTab = tmp;
2079
2080
48.2k
        tmp2 = xmlRealloc(ctxt->pushTab, newSize * sizeof(tmp2[0]));
2081
48.2k
        if (tmp2 == NULL)
2082
16
      goto mem_error;
2083
48.1k
  ctxt->pushTab = tmp2;
2084
2085
48.1k
        ctxt->nameMax = newSize;
2086
1.14M
    } else if (ctxt->pushTab == NULL) {
2087
24.0k
        ctxt->pushTab = xmlMalloc(ctxt->nameMax * sizeof(ctxt->pushTab[0]));
2088
24.0k
        if (ctxt->pushTab == NULL)
2089
38
            goto mem_error;
2090
24.0k
    }
2091
1.19M
    ctxt->nameTab[ctxt->nameNr] = value;
2092
1.19M
    ctxt->name = value;
2093
1.19M
    tag = &ctxt->pushTab[ctxt->nameNr];
2094
1.19M
    tag->prefix = prefix;
2095
1.19M
    tag->URI = URI;
2096
1.19M
    tag->line = line;
2097
1.19M
    tag->nsNr = nsNr;
2098
1.19M
    return (ctxt->nameNr++);
2099
65
mem_error:
2100
65
    xmlErrMemory(ctxt);
2101
65
    return (-1);
2102
1.19M
}
2103
#ifdef LIBXML_PUSH_ENABLED
2104
/**
2105
 * Pops the top element/prefix/URI name from the name stack
2106
 *
2107
 * @param ctxt  an XML parser context
2108
 * @returns the name just removed
2109
 */
2110
static const xmlChar *
2111
nameNsPop(xmlParserCtxtPtr ctxt)
2112
0
{
2113
0
    const xmlChar *ret;
2114
2115
0
    if (ctxt->nameNr <= 0)
2116
0
        return (NULL);
2117
0
    ctxt->nameNr--;
2118
0
    if (ctxt->nameNr > 0)
2119
0
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2120
0
    else
2121
0
        ctxt->name = NULL;
2122
0
    ret = ctxt->nameTab[ctxt->nameNr];
2123
0
    ctxt->nameTab[ctxt->nameNr] = NULL;
2124
0
    return (ret);
2125
0
}
2126
#endif /* LIBXML_PUSH_ENABLED */
2127
2128
/**
2129
 * Pops the top element name from the name stack
2130
 *
2131
 * @deprecated Internal function, do not use.
2132
 *
2133
 * @param ctxt  an XML parser context
2134
 * @returns the name just removed
2135
 */
2136
static const xmlChar *
2137
namePop(xmlParserCtxtPtr ctxt)
2138
1.17M
{
2139
1.17M
    const xmlChar *ret;
2140
2141
1.17M
    if ((ctxt == NULL) || (ctxt->nameNr <= 0))
2142
34
        return (NULL);
2143
1.17M
    ctxt->nameNr--;
2144
1.17M
    if (ctxt->nameNr > 0)
2145
1.17M
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2146
6.90k
    else
2147
6.90k
        ctxt->name = NULL;
2148
1.17M
    ret = ctxt->nameTab[ctxt->nameNr];
2149
1.17M
    ctxt->nameTab[ctxt->nameNr] = NULL;
2150
1.17M
    return (ret);
2151
1.17M
}
2152
2153
1.36M
static int spacePush(xmlParserCtxtPtr ctxt, int val) {
2154
1.36M
    if (ctxt->spaceNr >= ctxt->spaceMax) {
2155
50.3k
        int *tmp;
2156
50.3k
        int newSize;
2157
2158
50.3k
        newSize = xmlGrowCapacity(ctxt->spaceMax, sizeof(tmp[0]),
2159
50.3k
                                  10, XML_MAX_ITEMS);
2160
50.3k
        if (newSize < 0) {
2161
0
      xmlErrMemory(ctxt);
2162
0
      return(-1);
2163
0
        }
2164
2165
50.3k
        tmp = xmlRealloc(ctxt->spaceTab, newSize * sizeof(tmp[0]));
2166
50.3k
        if (tmp == NULL) {
2167
16
      xmlErrMemory(ctxt);
2168
16
      return(-1);
2169
16
  }
2170
50.3k
  ctxt->spaceTab = tmp;
2171
2172
50.3k
        ctxt->spaceMax = newSize;
2173
50.3k
    }
2174
1.36M
    ctxt->spaceTab[ctxt->spaceNr] = val;
2175
1.36M
    ctxt->space = &ctxt->spaceTab[ctxt->spaceNr];
2176
1.36M
    return(ctxt->spaceNr++);
2177
1.36M
}
2178
2179
1.34M
static int spacePop(xmlParserCtxtPtr ctxt) {
2180
1.34M
    int ret;
2181
1.34M
    if (ctxt->spaceNr <= 0) return(0);
2182
1.34M
    ctxt->spaceNr--;
2183
1.34M
    if (ctxt->spaceNr > 0)
2184
1.34M
  ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1];
2185
6.29k
    else
2186
6.29k
        ctxt->space = &ctxt->spaceTab[0];
2187
1.34M
    ret = ctxt->spaceTab[ctxt->spaceNr];
2188
1.34M
    ctxt->spaceTab[ctxt->spaceNr] = -1;
2189
1.34M
    return(ret);
2190
1.34M
}
2191
2192
/*
2193
 * Macros for accessing the content. Those should be used only by the parser,
2194
 * and not exported.
2195
 *
2196
 * Dirty macros, i.e. one often need to make assumption on the context to
2197
 * use them
2198
 *
2199
 *   CUR_PTR return the current pointer to the xmlChar to be parsed.
2200
 *           To be used with extreme caution since operations consuming
2201
 *           characters may move the input buffer to a different location !
2202
 *   CUR     returns the current xmlChar value, i.e. a 8 bit value if compiled
2203
 *           This should be used internally by the parser
2204
 *           only to compare to ASCII values otherwise it would break when
2205
 *           running with UTF-8 encoding.
2206
 *   RAW     same as CUR but in the input buffer, bypass any token
2207
 *           extraction that may have been done
2208
 *   NXT(n)  returns the n'th next xmlChar. Same as CUR is should be used only
2209
 *           to compare on ASCII based substring.
2210
 *   SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined
2211
 *           strings without newlines within the parser.
2212
 *   NEXT1(l) Skip 1 xmlChar, and must also be used only to skip 1 non-newline ASCII
2213
 *           defined char within the parser.
2214
 * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding
2215
 *
2216
 *   NEXT    Skip to the next character, this does the proper decoding
2217
 *           in UTF-8 mode. It also pop-up unfinished entities on the fly.
2218
 *   NEXTL(l) Skip the current unicode character of l xmlChars long.
2219
 *   COPY_BUF  copy the current unicode char to the target buffer, increment
2220
 *            the index
2221
 *   GROW, SHRINK  handling of input buffers
2222
 */
2223
2224
18.7M
#define RAW (*ctxt->input->cur)
2225
220M
#define CUR (*ctxt->input->cur)
2226
11.2M
#define NXT(val) ctxt->input->cur[(val)]
2227
391M
#define CUR_PTR ctxt->input->cur
2228
3.93M
#define BASE_PTR ctxt->input->base
2229
2230
#define CMP4( s, c1, c2, c3, c4 ) \
2231
10.8M
  ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \
2232
5.45M
    ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 )
2233
#define CMP5( s, c1, c2, c3, c4, c5 ) \
2234
10.4M
  ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 )
2235
#define CMP6( s, c1, c2, c3, c4, c5, c6 ) \
2236
9.80M
  ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 )
2237
#define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \
2238
9.26M
  ( CMP6( s, c1, c2, c3, c4, c5, c6 ) && ((unsigned char *) s)[ 6 ] == c7 )
2239
#define CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) \
2240
8.86M
  ( CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) && ((unsigned char *) s)[ 7 ] == c8 )
2241
#define CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) \
2242
4.33M
  ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \
2243
4.33M
    ((unsigned char *) s)[ 8 ] == c9 )
2244
#define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \
2245
1.52k
  ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \
2246
1.52k
    ((unsigned char *) s)[ 9 ] == c10 )
2247
2248
3.14M
#define SKIP(val) do {             \
2249
3.14M
    ctxt->input->cur += (val),ctxt->input->col+=(val);      \
2250
3.14M
    if (*ctxt->input->cur == 0)           \
2251
3.14M
        xmlParserGrow(ctxt);           \
2252
3.14M
  } while (0)
2253
2254
#define SKIPL(val) do {             \
2255
    int skipl;                \
2256
    for(skipl=0; skipl<val; skipl++) {          \
2257
  if (*(ctxt->input->cur) == '\n') {        \
2258
  ctxt->input->line++; ctxt->input->col = 1;      \
2259
  } else ctxt->input->col++;          \
2260
  ctxt->input->cur++;           \
2261
    }                 \
2262
    if (*ctxt->input->cur == 0)           \
2263
        xmlParserGrow(ctxt);            \
2264
  } while (0)
2265
2266
#define SHRINK \
2267
9.09M
    if (!PARSER_PROGRESSIVE(ctxt)) \
2268
9.09M
  xmlParserShrink(ctxt);
2269
2270
#define GROW \
2271
22.7M
    if ((!PARSER_PROGRESSIVE(ctxt)) && \
2272
22.7M
        (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \
2273
3.79M
  xmlParserGrow(ctxt);
2274
2275
3.99M
#define SKIP_BLANKS xmlSkipBlankChars(ctxt)
2276
2277
836k
#define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt)
2278
2279
49.9M
#define NEXT xmlNextChar(ctxt)
2280
2281
1.71M
#define NEXT1 {               \
2282
1.71M
  ctxt->input->col++;           \
2283
1.71M
  ctxt->input->cur++;           \
2284
1.71M
  if (*ctxt->input->cur == 0)         \
2285
1.71M
      xmlParserGrow(ctxt);           \
2286
1.71M
    }
2287
2288
251M
#define NEXTL(l) do {             \
2289
251M
    if (*(ctxt->input->cur) == '\n') {         \
2290
11.5M
  ctxt->input->line++; ctxt->input->col = 1;      \
2291
239M
    } else ctxt->input->col++;           \
2292
251M
    ctxt->input->cur += l;        \
2293
251M
  } while (0)
2294
2295
#define COPY_BUF(b, i, v)           \
2296
57.5M
    if (v < 0x80) b[i++] = v;           \
2297
57.5M
    else i += xmlCopyCharMultiByte(&b[i],v)
2298
2299
static int
2300
52.5M
xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) {
2301
52.5M
    int c = xmlCurrentChar(ctxt, len);
2302
2303
52.5M
    if (c == XML_INVALID_CHAR)
2304
15.8M
        c = 0xFFFD; /* replacement character */
2305
2306
52.5M
    return(c);
2307
52.5M
}
2308
2309
/**
2310
 * Skip whitespace in the input stream.
2311
 *
2312
 * @deprecated Internal function, do not use.
2313
 *
2314
 * @param ctxt  the XML parser context
2315
 * @returns the number of space chars skipped
2316
 */
2317
int
2318
4.13M
xmlSkipBlankChars(xmlParserCtxt *ctxt) {
2319
4.13M
    const xmlChar *cur;
2320
4.13M
    int res = 0;
2321
2322
4.13M
    cur = ctxt->input->cur;
2323
4.13M
    while (IS_BLANK_CH(*cur)) {
2324
2.73M
        if (*cur == '\n') {
2325
1.86M
            ctxt->input->line++; ctxt->input->col = 1;
2326
1.86M
        } else {
2327
869k
            ctxt->input->col++;
2328
869k
        }
2329
2.73M
        cur++;
2330
2.73M
        if (res < INT_MAX)
2331
2.73M
            res++;
2332
2.73M
        if (*cur == 0) {
2333
15.4k
            ctxt->input->cur = cur;
2334
15.4k
            xmlParserGrow(ctxt);
2335
15.4k
            cur = ctxt->input->cur;
2336
15.4k
        }
2337
2.73M
    }
2338
4.13M
    ctxt->input->cur = cur;
2339
2340
4.13M
    if (res > 4)
2341
85.5k
        GROW;
2342
2343
4.13M
    return(res);
2344
4.13M
}
2345
2346
static void
2347
79.3k
xmlPopPE(xmlParserCtxtPtr ctxt) {
2348
79.3k
    unsigned long consumed;
2349
79.3k
    xmlEntityPtr ent;
2350
2351
79.3k
    ent = ctxt->input->entity;
2352
2353
79.3k
    ent->flags &= ~XML_ENT_EXPANDING;
2354
2355
79.3k
    if ((ent->flags & XML_ENT_CHECKED) == 0) {
2356
4.86k
        int result;
2357
2358
        /*
2359
         * Read the rest of the stream in case of errors. We want
2360
         * to account for the whole entity size.
2361
         */
2362
5.29k
        do {
2363
5.29k
            ctxt->input->cur = ctxt->input->end;
2364
5.29k
            xmlParserShrink(ctxt);
2365
5.29k
            result = xmlParserGrow(ctxt);
2366
5.29k
        } while (result > 0);
2367
2368
4.86k
        consumed = ctxt->input->consumed;
2369
4.86k
        xmlSaturatedAddSizeT(&consumed,
2370
4.86k
                             ctxt->input->end - ctxt->input->base);
2371
2372
4.86k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
2373
2374
        /*
2375
         * Add to sizeentities when parsing an external entity
2376
         * for the first time.
2377
         */
2378
4.86k
        if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
2379
3.18k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
2380
3.18k
        }
2381
2382
4.86k
        ent->flags |= XML_ENT_CHECKED;
2383
4.86k
    }
2384
2385
79.3k
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
2386
2387
79.3k
    xmlParserEntityCheck(ctxt, ent->expandedSize);
2388
2389
79.3k
    GROW;
2390
79.3k
}
2391
2392
/**
2393
 * Skip whitespace in the input stream, also handling parameter
2394
 * entities.
2395
 *
2396
 * @param ctxt  the XML parser context
2397
 * @returns the number of space chars skipped
2398
 */
2399
static int
2400
836k
xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) {
2401
836k
    int res = 0;
2402
836k
    int inParam;
2403
836k
    int expandParam;
2404
2405
836k
    inParam = PARSER_IN_PE(ctxt);
2406
836k
    expandParam = PARSER_EXTERNAL(ctxt);
2407
2408
836k
    if (!inParam && !expandParam)
2409
137k
        return(xmlSkipBlankChars(ctxt));
2410
2411
    /*
2412
     * It's Okay to use CUR/NEXT here since all the blanks are on
2413
     * the ASCII range.
2414
     */
2415
1.61M
    while (PARSER_STOPPED(ctxt) == 0) {
2416
1.61M
        if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */
2417
894k
            NEXT;
2418
894k
        } else if (CUR == '%') {
2419
40.6k
            if ((expandParam == 0) ||
2420
40.5k
                (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0))
2421
21.9k
                break;
2422
2423
            /*
2424
             * Expand parameter entity. We continue to consume
2425
             * whitespace at the start of the entity and possible
2426
             * even consume the whole entity and pop it. We might
2427
             * even pop multiple PEs in this loop.
2428
             */
2429
18.6k
            xmlParsePERefInternal(ctxt, 0);
2430
2431
18.6k
            inParam = PARSER_IN_PE(ctxt);
2432
18.6k
            expandParam = PARSER_EXTERNAL(ctxt);
2433
681k
        } else if (CUR == 0) {
2434
32.5k
            if (inParam == 0)
2435
112
                break;
2436
2437
            /*
2438
             * Don't pop parameter entities that start a markup
2439
             * declaration to detect Well-formedness constraint:
2440
             * PE Between Declarations.
2441
             */
2442
32.4k
            if (ctxt->input->flags & XML_INPUT_MARKUP_DECL)
2443
27.9k
                break;
2444
2445
4.56k
            xmlPopPE(ctxt);
2446
2447
4.56k
            inParam = PARSER_IN_PE(ctxt);
2448
4.56k
            expandParam = PARSER_EXTERNAL(ctxt);
2449
648k
        } else {
2450
648k
            break;
2451
648k
        }
2452
2453
        /*
2454
         * Also increase the counter when entering or exiting a PERef.
2455
         * The spec says: "When a parameter-entity reference is recognized
2456
         * in the DTD and included, its replacement text MUST be enlarged
2457
         * by the attachment of one leading and one following space (#x20)
2458
         * character."
2459
         */
2460
917k
        if (res < INT_MAX)
2461
917k
            res++;
2462
917k
    }
2463
2464
698k
    return(res);
2465
836k
}
2466
2467
/************************************************************************
2468
 *                  *
2469
 *    Commodity functions to handle entities      *
2470
 *                  *
2471
 ************************************************************************/
2472
2473
/**
2474
 * @deprecated Internal function, don't use.
2475
 *
2476
 * @param ctxt  an XML parser context
2477
 * @returns the current xmlChar in the parser context
2478
 */
2479
xmlChar
2480
0
xmlPopInput(xmlParserCtxt *ctxt) {
2481
0
    xmlParserInputPtr input;
2482
2483
0
    if ((ctxt == NULL) || (ctxt->inputNr <= 1)) return(0);
2484
0
    input = xmlCtxtPopInput(ctxt);
2485
0
    xmlFreeInputStream(input);
2486
0
    if (*ctxt->input->cur == 0)
2487
0
        xmlParserGrow(ctxt);
2488
0
    return(CUR);
2489
0
}
2490
2491
/**
2492
 * Push an input stream onto the stack.
2493
 *
2494
 * @deprecated Internal function, don't use.
2495
 *
2496
 * @param ctxt  an XML parser context
2497
 * @param input  an XML parser input fragment (entity, XML fragment ...).
2498
 * @returns -1 in case of error or the index in the input stack
2499
 */
2500
int
2501
0
xmlPushInput(xmlParserCtxt *ctxt, xmlParserInput *input) {
2502
0
    int ret;
2503
2504
0
    if ((ctxt == NULL) || (input == NULL))
2505
0
        return(-1);
2506
2507
0
    ret = xmlCtxtPushInput(ctxt, input);
2508
0
    if (ret >= 0)
2509
0
        GROW;
2510
0
    return(ret);
2511
0
}
2512
2513
/**
2514
 * Parse a numeric character reference. Always consumes '&'.
2515
 *
2516
 * @deprecated Internal function, don't use.
2517
 *
2518
 *     [66] CharRef ::= '&#' [0-9]+ ';' |
2519
 *                      '&#x' [0-9a-fA-F]+ ';'
2520
 *
2521
 * [ WFC: Legal Character ]
2522
 * Characters referred to using character references must match the
2523
 * production for Char.
2524
 *
2525
 * @param ctxt  an XML parser context
2526
 * @returns the value parsed (as an int), 0 in case of error
2527
 */
2528
int
2529
83.1k
xmlParseCharRef(xmlParserCtxt *ctxt) {
2530
83.1k
    int val = 0;
2531
83.1k
    int count = 0;
2532
2533
    /*
2534
     * Using RAW/CUR/NEXT is okay since we are working on ASCII range here
2535
     */
2536
83.1k
    if ((RAW == '&') && (NXT(1) == '#') &&
2537
83.1k
        (NXT(2) == 'x')) {
2538
25.3k
  SKIP(3);
2539
25.3k
  GROW;
2540
69.7k
  while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) {
2541
54.9k
      if (count++ > 20) {
2542
527
    count = 0;
2543
527
    GROW;
2544
527
      }
2545
54.9k
      if ((RAW >= '0') && (RAW <= '9'))
2546
22.6k
          val = val * 16 + (CUR - '0');
2547
32.3k
      else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20))
2548
14.2k
          val = val * 16 + (CUR - 'a') + 10;
2549
18.1k
      else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20))
2550
7.61k
          val = val * 16 + (CUR - 'A') + 10;
2551
10.5k
      else {
2552
10.5k
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2553
10.5k
    val = 0;
2554
10.5k
    break;
2555
10.5k
      }
2556
44.4k
      if (val > 0x110000)
2557
10.1k
          val = 0x110000;
2558
2559
44.4k
      NEXT;
2560
44.4k
      count++;
2561
44.4k
  }
2562
25.3k
  if (RAW == ';') {
2563
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2564
14.8k
      ctxt->input->col++;
2565
14.8k
      ctxt->input->cur++;
2566
14.8k
  }
2567
57.7k
    } else if  ((RAW == '&') && (NXT(1) == '#')) {
2568
57.7k
  SKIP(2);
2569
57.7k
  GROW;
2570
187k
  while (RAW != ';') { /* loop blocked by count */
2571
135k
      if (count++ > 20) {
2572
1.41k
    count = 0;
2573
1.41k
    GROW;
2574
1.41k
      }
2575
135k
      if ((RAW >= '0') && (RAW <= '9'))
2576
129k
          val = val * 10 + (CUR - '0');
2577
6.09k
      else {
2578
6.09k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2579
6.09k
    val = 0;
2580
6.09k
    break;
2581
6.09k
      }
2582
129k
      if (val > 0x110000)
2583
13.2k
          val = 0x110000;
2584
2585
129k
      NEXT;
2586
129k
      count++;
2587
129k
  }
2588
57.7k
  if (RAW == ';') {
2589
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2590
51.6k
      ctxt->input->col++;
2591
51.6k
      ctxt->input->cur++;
2592
51.6k
  }
2593
57.7k
    } else {
2594
0
        if (RAW == '&')
2595
0
            SKIP(1);
2596
0
        xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2597
0
    }
2598
2599
    /*
2600
     * [ WFC: Legal Character ]
2601
     * Characters referred to using character references must match the
2602
     * production for Char.
2603
     */
2604
83.1k
    if (val >= 0x110000) {
2605
1.55k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2606
1.55k
                "xmlParseCharRef: character reference out of bounds\n",
2607
1.55k
          val);
2608
1.55k
        val = 0xFFFD;
2609
81.5k
    } else if (!IS_CHAR(val)) {
2610
19.0k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2611
19.0k
                          "xmlParseCharRef: invalid xmlChar value %d\n",
2612
19.0k
                    val);
2613
19.0k
    }
2614
83.1k
    return(val);
2615
83.1k
}
2616
2617
/**
2618
 * Parse Reference declarations, variant parsing from a string rather
2619
 * than an an input flow.
2620
 *
2621
 *     [66] CharRef ::= '&#' [0-9]+ ';' |
2622
 *                      '&#x' [0-9a-fA-F]+ ';'
2623
 *
2624
 * [ WFC: Legal Character ]
2625
 * Characters referred to using character references must match the
2626
 * production for Char.
2627
 *
2628
 * @param ctxt  an XML parser context
2629
 * @param str  a pointer to an index in the string
2630
 * @returns the value parsed (as an int), 0 in case of error, str will be
2631
 *         updated to the current value of the index
2632
 */
2633
static int
2634
135k
xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) {
2635
135k
    const xmlChar *ptr;
2636
135k
    xmlChar cur;
2637
135k
    int val = 0;
2638
2639
135k
    if ((str == NULL) || (*str == NULL)) return(0);
2640
135k
    ptr = *str;
2641
135k
    cur = *ptr;
2642
135k
    if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) {
2643
14.7k
  ptr += 3;
2644
14.7k
  cur = *ptr;
2645
66.0k
  while (cur != ';') { /* Non input consuming loop */
2646
52.2k
      if ((cur >= '0') && (cur <= '9'))
2647
35.9k
          val = val * 16 + (cur - '0');
2648
16.3k
      else if ((cur >= 'a') && (cur <= 'f'))
2649
13.8k
          val = val * 16 + (cur - 'a') + 10;
2650
2.45k
      else if ((cur >= 'A') && (cur <= 'F'))
2651
1.53k
          val = val * 16 + (cur - 'A') + 10;
2652
917
      else {
2653
917
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2654
917
    val = 0;
2655
917
    break;
2656
917
      }
2657
51.3k
      if (val > 0x110000)
2658
561
          val = 0x110000;
2659
2660
51.3k
      ptr++;
2661
51.3k
      cur = *ptr;
2662
51.3k
  }
2663
14.7k
  if (cur == ';')
2664
13.8k
      ptr++;
2665
121k
    } else if  ((cur == '&') && (ptr[1] == '#')){
2666
121k
  ptr += 2;
2667
121k
  cur = *ptr;
2668
409k
  while (cur != ';') { /* Non input consuming loops */
2669
290k
      if ((cur >= '0') && (cur <= '9'))
2670
288k
          val = val * 10 + (cur - '0');
2671
1.77k
      else {
2672
1.77k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2673
1.77k
    val = 0;
2674
1.77k
    break;
2675
1.77k
      }
2676
288k
      if (val > 0x110000)
2677
874
          val = 0x110000;
2678
2679
288k
      ptr++;
2680
288k
      cur = *ptr;
2681
288k
  }
2682
121k
  if (cur == ';')
2683
119k
      ptr++;
2684
121k
    } else {
2685
0
  xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2686
0
  return(0);
2687
0
    }
2688
135k
    *str = ptr;
2689
2690
    /*
2691
     * [ WFC: Legal Character ]
2692
     * Characters referred to using character references must match the
2693
     * production for Char.
2694
     */
2695
135k
    if (val >= 0x110000) {
2696
204
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2697
204
                "xmlParseStringCharRef: character reference out of bounds\n",
2698
204
                val);
2699
135k
    } else if (IS_CHAR(val)) {
2700
131k
        return(val);
2701
131k
    } else {
2702
3.95k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2703
3.95k
        "xmlParseStringCharRef: invalid xmlChar value %d\n",
2704
3.95k
        val);
2705
3.95k
    }
2706
4.16k
    return(0);
2707
135k
}
2708
2709
/**
2710
 *     [69] PEReference ::= '%' Name ';'
2711
 *
2712
 * @deprecated Internal function, do not use.
2713
 *
2714
 * [ WFC: No Recursion ]
2715
 * A parsed entity must not contain a recursive
2716
 * reference to itself, either directly or indirectly.
2717
 *
2718
 * [ WFC: Entity Declared ]
2719
 * In a document without any DTD, a document with only an internal DTD
2720
 * subset which contains no parameter entity references, or a document
2721
 * with "standalone='yes'", ...  ... The declaration of a parameter
2722
 * entity must precede any reference to it...
2723
 *
2724
 * [ VC: Entity Declared ]
2725
 * In a document with an external subset or external parameter entities
2726
 * with "standalone='no'", ...  ... The declaration of a parameter entity
2727
 * must precede any reference to it...
2728
 *
2729
 * [ WFC: In DTD ]
2730
 * Parameter-entity references may only appear in the DTD.
2731
 * NOTE: misleading but this is handled.
2732
 *
2733
 * A PEReference may have been detected in the current input stream
2734
 * the handling is done accordingly to
2735
 *      http://www.w3.org/TR/REC-xml#entproc
2736
 * i.e.
2737
 *   - Included in literal in entity values
2738
 *   - Included as Parameter Entity reference within DTDs
2739
 * @param ctxt  the parser context
2740
 */
2741
void
2742
0
xmlParserHandlePEReference(xmlParserCtxt *ctxt) {
2743
0
    xmlParsePERefInternal(ctxt, 0);
2744
0
}
2745
2746
/**
2747
 * @deprecated Internal function, don't use.
2748
 *
2749
 * @param ctxt  the parser context
2750
 * @param str  the input string
2751
 * @param len  the string length
2752
 * @param what  combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF
2753
 * @param end  an end marker xmlChar, 0 if none
2754
 * @param end2  an end marker xmlChar, 0 if none
2755
 * @param end3  an end marker xmlChar, 0 if none
2756
 * @returns A newly allocated string with the substitution done. The caller
2757
 *      must deallocate it !
2758
 */
2759
xmlChar *
2760
xmlStringLenDecodeEntities(xmlParserCtxt *ctxt, const xmlChar *str, int len,
2761
                           int what ATTRIBUTE_UNUSED,
2762
0
                           xmlChar end, xmlChar end2, xmlChar end3) {
2763
0
    if ((ctxt == NULL) || (str == NULL) || (len < 0))
2764
0
        return(NULL);
2765
2766
0
    if ((str[len] != 0) ||
2767
0
        (end != 0) || (end2 != 0) || (end3 != 0))
2768
0
        return(NULL);
2769
2770
0
    return(xmlExpandEntitiesInAttValue(ctxt, str, 0));
2771
0
}
2772
2773
/**
2774
 * @deprecated Internal function, don't use.
2775
 *
2776
 * @param ctxt  the parser context
2777
 * @param str  the input string
2778
 * @param what  combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF
2779
 * @param end  an end marker xmlChar, 0 if none
2780
 * @param end2  an end marker xmlChar, 0 if none
2781
 * @param end3  an end marker xmlChar, 0 if none
2782
 * @returns A newly allocated string with the substitution done. The caller
2783
 *      must deallocate it !
2784
 */
2785
xmlChar *
2786
xmlStringDecodeEntities(xmlParserCtxt *ctxt, const xmlChar *str,
2787
                        int what ATTRIBUTE_UNUSED,
2788
0
            xmlChar end, xmlChar  end2, xmlChar end3) {
2789
0
    if ((ctxt == NULL) || (str == NULL))
2790
0
        return(NULL);
2791
2792
0
    if ((end != 0) || (end2 != 0) || (end3 != 0))
2793
0
        return(NULL);
2794
2795
0
    return(xmlExpandEntitiesInAttValue(ctxt, str, 0));
2796
0
}
2797
2798
/************************************************************************
2799
 *                  *
2800
 *    Commodity functions, cleanup needed ?     *
2801
 *                  *
2802
 ************************************************************************/
2803
2804
/**
2805
 * Is this a sequence of blank chars that one can ignore ?
2806
 *
2807
 * @param ctxt  an XML parser context
2808
 * @param str  a xmlChar *
2809
 * @param len  the size of `str`
2810
 * @param blank_chars  we know the chars are blanks
2811
 * @returns 1 if ignorable 0 otherwise.
2812
 */
2813
2814
static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len,
2815
369k
                     int blank_chars) {
2816
369k
    int i;
2817
369k
    xmlNodePtr lastChild;
2818
2819
    /*
2820
     * Check for xml:space value.
2821
     */
2822
369k
    if ((ctxt->space == NULL) || (*(ctxt->space) == 1) ||
2823
368k
        (*(ctxt->space) == -2))
2824
313k
  return(0);
2825
2826
    /*
2827
     * Check that the string is made of blanks
2828
     */
2829
56.2k
    if (blank_chars == 0) {
2830
99.0k
  for (i = 0;i < len;i++)
2831
93.1k
      if (!(IS_BLANK_CH(str[i]))) return(0);
2832
46.0k
    }
2833
2834
    /*
2835
     * Look if the element is mixed content in the DTD if available
2836
     */
2837
16.1k
    if (ctxt->node == NULL) return(0);
2838
16.1k
    if (ctxt->myDoc != NULL) {
2839
16.1k
        xmlElementPtr elemDecl = NULL;
2840
16.1k
        xmlDocPtr doc = ctxt->myDoc;
2841
16.1k
        const xmlChar *prefix = NULL;
2842
2843
16.1k
        if (ctxt->node->ns)
2844
4.89k
            prefix = ctxt->node->ns->prefix;
2845
16.1k
        if (doc->intSubset != NULL)
2846
11.7k
            elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name,
2847
11.7k
                                      prefix);
2848
16.1k
        if ((elemDecl == NULL) && (doc->extSubset != NULL))
2849
889
            elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name,
2850
889
                                      prefix);
2851
16.1k
        if (elemDecl != NULL) {
2852
2.38k
            if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)
2853
535
                return(1);
2854
1.84k
            if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) ||
2855
1.65k
                (elemDecl->etype == XML_ELEMENT_TYPE_MIXED))
2856
402
                return(0);
2857
1.84k
        }
2858
16.1k
    }
2859
2860
    /*
2861
     * Otherwise, heuristic :-\
2862
     *
2863
     * When push parsing, we could be at the end of a chunk.
2864
     * This makes the look-ahead and consequently the NOBLANKS
2865
     * option unreliable.
2866
     */
2867
15.2k
    if ((RAW != '<') && (RAW != 0xD)) return(0);
2868
11.7k
    if ((ctxt->node->children == NULL) &&
2869
8.83k
  (RAW == '<') && (NXT(1) == '/')) return(0);
2870
2871
11.4k
    lastChild = xmlGetLastChild(ctxt->node);
2872
11.4k
    if (lastChild == NULL) {
2873
8.52k
        if ((ctxt->node->type != XML_ELEMENT_NODE) &&
2874
0
            (ctxt->node->content != NULL)) return(0);
2875
8.52k
    } else if (xmlNodeIsText(lastChild))
2876
329
        return(0);
2877
2.55k
    else if ((ctxt->node->children != NULL) &&
2878
2.55k
             (xmlNodeIsText(ctxt->node->children)))
2879
227
        return(0);
2880
10.8k
    return(1);
2881
11.4k
}
2882
2883
/************************************************************************
2884
 *                  *
2885
 *    Extra stuff for namespace support     *
2886
 *  Relates to http://www.w3.org/TR/WD-xml-names      *
2887
 *                  *
2888
 ************************************************************************/
2889
2890
/**
2891
 * Parse an UTF8 encoded XML qualified name string
2892
 *
2893
 * @deprecated Don't use.
2894
 *
2895
 * @param ctxt  an XML parser context
2896
 * @param name  an XML parser context
2897
 * @param prefixOut  a xmlChar **
2898
 * @returns the local part, and prefix is updated
2899
 *   to get the Prefix if any.
2900
 */
2901
2902
xmlChar *
2903
0
xmlSplitQName(xmlParserCtxt *ctxt, const xmlChar *name, xmlChar **prefixOut) {
2904
0
    xmlChar *ret;
2905
0
    const xmlChar *localname;
2906
2907
0
    localname = xmlSplitQName4(name, prefixOut);
2908
0
    if (localname == NULL) {
2909
0
        xmlCtxtErrMemory(ctxt);
2910
0
        return(NULL);
2911
0
    }
2912
2913
0
    ret = xmlStrdup(localname);
2914
0
    if (ret == NULL) {
2915
0
        xmlCtxtErrMemory(ctxt);
2916
0
        xmlFree(*prefixOut);
2917
0
    }
2918
2919
0
    return(ret);
2920
0
}
2921
2922
/************************************************************************
2923
 *                  *
2924
 *      The parser itself       *
2925
 *  Relates to http://www.w3.org/TR/REC-xml       *
2926
 *                  *
2927
 ************************************************************************/
2928
2929
/************************************************************************
2930
 *                  *
2931
 *  Routines to parse Name, NCName and NmToken      *
2932
 *                  *
2933
 ************************************************************************/
2934
2935
/*
2936
 * The two following functions are related to the change of accepted
2937
 * characters for Name and NmToken in the Revision 5 of XML-1.0
2938
 * They correspond to the modified production [4] and the new production [4a]
2939
 * changes in that revision. Also note that the macros used for the
2940
 * productions Letter, Digit, CombiningChar and Extender are not needed
2941
 * anymore.
2942
 * We still keep compatibility to pre-revision5 parsing semantic if the
2943
 * new XML_PARSE_OLD10 option is given to the parser.
2944
 */
2945
2946
static int
2947
1.43M
xmlIsNameStartCharNew(int c) {
2948
    /*
2949
     * Use the new checks of production [4] [4a] amd [5] of the
2950
     * Update 5 of XML-1.0
2951
     */
2952
1.43M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2953
1.42M
        (((c >= 'a') && (c <= 'z')) ||
2954
638k
         ((c >= 'A') && (c <= 'Z')) ||
2955
544k
         (c == '_') || (c == ':') ||
2956
537k
         ((c >= 0xC0) && (c <= 0xD6)) ||
2957
536k
         ((c >= 0xD8) && (c <= 0xF6)) ||
2958
535k
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2959
534k
         ((c >= 0x370) && (c <= 0x37D)) ||
2960
534k
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2961
527k
         ((c >= 0x200C) && (c <= 0x200D)) ||
2962
521k
         ((c >= 0x2070) && (c <= 0x218F)) ||
2963
520k
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2964
520k
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2965
519k
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2966
519k
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2967
507k
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2968
923k
        return(1);
2969
513k
    return(0);
2970
1.43M
}
2971
2972
static int
2973
35.7M
xmlIsNameCharNew(int c) {
2974
    /*
2975
     * Use the new checks of production [4] [4a] amd [5] of the
2976
     * Update 5 of XML-1.0
2977
     */
2978
35.7M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2979
35.3M
        (((c >= 'a') && (c <= 'z')) ||
2980
28.4M
         ((c >= 'A') && (c <= 'Z')) ||
2981
27.9M
         ((c >= '0') && (c <= '9')) || /* !start */
2982
27.6M
         (c == '_') || (c == ':') ||
2983
27.5M
         (c == '-') || (c == '.') || (c == 0xB7) || /* !start */
2984
26.1M
         ((c >= 0xC0) && (c <= 0xD6)) ||
2985
25.9M
         ((c >= 0xD8) && (c <= 0xF6)) ||
2986
25.9M
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2987
25.8M
         ((c >= 0x300) && (c <= 0x36F)) || /* !start */
2988
25.8M
         ((c >= 0x370) && (c <= 0x37D)) ||
2989
25.8M
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2990
25.6M
         ((c >= 0x200C) && (c <= 0x200D)) ||
2991
25.6M
         ((c >= 0x203F) && (c <= 0x2040)) || /* !start */
2992
25.6M
         ((c >= 0x2070) && (c <= 0x218F)) ||
2993
8.39M
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2994
8.39M
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2995
8.21M
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2996
8.21M
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2997
596k
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2998
34.7M
         return(1);
2999
931k
    return(0);
3000
35.7M
}
3001
3002
static int
3003
1.21M
xmlIsNameStartCharOld(int c) {
3004
1.21M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3005
1.20M
        ((IS_LETTER(c) || (c == '_') || (c == ':'))))
3006
618k
        return(1);
3007
596k
    return(0);
3008
1.21M
}
3009
3010
static int
3011
7.62M
xmlIsNameCharOld(int c) {
3012
7.62M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3013
7.61M
        ((IS_LETTER(c)) || (IS_DIGIT(c)) ||
3014
620k
         (c == '.') || (c == '-') ||
3015
615k
         (c == '_') || (c == ':') ||
3016
607k
         (IS_COMBINING(c)) ||
3017
606k
         (IS_EXTENDER(c))))
3018
7.01M
        return(1);
3019
610k
    return(0);
3020
7.62M
}
3021
3022
static int
3023
2.65M
xmlIsNameStartChar(int c, int old10) {
3024
2.65M
    if (!old10)
3025
1.43M
        return(xmlIsNameStartCharNew(c));
3026
1.21M
    else
3027
1.21M
        return(xmlIsNameStartCharOld(c));
3028
2.65M
}
3029
3030
static int
3031
43.3M
xmlIsNameChar(int c, int old10) {
3032
43.3M
    if (!old10)
3033
35.7M
        return(xmlIsNameCharNew(c));
3034
7.62M
    else
3035
7.62M
        return(xmlIsNameCharOld(c));
3036
43.3M
}
3037
3038
/*
3039
 * Scan an XML Name, NCName or Nmtoken.
3040
 *
3041
 * Returns a pointer to the end of the name on success. If the
3042
 * name is invalid, returns `ptr`. If the name is longer than
3043
 * `maxSize` bytes, returns NULL.
3044
 *
3045
 * @param ptr  pointer to the start of the name
3046
 * @param maxSize  maximum size in bytes
3047
 * @param flags  XML_SCAN_* flags
3048
 * @returns a pointer to the end of the name or NULL
3049
 */
3050
const xmlChar *
3051
1.41M
xmlScanName(const xmlChar *ptr, size_t maxSize, int flags) {
3052
1.41M
    int stop = flags & XML_SCAN_NC ? ':' : 0;
3053
1.41M
    int old10 = flags & XML_SCAN_OLD10 ? 1 : 0;
3054
3055
11.9M
    while (1) {
3056
11.9M
        int c, len;
3057
3058
11.9M
        c = *ptr;
3059
11.9M
        if (c < 0x80) {
3060
8.47M
            if (c == stop)
3061
15.5k
                break;
3062
8.45M
            len = 1;
3063
8.45M
        } else {
3064
3.48M
            len = 4;
3065
3.48M
            c = xmlGetUTF8Char(ptr, &len);
3066
3.48M
            if (c < 0)
3067
1.54k
                break;
3068
3.48M
        }
3069
3070
11.9M
        if (flags & XML_SCAN_NMTOKEN ?
3071
10.5M
                !xmlIsNameChar(c, old10) :
3072
11.9M
                !xmlIsNameStartChar(c, old10))
3073
1.39M
            break;
3074
3075
10.5M
        if ((size_t) len > maxSize)
3076
139
            return(NULL);
3077
10.5M
        ptr += len;
3078
10.5M
        maxSize -= len;
3079
10.5M
        flags |= XML_SCAN_NMTOKEN;
3080
10.5M
    }
3081
3082
1.41M
    return(ptr);
3083
1.41M
}
3084
3085
static const xmlChar *
3086
234k
xmlParseNameComplex(xmlParserCtxtPtr ctxt) {
3087
234k
    const xmlChar *ret;
3088
234k
    int len = 0, l;
3089
234k
    int c;
3090
234k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3091
107k
                    XML_MAX_TEXT_LENGTH :
3092
234k
                    XML_MAX_NAME_LENGTH;
3093
234k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3094
3095
    /*
3096
     * Handler for more complex cases
3097
     */
3098
234k
    c = xmlCurrentChar(ctxt, &l);
3099
234k
    if (!xmlIsNameStartChar(c, old10))
3100
186k
        return(NULL);
3101
48.3k
    len += l;
3102
48.3k
    NEXTL(l);
3103
48.3k
    c = xmlCurrentChar(ctxt, &l);
3104
11.4M
    while (xmlIsNameChar(c, old10)) {
3105
11.3M
        if (len <= INT_MAX - l)
3106
11.3M
            len += l;
3107
11.3M
        NEXTL(l);
3108
11.3M
        c = xmlCurrentChar(ctxt, &l);
3109
11.3M
    }
3110
48.3k
    if (len > maxLength) {
3111
106
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3112
106
        return(NULL);
3113
106
    }
3114
48.2k
    if (ctxt->input->cur - ctxt->input->base < len) {
3115
        /*
3116
         * There were a couple of bugs where PERefs lead to to a change
3117
         * of the buffer. Check the buffer size to avoid passing an invalid
3118
         * pointer to xmlDictLookup.
3119
         */
3120
0
        xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR,
3121
0
                    "unexpected change of input buffer");
3122
0
        return (NULL);
3123
0
    }
3124
48.2k
    if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r'))
3125
304
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len);
3126
47.9k
    else
3127
47.9k
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len);
3128
48.2k
    if (ret == NULL)
3129
1
        xmlErrMemory(ctxt);
3130
48.2k
    return(ret);
3131
48.2k
}
3132
3133
/**
3134
 * Parse an XML name.
3135
 *
3136
 * @deprecated Internal function, don't use.
3137
 *
3138
 *     [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' |
3139
 *                      CombiningChar | Extender
3140
 *
3141
 *     [5] Name ::= (Letter | '_' | ':') (NameChar)*
3142
 *
3143
 *     [6] Names ::= Name (#x20 Name)*
3144
 *
3145
 * @param ctxt  an XML parser context
3146
 * @returns the Name parsed or NULL
3147
 */
3148
3149
const xmlChar *
3150
1.39M
xmlParseName(xmlParserCtxt *ctxt) {
3151
1.39M
    const xmlChar *in;
3152
1.39M
    const xmlChar *ret;
3153
1.39M
    size_t count = 0;
3154
1.39M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3155
582k
                       XML_MAX_TEXT_LENGTH :
3156
1.39M
                       XML_MAX_NAME_LENGTH;
3157
3158
1.39M
    GROW;
3159
3160
    /*
3161
     * Accelerator for simple ASCII names
3162
     */
3163
1.39M
    in = ctxt->input->cur;
3164
1.39M
    if (((*in >= 0x61) && (*in <= 0x7A)) ||
3165
234k
  ((*in >= 0x41) && (*in <= 0x5A)) ||
3166
1.20M
  (*in == '_') || (*in == ':')) {
3167
1.20M
  in++;
3168
15.4M
  while (((*in >= 0x61) && (*in <= 0x7A)) ||
3169
1.81M
         ((*in >= 0x41) && (*in <= 0x5A)) ||
3170
1.68M
         ((*in >= 0x30) && (*in <= 0x39)) ||
3171
1.33M
         (*in == '_') || (*in == '-') ||
3172
1.30M
         (*in == ':') || (*in == '.'))
3173
14.2M
      in++;
3174
1.20M
  if ((*in > 0) && (*in < 0x80)) {
3175
1.16M
      count = in - ctxt->input->cur;
3176
1.16M
            if (count > maxLength) {
3177
106
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3178
106
                return(NULL);
3179
106
            }
3180
1.16M
      ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count);
3181
1.16M
      ctxt->input->cur = in;
3182
1.16M
      ctxt->input->col += count;
3183
1.16M
      if (ret == NULL)
3184
1
          xmlErrMemory(ctxt);
3185
1.16M
      return(ret);
3186
1.16M
  }
3187
1.20M
    }
3188
    /* accelerator for special cases */
3189
234k
    return(xmlParseNameComplex(ctxt));
3190
1.39M
}
3191
3192
static xmlHashedString
3193
1.03M
xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) {
3194
1.03M
    xmlHashedString ret;
3195
1.03M
    int len = 0, l;
3196
1.03M
    int c;
3197
1.03M
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3198
357k
                    XML_MAX_TEXT_LENGTH :
3199
1.03M
                    XML_MAX_NAME_LENGTH;
3200
1.03M
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3201
1.03M
    size_t startPosition = 0;
3202
3203
1.03M
    ret.name = NULL;
3204
1.03M
    ret.hashValue = 0;
3205
3206
    /*
3207
     * Handler for more complex cases
3208
     */
3209
1.03M
    startPosition = CUR_PTR - BASE_PTR;
3210
1.03M
    c = xmlCurrentChar(ctxt, &l);
3211
1.03M
    if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */
3212
1.00M
  (!xmlIsNameStartChar(c, old10) || (c == ':'))) {
3213
936k
  return(ret);
3214
936k
    }
3215
3216
13.8M
    while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */
3217
13.8M
     (xmlIsNameChar(c, old10) && (c != ':'))) {
3218
13.8M
        if (len <= INT_MAX - l)
3219
13.8M
      len += l;
3220
13.8M
  NEXTL(l);
3221
13.8M
  c = xmlCurrentChar(ctxt, &l);
3222
13.8M
    }
3223
94.6k
    if (len > maxLength) {
3224
158
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3225
158
        return(ret);
3226
158
    }
3227
94.5k
    ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len);
3228
94.5k
    if (ret.name == NULL)
3229
2
        xmlErrMemory(ctxt);
3230
94.5k
    return(ret);
3231
94.6k
}
3232
3233
/**
3234
 * Parse an XML name.
3235
 *
3236
 *     [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' |
3237
 *                          CombiningChar | Extender
3238
 *
3239
 *     [5NS] NCName ::= (Letter | '_') (NCNameChar)*
3240
 *
3241
 * @param ctxt  an XML parser context
3242
 * @returns the Name parsed or NULL
3243
 */
3244
3245
static xmlHashedString
3246
2.76M
xmlParseNCName(xmlParserCtxtPtr ctxt) {
3247
2.76M
    const xmlChar *in, *e;
3248
2.76M
    xmlHashedString ret;
3249
2.76M
    size_t count = 0;
3250
2.76M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3251
1.12M
                       XML_MAX_TEXT_LENGTH :
3252
2.76M
                       XML_MAX_NAME_LENGTH;
3253
3254
2.76M
    ret.name = NULL;
3255
3256
    /*
3257
     * Accelerator for simple ASCII names
3258
     */
3259
2.76M
    in = ctxt->input->cur;
3260
2.76M
    e = ctxt->input->end;
3261
2.76M
    if ((((*in >= 0x61) && (*in <= 0x7A)) ||
3262
1.01M
   ((*in >= 0x41) && (*in <= 0x5A)) ||
3263
1.79M
   (*in == '_')) && (in < e)) {
3264
1.79M
  in++;
3265
8.10M
  while ((((*in >= 0x61) && (*in <= 0x7A)) ||
3266
3.14M
          ((*in >= 0x41) && (*in <= 0x5A)) ||
3267
3.03M
          ((*in >= 0x30) && (*in <= 0x39)) ||
3268
2.99M
          (*in == '_') || (*in == '-') ||
3269
6.30M
          (*in == '.')) && (in < e))
3270
6.30M
      in++;
3271
1.79M
  if (in >= e)
3272
2.44k
      goto complex;
3273
1.79M
  if ((*in > 0) && (*in < 0x80)) {
3274
1.73M
      count = in - ctxt->input->cur;
3275
1.73M
            if (count > maxLength) {
3276
21
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3277
21
                return(ret);
3278
21
            }
3279
1.73M
      ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count);
3280
1.73M
      ctxt->input->cur = in;
3281
1.73M
      ctxt->input->col += count;
3282
1.73M
      if (ret.name == NULL) {
3283
3
          xmlErrMemory(ctxt);
3284
3
      }
3285
1.73M
      return(ret);
3286
1.73M
  }
3287
1.79M
    }
3288
1.03M
complex:
3289
1.03M
    return(xmlParseNCNameComplex(ctxt));
3290
2.76M
}
3291
3292
/**
3293
 * Parse an XML name and compares for match
3294
 * (specialized for endtag parsing)
3295
 *
3296
 * @param ctxt  an XML parser context
3297
 * @param other  the name to compare with
3298
 * @returns NULL for an illegal name, (xmlChar*) 1 for success
3299
 * and the name for mismatch
3300
 */
3301
3302
static const xmlChar *
3303
68.0k
xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) {
3304
68.0k
    register const xmlChar *cmp = other;
3305
68.0k
    register const xmlChar *in;
3306
68.0k
    const xmlChar *ret;
3307
3308
68.0k
    GROW;
3309
3310
68.0k
    in = ctxt->input->cur;
3311
231k
    while (*in != 0 && *in == *cmp) {
3312
163k
  ++in;
3313
163k
  ++cmp;
3314
163k
    }
3315
68.0k
    if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
3316
  /* success */
3317
53.4k
  ctxt->input->col += in - ctxt->input->cur;
3318
53.4k
  ctxt->input->cur = in;
3319
53.4k
  return (const xmlChar*) 1;
3320
53.4k
    }
3321
    /* failure (or end of input buffer), check with full function */
3322
14.6k
    ret = xmlParseName (ctxt);
3323
    /* strings coming from the dictionary direct compare possible */
3324
14.6k
    if (ret == other) {
3325
1.58k
  return (const xmlChar*) 1;
3326
1.58k
    }
3327
13.0k
    return ret;
3328
14.6k
}
3329
3330
/**
3331
 * Parse an XML name.
3332
 *
3333
 * @param ctxt  an XML parser context
3334
 * @param str  a pointer to the string pointer (IN/OUT)
3335
 * @returns the Name parsed or NULL. The `str` pointer
3336
 * is updated to the current location in the string.
3337
 */
3338
3339
static xmlChar *
3340
833k
xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) {
3341
833k
    xmlChar *ret;
3342
833k
    const xmlChar *cur = *str;
3343
833k
    int flags = 0;
3344
833k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3345
533k
                    XML_MAX_TEXT_LENGTH :
3346
833k
                    XML_MAX_NAME_LENGTH;
3347
3348
833k
    if (ctxt->options & XML_PARSE_OLD10)
3349
541k
        flags |= XML_SCAN_OLD10;
3350
3351
833k
    cur = xmlScanName(*str, maxLength, flags);
3352
833k
    if (cur == NULL) {
3353
75
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3354
75
        return(NULL);
3355
75
    }
3356
833k
    if (cur == *str)
3357
3.42k
        return(NULL);
3358
3359
829k
    ret = xmlStrndup(*str, cur - *str);
3360
829k
    if (ret == NULL)
3361
17
        xmlErrMemory(ctxt);
3362
829k
    *str = cur;
3363
829k
    return(ret);
3364
833k
}
3365
3366
/**
3367
 * Parse an XML Nmtoken.
3368
 *
3369
 * @deprecated Internal function, don't use.
3370
 *
3371
 *     [7] Nmtoken ::= (NameChar)+
3372
 *
3373
 *     [8] Nmtokens ::= Nmtoken (#x20 Nmtoken)*
3374
 *
3375
 * @param ctxt  an XML parser context
3376
 * @returns the Nmtoken parsed or NULL
3377
 */
3378
3379
xmlChar *
3380
47.1k
xmlParseNmtoken(xmlParserCtxt *ctxt) {
3381
47.1k
    xmlChar buf[XML_MAX_NAMELEN + 5];
3382
47.1k
    xmlChar *ret;
3383
47.1k
    int len = 0, l;
3384
47.1k
    int c;
3385
47.1k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3386
20.7k
                    XML_MAX_TEXT_LENGTH :
3387
47.1k
                    XML_MAX_NAME_LENGTH;
3388
47.1k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3389
3390
47.1k
    c = xmlCurrentChar(ctxt, &l);
3391
3392
186k
    while (xmlIsNameChar(c, old10)) {
3393
139k
  COPY_BUF(buf, len, c);
3394
139k
  NEXTL(l);
3395
139k
  c = xmlCurrentChar(ctxt, &l);
3396
139k
  if (len >= XML_MAX_NAMELEN) {
3397
      /*
3398
       * Okay someone managed to make a huge token, so he's ready to pay
3399
       * for the processing speed.
3400
       */
3401
883
      xmlChar *buffer;
3402
883
      int max = len * 2;
3403
3404
883
      buffer = xmlMalloc(max);
3405
883
      if (buffer == NULL) {
3406
1
          xmlErrMemory(ctxt);
3407
1
    return(NULL);
3408
1
      }
3409
882
      memcpy(buffer, buf, len);
3410
7.34M
      while (xmlIsNameChar(c, old10)) {
3411
7.34M
    if (len + 10 > max) {
3412
4.27k
        xmlChar *tmp;
3413
4.27k
                    int newSize;
3414
3415
4.27k
                    newSize = xmlGrowCapacity(max, 1, 1, maxLength);
3416
4.27k
                    if (newSize < 0) {
3417
125
                        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3418
125
                        xmlFree(buffer);
3419
125
                        return(NULL);
3420
125
                    }
3421
4.14k
        tmp = xmlRealloc(buffer, newSize);
3422
4.14k
        if (tmp == NULL) {
3423
1
      xmlErrMemory(ctxt);
3424
1
      xmlFree(buffer);
3425
1
      return(NULL);
3426
1
        }
3427
4.14k
        buffer = tmp;
3428
4.14k
                    max = newSize;
3429
4.14k
    }
3430
7.34M
    COPY_BUF(buffer, len, c);
3431
7.34M
    NEXTL(l);
3432
7.34M
    c = xmlCurrentChar(ctxt, &l);
3433
7.34M
      }
3434
756
      buffer[len] = 0;
3435
756
      return(buffer);
3436
882
  }
3437
139k
    }
3438
46.2k
    if (len == 0)
3439
12.1k
        return(NULL);
3440
34.0k
    if (len > maxLength) {
3441
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3442
0
        return(NULL);
3443
0
    }
3444
34.0k
    ret = xmlStrndup(buf, len);
3445
34.0k
    if (ret == NULL)
3446
4
        xmlErrMemory(ctxt);
3447
34.0k
    return(ret);
3448
34.0k
}
3449
3450
/**
3451
 * Validate an entity value and expand parameter entities.
3452
 *
3453
 * @param ctxt  parser context
3454
 * @param buf  string buffer
3455
 * @param str  entity value
3456
 * @param length  size of entity value
3457
 * @param depth  nesting depth
3458
 */
3459
static void
3460
xmlExpandPEsInEntityValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf,
3461
92.6k
                          const xmlChar *str, int length, int depth) {
3462
92.6k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3463
92.6k
    const xmlChar *end, *chunk;
3464
92.6k
    int c, l;
3465
3466
92.6k
    if (str == NULL)
3467
14.3k
        return;
3468
3469
78.3k
    depth += 1;
3470
78.3k
    if (depth > maxDepth) {
3471
0
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3472
0
                       "Maximum entity nesting depth exceeded");
3473
0
  return;
3474
0
    }
3475
3476
78.3k
    end = str + length;
3477
78.3k
    chunk = str;
3478
3479
109M
    while ((str < end) && (!PARSER_STOPPED(ctxt))) {
3480
109M
        c = *str;
3481
3482
109M
        if (c >= 0x80) {
3483
55.0M
            l = xmlUTF8MultibyteLen(ctxt, str,
3484
55.0M
                    "invalid character in entity value\n");
3485
55.0M
            if (l == 0) {
3486
38.1M
                if (chunk < str)
3487
284k
                    xmlSBufAddString(buf, chunk, str - chunk);
3488
38.1M
                xmlSBufAddReplChar(buf);
3489
38.1M
                str += 1;
3490
38.1M
                chunk = str;
3491
38.1M
            } else {
3492
16.8M
                str += l;
3493
16.8M
            }
3494
55.0M
        } else if (c == '&') {
3495
38.9k
            if (str[1] == '#') {
3496
18.0k
                if (chunk < str)
3497
5.89k
                    xmlSBufAddString(buf, chunk, str - chunk);
3498
3499
18.0k
                c = xmlParseStringCharRef(ctxt, &str);
3500
18.0k
                if (c == 0)
3501
4.14k
                    return;
3502
3503
13.9k
                xmlSBufAddChar(buf, c);
3504
3505
13.9k
                chunk = str;
3506
20.9k
            } else {
3507
20.9k
                xmlChar *name;
3508
3509
                /*
3510
                 * General entity references are checked for
3511
                 * syntactic validity.
3512
                 */
3513
20.9k
                str++;
3514
20.9k
                name = xmlParseStringName(ctxt, &str);
3515
3516
20.9k
                if ((name == NULL) || (*str++ != ';')) {
3517
5.81k
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR,
3518
5.81k
                            "EntityValue: '&' forbidden except for entities "
3519
5.81k
                            "references\n");
3520
5.81k
                    xmlFree(name);
3521
5.81k
                    return;
3522
5.81k
                }
3523
3524
15.0k
                xmlFree(name);
3525
15.0k
            }
3526
54.5M
        } else if (c == '%') {
3527
32.2k
            xmlEntityPtr ent;
3528
3529
32.2k
            if (chunk < str)
3530
14.5k
                xmlSBufAddString(buf, chunk, str - chunk);
3531
3532
32.2k
            ent = xmlParseStringPEReference(ctxt, &str);
3533
32.2k
            if (ent == NULL)
3534
7.63k
                return;
3535
3536
24.6k
            if (!PARSER_EXTERNAL(ctxt)) {
3537
2.15k
                xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL);
3538
2.15k
                return;
3539
2.15k
            }
3540
3541
22.5k
            if (ent->content == NULL) {
3542
                /*
3543
                 * Note: external parsed entities will not be loaded,
3544
                 * it is not required for a non-validating parser to
3545
                 * complete external PEReferences coming from the
3546
                 * internal subset
3547
                 */
3548
14.4k
                if (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
3549
14.4k
                    ((ctxt->replaceEntities) ||
3550
14.2k
                     (ctxt->validate))) {
3551
14.2k
                    xmlLoadEntityContent(ctxt, ent);
3552
14.2k
                } else {
3553
199
                    xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING,
3554
199
                                  "not validating will not read content for "
3555
199
                                  "PE entity %s\n", ent->name, NULL);
3556
199
                }
3557
14.4k
            }
3558
3559
            /*
3560
             * TODO: Skip if ent->content is still NULL.
3561
             */
3562
3563
22.5k
            if (xmlParserEntityCheck(ctxt, ent->length))
3564
14
                return;
3565
3566
22.4k
            if (ent->flags & XML_ENT_EXPANDING) {
3567
52
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3568
52
                return;
3569
52
            }
3570
3571
22.4k
            ent->flags |= XML_ENT_EXPANDING;
3572
22.4k
            xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length,
3573
22.4k
                                      depth);
3574
22.4k
            ent->flags &= ~XML_ENT_EXPANDING;
3575
3576
22.4k
            chunk = str;
3577
54.4M
        } else {
3578
            /* Normal ASCII char */
3579
54.4M
            if (!IS_BYTE_CHAR(c)) {
3580
1.79M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3581
1.79M
                        "invalid character in entity value\n");
3582
1.79M
                if (chunk < str)
3583
6.13k
                    xmlSBufAddString(buf, chunk, str - chunk);
3584
1.79M
                xmlSBufAddReplChar(buf);
3585
1.79M
                str += 1;
3586
1.79M
                chunk = str;
3587
52.6M
            } else {
3588
52.6M
                str += 1;
3589
52.6M
            }
3590
54.4M
        }
3591
109M
    }
3592
3593
58.5k
    if (chunk < str)
3594
50.1k
        xmlSBufAddString(buf, chunk, str - chunk);
3595
58.5k
}
3596
3597
/**
3598
 * Parse a value for ENTITY declarations
3599
 *
3600
 * @deprecated Internal function, don't use.
3601
 *
3602
 *     [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' |
3603
 *                         "'" ([^%&'] | PEReference | Reference)* "'"
3604
 *
3605
 * @param ctxt  an XML parser context
3606
 * @param orig  if non-NULL store a copy of the original entity value
3607
 * @returns the EntityValue parsed with reference substituted or NULL
3608
 */
3609
xmlChar *
3610
73.6k
xmlParseEntityValue(xmlParserCtxt *ctxt, xmlChar **orig) {
3611
73.6k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3612
18.8k
                         XML_MAX_HUGE_LENGTH :
3613
73.6k
                         XML_MAX_TEXT_LENGTH;
3614
73.6k
    xmlSBuf buf;
3615
73.6k
    const xmlChar *start;
3616
73.6k
    int quote, length;
3617
3618
73.6k
    xmlSBufInit(&buf, maxLength);
3619
3620
73.6k
    GROW;
3621
3622
73.6k
    quote = CUR;
3623
73.6k
    if ((quote != '"') && (quote != '\'')) {
3624
0
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
3625
0
  return(NULL);
3626
0
    }
3627
73.6k
    CUR_PTR++;
3628
3629
73.6k
    length = 0;
3630
3631
    /*
3632
     * Copy raw content of the entity into a buffer
3633
     */
3634
48.8M
    while (1) {
3635
48.8M
        int c;
3636
3637
48.8M
        if (PARSER_STOPPED(ctxt))
3638
4
            goto error;
3639
3640
48.8M
        if (CUR_PTR >= ctxt->input->end) {
3641
3.40k
            xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL);
3642
3.40k
            goto error;
3643
3.40k
        }
3644
3645
48.8M
        c = CUR;
3646
3647
48.8M
        if (c == 0) {
3648
15
            xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3649
15
                    "invalid character in entity value\n");
3650
15
            goto error;
3651
15
        }
3652
48.8M
        if (c == quote)
3653
70.2k
            break;
3654
48.8M
        NEXTL(1);
3655
48.8M
        length += 1;
3656
3657
        /*
3658
         * TODO: Check growth threshold
3659
         */
3660
48.8M
        if (ctxt->input->end - CUR_PTR < 10)
3661
64.7k
            GROW;
3662
48.8M
    }
3663
3664
70.2k
    start = CUR_PTR - length;
3665
3666
70.2k
    if (orig != NULL) {
3667
70.2k
        *orig = xmlStrndup(start, length);
3668
70.2k
        if (*orig == NULL)
3669
4
            xmlErrMemory(ctxt);
3670
70.2k
    }
3671
3672
70.2k
    xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr);
3673
3674
70.2k
    NEXTL(1);
3675
3676
70.2k
    return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long"));
3677
3678
3.42k
error:
3679
3.42k
    xmlSBufCleanup(&buf, ctxt, "entity length too long");
3680
3.42k
    return(NULL);
3681
73.6k
}
3682
3683
/**
3684
 * Check an entity reference in an attribute value for validity
3685
 * without expanding it.
3686
 *
3687
 * @param ctxt  parser context
3688
 * @param pent  entity
3689
 * @param depth  nesting depth
3690
 */
3691
static void
3692
3.71k
xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) {
3693
3.71k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3694
3.71k
    const xmlChar *str;
3695
3.71k
    unsigned long expandedSize = pent->length;
3696
3.71k
    int c, flags;
3697
3698
3.71k
    depth += 1;
3699
3.71k
    if (depth > maxDepth) {
3700
1
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3701
1
                       "Maximum entity nesting depth exceeded");
3702
1
  return;
3703
1
    }
3704
3705
3.71k
    if (pent->flags & XML_ENT_EXPANDING) {
3706
14
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3707
14
        return;
3708
14
    }
3709
3710
    /*
3711
     * If we're parsing a default attribute value in DTD content,
3712
     * the entity might reference other entities which weren't
3713
     * defined yet, so the check isn't reliable.
3714
     */
3715
3.69k
    if (ctxt->inSubset == 0)
3716
3.63k
        flags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
3717
68
    else
3718
68
        flags = XML_ENT_VALIDATED;
3719
3720
3.69k
    str = pent->content;
3721
3.69k
    if (str == NULL)
3722
3
        goto done;
3723
3724
    /*
3725
     * Note that entity values are already validated. We only check
3726
     * for illegal less-than signs and compute the expanded size
3727
     * of the entity. No special handling for multi-byte characters
3728
     * is needed.
3729
     */
3730
45.0M
    while (!PARSER_STOPPED(ctxt)) {
3731
45.0M
        c = *str;
3732
3733
45.0M
  if (c != '&') {
3734
45.0M
            if (c == 0)
3735
3.55k
                break;
3736
3737
45.0M
            if (c == '<')
3738
2.18k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3739
2.18k
                        "'<' in entity '%s' is not allowed in attributes "
3740
2.18k
                        "values\n", pent->name);
3741
3742
45.0M
            str += 1;
3743
45.0M
        } else if (str[1] == '#') {
3744
1.30k
            int val;
3745
3746
1.30k
      val = xmlParseStringCharRef(ctxt, &str);
3747
1.30k
      if (val == 0) {
3748
11
                pent->content[0] = 0;
3749
11
                break;
3750
11
            }
3751
7.84k
  } else {
3752
7.84k
            xmlChar *name;
3753
7.84k
            xmlEntityPtr ent;
3754
3755
7.84k
      name = xmlParseStringEntityRef(ctxt, &str);
3756
7.84k
      if (name == NULL) {
3757
13
                pent->content[0] = 0;
3758
13
                break;
3759
13
            }
3760
3761
7.82k
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3762
7.82k
            xmlFree(name);
3763
3764
7.82k
            if ((ent != NULL) &&
3765
7.13k
                (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) {
3766
6.42k
                if ((ent->flags & flags) != flags) {
3767
2.98k
                    pent->flags |= XML_ENT_EXPANDING;
3768
2.98k
                    xmlCheckEntityInAttValue(ctxt, ent, depth);
3769
2.98k
                    pent->flags &= ~XML_ENT_EXPANDING;
3770
2.98k
                }
3771
3772
6.42k
                xmlSaturatedAdd(&expandedSize, ent->expandedSize);
3773
6.42k
                xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST);
3774
6.42k
            }
3775
7.82k
        }
3776
45.0M
    }
3777
3778
3.69k
done:
3779
3.69k
    if (ctxt->inSubset == 0)
3780
3.63k
        pent->expandedSize = expandedSize;
3781
3782
3.69k
    pent->flags |= flags;
3783
3.69k
}
3784
3785
/**
3786
 * Expand general entity references in an entity or attribute value.
3787
 * Perform attribute value normalization.
3788
 *
3789
 * @param ctxt  parser context
3790
 * @param buf  string buffer
3791
 * @param str  entity or attribute value
3792
 * @param pent  entity for entity value, NULL for attribute values
3793
 * @param normalize  whether to collapse whitespace
3794
 * @param inSpace  whitespace state
3795
 * @param depth  nesting depth
3796
 * @param check  whether to check for amplification
3797
 * @returns  whether there was a normalization change
3798
 */
3799
static int
3800
xmlExpandEntityInAttValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf,
3801
                          const xmlChar *str, xmlEntityPtr pent, int normalize,
3802
678k
                          int *inSpace, int depth, int check) {
3803
678k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3804
678k
    int c, chunkSize;
3805
678k
    int normChange = 0;
3806
3807
678k
    if (str == NULL)
3808
194
        return(0);
3809
3810
678k
    depth += 1;
3811
678k
    if (depth > maxDepth) {
3812
1
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3813
1
                       "Maximum entity nesting depth exceeded");
3814
1
  return(0);
3815
1
    }
3816
3817
678k
    if (pent != NULL) {
3818
678k
        if (pent->flags & XML_ENT_EXPANDING) {
3819
5
            xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3820
5
            return(0);
3821
5
        }
3822
3823
678k
        if (check) {
3824
678k
            if (xmlParserEntityCheck(ctxt, pent->length))
3825
48
                return(0);
3826
678k
        }
3827
678k
    }
3828
3829
678k
    chunkSize = 0;
3830
3831
    /*
3832
     * Note that entity values are already validated. No special
3833
     * handling for multi-byte characters is needed.
3834
     */
3835
200M
    while (!PARSER_STOPPED(ctxt)) {
3836
200M
        c = *str;
3837
3838
200M
  if (c != '&') {
3839
199M
            if (c == 0)
3840
661k
                break;
3841
3842
            /*
3843
             * If this function is called without an entity, it is used to
3844
             * expand entities in an attribute content where less-than was
3845
             * already unscaped and is allowed.
3846
             */
3847
198M
            if ((pent != NULL) && (c == '<')) {
3848
15.8k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3849
15.8k
                        "'<' in entity '%s' is not allowed in attributes "
3850
15.8k
                        "values\n", pent->name);
3851
15.8k
                break;
3852
15.8k
            }
3853
3854
198M
            if (c <= 0x20) {
3855
2.65M
                if ((normalize) && (*inSpace)) {
3856
                    /* Skip char */
3857
109k
                    if (chunkSize > 0) {
3858
105k
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3859
105k
                        chunkSize = 0;
3860
105k
                    }
3861
109k
                    normChange = 1;
3862
2.54M
                } else if (c < 0x20) {
3863
317k
                    if (chunkSize > 0) {
3864
196k
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3865
196k
                        chunkSize = 0;
3866
196k
                    }
3867
3868
317k
                    xmlSBufAddCString(buf, " ", 1);
3869
2.22M
                } else {
3870
2.22M
                    chunkSize += 1;
3871
2.22M
                }
3872
3873
2.65M
                *inSpace = 1;
3874
196M
            } else {
3875
196M
                chunkSize += 1;
3876
196M
                *inSpace = 0;
3877
196M
            }
3878
3879
198M
            str += 1;
3880
198M
        } else if (str[1] == '#') {
3881
116k
            int val;
3882
3883
116k
            if (chunkSize > 0) {
3884
114k
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3885
114k
                chunkSize = 0;
3886
114k
            }
3887
3888
116k
      val = xmlParseStringCharRef(ctxt, &str);
3889
116k
      if (val == 0) {
3890
11
                if (pent != NULL)
3891
11
                    pent->content[0] = 0;
3892
11
                break;
3893
11
            }
3894
3895
116k
            if (val == ' ') {
3896
2.87k
                if ((normalize) && (*inSpace))
3897
214
                    normChange = 1;
3898
2.65k
                else
3899
2.65k
                    xmlSBufAddCString(buf, " ", 1);
3900
2.87k
                *inSpace = 1;
3901
113k
            } else {
3902
113k
                xmlSBufAddChar(buf, val);
3903
113k
                *inSpace = 0;
3904
113k
            }
3905
772k
  } else {
3906
772k
            xmlChar *name;
3907
772k
            xmlEntityPtr ent;
3908
3909
772k
            if (chunkSize > 0) {
3910
274k
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3911
274k
                chunkSize = 0;
3912
274k
            }
3913
3914
772k
      name = xmlParseStringEntityRef(ctxt, &str);
3915
772k
            if (name == NULL) {
3916
22
                if (pent != NULL)
3917
22
                    pent->content[0] = 0;
3918
22
                break;
3919
22
            }
3920
3921
772k
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3922
772k
            xmlFree(name);
3923
3924
772k
      if ((ent != NULL) &&
3925
653k
    (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
3926
193k
    if (ent->content == NULL) {
3927
0
        xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
3928
0
          "predefined entity has no content\n");
3929
0
                    break;
3930
0
                }
3931
3932
193k
                xmlSBufAddString(buf, ent->content, ent->length);
3933
3934
193k
                *inSpace = 0;
3935
579k
      } else if ((ent != NULL) && (ent->content != NULL)) {
3936
459k
                if (pent != NULL)
3937
459k
                    pent->flags |= XML_ENT_EXPANDING;
3938
459k
    normChange |= xmlExpandEntityInAttValue(ctxt, buf,
3939
459k
                        ent->content, ent, normalize, inSpace, depth, check);
3940
459k
                if (pent != NULL)
3941
459k
                    pent->flags &= ~XML_ENT_EXPANDING;
3942
459k
      }
3943
772k
        }
3944
200M
    }
3945
3946
678k
    if (chunkSize > 0)
3947
218k
        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3948
3949
678k
    return(normChange);
3950
678k
}
3951
3952
/**
3953
 * Expand general entity references in an entity or attribute value.
3954
 * Perform attribute value normalization.
3955
 *
3956
 * @param ctxt  parser context
3957
 * @param str  entity or attribute value
3958
 * @param normalize  whether to collapse whitespace
3959
 * @returns the expanded attribtue value.
3960
 */
3961
xmlChar *
3962
xmlExpandEntitiesInAttValue(xmlParserCtxt *ctxt, const xmlChar *str,
3963
0
                            int normalize) {
3964
0
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3965
0
                         XML_MAX_HUGE_LENGTH :
3966
0
                         XML_MAX_TEXT_LENGTH;
3967
0
    xmlSBuf buf;
3968
0
    int inSpace = 1;
3969
3970
0
    xmlSBufInit(&buf, maxLength);
3971
3972
0
    xmlExpandEntityInAttValue(ctxt, &buf, str, NULL, normalize, &inSpace,
3973
0
                              ctxt->inputNr, /* check */ 0);
3974
3975
0
    if ((normalize) && (inSpace) && (buf.size > 0))
3976
0
        buf.size--;
3977
3978
0
    return(xmlSBufFinish(&buf, NULL, ctxt, "AttValue length too long"));
3979
0
}
3980
3981
/**
3982
 * Parse a value for an attribute.
3983
 *
3984
 * NOTE: if no normalization is needed, the routine will return pointers
3985
 * directly from the data buffer.
3986
 *
3987
 * 3.3.3 Attribute-Value Normalization:
3988
 *
3989
 * Before the value of an attribute is passed to the application or
3990
 * checked for validity, the XML processor must normalize it as follows:
3991
 *
3992
 * - a character reference is processed by appending the referenced
3993
 *   character to the attribute value
3994
 * - an entity reference is processed by recursively processing the
3995
 *   replacement text of the entity
3996
 * - a whitespace character (\#x20, \#xD, \#xA, \#x9) is processed by
3997
 *   appending \#x20 to the normalized value, except that only a single
3998
 *   \#x20 is appended for a "#xD#xA" sequence that is part of an external
3999
 *   parsed entity or the literal entity value of an internal parsed entity
4000
 * - other characters are processed by appending them to the normalized value
4001
 *
4002
 * If the declared value is not CDATA, then the XML processor must further
4003
 * process the normalized attribute value by discarding any leading and
4004
 * trailing space (\#x20) characters, and by replacing sequences of space
4005
 * (\#x20) characters by a single space (\#x20) character.
4006
 * All attributes for which no declaration has been read should be treated
4007
 * by a non-validating parser as if declared CDATA.
4008
 *
4009
 * @param ctxt  an XML parser context
4010
 * @param attlen  attribute len result
4011
 * @param outFlags  resulting XML_ATTVAL_* flags
4012
 * @param special  value from attsSpecial
4013
 * @param isNamespace  whether this is a namespace declaration
4014
 * @returns the AttValue parsed or NULL. The value has to be freed by the
4015
 *     caller if it was copied, this can be detected by val[*len] == 0.
4016
 */
4017
static xmlChar *
4018
xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *attlen, int *outFlags,
4019
333k
                         int special, int isNamespace) {
4020
333k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4021
148k
                         XML_MAX_HUGE_LENGTH :
4022
333k
                         XML_MAX_TEXT_LENGTH;
4023
333k
    xmlSBuf buf;
4024
333k
    xmlChar *ret;
4025
333k
    int c, l, quote, entFlags, chunkSize;
4026
333k
    int inSpace = 1;
4027
333k
    int replaceEntities;
4028
333k
    int normalize = (special & XML_SPECIAL_TYPE_MASK) > XML_ATTRIBUTE_CDATA;
4029
333k
    int attvalFlags = 0;
4030
4031
    /* Always expand namespace URIs */
4032
333k
    replaceEntities = (ctxt->replaceEntities) || (isNamespace);
4033
4034
333k
    xmlSBufInit(&buf, maxLength);
4035
4036
333k
    GROW;
4037
4038
333k
    quote = CUR;
4039
333k
    if ((quote != '"') && (quote != '\'')) {
4040
3.80k
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
4041
3.80k
  return(NULL);
4042
3.80k
    }
4043
329k
    NEXTL(1);
4044
4045
329k
    if (ctxt->inSubset == 0)
4046
295k
        entFlags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
4047
34.8k
    else
4048
34.8k
        entFlags = XML_ENT_VALIDATED;
4049
4050
329k
    inSpace = 1;
4051
329k
    chunkSize = 0;
4052
4053
117M
    while (1) {
4054
117M
        if (PARSER_STOPPED(ctxt))
4055
139
            goto error;
4056
4057
117M
        if (CUR_PTR >= ctxt->input->end) {
4058
11.0k
            xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED,
4059
11.0k
                           "AttValue: ' expected\n");
4060
11.0k
            goto error;
4061
11.0k
        }
4062
4063
        /*
4064
         * TODO: Check growth threshold
4065
         */
4066
117M
        if (ctxt->input->end - CUR_PTR < 10)
4067
154k
            GROW;
4068
4069
117M
        c = CUR;
4070
4071
117M
        if (c >= 0x80) {
4072
54.2M
            l = xmlUTF8MultibyteLen(ctxt, CUR_PTR,
4073
54.2M
                    "invalid character in attribute value\n");
4074
54.2M
            if (l == 0) {
4075
43.5M
                if (chunkSize > 0) {
4076
161k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4077
161k
                    chunkSize = 0;
4078
161k
                }
4079
43.5M
                xmlSBufAddReplChar(&buf);
4080
43.5M
                NEXTL(1);
4081
43.5M
            } else {
4082
10.7M
                chunkSize += l;
4083
10.7M
                NEXTL(l);
4084
10.7M
            }
4085
4086
54.2M
            inSpace = 0;
4087
63.4M
        } else if (c != '&') {
4088
62.7M
            if (c > 0x20) {
4089
33.2M
                if (c == quote)
4090
317k
                    break;
4091
4092
32.9M
                if (c == '<')
4093
494k
                    xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL);
4094
4095
32.9M
                chunkSize += 1;
4096
32.9M
                inSpace = 0;
4097
32.9M
            } else if (!IS_BYTE_CHAR(c)) {
4098
20.0M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
4099
20.0M
                        "invalid character in attribute value\n");
4100
20.0M
                if (chunkSize > 0) {
4101
57.2k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4102
57.2k
                    chunkSize = 0;
4103
57.2k
                }
4104
20.0M
                xmlSBufAddReplChar(&buf);
4105
20.0M
                inSpace = 0;
4106
20.0M
            } else {
4107
                /* Whitespace */
4108
9.43M
                if ((normalize) && (inSpace)) {
4109
                    /* Skip char */
4110
512k
                    if (chunkSize > 0) {
4111
8.18k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4112
8.18k
                        chunkSize = 0;
4113
8.18k
                    }
4114
512k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4115
8.92M
                } else if (c < 0x20) {
4116
                    /* Convert to space */
4117
7.87M
                    if (chunkSize > 0) {
4118
258k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4119
258k
                        chunkSize = 0;
4120
258k
                    }
4121
4122
7.87M
                    xmlSBufAddCString(&buf, " ", 1);
4123
7.87M
                } else {
4124
1.04M
                    chunkSize += 1;
4125
1.04M
                }
4126
4127
9.43M
                inSpace = 1;
4128
4129
9.43M
                if ((c == 0xD) && (NXT(1) == 0xA))
4130
7.65k
                    CUR_PTR++;
4131
9.43M
            }
4132
4133
62.4M
            NEXTL(1);
4134
62.4M
        } else if (NXT(1) == '#') {
4135
58.5k
            int val;
4136
4137
58.5k
            if (chunkSize > 0) {
4138
42.5k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4139
42.5k
                chunkSize = 0;
4140
42.5k
            }
4141
4142
58.5k
            val = xmlParseCharRef(ctxt);
4143
58.5k
            if (val == 0)
4144
1.15k
                goto error;
4145
4146
57.4k
            if ((val == '&') && (!replaceEntities)) {
4147
                /*
4148
                 * The reparsing will be done in xmlNodeParseContent()
4149
                 * called from SAX2.c
4150
                 */
4151
1.83k
                xmlSBufAddCString(&buf, "&#38;", 5);
4152
1.83k
                inSpace = 0;
4153
55.5k
            } else if (val == ' ') {
4154
22.4k
                if ((normalize) && (inSpace))
4155
1.91k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4156
20.5k
                else
4157
20.5k
                    xmlSBufAddCString(&buf, " ", 1);
4158
22.4k
                inSpace = 1;
4159
33.1k
            } else {
4160
33.1k
                xmlSBufAddChar(&buf, val);
4161
33.1k
                inSpace = 0;
4162
33.1k
            }
4163
613k
        } else {
4164
613k
            const xmlChar *name;
4165
613k
            xmlEntityPtr ent;
4166
4167
613k
            if (chunkSize > 0) {
4168
214k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4169
214k
                chunkSize = 0;
4170
214k
            }
4171
4172
613k
            name = xmlParseEntityRefInternal(ctxt);
4173
613k
            if (name == NULL) {
4174
                /*
4175
                 * Probably a literal '&' which wasn't escaped.
4176
                 * TODO: Handle gracefully in recovery mode.
4177
                 */
4178
189k
                continue;
4179
189k
            }
4180
4181
424k
            ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1);
4182
424k
            if (ent == NULL)
4183
125k
                continue;
4184
4185
299k
            if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) {
4186
8.56k
                if ((ent->content[0] == '&') && (!replaceEntities))
4187
1.71k
                    xmlSBufAddCString(&buf, "&#38;", 5);
4188
6.84k
                else
4189
6.84k
                    xmlSBufAddString(&buf, ent->content, ent->length);
4190
8.56k
                inSpace = 0;
4191
290k
            } else if (replaceEntities) {
4192
218k
                if (xmlExpandEntityInAttValue(ctxt, &buf,
4193
218k
                        ent->content, ent, normalize, &inSpace, ctxt->inputNr,
4194
218k
                        /* check */ 1) > 0)
4195
104k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4196
218k
            } else {
4197
72.2k
                if ((ent->flags & entFlags) != entFlags)
4198
733
                    xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr);
4199
4200
72.2k
                if (xmlParserEntityCheck(ctxt, ent->expandedSize)) {
4201
109
                    ent->content[0] = 0;
4202
109
                    goto error;
4203
109
                }
4204
4205
                /*
4206
                 * Just output the reference
4207
                 */
4208
72.1k
                xmlSBufAddCString(&buf, "&", 1);
4209
72.1k
                xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name));
4210
72.1k
                xmlSBufAddCString(&buf, ";", 1);
4211
4212
72.1k
                inSpace = 0;
4213
72.1k
            }
4214
299k
  }
4215
117M
    }
4216
4217
317k
    if ((buf.mem == NULL) && (outFlags != NULL)) {
4218
239k
        ret = (xmlChar *) CUR_PTR - chunkSize;
4219
4220
239k
        if (attlen != NULL)
4221
239k
            *attlen = chunkSize;
4222
239k
        if ((normalize) && (inSpace) && (chunkSize > 0)) {
4223
230
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4224
230
            *attlen -= 1;
4225
230
        }
4226
4227
        /* Report potential error */
4228
239k
        xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4229
239k
    } else {
4230
78.3k
        if (chunkSize > 0)
4231
59.3k
            xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4232
4233
78.3k
        if ((normalize) && (inSpace) && (buf.size > 0)) {
4234
1.05k
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4235
1.05k
            buf.size--;
4236
1.05k
        }
4237
4238
78.3k
        ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long");
4239
78.3k
        attvalFlags |= XML_ATTVAL_ALLOC;
4240
4241
78.3k
        if (ret != NULL) {
4242
78.3k
            if (attlen != NULL)
4243
53.2k
                *attlen = buf.size;
4244
78.3k
        }
4245
78.3k
    }
4246
4247
317k
    if (outFlags != NULL)
4248
292k
        *outFlags = attvalFlags;
4249
4250
317k
    NEXTL(1);
4251
4252
317k
    return(ret);
4253
4254
12.4k
error:
4255
12.4k
    xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4256
12.4k
    return(NULL);
4257
329k
}
4258
4259
/**
4260
 * Parse a value for an attribute
4261
 * Note: the parser won't do substitution of entities here, this
4262
 * will be handled later in #xmlStringGetNodeList
4263
 *
4264
 * @deprecated Internal function, don't use.
4265
 *
4266
 *     [10] AttValue ::= '"' ([^<&"] | Reference)* '"' |
4267
 *                       "'" ([^<&'] | Reference)* "'"
4268
 *
4269
 * 3.3.3 Attribute-Value Normalization:
4270
 *
4271
 * Before the value of an attribute is passed to the application or
4272
 * checked for validity, the XML processor must normalize it as follows:
4273
 *
4274
 * - a character reference is processed by appending the referenced
4275
 *   character to the attribute value
4276
 * - an entity reference is processed by recursively processing the
4277
 *   replacement text of the entity
4278
 * - a whitespace character (\#x20, \#xD, \#xA, \#x9) is processed by
4279
 *   appending \#x20 to the normalized value, except that only a single
4280
 *   \#x20 is appended for a "#xD#xA" sequence that is part of an external
4281
 *   parsed entity or the literal entity value of an internal parsed entity
4282
 * - other characters are processed by appending them to the normalized value
4283
 *
4284
 * If the declared value is not CDATA, then the XML processor must further
4285
 * process the normalized attribute value by discarding any leading and
4286
 * trailing space (\#x20) characters, and by replacing sequences of space
4287
 * (\#x20) characters by a single space (\#x20) character.
4288
 * All attributes for which no declaration has been read should be treated
4289
 * by a non-validating parser as if declared CDATA.
4290
 *
4291
 * @param ctxt  an XML parser context
4292
 * @returns the AttValue parsed or NULL. The value has to be freed by the
4293
 * caller.
4294
 */
4295
xmlChar *
4296
35.7k
xmlParseAttValue(xmlParserCtxt *ctxt) {
4297
35.7k
    if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL);
4298
35.7k
    return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0));
4299
35.7k
}
4300
4301
/**
4302
 * Parse an XML Literal
4303
 *
4304
 * @deprecated Internal function, don't use.
4305
 *
4306
 *     [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'")
4307
 *
4308
 * @param ctxt  an XML parser context
4309
 * @returns the SystemLiteral parsed or NULL
4310
 */
4311
4312
xmlChar *
4313
27.6k
xmlParseSystemLiteral(xmlParserCtxt *ctxt) {
4314
27.6k
    xmlChar *buf = NULL;
4315
27.6k
    int len = 0;
4316
27.6k
    int size = XML_PARSER_BUFFER_SIZE;
4317
27.6k
    int cur, l;
4318
27.6k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4319
9.00k
                    XML_MAX_TEXT_LENGTH :
4320
27.6k
                    XML_MAX_NAME_LENGTH;
4321
27.6k
    xmlChar stop;
4322
4323
27.6k
    if (RAW == '"') {
4324
11.8k
        NEXT;
4325
11.8k
  stop = '"';
4326
15.7k
    } else if (RAW == '\'') {
4327
14.2k
        NEXT;
4328
14.2k
  stop = '\'';
4329
14.2k
    } else {
4330
1.57k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4331
1.57k
  return(NULL);
4332
1.57k
    }
4333
4334
26.0k
    buf = xmlMalloc(size);
4335
26.0k
    if (buf == NULL) {
4336
2
        xmlErrMemory(ctxt);
4337
2
  return(NULL);
4338
2
    }
4339
26.0k
    cur = xmlCurrentCharRecover(ctxt, &l);
4340
2.73M
    while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */
4341
2.70M
  if (len + 5 >= size) {
4342
1.45k
      xmlChar *tmp;
4343
1.45k
            int newSize;
4344
4345
1.45k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4346
1.45k
            if (newSize < 0) {
4347
8
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral");
4348
8
                xmlFree(buf);
4349
8
                return(NULL);
4350
8
            }
4351
1.44k
      tmp = xmlRealloc(buf, newSize);
4352
1.44k
      if (tmp == NULL) {
4353
3
          xmlFree(buf);
4354
3
    xmlErrMemory(ctxt);
4355
3
    return(NULL);
4356
3
      }
4357
1.44k
      buf = tmp;
4358
1.44k
            size = newSize;
4359
1.44k
  }
4360
2.70M
  COPY_BUF(buf, len, cur);
4361
2.70M
  NEXTL(l);
4362
2.70M
  cur = xmlCurrentCharRecover(ctxt, &l);
4363
2.70M
    }
4364
26.0k
    buf[len] = 0;
4365
26.0k
    if (!IS_CHAR(cur)) {
4366
1.98k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4367
24.0k
    } else {
4368
24.0k
  NEXT;
4369
24.0k
    }
4370
26.0k
    return(buf);
4371
26.0k
}
4372
4373
/**
4374
 * Parse an XML public literal
4375
 *
4376
 * @deprecated Internal function, don't use.
4377
 *
4378
 *     [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'"
4379
 *
4380
 * @param ctxt  an XML parser context
4381
 * @returns the PubidLiteral parsed or NULL.
4382
 */
4383
4384
xmlChar *
4385
7.23k
xmlParsePubidLiteral(xmlParserCtxt *ctxt) {
4386
7.23k
    xmlChar *buf = NULL;
4387
7.23k
    int len = 0;
4388
7.23k
    int size = XML_PARSER_BUFFER_SIZE;
4389
7.23k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4390
3.62k
                    XML_MAX_TEXT_LENGTH :
4391
7.23k
                    XML_MAX_NAME_LENGTH;
4392
7.23k
    xmlChar cur;
4393
7.23k
    xmlChar stop;
4394
4395
7.23k
    if (RAW == '"') {
4396
262
        NEXT;
4397
262
  stop = '"';
4398
6.96k
    } else if (RAW == '\'') {
4399
6.21k
        NEXT;
4400
6.21k
  stop = '\'';
4401
6.21k
    } else {
4402
751
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4403
751
  return(NULL);
4404
751
    }
4405
6.47k
    buf = xmlMalloc(size);
4406
6.47k
    if (buf == NULL) {
4407
2
  xmlErrMemory(ctxt);
4408
2
  return(NULL);
4409
2
    }
4410
6.47k
    cur = CUR;
4411
108k
    while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) &&
4412
102k
           (PARSER_STOPPED(ctxt) == 0)) { /* checked */
4413
102k
  if (len + 1 >= size) {
4414
372
      xmlChar *tmp;
4415
372
            int newSize;
4416
4417
372
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4418
372
            if (newSize < 0) {
4419
1
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID");
4420
1
                xmlFree(buf);
4421
1
                return(NULL);
4422
1
            }
4423
371
      tmp = xmlRealloc(buf, newSize);
4424
371
      if (tmp == NULL) {
4425
1
    xmlErrMemory(ctxt);
4426
1
    xmlFree(buf);
4427
1
    return(NULL);
4428
1
      }
4429
370
      buf = tmp;
4430
370
            size = newSize;
4431
370
  }
4432
102k
  buf[len++] = cur;
4433
102k
  NEXT;
4434
102k
  cur = CUR;
4435
102k
    }
4436
6.47k
    buf[len] = 0;
4437
6.47k
    if (cur != stop) {
4438
2.18k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4439
4.28k
    } else {
4440
4.28k
  NEXTL(1);
4441
4.28k
    }
4442
6.47k
    return(buf);
4443
6.47k
}
4444
4445
static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial);
4446
4447
/*
4448
 * used for the test in the inner loop of the char data testing
4449
 */
4450
static const unsigned char test_char_data[256] = {
4451
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4452
    0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */
4453
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4454
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4455
    0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */
4456
    0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F,
4457
    0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37,
4458
    0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */
4459
    0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47,
4460
    0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F,
4461
    0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57,
4462
    0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */
4463
    0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67,
4464
    0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F,
4465
    0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77,
4466
    0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F,
4467
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */
4468
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4469
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4470
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4471
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4472
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4473
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4474
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4475
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4476
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4477
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4478
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4479
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4480
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4481
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4482
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
4483
};
4484
4485
static void
4486
xmlCharacters(xmlParserCtxtPtr ctxt, const xmlChar *buf, int size,
4487
1.02M
              int isBlank) {
4488
1.02M
    int checkBlanks;
4489
4490
1.02M
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
4491
134k
        return;
4492
4493
892k
    checkBlanks = (!ctxt->keepBlanks) ||
4494
522k
                  (ctxt->sax->ignorableWhitespace != ctxt->sax->characters);
4495
4496
    /*
4497
     * Calling areBlanks with only parts of a text node
4498
     * is fundamentally broken, making the NOBLANKS option
4499
     * essentially unusable.
4500
     */
4501
892k
    if ((checkBlanks) &&
4502
369k
        (areBlanks(ctxt, buf, size, isBlank))) {
4503
11.3k
        if ((ctxt->sax->ignorableWhitespace != NULL) &&
4504
11.3k
            (ctxt->keepBlanks))
4505
0
            ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size);
4506
880k
    } else {
4507
880k
        if (ctxt->sax->characters != NULL)
4508
880k
            ctxt->sax->characters(ctxt->userData, buf, size);
4509
4510
        /*
4511
         * The old code used to update this value for "complex" data
4512
         * even if checkBlanks was false. This was probably a bug.
4513
         */
4514
880k
        if ((checkBlanks) && (*ctxt->space == -1))
4515
43.8k
            *ctxt->space = -2;
4516
880k
    }
4517
892k
}
4518
4519
/**
4520
 * Parse character data. Always makes progress if the first char isn't
4521
 * '<' or '&'.
4522
 *
4523
 * The right angle bracket (>) may be represented using the string "&gt;",
4524
 * and must, for compatibility, be escaped using "&gt;" or a character
4525
 * reference when it appears in the string "]]>" in content, when that
4526
 * string is not marking the end of a CDATA section.
4527
 *
4528
 *     [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*)
4529
 * @param ctxt  an XML parser context
4530
 * @param partial  buffer may contain partial UTF-8 sequences
4531
 */
4532
static void
4533
2.64M
xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) {
4534
2.64M
    const xmlChar *in;
4535
2.64M
    int line = ctxt->input->line;
4536
2.64M
    int col = ctxt->input->col;
4537
2.64M
    int ccol;
4538
2.64M
    int terminate = 0;
4539
4540
2.64M
    GROW;
4541
    /*
4542
     * Accelerated common case where input don't need to be
4543
     * modified before passing it to the handler.
4544
     */
4545
2.64M
    in = ctxt->input->cur;
4546
2.65M
    do {
4547
2.73M
get_more_space:
4548
2.84M
        while (*in == 0x20) { in++; ctxt->input->col++; }
4549
2.73M
        if (*in == 0xA) {
4550
3.42M
            do {
4551
3.42M
                ctxt->input->line++; ctxt->input->col = 1;
4552
3.42M
                in++;
4553
3.42M
            } while (*in == 0xA);
4554
78.6k
            goto get_more_space;
4555
78.6k
        }
4556
2.65M
        if (*in == '<') {
4557
105k
            while (in > ctxt->input->cur) {
4558
52.6k
                const xmlChar *tmp = ctxt->input->cur;
4559
52.6k
                size_t nbchar = in - tmp;
4560
4561
52.6k
                if (nbchar > XML_MAX_ITEMS)
4562
0
                    nbchar = XML_MAX_ITEMS;
4563
52.6k
                ctxt->input->cur += nbchar;
4564
4565
52.6k
                xmlCharacters(ctxt, tmp, nbchar, 1);
4566
52.6k
            }
4567
52.6k
            return;
4568
52.6k
        }
4569
4570
2.70M
get_more:
4571
2.70M
        ccol = ctxt->input->col;
4572
13.8M
        while (test_char_data[*in]) {
4573
11.1M
            in++;
4574
11.1M
            ccol++;
4575
11.1M
        }
4576
2.70M
        ctxt->input->col = ccol;
4577
2.70M
        if (*in == 0xA) {
4578
2.06M
            do {
4579
2.06M
                ctxt->input->line++; ctxt->input->col = 1;
4580
2.06M
                in++;
4581
2.06M
            } while (*in == 0xA);
4582
69.8k
            goto get_more;
4583
69.8k
        }
4584
2.63M
        if (*in == ']') {
4585
27.7k
            size_t avail = ctxt->input->end - in;
4586
4587
27.7k
            if (partial && avail < 2) {
4588
0
                terminate = 1;
4589
0
                goto invoke_callback;
4590
0
            }
4591
27.7k
            if (in[1] == ']') {
4592
3.56k
                if (partial && avail < 3) {
4593
0
                    terminate = 1;
4594
0
                    goto invoke_callback;
4595
0
                }
4596
3.56k
                if (in[2] == '>')
4597
2.54k
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4598
3.56k
            }
4599
4600
27.7k
            in++;
4601
27.7k
            ctxt->input->col++;
4602
27.7k
            goto get_more;
4603
27.7k
        }
4604
4605
2.60M
invoke_callback:
4606
3.07M
        while (in > ctxt->input->cur) {
4607
470k
            const xmlChar *tmp = ctxt->input->cur;
4608
470k
            size_t nbchar = in - tmp;
4609
4610
470k
            if (nbchar > XML_MAX_ITEMS)
4611
0
                nbchar = XML_MAX_ITEMS;
4612
470k
            ctxt->input->cur += nbchar;
4613
4614
470k
            xmlCharacters(ctxt, tmp, nbchar, 0);
4615
4616
470k
            line = ctxt->input->line;
4617
470k
            col = ctxt->input->col;
4618
470k
        }
4619
2.60M
        ctxt->input->cur = in;
4620
2.60M
        if (*in == 0xD) {
4621
16.7k
            in++;
4622
16.7k
            if (*in == 0xA) {
4623
10.5k
                ctxt->input->cur = in;
4624
10.5k
                in++;
4625
10.5k
                ctxt->input->line++; ctxt->input->col = 1;
4626
10.5k
                continue; /* while */
4627
10.5k
            }
4628
6.20k
            in--;
4629
6.20k
        }
4630
2.59M
        if (*in == '<') {
4631
228k
            return;
4632
228k
        }
4633
2.36M
        if (*in == '&') {
4634
49.5k
            return;
4635
49.5k
        }
4636
2.31M
        if (terminate) {
4637
0
            return;
4638
0
        }
4639
2.31M
        SHRINK;
4640
2.31M
        GROW;
4641
2.31M
        in = ctxt->input->cur;
4642
2.32M
    } while (((*in >= 0x20) && (*in <= 0x7F)) ||
4643
2.31M
             (*in == 0x09) || (*in == 0x0a));
4644
2.31M
    ctxt->input->line = line;
4645
2.31M
    ctxt->input->col = col;
4646
2.31M
    xmlParseCharDataComplex(ctxt, partial);
4647
2.31M
}
4648
4649
/**
4650
 * Always makes progress if the first char isn't '<' or '&'.
4651
 *
4652
 * parse a CharData section.this is the fallback function
4653
 * of #xmlParseCharData when the parsing requires handling
4654
 * of non-ASCII characters.
4655
 *
4656
 * @param ctxt  an XML parser context
4657
 * @param partial  whether the input can end with truncated UTF-8
4658
 */
4659
static void
4660
2.31M
xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) {
4661
2.31M
    xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5];
4662
2.31M
    int nbchar = 0;
4663
2.31M
    int cur, l;
4664
4665
2.31M
    cur = xmlCurrentCharRecover(ctxt, &l);
4666
30.3M
    while ((cur != '<') && /* checked */
4667
30.1M
           (cur != '&') &&
4668
30.1M
     (IS_CHAR(cur))) {
4669
28.0M
        if (cur == ']') {
4670
13.2k
            size_t avail = ctxt->input->end - ctxt->input->cur;
4671
4672
13.2k
            if (partial && avail < 2)
4673
0
                break;
4674
13.2k
            if (NXT(1) == ']') {
4675
2.25k
                if (partial && avail < 3)
4676
0
                    break;
4677
2.25k
                if (NXT(2) == '>')
4678
1.43k
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4679
2.25k
            }
4680
13.2k
        }
4681
4682
28.0M
  COPY_BUF(buf, nbchar, cur);
4683
  /* move current position before possible calling of ctxt->sax->characters */
4684
28.0M
  NEXTL(l);
4685
28.0M
  if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) {
4686
209k
      buf[nbchar] = 0;
4687
4688
209k
            xmlCharacters(ctxt, buf, nbchar, 0);
4689
209k
      nbchar = 0;
4690
209k
            SHRINK;
4691
209k
  }
4692
28.0M
  cur = xmlCurrentCharRecover(ctxt, &l);
4693
28.0M
    }
4694
2.31M
    if (nbchar != 0) {
4695
293k
        buf[nbchar] = 0;
4696
4697
293k
        xmlCharacters(ctxt, buf, nbchar, 0);
4698
293k
    }
4699
    /*
4700
     * cur == 0 can mean
4701
     *
4702
     * - End of buffer.
4703
     * - An actual 0 character.
4704
     * - An incomplete UTF-8 sequence. This is allowed if partial is set.
4705
     */
4706
2.31M
    if (ctxt->input->cur < ctxt->input->end) {
4707
2.30M
        if ((cur == 0) && (CUR != 0)) {
4708
1.74k
            if (partial == 0) {
4709
1.74k
                xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4710
1.74k
                        "Incomplete UTF-8 sequence starting with %02X\n", CUR);
4711
1.74k
                NEXTL(1);
4712
1.74k
            }
4713
2.30M
        } else if ((cur != '<') && (cur != '&') && (cur != ']')) {
4714
            /* Generate the error and skip the offending character */
4715
2.12M
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4716
2.12M
                              "PCDATA invalid Char value %d\n", cur);
4717
2.12M
            NEXTL(l);
4718
2.12M
        }
4719
2.30M
    }
4720
2.31M
}
4721
4722
/**
4723
 * @deprecated Internal function, don't use.
4724
 * @param ctxt  an XML parser context
4725
 * @param cdata  unused
4726
 */
4727
void
4728
0
xmlParseCharData(xmlParserCtxt *ctxt, ATTRIBUTE_UNUSED int cdata) {
4729
0
    xmlParseCharDataInternal(ctxt, 0);
4730
0
}
4731
4732
/**
4733
 * Parse an External ID or a Public ID
4734
 *
4735
 * @deprecated Internal function, don't use.
4736
 *
4737
 * NOTE: Productions [75] and [83] interact badly since [75] can generate
4738
 * `'PUBLIC' S PubidLiteral S SystemLiteral`
4739
 *
4740
 *     [75] ExternalID ::= 'SYSTEM' S SystemLiteral
4741
 *                       | 'PUBLIC' S PubidLiteral S SystemLiteral
4742
 *
4743
 *     [83] PublicID ::= 'PUBLIC' S PubidLiteral
4744
 *
4745
 * @param ctxt  an XML parser context
4746
 * @param publicId  a xmlChar** receiving PubidLiteral
4747
 * @param strict  indicate whether we should restrict parsing to only
4748
 *          production [75], see NOTE below
4749
 * @returns the function returns SystemLiteral and in the second
4750
 *                case publicID receives PubidLiteral, is strict is off
4751
 *                it is possible to return NULL and have publicID set.
4752
 */
4753
4754
xmlChar *
4755
43.8k
xmlParseExternalID(xmlParserCtxt *ctxt, xmlChar **publicId, int strict) {
4756
43.8k
    xmlChar *URI = NULL;
4757
4758
43.8k
    *publicId = NULL;
4759
43.8k
    if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) {
4760
21.6k
        SKIP(6);
4761
21.6k
  if (SKIP_BLANKS == 0) {
4762
427
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4763
427
                     "Space required after 'SYSTEM'\n");
4764
427
  }
4765
21.6k
  URI = xmlParseSystemLiteral(ctxt);
4766
21.6k
  if (URI == NULL) {
4767
312
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4768
312
        }
4769
22.2k
    } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) {
4770
7.23k
        SKIP(6);
4771
7.23k
  if (SKIP_BLANKS == 0) {
4772
1.30k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4773
1.30k
        "Space required after 'PUBLIC'\n");
4774
1.30k
  }
4775
7.23k
  *publicId = xmlParsePubidLiteral(ctxt);
4776
7.23k
  if (*publicId == NULL) {
4777
755
      xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL);
4778
755
  }
4779
7.23k
  if (strict) {
4780
      /*
4781
       * We don't handle [83] so "S SystemLiteral" is required.
4782
       */
4783
5.61k
      if (SKIP_BLANKS == 0) {
4784
1.81k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4785
1.81k
      "Space required after the Public Identifier\n");
4786
1.81k
      }
4787
5.61k
  } else {
4788
      /*
4789
       * We handle [83] so we return immediately, if
4790
       * "S SystemLiteral" is not detected. We skip blanks if no
4791
             * system literal was found, but this is harmless since we must
4792
             * be at the end of a NotationDecl.
4793
       */
4794
1.61k
      if (SKIP_BLANKS == 0) return(NULL);
4795
409
      if ((CUR != '\'') && (CUR != '"')) return(NULL);
4796
409
  }
4797
5.95k
  URI = xmlParseSystemLiteral(ctxt);
4798
5.95k
  if (URI == NULL) {
4799
1.27k
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4800
1.27k
        }
4801
5.95k
    }
4802
42.5k
    return(URI);
4803
43.8k
}
4804
4805
/**
4806
 * Skip an XML (SGML) comment <!-- .... -->
4807
 *  The spec says that "For compatibility, the string "--" (double-hyphen)
4808
 *  must not occur within comments. "
4809
 * This is the slow routine in case the accelerator for ascii didn't work
4810
 *
4811
 *     [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->'
4812
 * @param ctxt  an XML parser context
4813
 * @param buf  the already parsed part of the buffer
4814
 * @param len  number of bytes in the buffer
4815
 * @param size  allocated size of the buffer
4816
 */
4817
static void
4818
xmlParseCommentComplex(xmlParserCtxtPtr ctxt, xmlChar *buf,
4819
27.3k
                       size_t len, size_t size) {
4820
27.3k
    int q, ql;
4821
27.3k
    int r, rl;
4822
27.3k
    int cur, l;
4823
27.3k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4824
8.86k
                    XML_MAX_HUGE_LENGTH :
4825
27.3k
                    XML_MAX_TEXT_LENGTH;
4826
4827
27.3k
    if (buf == NULL) {
4828
6.38k
        len = 0;
4829
6.38k
  size = XML_PARSER_BUFFER_SIZE;
4830
6.38k
  buf = xmlMalloc(size);
4831
6.38k
  if (buf == NULL) {
4832
2
      xmlErrMemory(ctxt);
4833
2
      return;
4834
2
  }
4835
6.38k
    }
4836
27.3k
    q = xmlCurrentCharRecover(ctxt, &ql);
4837
27.3k
    if (q == 0)
4838
3.70k
        goto not_terminated;
4839
23.6k
    if (!IS_CHAR(q)) {
4840
824
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4841
824
                          "xmlParseComment: invalid xmlChar value %d\n",
4842
824
                    q);
4843
824
  xmlFree (buf);
4844
824
  return;
4845
824
    }
4846
22.8k
    NEXTL(ql);
4847
22.8k
    r = xmlCurrentCharRecover(ctxt, &rl);
4848
22.8k
    if (r == 0)
4849
399
        goto not_terminated;
4850
22.4k
    if (!IS_CHAR(r)) {
4851
2.82k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4852
2.82k
                          "xmlParseComment: invalid xmlChar value %d\n",
4853
2.82k
                    r);
4854
2.82k
  xmlFree (buf);
4855
2.82k
  return;
4856
2.82k
    }
4857
19.6k
    NEXTL(rl);
4858
19.6k
    cur = xmlCurrentCharRecover(ctxt, &l);
4859
19.6k
    if (cur == 0)
4860
786
        goto not_terminated;
4861
10.6M
    while (IS_CHAR(cur) && /* checked */
4862
10.5M
           ((cur != '>') ||
4863
10.5M
      (r != '-') || (q != '-'))) {
4864
10.5M
  if ((r == '-') && (q == '-')) {
4865
160k
      xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL);
4866
160k
  }
4867
10.5M
  if (len + 5 >= size) {
4868
6.21k
      xmlChar *tmp;
4869
6.21k
            int newSize;
4870
4871
6.21k
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4872
6.21k
            if (newSize < 0) {
4873
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4874
0
                             "Comment too big found", NULL);
4875
0
                xmlFree (buf);
4876
0
                return;
4877
0
            }
4878
6.21k
      tmp = xmlRealloc(buf, newSize);
4879
6.21k
      if (tmp == NULL) {
4880
7
    xmlErrMemory(ctxt);
4881
7
    xmlFree(buf);
4882
7
    return;
4883
7
      }
4884
6.21k
      buf = tmp;
4885
6.21k
            size = newSize;
4886
6.21k
  }
4887
10.5M
  COPY_BUF(buf, len, q);
4888
4889
10.5M
  q = r;
4890
10.5M
  ql = rl;
4891
10.5M
  r = cur;
4892
10.5M
  rl = l;
4893
4894
10.5M
  NEXTL(l);
4895
10.5M
  cur = xmlCurrentCharRecover(ctxt, &l);
4896
4897
10.5M
    }
4898
18.8k
    buf[len] = 0;
4899
18.8k
    if (cur == 0) {
4900
3.08k
  xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4901
3.08k
                       "Comment not terminated \n<!--%.50s\n", buf);
4902
15.7k
    } else if (!IS_CHAR(cur)) {
4903
719
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4904
719
                          "xmlParseComment: invalid xmlChar value %d\n",
4905
719
                    cur);
4906
15.0k
    } else {
4907
15.0k
        NEXT;
4908
15.0k
  if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
4909
15.0k
      (!ctxt->disableSAX))
4910
12.0k
      ctxt->sax->comment(ctxt->userData, buf);
4911
15.0k
    }
4912
18.8k
    xmlFree(buf);
4913
18.8k
    return;
4914
4.88k
not_terminated:
4915
4.88k
    xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4916
4.88k
       "Comment not terminated\n", NULL);
4917
4.88k
    xmlFree(buf);
4918
4.88k
}
4919
4920
/**
4921
 * Parse an XML (SGML) comment. Always consumes '<!'.
4922
 *
4923
 * @deprecated Internal function, don't use.
4924
 *
4925
 *  The spec says that "For compatibility, the string "--" (double-hyphen)
4926
 *  must not occur within comments. "
4927
 *
4928
 *     [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->'
4929
 * @param ctxt  an XML parser context
4930
 */
4931
void
4932
747k
xmlParseComment(xmlParserCtxt *ctxt) {
4933
747k
    xmlChar *buf = NULL;
4934
747k
    size_t size = XML_PARSER_BUFFER_SIZE;
4935
747k
    size_t len = 0;
4936
747k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4937
25.5k
                       XML_MAX_HUGE_LENGTH :
4938
747k
                       XML_MAX_TEXT_LENGTH;
4939
747k
    const xmlChar *in;
4940
747k
    size_t nbchar = 0;
4941
747k
    int ccol;
4942
4943
    /*
4944
     * Check that there is a comment right here.
4945
     */
4946
747k
    if ((RAW != '<') || (NXT(1) != '!'))
4947
0
        return;
4948
747k
    SKIP(2);
4949
747k
    if ((RAW != '-') || (NXT(1) != '-'))
4950
27
        return;
4951
747k
    SKIP(2);
4952
747k
    GROW;
4953
4954
    /*
4955
     * Accelerated common case where input don't need to be
4956
     * modified before passing it to the handler.
4957
     */
4958
747k
    in = ctxt->input->cur;
4959
747k
    do {
4960
747k
  if (*in == 0xA) {
4961
8.04k
      do {
4962
8.04k
    ctxt->input->line++; ctxt->input->col = 1;
4963
8.04k
    in++;
4964
8.04k
      } while (*in == 0xA);
4965
2.93k
  }
4966
1.10M
get_more:
4967
1.10M
        ccol = ctxt->input->col;
4968
7.12M
  while (((*in > '-') && (*in <= 0x7F)) ||
4969
2.83M
         ((*in >= 0x20) && (*in < '-')) ||
4970
6.01M
         (*in == 0x09)) {
4971
6.01M
        in++;
4972
6.01M
        ccol++;
4973
6.01M
  }
4974
1.10M
  ctxt->input->col = ccol;
4975
1.10M
  if (*in == 0xA) {
4976
92.5k
      do {
4977
92.5k
    ctxt->input->line++; ctxt->input->col = 1;
4978
92.5k
    in++;
4979
92.5k
      } while (*in == 0xA);
4980
25.4k
      goto get_more;
4981
25.4k
  }
4982
1.07M
  nbchar = in - ctxt->input->cur;
4983
  /*
4984
   * save current set of data
4985
   */
4986
1.07M
  if (nbchar > 0) {
4987
358k
            if (nbchar > maxLength - len) {
4988
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4989
0
                                  "Comment too big found", NULL);
4990
0
                xmlFree(buf);
4991
0
                return;
4992
0
            }
4993
358k
            if (buf == NULL) {
4994
35.3k
                if ((*in == '-') && (in[1] == '-'))
4995
12.3k
                    size = nbchar + 1;
4996
23.0k
                else
4997
23.0k
                    size = XML_PARSER_BUFFER_SIZE + nbchar;
4998
35.3k
                buf = xmlMalloc(size);
4999
35.3k
                if (buf == NULL) {
5000
5
                    xmlErrMemory(ctxt);
5001
5
                    return;
5002
5
                }
5003
35.3k
                len = 0;
5004
323k
            } else if (len + nbchar + 1 >= size) {
5005
10.0k
                xmlChar *new_buf;
5006
10.0k
                size += len + nbchar + XML_PARSER_BUFFER_SIZE;
5007
10.0k
                new_buf = xmlRealloc(buf, size);
5008
10.0k
                if (new_buf == NULL) {
5009
2
                    xmlErrMemory(ctxt);
5010
2
                    xmlFree(buf);
5011
2
                    return;
5012
2
                }
5013
10.0k
                buf = new_buf;
5014
10.0k
            }
5015
358k
            memcpy(&buf[len], ctxt->input->cur, nbchar);
5016
358k
            len += nbchar;
5017
358k
            buf[len] = 0;
5018
358k
  }
5019
1.07M
  ctxt->input->cur = in;
5020
1.07M
  if (*in == 0xA) {
5021
0
      in++;
5022
0
      ctxt->input->line++; ctxt->input->col = 1;
5023
0
  }
5024
1.07M
  if (*in == 0xD) {
5025
18.0k
      in++;
5026
18.0k
      if (*in == 0xA) {
5027
15.1k
    ctxt->input->cur = in;
5028
15.1k
    in++;
5029
15.1k
    ctxt->input->line++; ctxt->input->col = 1;
5030
15.1k
    goto get_more;
5031
15.1k
      }
5032
2.82k
      in--;
5033
2.82k
  }
5034
1.06M
  SHRINK;
5035
1.06M
  GROW;
5036
1.06M
  in = ctxt->input->cur;
5037
1.06M
  if (*in == '-') {
5038
1.03M
      if (in[1] == '-') {
5039
980k
          if (in[2] == '>') {
5040
719k
        SKIP(3);
5041
719k
        if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
5042
719k
            (!ctxt->disableSAX)) {
5043
285k
      if (buf != NULL)
5044
12.6k
          ctxt->sax->comment(ctxt->userData, buf);
5045
272k
      else
5046
272k
          ctxt->sax->comment(ctxt->userData, BAD_CAST "");
5047
285k
        }
5048
719k
        if (buf != NULL)
5049
14.4k
            xmlFree(buf);
5050
719k
        return;
5051
719k
    }
5052
260k
    if (buf != NULL) {
5053
256k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5054
256k
                          "Double hyphen within comment: "
5055
256k
                                      "<!--%.50s\n",
5056
256k
              buf);
5057
256k
    } else
5058
3.57k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5059
3.57k
                          "Double hyphen within comment\n", NULL);
5060
260k
    in++;
5061
260k
    ctxt->input->col++;
5062
260k
      }
5063
316k
      in++;
5064
316k
      ctxt->input->col++;
5065
316k
      goto get_more;
5066
1.03M
  }
5067
1.06M
    } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a));
5068
27.3k
    xmlParseCommentComplex(ctxt, buf, len, size);
5069
27.3k
}
5070
5071
5072
/**
5073
 * Parse the name of a PI
5074
 *
5075
 * @deprecated Internal function, don't use.
5076
 *
5077
 *     [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l'))
5078
 *
5079
 * @param ctxt  an XML parser context
5080
 * @returns the PITarget name or NULL
5081
 */
5082
5083
const xmlChar *
5084
46.3k
xmlParsePITarget(xmlParserCtxt *ctxt) {
5085
46.3k
    const xmlChar *name;
5086
5087
46.3k
    name = xmlParseName(ctxt);
5088
46.3k
    if ((name != NULL) &&
5089
41.5k
        ((name[0] == 'x') || (name[0] == 'X')) &&
5090
29.0k
        ((name[1] == 'm') || (name[1] == 'M')) &&
5091
27.0k
        ((name[2] == 'l') || (name[2] == 'L'))) {
5092
22.5k
  int i;
5093
22.5k
  if ((name[0] == 'x') && (name[1] == 'm') &&
5094
21.1k
      (name[2] == 'l') && (name[3] == 0)) {
5095
19.6k
      xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5096
19.6k
     "XML declaration allowed only at the start of the document\n");
5097
19.6k
      return(name);
5098
19.6k
  } else if (name[3] == 0) {
5099
907
      xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL);
5100
907
      return(name);
5101
907
  }
5102
5.54k
  for (i = 0;;i++) {
5103
5.54k
      if (xmlW3CPIs[i] == NULL) break;
5104
3.82k
      if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i]))
5105
327
          return(name);
5106
3.82k
  }
5107
1.71k
  xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5108
1.71k
          "xmlParsePITarget: invalid name prefix 'xml'\n",
5109
1.71k
          NULL, NULL);
5110
1.71k
    }
5111
25.5k
    if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) {
5112
2.10k
  xmlNsErr(ctxt, XML_NS_ERR_COLON,
5113
2.10k
     "colons are forbidden from PI names '%s'\n", name, NULL, NULL);
5114
2.10k
    }
5115
25.5k
    return(name);
5116
46.3k
}
5117
5118
#ifdef LIBXML_CATALOG_ENABLED
5119
/**
5120
 * Parse an XML Catalog Processing Instruction.
5121
 *
5122
 * <?oasis-xml-catalog catalog="http://example.com/catalog.xml"?>
5123
 *
5124
 * Occurs only if allowed by the user and if happening in the Misc
5125
 * part of the document before any doctype information
5126
 * This will add the given catalog to the parsing context in order
5127
 * to be used if there is a resolution need further down in the document
5128
 *
5129
 * @param ctxt  an XML parser context
5130
 * @param catalog  the PI value string
5131
 */
5132
5133
static void
5134
0
xmlParseCatalogPI(xmlParserCtxtPtr ctxt, const xmlChar *catalog) {
5135
0
    xmlChar *URL = NULL;
5136
0
    const xmlChar *tmp, *base;
5137
0
    xmlChar marker;
5138
5139
0
    tmp = catalog;
5140
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5141
0
    if (xmlStrncmp(tmp, BAD_CAST"catalog", 7))
5142
0
  goto error;
5143
0
    tmp += 7;
5144
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5145
0
    if (*tmp != '=') {
5146
0
  return;
5147
0
    }
5148
0
    tmp++;
5149
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5150
0
    marker = *tmp;
5151
0
    if ((marker != '\'') && (marker != '"'))
5152
0
  goto error;
5153
0
    tmp++;
5154
0
    base = tmp;
5155
0
    while ((*tmp != 0) && (*tmp != marker)) tmp++;
5156
0
    if (*tmp == 0)
5157
0
  goto error;
5158
0
    URL = xmlStrndup(base, tmp - base);
5159
0
    tmp++;
5160
0
    while (IS_BLANK_CH(*tmp)) tmp++;
5161
0
    if (*tmp != 0)
5162
0
  goto error;
5163
5164
0
    if (URL != NULL) {
5165
        /*
5166
         * Unfortunately, the catalog API doesn't report OOM errors.
5167
         * xmlGetLastError isn't very helpful since we don't know
5168
         * where the last error came from. We'd have to reset it
5169
         * before this call and restore it afterwards.
5170
         */
5171
0
  ctxt->catalogs = xmlCatalogAddLocal(ctxt->catalogs, URL);
5172
0
  xmlFree(URL);
5173
0
    }
5174
0
    return;
5175
5176
0
error:
5177
0
    xmlWarningMsg(ctxt, XML_WAR_CATALOG_PI,
5178
0
            "Catalog PI syntax error: %s\n",
5179
0
      catalog, NULL);
5180
0
    if (URL != NULL)
5181
0
  xmlFree(URL);
5182
0
}
5183
#endif
5184
5185
/**
5186
 * Parse an XML Processing Instruction.
5187
 *
5188
 * @deprecated Internal function, don't use.
5189
 *
5190
 *     [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>'
5191
 *
5192
 * The processing is transferred to SAX once parsed.
5193
 *
5194
 * @param ctxt  an XML parser context
5195
 */
5196
5197
void
5198
46.3k
xmlParsePI(xmlParserCtxt *ctxt) {
5199
46.3k
    xmlChar *buf = NULL;
5200
46.3k
    size_t len = 0;
5201
46.3k
    size_t size = XML_PARSER_BUFFER_SIZE;
5202
46.3k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
5203
13.1k
                       XML_MAX_HUGE_LENGTH :
5204
46.3k
                       XML_MAX_TEXT_LENGTH;
5205
46.3k
    int cur, l;
5206
46.3k
    const xmlChar *target;
5207
5208
46.3k
    if ((RAW == '<') && (NXT(1) == '?')) {
5209
  /*
5210
   * this is a Processing Instruction.
5211
   */
5212
46.3k
  SKIP(2);
5213
5214
  /*
5215
   * Parse the target name and check for special support like
5216
   * namespace.
5217
   */
5218
46.3k
        target = xmlParsePITarget(ctxt);
5219
46.3k
  if (target != NULL) {
5220
41.5k
      if ((RAW == '?') && (NXT(1) == '>')) {
5221
4.96k
    SKIP(2);
5222
5223
    /*
5224
     * SAX: PI detected.
5225
     */
5226
4.96k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5227
4.41k
        (ctxt->sax->processingInstruction != NULL))
5228
4.41k
        ctxt->sax->processingInstruction(ctxt->userData,
5229
4.41k
                                         target, NULL);
5230
4.96k
    return;
5231
4.96k
      }
5232
36.6k
      buf = xmlMalloc(size);
5233
36.6k
      if (buf == NULL) {
5234
8
    xmlErrMemory(ctxt);
5235
8
    return;
5236
8
      }
5237
36.5k
      if (SKIP_BLANKS == 0) {
5238
10.7k
    xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED,
5239
10.7k
        "ParsePI: PI %s space expected\n", target);
5240
10.7k
      }
5241
36.5k
      cur = xmlCurrentCharRecover(ctxt, &l);
5242
6.05M
      while (IS_CHAR(cur) && /* checked */
5243
6.04M
       ((cur != '?') || (NXT(1) != '>'))) {
5244
6.02M
    if (len + 5 >= size) {
5245
9.06k
        xmlChar *tmp;
5246
9.06k
                    int newSize;
5247
5248
9.06k
                    newSize = xmlGrowCapacity(size, 1, 1, maxLength);
5249
9.06k
                    if (newSize < 0) {
5250
0
                        xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5251
0
                                          "PI %s too big found", target);
5252
0
                        xmlFree(buf);
5253
0
                        return;
5254
0
                    }
5255
9.06k
        tmp = xmlRealloc(buf, newSize);
5256
9.06k
        if (tmp == NULL) {
5257
7
      xmlErrMemory(ctxt);
5258
7
      xmlFree(buf);
5259
7
      return;
5260
7
        }
5261
9.05k
        buf = tmp;
5262
9.05k
                    size = newSize;
5263
9.05k
    }
5264
6.02M
    COPY_BUF(buf, len, cur);
5265
6.02M
    NEXTL(l);
5266
6.02M
    cur = xmlCurrentCharRecover(ctxt, &l);
5267
6.02M
      }
5268
36.5k
      buf[len] = 0;
5269
36.5k
      if (cur != '?') {
5270
13.0k
    xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5271
13.0k
          "ParsePI: PI %s never end ...\n", target);
5272
23.5k
      } else {
5273
23.5k
    SKIP(2);
5274
5275
23.5k
#ifdef LIBXML_CATALOG_ENABLED
5276
23.5k
    if ((ctxt->inSubset == 0) &&
5277
6.18k
        (xmlStrEqual(target, XML_CATALOG_PI))) {
5278
388
        xmlCatalogAllow allow = xmlCatalogGetDefaults();
5279
5280
388
        if ((ctxt->options & XML_PARSE_CATALOG_PI) &&
5281
194
                        ((allow == XML_CATA_ALLOW_DOCUMENT) ||
5282
194
       (allow == XML_CATA_ALLOW_ALL)))
5283
0
      xmlParseCatalogPI(ctxt, buf);
5284
388
    }
5285
23.5k
#endif
5286
5287
    /*
5288
     * SAX: PI detected.
5289
     */
5290
23.5k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5291
20.1k
        (ctxt->sax->processingInstruction != NULL))
5292
20.1k
        ctxt->sax->processingInstruction(ctxt->userData,
5293
20.1k
                                         target, buf);
5294
23.5k
      }
5295
36.5k
      xmlFree(buf);
5296
36.5k
  } else {
5297
4.79k
      xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL);
5298
4.79k
  }
5299
46.3k
    }
5300
46.3k
}
5301
5302
/**
5303
 * Parse a notation declaration. Always consumes '<!'.
5304
 *
5305
 * @deprecated Internal function, don't use.
5306
 *
5307
 *     [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID |  PublicID)
5308
 *                           S? '>'
5309
 *
5310
 * Hence there is actually 3 choices:
5311
 *
5312
 *     'PUBLIC' S PubidLiteral
5313
 *     'PUBLIC' S PubidLiteral S SystemLiteral
5314
 *     'SYSTEM' S SystemLiteral
5315
 *
5316
 * See the NOTE on #xmlParseExternalID.
5317
 *
5318
 * @param ctxt  an XML parser context
5319
 */
5320
5321
void
5322
4.52k
xmlParseNotationDecl(xmlParserCtxt *ctxt) {
5323
4.52k
    const xmlChar *name;
5324
4.52k
    xmlChar *Pubid;
5325
4.52k
    xmlChar *Systemid;
5326
5327
4.52k
    if ((CUR != '<') || (NXT(1) != '!'))
5328
0
        return;
5329
4.52k
    SKIP(2);
5330
5331
4.52k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5332
4.43k
#ifdef LIBXML_VALID_ENABLED
5333
4.43k
  int oldInputNr = ctxt->inputNr;
5334
4.43k
#endif
5335
5336
4.43k
  SKIP(8);
5337
4.43k
  if (SKIP_BLANKS_PE == 0) {
5338
257
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5339
257
         "Space required after '<!NOTATION'\n");
5340
257
      return;
5341
257
  }
5342
5343
4.17k
        name = xmlParseName(ctxt);
5344
4.17k
  if (name == NULL) {
5345
392
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5346
392
      return;
5347
392
  }
5348
3.78k
  if (xmlStrchr(name, ':') != NULL) {
5349
491
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5350
491
         "colons are forbidden from notation names '%s'\n",
5351
491
         name, NULL, NULL);
5352
491
  }
5353
3.78k
  if (SKIP_BLANKS_PE == 0) {
5354
228
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5355
228
         "Space required after the NOTATION name'\n");
5356
228
      return;
5357
228
  }
5358
5359
  /*
5360
   * Parse the IDs.
5361
   */
5362
3.55k
  Systemid = xmlParseExternalID(ctxt, &Pubid, 0);
5363
3.55k
  SKIP_BLANKS_PE;
5364
5365
3.55k
  if (RAW == '>') {
5366
2.09k
#ifdef LIBXML_VALID_ENABLED
5367
2.09k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
5368
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
5369
0
                           "Notation declaration doesn't start and stop"
5370
0
                                 " in the same entity\n",
5371
0
                                 NULL, NULL);
5372
0
      }
5373
2.09k
#endif
5374
2.09k
      NEXT;
5375
2.09k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5376
1.74k
    (ctxt->sax->notationDecl != NULL))
5377
1.74k
    ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid);
5378
2.09k
  } else {
5379
1.46k
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5380
1.46k
  }
5381
3.55k
  if (Systemid != NULL) xmlFree(Systemid);
5382
3.55k
  if (Pubid != NULL) xmlFree(Pubid);
5383
3.55k
    }
5384
4.52k
}
5385
5386
/**
5387
 * Parse an entity declaration. Always consumes '<!'.
5388
 *
5389
 * @deprecated Internal function, don't use.
5390
 *
5391
 *     [70] EntityDecl ::= GEDecl | PEDecl
5392
 *
5393
 *     [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>'
5394
 *
5395
 *     [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
5396
 *
5397
 *     [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?)
5398
 *
5399
 *     [74] PEDef ::= EntityValue | ExternalID
5400
 *
5401
 *     [76] NDataDecl ::= S 'NDATA' S Name
5402
 *
5403
 * [ VC: Notation Declared ]
5404
 * The Name must match the declared name of a notation.
5405
 *
5406
 * @param ctxt  an XML parser context
5407
 */
5408
5409
void
5410
101k
xmlParseEntityDecl(xmlParserCtxt *ctxt) {
5411
101k
    const xmlChar *name = NULL;
5412
101k
    xmlChar *value = NULL;
5413
101k
    xmlChar *URI = NULL, *literal = NULL;
5414
101k
    const xmlChar *ndata = NULL;
5415
101k
    int isParameter = 0;
5416
101k
    xmlChar *orig = NULL;
5417
5418
101k
    if ((CUR != '<') || (NXT(1) != '!'))
5419
0
        return;
5420
101k
    SKIP(2);
5421
5422
    /* GROW; done in the caller */
5423
101k
    if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5424
101k
#ifdef LIBXML_VALID_ENABLED
5425
101k
  int oldInputNr = ctxt->inputNr;
5426
101k
#endif
5427
5428
101k
  SKIP(6);
5429
101k
  if (SKIP_BLANKS_PE == 0) {
5430
7.61k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5431
7.61k
         "Space required after '<!ENTITY'\n");
5432
7.61k
  }
5433
5434
101k
  if (RAW == '%') {
5435
29.9k
      NEXT;
5436
29.9k
      if (SKIP_BLANKS_PE == 0) {
5437
1.13k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5438
1.13k
             "Space required after '%%'\n");
5439
1.13k
      }
5440
29.9k
      isParameter = 1;
5441
29.9k
  }
5442
5443
101k
        name = xmlParseName(ctxt);
5444
101k
  if (name == NULL) {
5445
1.52k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5446
1.52k
                     "xmlParseEntityDecl: no name\n");
5447
1.52k
            return;
5448
1.52k
  }
5449
99.5k
  if (xmlStrchr(name, ':') != NULL) {
5450
1.34k
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5451
1.34k
         "colons are forbidden from entities names '%s'\n",
5452
1.34k
         name, NULL, NULL);
5453
1.34k
  }
5454
99.5k
  if (SKIP_BLANKS_PE == 0) {
5455
5.39k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5456
5.39k
         "Space required after the entity name\n");
5457
5.39k
  }
5458
5459
  /*
5460
   * handle the various case of definitions...
5461
   */
5462
99.5k
  if (isParameter) {
5463
29.6k
      if ((RAW == '"') || (RAW == '\'')) {
5464
14.9k
          value = xmlParseEntityValue(ctxt, &orig);
5465
14.9k
    if (value) {
5466
14.6k
        if ((ctxt->sax != NULL) &&
5467
14.6k
      (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5468
12.5k
      ctxt->sax->entityDecl(ctxt->userData, name,
5469
12.5k
                        XML_INTERNAL_PARAMETER_ENTITY,
5470
12.5k
            NULL, NULL, value);
5471
14.6k
    }
5472
14.9k
      } else {
5473
14.7k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5474
14.7k
    if ((URI == NULL) && (literal == NULL)) {
5475
673
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5476
673
    }
5477
14.7k
    if (URI) {
5478
13.8k
                    if (xmlStrchr(URI, '#')) {
5479
207
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5480
13.6k
                    } else {
5481
13.6k
                        if ((ctxt->sax != NULL) &&
5482
13.6k
                            (!ctxt->disableSAX) &&
5483
13.3k
                            (ctxt->sax->entityDecl != NULL))
5484
13.3k
                            ctxt->sax->entityDecl(ctxt->userData, name,
5485
13.3k
                                        XML_EXTERNAL_PARAMETER_ENTITY,
5486
13.3k
                                        literal, URI, NULL);
5487
13.6k
                    }
5488
13.8k
    }
5489
14.7k
      }
5490
69.9k
  } else {
5491
69.9k
      if ((RAW == '"') || (RAW == '\'')) {
5492
58.6k
          value = xmlParseEntityValue(ctxt, &orig);
5493
58.6k
    if ((ctxt->sax != NULL) &&
5494
58.6k
        (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5495
52.7k
        ctxt->sax->entityDecl(ctxt->userData, name,
5496
52.7k
        XML_INTERNAL_GENERAL_ENTITY,
5497
52.7k
        NULL, NULL, value);
5498
    /*
5499
     * For expat compatibility in SAX mode.
5500
     */
5501
58.6k
    if ((ctxt->myDoc == NULL) ||
5502
58.5k
        (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) {
5503
2.46k
        if (ctxt->myDoc == NULL) {
5504
86
      ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5505
86
      if (ctxt->myDoc == NULL) {
5506
1
          xmlErrMemory(ctxt);
5507
1
          goto done;
5508
1
      }
5509
85
      ctxt->myDoc->properties = XML_DOC_INTERNAL;
5510
85
        }
5511
2.46k
        if (ctxt->myDoc->intSubset == NULL) {
5512
85
      ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5513
85
              BAD_CAST "fake", NULL, NULL);
5514
85
                        if (ctxt->myDoc->intSubset == NULL) {
5515
1
                            xmlErrMemory(ctxt);
5516
1
                            goto done;
5517
1
                        }
5518
85
                    }
5519
5520
2.46k
        xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY,
5521
2.46k
                    NULL, NULL, value);
5522
2.46k
    }
5523
58.6k
      } else {
5524
11.2k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5525
11.2k
    if ((URI == NULL) && (literal == NULL)) {
5526
3.17k
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5527
3.17k
    }
5528
11.2k
    if (URI) {
5529
7.47k
                    if (xmlStrchr(URI, '#')) {
5530
350
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5531
350
                    }
5532
7.47k
    }
5533
11.2k
    if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) {
5534
3.74k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5535
3.74k
           "Space required before 'NDATA'\n");
5536
3.74k
    }
5537
11.2k
    if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) {
5538
2.19k
        SKIP(5);
5539
2.19k
        if (SKIP_BLANKS_PE == 0) {
5540
442
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5541
442
               "Space required after 'NDATA'\n");
5542
442
        }
5543
2.19k
        ndata = xmlParseName(ctxt);
5544
2.19k
        if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5545
1.58k
            (ctxt->sax->unparsedEntityDecl != NULL))
5546
1.58k
      ctxt->sax->unparsedEntityDecl(ctxt->userData, name,
5547
1.58k
            literal, URI, ndata);
5548
9.05k
    } else {
5549
9.05k
        if ((ctxt->sax != NULL) &&
5550
9.05k
            (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5551
8.29k
      ctxt->sax->entityDecl(ctxt->userData, name,
5552
8.29k
            XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5553
8.29k
            literal, URI, NULL);
5554
        /*
5555
         * For expat compatibility in SAX mode.
5556
         * assuming the entity replacement was asked for
5557
         */
5558
9.05k
        if ((ctxt->replaceEntities != 0) &&
5559
6.75k
      ((ctxt->myDoc == NULL) ||
5560
6.74k
      (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) {
5561
211
      if (ctxt->myDoc == NULL) {
5562
14
          ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5563
14
          if (ctxt->myDoc == NULL) {
5564
1
              xmlErrMemory(ctxt);
5565
1
        goto done;
5566
1
          }
5567
13
          ctxt->myDoc->properties = XML_DOC_INTERNAL;
5568
13
      }
5569
5570
210
      if (ctxt->myDoc->intSubset == NULL) {
5571
13
          ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5572
13
            BAD_CAST "fake", NULL, NULL);
5573
13
                            if (ctxt->myDoc->intSubset == NULL) {
5574
1
                                xmlErrMemory(ctxt);
5575
1
                                goto done;
5576
1
                            }
5577
13
                        }
5578
209
      xmlSAX2EntityDecl(ctxt, name,
5579
209
                  XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5580
209
                  literal, URI, NULL);
5581
209
        }
5582
9.05k
    }
5583
11.2k
      }
5584
69.9k
  }
5585
99.5k
  SKIP_BLANKS_PE;
5586
99.5k
  if (RAW != '>') {
5587
9.23k
      xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED,
5588
9.23k
              "xmlParseEntityDecl: entity %s not terminated\n", name);
5589
90.3k
  } else {
5590
90.3k
#ifdef LIBXML_VALID_ENABLED
5591
90.3k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
5592
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
5593
0
                           "Entity declaration doesn't start and stop in"
5594
0
                                 " the same entity\n",
5595
0
                                 NULL, NULL);
5596
0
      }
5597
90.3k
#endif
5598
90.3k
      NEXT;
5599
90.3k
  }
5600
99.5k
  if (orig != NULL) {
5601
      /*
5602
       * Ugly mechanism to save the raw entity value.
5603
       */
5604
70.2k
      xmlEntityPtr cur = NULL;
5605
5606
70.2k
      if (isParameter) {
5607
14.6k
          if ((ctxt->sax != NULL) &&
5608
14.6k
        (ctxt->sax->getParameterEntity != NULL))
5609
14.6k
        cur = ctxt->sax->getParameterEntity(ctxt->userData, name);
5610
55.5k
      } else {
5611
55.5k
          if ((ctxt->sax != NULL) &&
5612
55.5k
        (ctxt->sax->getEntity != NULL))
5613
55.5k
        cur = ctxt->sax->getEntity(ctxt->userData, name);
5614
55.5k
    if ((cur == NULL) && (ctxt->userData==ctxt)) {
5615
3.29k
        cur = xmlSAX2GetEntity(ctxt, name);
5616
3.29k
    }
5617
55.5k
      }
5618
70.2k
            if ((cur != NULL) && (cur->orig == NULL)) {
5619
12.7k
    cur->orig = orig;
5620
12.7k
                orig = NULL;
5621
12.7k
      }
5622
70.2k
  }
5623
5624
99.5k
done:
5625
99.5k
  if (value != NULL) xmlFree(value);
5626
99.5k
  if (URI != NULL) xmlFree(URI);
5627
99.5k
  if (literal != NULL) xmlFree(literal);
5628
99.5k
        if (orig != NULL) xmlFree(orig);
5629
99.5k
    }
5630
101k
}
5631
5632
/**
5633
 * Parse an attribute default declaration
5634
 *
5635
 * @deprecated Internal function, don't use.
5636
 *
5637
 *     [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue)
5638
 *
5639
 * [ VC: Required Attribute ]
5640
 * if the default declaration is the keyword \#REQUIRED, then the
5641
 * attribute must be specified for all elements of the type in the
5642
 * attribute-list declaration.
5643
 *
5644
 * [ VC: Attribute Default Legal ]
5645
 * The declared default value must meet the lexical constraints of
5646
 * the declared attribute type c.f. #xmlValidateAttributeDecl
5647
 *
5648
 * [ VC: Fixed Attribute Default ]
5649
 * if an attribute has a default value declared with the \#FIXED
5650
 * keyword, instances of that attribute must match the default value.
5651
 *
5652
 * [ WFC: No < in Attribute Values ]
5653
 * handled in #xmlParseAttValue
5654
 *
5655
 * @param ctxt  an XML parser context
5656
 * @param value  Receive a possible fixed default value for the attribute
5657
 * @returns XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED
5658
 *          or XML_ATTRIBUTE_FIXED.
5659
 */
5660
5661
int
5662
59.1k
xmlParseDefaultDecl(xmlParserCtxt *ctxt, xmlChar **value) {
5663
59.1k
    int val;
5664
59.1k
    xmlChar *ret;
5665
5666
59.1k
    *value = NULL;
5667
59.1k
    if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) {
5668
10.2k
  SKIP(9);
5669
10.2k
  return(XML_ATTRIBUTE_REQUIRED);
5670
10.2k
    }
5671
48.9k
    if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) {
5672
13.1k
  SKIP(8);
5673
13.1k
  return(XML_ATTRIBUTE_IMPLIED);
5674
13.1k
    }
5675
35.7k
    val = XML_ATTRIBUTE_NONE;
5676
35.7k
    if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) {
5677
15.0k
  SKIP(6);
5678
15.0k
  val = XML_ATTRIBUTE_FIXED;
5679
15.0k
  if (SKIP_BLANKS_PE == 0) {
5680
7.38k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5681
7.38k
         "Space required after '#FIXED'\n");
5682
7.38k
  }
5683
15.0k
    }
5684
35.7k
    ret = xmlParseAttValue(ctxt);
5685
35.7k
    if (ret == NULL) {
5686
10.5k
  xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo,
5687
10.5k
           "Attribute default value declaration error\n");
5688
10.5k
    } else
5689
25.1k
        *value = ret;
5690
35.7k
    return(val);
5691
48.9k
}
5692
5693
/**
5694
 * Parse an Notation attribute type.
5695
 *
5696
 * @deprecated Internal function, don't use.
5697
 *
5698
 * Note: the leading 'NOTATION' S part has already being parsed...
5699
 *
5700
 *     [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
5701
 *
5702
 * [ VC: Notation Attributes ]
5703
 * Values of this type must match one of the notation names included
5704
 * in the declaration; all notation names in the declaration must be declared.
5705
 *
5706
 * @param ctxt  an XML parser context
5707
 * @returns the notation attribute tree built while parsing
5708
 */
5709
5710
xmlEnumeration *
5711
1.62k
xmlParseNotationType(xmlParserCtxt *ctxt) {
5712
1.62k
    const xmlChar *name;
5713
1.62k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5714
5715
1.62k
    if (RAW != '(') {
5716
217
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5717
217
  return(NULL);
5718
217
    }
5719
1.60k
    do {
5720
1.60k
        NEXT;
5721
1.60k
  SKIP_BLANKS_PE;
5722
1.60k
        name = xmlParseName(ctxt);
5723
1.60k
  if (name == NULL) {
5724
209
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5725
209
         "Name expected in NOTATION declaration\n");
5726
209
            xmlFreeEnumeration(ret);
5727
209
      return(NULL);
5728
209
  }
5729
1.39k
        tmp = NULL;
5730
1.39k
#ifdef LIBXML_VALID_ENABLED
5731
1.39k
        if (ctxt->validate) {
5732
0
            tmp = ret;
5733
0
            while (tmp != NULL) {
5734
0
                if (xmlStrEqual(name, tmp->name)) {
5735
0
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5736
0
              "standalone: attribute notation value token %s duplicated\n",
5737
0
                                     name, NULL);
5738
0
                    if (!xmlDictOwns(ctxt->dict, name))
5739
0
                        xmlFree((xmlChar *) name);
5740
0
                    break;
5741
0
                }
5742
0
                tmp = tmp->next;
5743
0
            }
5744
0
        }
5745
1.39k
#endif /* LIBXML_VALID_ENABLED */
5746
1.39k
  if (tmp == NULL) {
5747
1.39k
      cur = xmlCreateEnumeration(name);
5748
1.39k
      if (cur == NULL) {
5749
1
                xmlErrMemory(ctxt);
5750
1
                xmlFreeEnumeration(ret);
5751
1
                return(NULL);
5752
1
            }
5753
1.39k
      if (last == NULL) ret = last = cur;
5754
202
      else {
5755
202
    last->next = cur;
5756
202
    last = cur;
5757
202
      }
5758
1.39k
  }
5759
1.39k
  SKIP_BLANKS_PE;
5760
1.39k
    } while (RAW == '|');
5761
1.19k
    if (RAW != ')') {
5762
254
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5763
254
        xmlFreeEnumeration(ret);
5764
254
  return(NULL);
5765
254
    }
5766
940
    NEXT;
5767
940
    return(ret);
5768
1.19k
}
5769
5770
/**
5771
 * Parse an Enumeration attribute type.
5772
 *
5773
 * @deprecated Internal function, don't use.
5774
 *
5775
 *     [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')'
5776
 *
5777
 * [ VC: Enumeration ]
5778
 * Values of this type must match one of the Nmtoken tokens in
5779
 * the declaration
5780
 *
5781
 * @param ctxt  an XML parser context
5782
 * @returns the enumeration attribute tree built while parsing
5783
 */
5784
5785
xmlEnumeration *
5786
14.1k
xmlParseEnumerationType(xmlParserCtxt *ctxt) {
5787
14.1k
    xmlChar *name;
5788
14.1k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5789
5790
14.1k
    if (RAW != '(') {
5791
1.79k
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL);
5792
1.79k
  return(NULL);
5793
1.79k
    }
5794
18.6k
    do {
5795
18.6k
        NEXT;
5796
18.6k
  SKIP_BLANKS_PE;
5797
18.6k
        name = xmlParseNmtoken(ctxt);
5798
18.6k
  if (name == NULL) {
5799
358
      xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL);
5800
358
      return(ret);
5801
358
  }
5802
18.2k
        tmp = NULL;
5803
18.2k
#ifdef LIBXML_VALID_ENABLED
5804
18.2k
        if (ctxt->validate) {
5805
0
            tmp = ret;
5806
0
            while (tmp != NULL) {
5807
0
                if (xmlStrEqual(name, tmp->name)) {
5808
0
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5809
0
              "standalone: attribute enumeration value token %s duplicated\n",
5810
0
                                     name, NULL);
5811
0
                    if (!xmlDictOwns(ctxt->dict, name))
5812
0
                        xmlFree(name);
5813
0
                    break;
5814
0
                }
5815
0
                tmp = tmp->next;
5816
0
            }
5817
0
        }
5818
18.2k
#endif /* LIBXML_VALID_ENABLED */
5819
18.2k
  if (tmp == NULL) {
5820
18.2k
      cur = xmlCreateEnumeration(name);
5821
18.2k
      if (!xmlDictOwns(ctxt->dict, name))
5822
18.2k
    xmlFree(name);
5823
18.2k
      if (cur == NULL) {
5824
2
                xmlErrMemory(ctxt);
5825
2
                xmlFreeEnumeration(ret);
5826
2
                return(NULL);
5827
2
            }
5828
18.2k
      if (last == NULL) ret = last = cur;
5829
6.20k
      else {
5830
6.20k
    last->next = cur;
5831
6.20k
    last = cur;
5832
6.20k
      }
5833
18.2k
  }
5834
18.2k
  SKIP_BLANKS_PE;
5835
18.2k
    } while (RAW == '|');
5836
12.0k
    if (RAW != ')') {
5837
472
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL);
5838
472
  return(ret);
5839
472
    }
5840
11.5k
    NEXT;
5841
11.5k
    return(ret);
5842
12.0k
}
5843
5844
/**
5845
 * Parse an Enumerated attribute type.
5846
 *
5847
 * @deprecated Internal function, don't use.
5848
 *
5849
 *     [57] EnumeratedType ::= NotationType | Enumeration
5850
 *
5851
 *     [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
5852
 *
5853
 * @param ctxt  an XML parser context
5854
 * @param tree  the enumeration tree built while parsing
5855
 * @returns XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION
5856
 */
5857
5858
int
5859
20.0k
xmlParseEnumeratedType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5860
20.0k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5861
5.88k
  SKIP(8);
5862
5.88k
  if (SKIP_BLANKS_PE == 0) {
5863
4.26k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5864
4.26k
         "Space required after 'NOTATION'\n");
5865
4.26k
      return(0);
5866
4.26k
  }
5867
1.62k
  *tree = xmlParseNotationType(ctxt);
5868
1.62k
  if (*tree == NULL) return(0);
5869
940
  return(XML_ATTRIBUTE_NOTATION);
5870
1.62k
    }
5871
14.1k
    *tree = xmlParseEnumerationType(ctxt);
5872
14.1k
    if (*tree == NULL) return(0);
5873
12.0k
    return(XML_ATTRIBUTE_ENUMERATION);
5874
14.1k
}
5875
5876
/**
5877
 * Parse the Attribute list def for an element
5878
 *
5879
 * @deprecated Internal function, don't use.
5880
 *
5881
 *     [54] AttType ::= StringType | TokenizedType | EnumeratedType
5882
 *
5883
 *     [55] StringType ::= 'CDATA'
5884
 *
5885
 *     [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' |
5886
 *                            'ENTITIES' | 'NMTOKEN' | 'NMTOKENS'
5887
 *
5888
 * Validity constraints for attribute values syntax are checked in
5889
 * #xmlValidateAttributeValue
5890
 *
5891
 * [ VC: ID ]
5892
 * Values of type ID must match the Name production. A name must not
5893
 * appear more than once in an XML document as a value of this type;
5894
 * i.e., ID values must uniquely identify the elements which bear them.
5895
 *
5896
 * [ VC: One ID per Element Type ]
5897
 * No element type may have more than one ID attribute specified.
5898
 *
5899
 * [ VC: ID Attribute Default ]
5900
 * An ID attribute must have a declared default of \#IMPLIED or \#REQUIRED.
5901
 *
5902
 * [ VC: IDREF ]
5903
 * Values of type IDREF must match the Name production, and values
5904
 * of type IDREFS must match Names; each IDREF Name must match the value
5905
 * of an ID attribute on some element in the XML document; i.e. IDREF
5906
 * values must match the value of some ID attribute.
5907
 *
5908
 * [ VC: Entity Name ]
5909
 * Values of type ENTITY must match the Name production, values
5910
 * of type ENTITIES must match Names; each Entity Name must match the
5911
 * name of an unparsed entity declared in the DTD.
5912
 *
5913
 * [ VC: Name Token ]
5914
 * Values of type NMTOKEN must match the Nmtoken production; values
5915
 * of type NMTOKENS must match Nmtokens.
5916
 *
5917
 * @param ctxt  an XML parser context
5918
 * @param tree  the enumeration tree built while parsing
5919
 * @returns the attribute type
5920
 */
5921
int
5922
68.5k
xmlParseAttributeType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5923
68.5k
    if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) {
5924
6.45k
  SKIP(5);
5925
6.45k
  return(XML_ATTRIBUTE_CDATA);
5926
62.0k
     } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) {
5927
4.58k
  SKIP(6);
5928
4.58k
  return(XML_ATTRIBUTE_IDREFS);
5929
57.4k
     } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) {
5930
1.42k
  SKIP(5);
5931
1.42k
  return(XML_ATTRIBUTE_IDREF);
5932
56.0k
     } else if ((RAW == 'I') && (NXT(1) == 'D')) {
5933
9.33k
        SKIP(2);
5934
9.33k
  return(XML_ATTRIBUTE_ID);
5935
46.7k
     } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5936
1.45k
  SKIP(6);
5937
1.45k
  return(XML_ATTRIBUTE_ENTITY);
5938
45.2k
     } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) {
5939
8.10k
  SKIP(8);
5940
8.10k
  return(XML_ATTRIBUTE_ENTITIES);
5941
37.1k
     } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) {
5942
1.30k
  SKIP(8);
5943
1.30k
  return(XML_ATTRIBUTE_NMTOKENS);
5944
35.8k
     } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) {
5945
15.8k
  SKIP(7);
5946
15.8k
  return(XML_ATTRIBUTE_NMTOKEN);
5947
15.8k
     }
5948
20.0k
     return(xmlParseEnumeratedType(ctxt, tree));
5949
68.5k
}
5950
5951
/**
5952
 * Parse an attribute list declaration for an element. Always consumes '<!'.
5953
 *
5954
 * @deprecated Internal function, don't use.
5955
 *
5956
 *     [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>'
5957
 *
5958
 *     [53] AttDef ::= S Name S AttType S DefaultDecl
5959
 * @param ctxt  an XML parser context
5960
 */
5961
void
5962
63.3k
xmlParseAttributeListDecl(xmlParserCtxt *ctxt) {
5963
63.3k
    const xmlChar *elemName;
5964
63.3k
    const xmlChar *attrName;
5965
63.3k
    xmlEnumerationPtr tree;
5966
5967
63.3k
    if ((CUR != '<') || (NXT(1) != '!'))
5968
0
        return;
5969
63.3k
    SKIP(2);
5970
5971
63.3k
    if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) {
5972
63.2k
#ifdef LIBXML_VALID_ENABLED
5973
63.2k
  int oldInputNr = ctxt->inputNr;
5974
63.2k
#endif
5975
5976
63.2k
  SKIP(7);
5977
63.2k
  if (SKIP_BLANKS_PE == 0) {
5978
6.04k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5979
6.04k
                     "Space required after '<!ATTLIST'\n");
5980
6.04k
  }
5981
63.2k
        elemName = xmlParseName(ctxt);
5982
63.2k
  if (elemName == NULL) {
5983
7.84k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5984
7.84k
         "ATTLIST: no name for Element\n");
5985
7.84k
      return;
5986
7.84k
  }
5987
55.3k
  SKIP_BLANKS_PE;
5988
55.3k
  GROW;
5989
103k
  while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) {
5990
77.2k
      int type;
5991
77.2k
      int def;
5992
77.2k
      xmlChar *defaultValue = NULL;
5993
5994
77.2k
      GROW;
5995
77.2k
            tree = NULL;
5996
77.2k
      attrName = xmlParseName(ctxt);
5997
77.2k
      if (attrName == NULL) {
5998
7.75k
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5999
7.75k
             "ATTLIST: no name for Attribute\n");
6000
7.75k
    break;
6001
7.75k
      }
6002
69.4k
      GROW;
6003
69.4k
      if (SKIP_BLANKS_PE == 0) {
6004
905
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6005
905
            "Space required after the attribute name\n");
6006
905
    break;
6007
905
      }
6008
6009
68.5k
      type = xmlParseAttributeType(ctxt, &tree);
6010
68.5k
      if (type <= 0) {
6011
7.10k
          break;
6012
7.10k
      }
6013
6014
61.4k
      GROW;
6015
61.4k
      if (SKIP_BLANKS_PE == 0) {
6016
2.24k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6017
2.24k
             "Space required after the attribute type\n");
6018
2.24k
          if (tree != NULL)
6019
494
        xmlFreeEnumeration(tree);
6020
2.24k
    break;
6021
2.24k
      }
6022
6023
59.1k
      def = xmlParseDefaultDecl(ctxt, &defaultValue);
6024
59.1k
      if (def <= 0) {
6025
0
                if (defaultValue != NULL)
6026
0
        xmlFree(defaultValue);
6027
0
          if (tree != NULL)
6028
0
        xmlFreeEnumeration(tree);
6029
0
          break;
6030
0
      }
6031
59.1k
      if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL))
6032
24.0k
          xmlAttrNormalizeSpace(defaultValue, defaultValue);
6033
6034
59.1k
      GROW;
6035
59.1k
            if (RAW != '>') {
6036
37.0k
    if (SKIP_BLANKS_PE == 0) {
6037
11.5k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6038
11.5k
      "Space required after the attribute default value\n");
6039
11.5k
        if (defaultValue != NULL)
6040
1.06k
      xmlFree(defaultValue);
6041
11.5k
        if (tree != NULL)
6042
966
      xmlFreeEnumeration(tree);
6043
11.5k
        break;
6044
11.5k
    }
6045
37.0k
      }
6046
47.6k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6047
39.9k
    (ctxt->sax->attributeDecl != NULL))
6048
39.9k
    ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName,
6049
39.9k
                          type, def, defaultValue, tree);
6050
7.77k
      else if (tree != NULL)
6051
1.09k
    xmlFreeEnumeration(tree);
6052
6053
47.6k
      if ((ctxt->sax2) && (defaultValue != NULL) &&
6054
24.0k
          (def != XML_ATTRIBUTE_IMPLIED) &&
6055
24.0k
    (def != XML_ATTRIBUTE_REQUIRED)) {
6056
24.0k
    xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue);
6057
24.0k
      }
6058
47.6k
      if (ctxt->sax2) {
6059
47.6k
    xmlAddSpecialAttr(ctxt, elemName, attrName, type);
6060
47.6k
      }
6061
47.6k
      if (defaultValue != NULL)
6062
24.0k
          xmlFree(defaultValue);
6063
47.6k
      GROW;
6064
47.6k
  }
6065
55.3k
  if (RAW == '>') {
6066
26.0k
#ifdef LIBXML_VALID_ENABLED
6067
26.0k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6068
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6069
0
                                 "Attribute list declaration doesn't start and"
6070
0
                                 " stop in the same entity\n",
6071
0
                                 NULL, NULL);
6072
0
      }
6073
26.0k
#endif
6074
26.0k
      NEXT;
6075
26.0k
  }
6076
55.3k
    }
6077
63.3k
}
6078
6079
/**
6080
 * Handle PEs and check that we don't pop the entity that started
6081
 * a balanced group.
6082
 *
6083
 * @param ctxt  parser context
6084
 * @param openInputNr  input nr of the entity with opening '('
6085
 */
6086
static void
6087
211k
xmlSkipBlankCharsPEBalanced(xmlParserCtxt *ctxt, int openInputNr) {
6088
211k
    SKIP_BLANKS;
6089
211k
    GROW;
6090
6091
211k
    (void) openInputNr;
6092
6093
211k
    if (!PARSER_EXTERNAL(ctxt) && !PARSER_IN_PE(ctxt))
6094
87.0k
        return;
6095
6096
132k
    while (!PARSER_STOPPED(ctxt)) {
6097
132k
        if (ctxt->input->cur >= ctxt->input->end) {
6098
3.70k
#ifdef LIBXML_VALID_ENABLED
6099
3.70k
            if ((ctxt->validate) && (ctxt->inputNr <= openInputNr)) {
6100
0
                xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6101
0
                                 "Element content declaration doesn't start "
6102
0
                                 "and stop in the same entity\n",
6103
0
                                 NULL, NULL);
6104
0
            }
6105
3.70k
#endif
6106
3.70k
            if (PARSER_IN_PE(ctxt))
6107
3.67k
                xmlPopPE(ctxt);
6108
37
            else
6109
37
                break;
6110
128k
        } else if (RAW == '%') {
6111
4.82k
            xmlParsePERefInternal(ctxt, 0);
6112
123k
        } else {
6113
123k
            break;
6114
123k
        }
6115
6116
8.49k
        SKIP_BLANKS;
6117
8.49k
        GROW;
6118
8.49k
    }
6119
124k
}
6120
6121
/**
6122
 * Parse the declaration for a Mixed Element content
6123
 * The leading '(' and spaces have been skipped in #xmlParseElementContentDecl
6124
 *
6125
 * @deprecated Internal function, don't use.
6126
 *
6127
 *     [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' |
6128
 *                    '(' S? '#PCDATA' S? ')'
6129
 *
6130
 * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49])
6131
 *
6132
 * [ VC: No Duplicate Types ]
6133
 * The same name must not appear more than once in a single
6134
 * mixed-content declaration.
6135
 *
6136
 * @param ctxt  an XML parser context
6137
 * @param openInputNr  the input used for the current entity, needed for
6138
 * boundary checks
6139
 * @returns the list of the xmlElementContent describing the element choices
6140
 */
6141
xmlElementContent *
6142
6.23k
xmlParseElementMixedContentDecl(xmlParserCtxt *ctxt, int openInputNr) {
6143
6.23k
    xmlElementContentPtr ret = NULL, cur = NULL, n;
6144
6.23k
    const xmlChar *elem = NULL;
6145
6146
6.23k
    GROW;
6147
6.23k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6148
6.23k
  SKIP(7);
6149
6.23k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6150
6.23k
  if (RAW == ')') {
6151
4.88k
#ifdef LIBXML_VALID_ENABLED
6152
4.88k
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6153
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6154
0
                                 "Element content declaration doesn't start "
6155
0
                                 "and stop in the same entity\n",
6156
0
                                 NULL, NULL);
6157
0
      }
6158
4.88k
#endif
6159
4.88k
      NEXT;
6160
4.88k
      ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6161
4.88k
      if (ret == NULL)
6162
1
                goto mem_error;
6163
4.88k
      if (RAW == '*') {
6164
400
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6165
400
    NEXT;
6166
400
      }
6167
4.88k
      return(ret);
6168
4.88k
  }
6169
1.34k
  if ((RAW == '(') || (RAW == '|')) {
6170
1.12k
      ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6171
1.12k
      if (ret == NULL)
6172
1
                goto mem_error;
6173
1.12k
  }
6174
13.1k
  while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) {
6175
11.7k
      NEXT;
6176
11.7k
            n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6177
11.7k
            if (n == NULL)
6178
1
                goto mem_error;
6179
11.7k
      if (elem == NULL) {
6180
1.11k
    n->c1 = cur;
6181
1.11k
    if (cur != NULL)
6182
1.11k
        cur->parent = n;
6183
1.11k
    ret = cur = n;
6184
10.6k
      } else {
6185
10.6k
          cur->c2 = n;
6186
10.6k
    n->parent = cur;
6187
10.6k
    n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6188
10.6k
                if (n->c1 == NULL)
6189
1
                    goto mem_error;
6190
10.6k
    n->c1->parent = n;
6191
10.6k
    cur = n;
6192
10.6k
      }
6193
11.7k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6194
11.7k
      elem = xmlParseName(ctxt);
6195
11.7k
      if (elem == NULL) {
6196
28
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6197
28
      "xmlParseElementMixedContentDecl : Name expected\n");
6198
28
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6199
28
    return(NULL);
6200
28
      }
6201
11.7k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6202
11.7k
  }
6203
1.31k
  if ((RAW == ')') && (NXT(1) == '*')) {
6204
995
      if (elem != NULL) {
6205
995
    cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem,
6206
995
                                   XML_ELEMENT_CONTENT_ELEMENT);
6207
995
    if (cur->c2 == NULL)
6208
1
                    goto mem_error;
6209
994
    cur->c2->parent = cur;
6210
994
            }
6211
994
            if (ret != NULL)
6212
994
                ret->ocur = XML_ELEMENT_CONTENT_MULT;
6213
994
#ifdef LIBXML_VALID_ENABLED
6214
994
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6215
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6216
0
                                 "Element content declaration doesn't start "
6217
0
                                 "and stop in the same entity\n",
6218
0
                                 NULL, NULL);
6219
0
      }
6220
994
#endif
6221
994
      SKIP(2);
6222
994
  } else {
6223
319
      xmlFreeDocElementContent(ctxt->myDoc, ret);
6224
319
      xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL);
6225
319
      return(NULL);
6226
319
  }
6227
6228
1.31k
    } else {
6229
0
  xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL);
6230
0
    }
6231
994
    return(ret);
6232
6233
5
mem_error:
6234
5
    xmlErrMemory(ctxt);
6235
5
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6236
5
    return(NULL);
6237
6.23k
}
6238
6239
/**
6240
 * Parse the declaration for a Mixed Element content
6241
 * The leading '(' and spaces have been skipped in #xmlParseElementContentDecl
6242
 *
6243
 *     [47] children ::= (choice | seq) ('?' | '*' | '+')?
6244
 *
6245
 *     [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')?
6246
 *
6247
 *     [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')'
6248
 *
6249
 *     [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
6250
 *
6251
 * [ VC: Proper Group/PE Nesting ] applies to [49] and [50]
6252
 * TODO Parameter-entity replacement text must be properly nested
6253
 *  with parenthesized groups. That is to say, if either of the
6254
 *  opening or closing parentheses in a choice, seq, or Mixed
6255
 *  construct is contained in the replacement text for a parameter
6256
 *  entity, both must be contained in the same replacement text. For
6257
 *  interoperability, if a parameter-entity reference appears in a
6258
 *  choice, seq, or Mixed construct, its replacement text should not
6259
 *  be empty, and neither the first nor last non-blank character of
6260
 *  the replacement text should be a connector (| or ,).
6261
 *
6262
 * @param ctxt  an XML parser context
6263
 * @param openInputNr  the input used for the current entity, needed for
6264
 * boundary checks
6265
 * @param depth  the level of recursion
6266
 * @returns the tree of xmlElementContent describing the element
6267
 *          hierarchy.
6268
 */
6269
static xmlElementContentPtr
6270
xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int openInputNr,
6271
59.7k
                                       int depth) {
6272
59.7k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
6273
59.7k
    xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL;
6274
59.7k
    const xmlChar *elem;
6275
59.7k
    xmlChar type = 0;
6276
6277
59.7k
    if (depth > maxDepth) {
6278
7
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
6279
7
                "xmlParseElementChildrenContentDecl : depth %d too deep, "
6280
7
                "use XML_PARSE_HUGE\n", depth);
6281
7
  return(NULL);
6282
7
    }
6283
59.7k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6284
59.7k
    if (RAW == '(') {
6285
29.4k
        int newInputNr = ctxt->inputNr;
6286
6287
        /* Recurse on first child */
6288
29.4k
  NEXT;
6289
29.4k
        cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6290
29.4k
                                                           depth + 1);
6291
29.4k
        if (cur == NULL)
6292
28.6k
            return(NULL);
6293
30.3k
    } else {
6294
30.3k
  elem = xmlParseName(ctxt);
6295
30.3k
  if (elem == NULL) {
6296
546
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6297
546
      return(NULL);
6298
546
  }
6299
29.8k
        cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6300
29.8k
  if (cur == NULL) {
6301
3
      xmlErrMemory(ctxt);
6302
3
      return(NULL);
6303
3
  }
6304
29.8k
  GROW;
6305
29.8k
  if (RAW == '?') {
6306
2.15k
      cur->ocur = XML_ELEMENT_CONTENT_OPT;
6307
2.15k
      NEXT;
6308
27.6k
  } else if (RAW == '*') {
6309
1.77k
      cur->ocur = XML_ELEMENT_CONTENT_MULT;
6310
1.77k
      NEXT;
6311
25.8k
  } else if (RAW == '+') {
6312
752
      cur->ocur = XML_ELEMENT_CONTENT_PLUS;
6313
752
      NEXT;
6314
25.1k
  } else {
6315
25.1k
      cur->ocur = XML_ELEMENT_CONTENT_ONCE;
6316
25.1k
  }
6317
29.8k
  GROW;
6318
29.8k
    }
6319
60.0k
    while (!PARSER_STOPPED(ctxt)) {
6320
59.7k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6321
59.7k
        if (RAW == ')')
6322
27.1k
            break;
6323
        /*
6324
   * Each loop we parse one separator and one element.
6325
   */
6326
32.5k
        if (RAW == ',') {
6327
12.8k
      if (type == 0) type = CUR;
6328
6329
      /*
6330
       * Detect "Name | Name , Name" error
6331
       */
6332
9.78k
      else if (type != CUR) {
6333
1
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6334
1
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6335
1
                      type);
6336
1
    if ((last != NULL) && (last != ret))
6337
1
        xmlFreeDocElementContent(ctxt->myDoc, last);
6338
1
    if (ret != NULL)
6339
1
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6340
1
    return(NULL);
6341
1
      }
6342
12.8k
      NEXT;
6343
6344
12.8k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ);
6345
12.8k
      if (op == NULL) {
6346
3
                xmlErrMemory(ctxt);
6347
3
    if ((last != NULL) && (last != ret))
6348
2
        xmlFreeDocElementContent(ctxt->myDoc, last);
6349
3
          xmlFreeDocElementContent(ctxt->myDoc, ret);
6350
3
    return(NULL);
6351
3
      }
6352
12.8k
      if (last == NULL) {
6353
3.07k
    op->c1 = ret;
6354
3.07k
    if (ret != NULL)
6355
3.07k
        ret->parent = op;
6356
3.07k
    ret = cur = op;
6357
9.78k
      } else {
6358
9.78k
          cur->c2 = op;
6359
9.78k
    if (op != NULL)
6360
9.78k
        op->parent = cur;
6361
9.78k
    op->c1 = last;
6362
9.78k
    if (last != NULL)
6363
9.78k
        last->parent = op;
6364
9.78k
    cur =op;
6365
9.78k
    last = NULL;
6366
9.78k
      }
6367
19.6k
  } else if (RAW == '|') {
6368
17.7k
      if (type == 0) type = CUR;
6369
6370
      /*
6371
       * Detect "Name , Name | Name" error
6372
       */
6373
11.1k
      else if (type != CUR) {
6374
1
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6375
1
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6376
1
          type);
6377
1
    if ((last != NULL) && (last != ret))
6378
1
        xmlFreeDocElementContent(ctxt->myDoc, last);
6379
1
    if (ret != NULL)
6380
1
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6381
1
    return(NULL);
6382
1
      }
6383
17.7k
      NEXT;
6384
6385
17.7k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6386
17.7k
      if (op == NULL) {
6387
3
                xmlErrMemory(ctxt);
6388
3
    if ((last != NULL) && (last != ret))
6389
2
        xmlFreeDocElementContent(ctxt->myDoc, last);
6390
3
    if (ret != NULL)
6391
3
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6392
3
    return(NULL);
6393
3
      }
6394
17.7k
      if (last == NULL) {
6395
6.61k
    op->c1 = ret;
6396
6.61k
    if (ret != NULL)
6397
6.61k
        ret->parent = op;
6398
6.61k
    ret = cur = op;
6399
11.1k
      } else {
6400
11.1k
          cur->c2 = op;
6401
11.1k
    if (op != NULL)
6402
11.1k
        op->parent = cur;
6403
11.1k
    op->c1 = last;
6404
11.1k
    if (last != NULL)
6405
11.1k
        last->parent = op;
6406
11.1k
    cur =op;
6407
11.1k
    last = NULL;
6408
11.1k
      }
6409
17.7k
  } else {
6410
1.92k
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL);
6411
1.92k
      if ((last != NULL) && (last != ret))
6412
1.00k
          xmlFreeDocElementContent(ctxt->myDoc, last);
6413
1.92k
      if (ret != NULL)
6414
1.92k
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6415
1.92k
      return(NULL);
6416
1.92k
  }
6417
30.5k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6418
30.5k
        if (RAW == '(') {
6419
5.18k
            int newInputNr = ctxt->inputNr;
6420
6421
      /* Recurse on second child */
6422
5.18k
      NEXT;
6423
5.18k
      last = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6424
5.18k
                                                          depth + 1);
6425
5.18k
            if (last == NULL) {
6426
808
    if (ret != NULL)
6427
808
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6428
808
    return(NULL);
6429
808
            }
6430
25.3k
  } else {
6431
25.3k
      elem = xmlParseName(ctxt);
6432
25.3k
      if (elem == NULL) {
6433
288
    xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6434
288
    if (ret != NULL)
6435
288
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6436
288
    return(NULL);
6437
288
      }
6438
25.1k
      last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6439
25.1k
      if (last == NULL) {
6440
4
                xmlErrMemory(ctxt);
6441
4
    if (ret != NULL)
6442
4
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6443
4
    return(NULL);
6444
4
      }
6445
25.1k
      if (RAW == '?') {
6446
3.61k
    last->ocur = XML_ELEMENT_CONTENT_OPT;
6447
3.61k
    NEXT;
6448
21.4k
      } else if (RAW == '*') {
6449
4.52k
    last->ocur = XML_ELEMENT_CONTENT_MULT;
6450
4.52k
    NEXT;
6451
16.9k
      } else if (RAW == '+') {
6452
1.52k
    last->ocur = XML_ELEMENT_CONTENT_PLUS;
6453
1.52k
    NEXT;
6454
15.4k
      } else {
6455
15.4k
    last->ocur = XML_ELEMENT_CONTENT_ONCE;
6456
15.4k
      }
6457
25.1k
  }
6458
30.5k
    }
6459
27.5k
    if ((cur != NULL) && (last != NULL)) {
6460
7.57k
        cur->c2 = last;
6461
7.57k
  if (last != NULL)
6462
7.57k
      last->parent = cur;
6463
7.57k
    }
6464
27.5k
#ifdef LIBXML_VALID_ENABLED
6465
27.5k
    if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6466
0
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6467
0
                         "Element content declaration doesn't start "
6468
0
                         "and stop in the same entity\n",
6469
0
                         NULL, NULL);
6470
0
    }
6471
27.5k
#endif
6472
27.5k
    NEXT;
6473
27.5k
    if (RAW == '?') {
6474
3.70k
  if (ret != NULL) {
6475
3.70k
      if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) ||
6476
3.46k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6477
682
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6478
3.02k
      else
6479
3.02k
          ret->ocur = XML_ELEMENT_CONTENT_OPT;
6480
3.70k
  }
6481
3.70k
  NEXT;
6482
23.8k
    } else if (RAW == '*') {
6483
2.01k
  if (ret != NULL) {
6484
2.01k
      ret->ocur = XML_ELEMENT_CONTENT_MULT;
6485
2.01k
      cur = ret;
6486
      /*
6487
       * Some normalization:
6488
       * (a | b* | c?)* == (a | b | c)*
6489
       */
6490
6.48k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6491
4.46k
    if ((cur->c1 != NULL) &&
6492
4.46k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6493
4.24k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT)))
6494
426
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6495
4.46k
    if ((cur->c2 != NULL) &&
6496
4.46k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6497
4.06k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT)))
6498
472
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6499
4.46k
    cur = cur->c2;
6500
4.46k
      }
6501
2.01k
  }
6502
2.01k
  NEXT;
6503
21.8k
    } else if (RAW == '+') {
6504
3.09k
  if (ret != NULL) {
6505
3.09k
      int found = 0;
6506
6507
3.09k
      if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) ||
6508
2.51k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6509
659
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6510
2.43k
      else
6511
2.43k
          ret->ocur = XML_ELEMENT_CONTENT_PLUS;
6512
      /*
6513
       * Some normalization:
6514
       * (a | b*)+ == (a | b)*
6515
       * (a | b?)+ == (a | b)*
6516
       */
6517
5.06k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6518
1.96k
    if ((cur->c1 != NULL) &&
6519
1.96k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6520
1.61k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) {
6521
865
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6522
865
        found = 1;
6523
865
    }
6524
1.96k
    if ((cur->c2 != NULL) &&
6525
1.96k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6526
1.38k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) {
6527
653
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6528
653
        found = 1;
6529
653
    }
6530
1.96k
    cur = cur->c2;
6531
1.96k
      }
6532
3.09k
      if (found)
6533
1.17k
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6534
3.09k
  }
6535
3.09k
  NEXT;
6536
3.09k
    }
6537
27.5k
    return(ret);
6538
30.5k
}
6539
6540
/**
6541
 * Parse the declaration for a Mixed Element content
6542
 * The leading '(' and spaces have been skipped in #xmlParseElementContentDecl
6543
 *
6544
 * @deprecated Internal function, don't use.
6545
 *
6546
 *     [47] children ::= (choice | seq) ('?' | '*' | '+')?
6547
 *
6548
 *     [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')?
6549
 *
6550
 *     [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')'
6551
 *
6552
 *     [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
6553
 *
6554
 * [ VC: Proper Group/PE Nesting ] applies to [49] and [50]
6555
 * TODO Parameter-entity replacement text must be properly nested
6556
 *  with parenthesized groups. That is to say, if either of the
6557
 *  opening or closing parentheses in a choice, seq, or Mixed
6558
 *  construct is contained in the replacement text for a parameter
6559
 *  entity, both must be contained in the same replacement text. For
6560
 *  interoperability, if a parameter-entity reference appears in a
6561
 *  choice, seq, or Mixed construct, its replacement text should not
6562
 *  be empty, and neither the first nor last non-blank character of
6563
 *  the replacement text should be a connector (| or ,).
6564
 *
6565
 * @param ctxt  an XML parser context
6566
 * @param inputchk  the input used for the current entity, needed for boundary checks
6567
 * @returns the tree of xmlElementContent describing the element
6568
 *          hierarchy.
6569
 */
6570
xmlElementContent *
6571
0
xmlParseElementChildrenContentDecl(xmlParserCtxt *ctxt, int inputchk) {
6572
    /* stub left for API/ABI compat */
6573
0
    return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1));
6574
0
}
6575
6576
/**
6577
 * Parse the declaration for an Element content either Mixed or Children,
6578
 * the cases EMPTY and ANY are handled directly in #xmlParseElementDecl
6579
 *
6580
 * @deprecated Internal function, don't use.
6581
 *
6582
 *     [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children
6583
 *
6584
 * @param ctxt  an XML parser context
6585
 * @param name  the name of the element being defined.
6586
 * @param result  the Element Content pointer will be stored here if any
6587
 * @returns an xmlElementTypeVal value or -1 on error
6588
 */
6589
6590
int
6591
xmlParseElementContentDecl(xmlParserCtxt *ctxt, const xmlChar *name,
6592
31.4k
                           xmlElementContent **result) {
6593
6594
31.4k
    xmlElementContentPtr tree = NULL;
6595
31.4k
    int openInputNr = ctxt->inputNr;
6596
31.4k
    int res;
6597
6598
31.4k
    *result = NULL;
6599
6600
31.4k
    if (RAW != '(') {
6601
0
  xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6602
0
    "xmlParseElementContentDecl : %s '(' expected\n", name);
6603
0
  return(-1);
6604
0
    }
6605
31.4k
    NEXT;
6606
31.4k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6607
31.4k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6608
6.23k
        tree = xmlParseElementMixedContentDecl(ctxt, openInputNr);
6609
6.23k
  res = XML_ELEMENT_TYPE_MIXED;
6610
25.1k
    } else {
6611
25.1k
        tree = xmlParseElementChildrenContentDeclPriv(ctxt, openInputNr, 1);
6612
25.1k
  res = XML_ELEMENT_TYPE_ELEMENT;
6613
25.1k
    }
6614
31.4k
    if (tree == NULL)
6615
3.13k
        return(-1);
6616
28.2k
    SKIP_BLANKS_PE;
6617
28.2k
    *result = tree;
6618
28.2k
    return(res);
6619
31.4k
}
6620
6621
/**
6622
 * Parse an element declaration. Always consumes '<!'.
6623
 *
6624
 * @deprecated Internal function, don't use.
6625
 *
6626
 *     [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>'
6627
 *
6628
 * [ VC: Unique Element Type Declaration ]
6629
 * No element type may be declared more than once
6630
 *
6631
 * @param ctxt  an XML parser context
6632
 * @returns the type of the element, or -1 in case of error
6633
 */
6634
int
6635
36.5k
xmlParseElementDecl(xmlParserCtxt *ctxt) {
6636
36.5k
    const xmlChar *name;
6637
36.5k
    int ret = -1;
6638
36.5k
    xmlElementContentPtr content  = NULL;
6639
6640
36.5k
    if ((CUR != '<') || (NXT(1) != '!'))
6641
0
        return(ret);
6642
36.5k
    SKIP(2);
6643
6644
    /* GROW; done in the caller */
6645
36.5k
    if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) {
6646
36.4k
#ifdef LIBXML_VALID_ENABLED
6647
36.4k
  int oldInputNr = ctxt->inputNr;
6648
36.4k
#endif
6649
6650
36.4k
  SKIP(7);
6651
36.4k
  if (SKIP_BLANKS_PE == 0) {
6652
244
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6653
244
               "Space required after 'ELEMENT'\n");
6654
244
      return(-1);
6655
244
  }
6656
36.2k
        name = xmlParseName(ctxt);
6657
36.2k
  if (name == NULL) {
6658
318
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6659
318
         "xmlParseElementDecl: no name for Element\n");
6660
318
      return(-1);
6661
318
  }
6662
35.8k
  if (SKIP_BLANKS_PE == 0) {
6663
1.52k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6664
1.52k
         "Space required after the element name\n");
6665
1.52k
  }
6666
35.8k
  if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) {
6667
1.49k
      SKIP(5);
6668
      /*
6669
       * Element must always be empty.
6670
       */
6671
1.49k
      ret = XML_ELEMENT_TYPE_EMPTY;
6672
34.4k
  } else if ((RAW == 'A') && (NXT(1) == 'N') &&
6673
1.92k
             (NXT(2) == 'Y')) {
6674
1.91k
      SKIP(3);
6675
      /*
6676
       * Element is a generic container.
6677
       */
6678
1.91k
      ret = XML_ELEMENT_TYPE_ANY;
6679
32.4k
  } else if (RAW == '(') {
6680
31.4k
      ret = xmlParseElementContentDecl(ctxt, name, &content);
6681
31.4k
            if (ret <= 0)
6682
3.13k
                return(-1);
6683
31.4k
  } else {
6684
      /*
6685
       * [ WFC: PEs in Internal Subset ] error handling.
6686
       */
6687
1.06k
            xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6688
1.06k
                  "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n");
6689
1.06k
      return(-1);
6690
1.06k
  }
6691
6692
31.6k
  SKIP_BLANKS_PE;
6693
6694
31.6k
  if (RAW != '>') {
6695
3.35k
      xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
6696
3.35k
      if (content != NULL) {
6697
2.95k
    xmlFreeDocElementContent(ctxt->myDoc, content);
6698
2.95k
      }
6699
28.3k
  } else {
6700
28.3k
#ifdef LIBXML_VALID_ENABLED
6701
28.3k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6702
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6703
0
                                 "Element declaration doesn't start and stop in"
6704
0
                                 " the same entity\n",
6705
0
                                 NULL, NULL);
6706
0
      }
6707
28.3k
#endif
6708
6709
28.3k
      NEXT;
6710
28.3k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6711
13.0k
    (ctxt->sax->elementDecl != NULL)) {
6712
13.0k
    if (content != NULL)
6713
10.3k
        content->parent = NULL;
6714
13.0k
          ctxt->sax->elementDecl(ctxt->userData, name, ret,
6715
13.0k
                           content);
6716
13.0k
    if ((content != NULL) && (content->parent == NULL)) {
6717
        /*
6718
         * this is a trick: if xmlAddElementDecl is called,
6719
         * instead of copying the full tree it is plugged directly
6720
         * if called from the parser. Avoid duplicating the
6721
         * interfaces or change the API/ABI
6722
         */
6723
5.76k
        xmlFreeDocElementContent(ctxt->myDoc, content);
6724
5.76k
    }
6725
15.3k
      } else if (content != NULL) {
6726
14.9k
    xmlFreeDocElementContent(ctxt->myDoc, content);
6727
14.9k
      }
6728
28.3k
  }
6729
31.6k
    }
6730
31.7k
    return(ret);
6731
36.5k
}
6732
6733
/**
6734
 * Parse a conditional section. Always consumes '<!['.
6735
 *
6736
 *     [61] conditionalSect ::= includeSect | ignoreSect
6737
 *     [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>'
6738
 *     [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>'
6739
 *     [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>'
6740
 *                                 Ignore)*
6741
 *     [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*)
6742
 * @param ctxt  an XML parser context
6743
 */
6744
6745
static void
6746
3.41k
xmlParseConditionalSections(xmlParserCtxtPtr ctxt) {
6747
3.41k
    size_t depth = 0;
6748
3.41k
    int isFreshPE = 0;
6749
3.41k
    int oldInputNr = ctxt->inputNr;
6750
3.41k
    int declInputNr = ctxt->inputNr;
6751
6752
32.8k
    while (!PARSER_STOPPED(ctxt)) {
6753
32.8k
        if (ctxt->input->cur >= ctxt->input->end) {
6754
950
            if (ctxt->inputNr <= oldInputNr) {
6755
436
                xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6756
436
                return;
6757
436
            }
6758
6759
514
            xmlPopPE(ctxt);
6760
514
            declInputNr = ctxt->inputNr;
6761
31.8k
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6762
3.80k
            SKIP(3);
6763
3.80k
            SKIP_BLANKS_PE;
6764
6765
3.80k
            isFreshPE = 0;
6766
6767
3.80k
            if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) {
6768
1.95k
                SKIP(7);
6769
1.95k
                SKIP_BLANKS_PE;
6770
1.95k
                if (RAW != '[') {
6771
375
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6772
375
                    return;
6773
375
                }
6774
1.57k
#ifdef LIBXML_VALID_ENABLED
6775
1.57k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6776
0
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6777
0
                                     "All markup of the conditional section is"
6778
0
                                     " not in the same entity\n",
6779
0
                                     NULL, NULL);
6780
0
                }
6781
1.57k
#endif
6782
1.57k
                NEXT;
6783
6784
1.57k
                depth++;
6785
1.85k
            } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) {
6786
1.34k
                size_t ignoreDepth = 0;
6787
6788
1.34k
                SKIP(6);
6789
1.34k
                SKIP_BLANKS_PE;
6790
1.34k
                if (RAW != '[') {
6791
64
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6792
64
                    return;
6793
64
                }
6794
1.28k
#ifdef LIBXML_VALID_ENABLED
6795
1.28k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6796
0
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6797
0
                                     "All markup of the conditional section is"
6798
0
                                     " not in the same entity\n",
6799
0
                                     NULL, NULL);
6800
0
                }
6801
1.28k
#endif
6802
1.28k
                NEXT;
6803
6804
751k
                while (PARSER_STOPPED(ctxt) == 0) {
6805
751k
                    if (RAW == 0) {
6806
467
                        xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL);
6807
467
                        return;
6808
467
                    }
6809
751k
                    if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6810
3.50k
                        SKIP(3);
6811
3.50k
                        ignoreDepth++;
6812
                        /* Check for integer overflow */
6813
3.50k
                        if (ignoreDepth == 0) {
6814
0
                            xmlErrMemory(ctxt);
6815
0
                            return;
6816
0
                        }
6817
747k
                    } else if ((RAW == ']') && (NXT(1) == ']') &&
6818
2.06k
                               (NXT(2) == '>')) {
6819
1.70k
                        SKIP(3);
6820
1.70k
                        if (ignoreDepth == 0)
6821
813
                            break;
6822
896
                        ignoreDepth--;
6823
746k
                    } else {
6824
746k
                        NEXT;
6825
746k
                    }
6826
751k
                }
6827
6828
816
#ifdef LIBXML_VALID_ENABLED
6829
816
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6830
0
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6831
0
                                     "All markup of the conditional section is"
6832
0
                                     " not in the same entity\n",
6833
0
                                     NULL, NULL);
6834
0
                }
6835
816
#endif
6836
816
            } else {
6837
506
                xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL);
6838
506
                return;
6839
506
            }
6840
28.0k
        } else if ((depth > 0) &&
6841
28.0k
                   (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) {
6842
702
            if (isFreshPE) {
6843
1
                xmlFatalErrMsg(ctxt, XML_ERR_CONDSEC_INVALID,
6844
1
                               "Parameter entity must match "
6845
1
                               "extSubsetDecl\n");
6846
1
                return;
6847
1
            }
6848
6849
701
            depth--;
6850
701
#ifdef LIBXML_VALID_ENABLED
6851
701
            if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6852
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6853
0
                                 "All markup of the conditional section is not"
6854
0
                                 " in the same entity\n",
6855
0
                                 NULL, NULL);
6856
0
            }
6857
701
#endif
6858
701
            SKIP(3);
6859
27.3k
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
6860
26.2k
            isFreshPE = 0;
6861
26.2k
            xmlParseMarkupDecl(ctxt);
6862
26.2k
        } else if (RAW == '%') {
6863
1.03k
            xmlParsePERefInternal(ctxt, 1);
6864
1.03k
            if (ctxt->inputNr > declInputNr) {
6865
512
                isFreshPE = 1;
6866
512
                declInputNr = ctxt->inputNr;
6867
512
            }
6868
1.03k
        } else {
6869
47
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6870
47
            return;
6871
47
        }
6872
6873
30.9k
        if (depth == 0)
6874
1.51k
            break;
6875
6876
29.3k
        SKIP_BLANKS;
6877
29.3k
        SHRINK;
6878
29.3k
        GROW;
6879
29.3k
    }
6880
3.41k
}
6881
6882
/**
6883
 * Parse markup declarations. Always consumes '<!' or '<?'.
6884
 *
6885
 * @deprecated Internal function, don't use.
6886
 *
6887
 *     [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl |
6888
 *                         NotationDecl | PI | Comment
6889
 *
6890
 * [ VC: Proper Declaration/PE Nesting ]
6891
 * Parameter-entity replacement text must be properly nested with
6892
 * markup declarations. That is to say, if either the first character
6893
 * or the last character of a markup declaration (markupdecl above) is
6894
 * contained in the replacement text for a parameter-entity reference,
6895
 * both must be contained in the same replacement text.
6896
 *
6897
 * [ WFC: PEs in Internal Subset ]
6898
 * In the internal DTD subset, parameter-entity references can occur
6899
 * only where markup declarations can occur, not within markup declarations.
6900
 * (This does not apply to references that occur in external parameter
6901
 * entities or to the external subset.)
6902
 *
6903
 * @param ctxt  an XML parser context
6904
 */
6905
void
6906
984k
xmlParseMarkupDecl(xmlParserCtxt *ctxt) {
6907
984k
    GROW;
6908
984k
    if (CUR == '<') {
6909
984k
        if (NXT(1) == '!') {
6910
952k
      switch (NXT(2)) {
6911
137k
          case 'E':
6912
137k
        if (NXT(3) == 'L')
6913
36.5k
      xmlParseElementDecl(ctxt);
6914
101k
        else if (NXT(3) == 'N')
6915
101k
      xmlParseEntityDecl(ctxt);
6916
31
                    else
6917
31
                        SKIP(2);
6918
137k
        break;
6919
63.3k
          case 'A':
6920
63.3k
        xmlParseAttributeListDecl(ctxt);
6921
63.3k
        break;
6922
4.52k
          case 'N':
6923
4.52k
        xmlParseNotationDecl(ctxt);
6924
4.52k
        break;
6925
733k
          case '-':
6926
733k
        xmlParseComment(ctxt);
6927
733k
        break;
6928
12.9k
    default:
6929
12.9k
                    xmlFatalErr(ctxt,
6930
12.9k
                                ctxt->inSubset == 2 ?
6931
685
                                    XML_ERR_EXT_SUBSET_NOT_FINISHED :
6932
12.9k
                                    XML_ERR_INT_SUBSET_NOT_FINISHED,
6933
12.9k
                                NULL);
6934
12.9k
                    SKIP(2);
6935
12.9k
        break;
6936
952k
      }
6937
952k
  } else if (NXT(1) == '?') {
6938
31.9k
      xmlParsePI(ctxt);
6939
31.9k
  }
6940
984k
    }
6941
984k
}
6942
6943
/**
6944
 * Parse an XML declaration header for external entities
6945
 *
6946
 * @deprecated Internal function, don't use.
6947
 *
6948
 *     [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>'
6949
 * @param ctxt  an XML parser context
6950
 */
6951
6952
void
6953
30.3k
xmlParseTextDecl(xmlParserCtxt *ctxt) {
6954
30.3k
    xmlChar *version;
6955
6956
    /*
6957
     * We know that '<?xml' is here.
6958
     */
6959
30.3k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
6960
30.3k
  SKIP(5);
6961
30.3k
    } else {
6962
14
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL);
6963
14
  return;
6964
14
    }
6965
6966
30.3k
    if (SKIP_BLANKS == 0) {
6967
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6968
0
           "Space needed after '<?xml'\n");
6969
0
    }
6970
6971
    /*
6972
     * We may have the VersionInfo here.
6973
     */
6974
30.3k
    version = xmlParseVersionInfo(ctxt);
6975
30.3k
    if (version == NULL) {
6976
25.4k
  version = xmlCharStrdup(XML_DEFAULT_VERSION);
6977
25.4k
        if (version == NULL) {
6978
4
            xmlErrMemory(ctxt);
6979
4
            return;
6980
4
        }
6981
25.4k
    } else {
6982
4.90k
  if (SKIP_BLANKS == 0) {
6983
807
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6984
807
               "Space needed here\n");
6985
807
  }
6986
4.90k
    }
6987
30.3k
    ctxt->input->version = version;
6988
6989
    /*
6990
     * We must have the encoding declaration
6991
     */
6992
30.3k
    xmlParseEncodingDecl(ctxt);
6993
6994
30.3k
    SKIP_BLANKS;
6995
30.3k
    if ((RAW == '?') && (NXT(1) == '>')) {
6996
1.51k
        SKIP(2);
6997
28.8k
    } else if (RAW == '>') {
6998
        /* Deprecated old WD ... */
6999
201
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7000
201
  NEXT;
7001
28.6k
    } else {
7002
28.6k
        int c;
7003
7004
28.6k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7005
45.1M
        while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) {
7006
45.0M
            NEXT;
7007
45.0M
            if (c == '>')
7008
19.2k
                break;
7009
45.0M
        }
7010
28.6k
    }
7011
30.3k
}
7012
7013
/**
7014
 * Parse Markup declarations from an external subset
7015
 *
7016
 * @deprecated Internal function, don't use.
7017
 *
7018
 *     [30] extSubset ::= textDecl? extSubsetDecl
7019
 *
7020
 *     [31] extSubsetDecl ::= (markupdecl | conditionalSect |
7021
 *                             PEReference | S) *
7022
 * @param ctxt  an XML parser context
7023
 * @param publicId  the public identifier
7024
 * @param systemId  the system identifier (URL)
7025
 */
7026
void
7027
xmlParseExternalSubset(xmlParserCtxt *ctxt, const xmlChar *publicId,
7028
2.07k
                       const xmlChar *systemId) {
7029
2.07k
    int oldInputNr;
7030
7031
2.07k
    xmlCtxtInitializeLate(ctxt);
7032
7033
2.07k
    xmlDetectEncoding(ctxt);
7034
7035
2.07k
    if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) {
7036
67
  xmlParseTextDecl(ctxt);
7037
67
    }
7038
2.07k
    if (ctxt->myDoc == NULL) {
7039
0
        ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0");
7040
0
  if (ctxt->myDoc == NULL) {
7041
0
      xmlErrMemory(ctxt);
7042
0
      return;
7043
0
  }
7044
0
  ctxt->myDoc->properties = XML_DOC_INTERNAL;
7045
0
    }
7046
2.07k
    if ((ctxt->myDoc->intSubset == NULL) &&
7047
0
        (xmlCreateIntSubset(ctxt->myDoc, NULL, publicId, systemId) == NULL)) {
7048
0
        xmlErrMemory(ctxt);
7049
0
    }
7050
7051
2.07k
    ctxt->inSubset = 2;
7052
2.07k
    oldInputNr = ctxt->inputNr;
7053
7054
2.07k
    SKIP_BLANKS;
7055
739k
    while (!PARSER_STOPPED(ctxt)) {
7056
738k
        if (ctxt->input->cur >= ctxt->input->end) {
7057
934
            if (ctxt->inputNr <= oldInputNr) {
7058
358
                xmlParserCheckEOF(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED);
7059
358
                break;
7060
358
            }
7061
7062
576
            xmlPopPE(ctxt);
7063
737k
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
7064
1.13k
            xmlParseConditionalSections(ctxt);
7065
736k
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
7066
734k
            xmlParseMarkupDecl(ctxt);
7067
734k
        } else if (RAW == '%') {
7068
1.26k
            xmlParsePERefInternal(ctxt, 1);
7069
1.58k
        } else {
7070
1.58k
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
7071
7072
1.68k
            while (ctxt->inputNr > oldInputNr)
7073
102
                xmlPopPE(ctxt);
7074
1.58k
            break;
7075
1.58k
        }
7076
736k
        SKIP_BLANKS;
7077
736k
        SHRINK;
7078
736k
        GROW;
7079
736k
    }
7080
2.07k
}
7081
7082
/**
7083
 * Parse and handle entity references in content, depending on the SAX
7084
 * interface, this may end-up in a call to character() if this is a
7085
 * CharRef, a predefined entity, if there is no reference() callback.
7086
 * or if the parser was asked to switch to that mode.
7087
 *
7088
 * @deprecated Internal function, don't use.
7089
 *
7090
 * Always consumes '&'.
7091
 *
7092
 *     [67] Reference ::= EntityRef | CharRef
7093
 * @param ctxt  an XML parser context
7094
 */
7095
void
7096
159k
xmlParseReference(xmlParserCtxt *ctxt) {
7097
159k
    xmlEntityPtr ent = NULL;
7098
159k
    const xmlChar *name;
7099
159k
    xmlChar *val;
7100
7101
159k
    if (RAW != '&')
7102
0
        return;
7103
7104
    /*
7105
     * Simple case of a CharRef
7106
     */
7107
159k
    if (NXT(1) == '#') {
7108
24.5k
  int i = 0;
7109
24.5k
  xmlChar out[16];
7110
24.5k
  int value = xmlParseCharRef(ctxt);
7111
7112
24.5k
  if (value == 0)
7113
16.0k
      return;
7114
7115
        /*
7116
         * Just encode the value in UTF-8
7117
         */
7118
8.52k
        COPY_BUF(out, i, value);
7119
8.52k
        out[i] = 0;
7120
8.52k
        if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7121
8.52k
            (!ctxt->disableSAX))
7122
6.63k
            ctxt->sax->characters(ctxt->userData, out, i);
7123
8.52k
  return;
7124
24.5k
    }
7125
7126
    /*
7127
     * We are seeing an entity reference
7128
     */
7129
135k
    name = xmlParseEntityRefInternal(ctxt);
7130
135k
    if (name == NULL)
7131
49.0k
        return;
7132
86.4k
    ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0);
7133
86.4k
    if (ent == NULL) {
7134
        /*
7135
         * Create a reference for undeclared entities.
7136
         */
7137
51.5k
        if ((ctxt->replaceEntities == 0) &&
7138
21.5k
            (ctxt->sax != NULL) &&
7139
21.5k
            (ctxt->disableSAX == 0) &&
7140
20.3k
            (ctxt->sax->reference != NULL)) {
7141
20.3k
            ctxt->sax->reference(ctxt->userData, name);
7142
20.3k
        }
7143
51.5k
        return;
7144
51.5k
    }
7145
34.8k
    if (!ctxt->wellFormed)
7146
24.1k
  return;
7147
7148
    /* special case of predefined entities */
7149
10.7k
    if ((ent->name == NULL) ||
7150
10.7k
        (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
7151
267
  val = ent->content;
7152
267
  if (val == NULL) return;
7153
  /*
7154
   * inline the entity.
7155
   */
7156
267
  if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7157
267
      (!ctxt->disableSAX))
7158
267
      ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val));
7159
267
  return;
7160
267
    }
7161
7162
    /*
7163
     * Some users try to parse entities on their own and used to set
7164
     * the renamed "checked" member. Fix the flags to cover this
7165
     * case.
7166
     */
7167
10.4k
    if (((ent->flags & XML_ENT_PARSED) == 0) && (ent->children != NULL))
7168
0
        ent->flags |= XML_ENT_PARSED;
7169
7170
    /*
7171
     * The first reference to the entity trigger a parsing phase
7172
     * where the ent->children is filled with the result from
7173
     * the parsing.
7174
     * Note: external parsed entities will not be loaded, it is not
7175
     * required for a non-validating parser, unless the parsing option
7176
     * of validating, or substituting entities were given. Doing so is
7177
     * far more secure as the parser will only process data coming from
7178
     * the document entity by default.
7179
     *
7180
     * FIXME: This doesn't work correctly since entities can be
7181
     * expanded with different namespace declarations in scope.
7182
     * For example:
7183
     *
7184
     * <!DOCTYPE doc [
7185
     *   <!ENTITY ent "<ns:elem/>">
7186
     * ]>
7187
     * <doc>
7188
     *   <decl1 xmlns:ns="urn:ns1">
7189
     *     &ent;
7190
     *   </decl1>
7191
     *   <decl2 xmlns:ns="urn:ns2">
7192
     *     &ent;
7193
     *   </decl2>
7194
     * </doc>
7195
     *
7196
     * Proposed fix:
7197
     *
7198
     * - Ignore current namespace declarations when parsing the
7199
     *   entity. If a prefix can't be resolved, don't report an error
7200
     *   but mark it as unresolved.
7201
     * - Try to resolve these prefixes when expanding the entity.
7202
     *   This will require a specialized version of xmlStaticCopyNode
7203
     *   which can also make use of the namespace hash table to avoid
7204
     *   quadratic behavior.
7205
     *
7206
     * Alternatively, we could simply reparse the entity on each
7207
     * expansion like we already do with custom SAX callbacks.
7208
     * External entity content should be cached in this case.
7209
     */
7210
10.4k
    if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) ||
7211
7.57k
        (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
7212
7.26k
         ((ctxt->replaceEntities) ||
7213
9.70k
          (ctxt->validate)))) {
7214
9.70k
        if ((ent->flags & XML_ENT_PARSED) == 0) {
7215
3.77k
            xmlCtxtParseEntity(ctxt, ent);
7216
5.93k
        } else if (ent->children == NULL) {
7217
            /*
7218
             * Probably running in SAX mode and the callbacks don't
7219
             * build the entity content. Parse the entity again.
7220
             *
7221
             * This will also be triggered in normal tree builder mode
7222
             * if an entity happens to be empty, causing unnecessary
7223
             * reloads. It's hard to come up with a reliable check in
7224
             * which mode we're running.
7225
             */
7226
3.47k
            xmlCtxtParseEntity(ctxt, ent);
7227
3.47k
        }
7228
9.70k
    }
7229
7230
    /*
7231
     * We also check for amplification if entities aren't substituted.
7232
     * They might be expanded later.
7233
     */
7234
10.4k
    if (xmlParserEntityCheck(ctxt, ent->expandedSize))
7235
58
        return;
7236
7237
10.4k
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
7238
670
        return;
7239
7240
9.73k
    if (ctxt->replaceEntities == 0) {
7241
  /*
7242
   * Create a reference
7243
   */
7244
1.35k
        if (ctxt->sax->reference != NULL)
7245
1.35k
      ctxt->sax->reference(ctxt->userData, ent->name);
7246
8.37k
    } else if ((ent->children != NULL) && (ctxt->node != NULL)) {
7247
4.74k
        xmlNodePtr copy, cur;
7248
7249
        /*
7250
         * Seems we are generating the DOM content, copy the tree
7251
   */
7252
4.74k
        cur = ent->children;
7253
7254
        /*
7255
         * Handle first text node with SAX to coalesce text efficiently
7256
         */
7257
4.74k
        if ((cur->type == XML_TEXT_NODE) ||
7258
4.29k
            (cur->type == XML_CDATA_SECTION_NODE)) {
7259
4.29k
            int len = xmlStrlen(cur->content);
7260
7261
4.29k
            if ((cur->type == XML_TEXT_NODE) ||
7262
4.25k
                (ctxt->options & XML_PARSE_NOCDATA)) {
7263
4.25k
                if (ctxt->sax->characters != NULL)
7264
4.25k
                    ctxt->sax->characters(ctxt, cur->content, len);
7265
4.25k
            } else {
7266
35
                if (ctxt->sax->cdataBlock != NULL)
7267
35
                    ctxt->sax->cdataBlock(ctxt, cur->content, len);
7268
35
            }
7269
7270
4.29k
            cur = cur->next;
7271
4.29k
        }
7272
7273
11.8k
        while (cur != NULL) {
7274
7.63k
            xmlNodePtr last;
7275
7276
            /*
7277
             * Handle last text node with SAX to coalesce text efficiently
7278
             */
7279
7.63k
            if ((cur->next == NULL) &&
7280
3.32k
                ((cur->type == XML_TEXT_NODE) ||
7281
2.86k
                 (cur->type == XML_CDATA_SECTION_NODE))) {
7282
489
                int len = xmlStrlen(cur->content);
7283
7284
489
                if ((cur->type == XML_TEXT_NODE) ||
7285
454
                    (ctxt->options & XML_PARSE_NOCDATA)) {
7286
454
                    if (ctxt->sax->characters != NULL)
7287
454
                        ctxt->sax->characters(ctxt, cur->content, len);
7288
454
                } else {
7289
35
                    if (ctxt->sax->cdataBlock != NULL)
7290
35
                        ctxt->sax->cdataBlock(ctxt, cur->content, len);
7291
35
                }
7292
7293
489
                break;
7294
489
            }
7295
7296
            /*
7297
             * Reset coalesce buffer stats only for non-text nodes.
7298
             */
7299
7.14k
            ctxt->nodemem = 0;
7300
7.14k
            ctxt->nodelen = 0;
7301
7302
7.14k
            copy = xmlDocCopyNode(cur, ctxt->myDoc, 1);
7303
7304
7.14k
            if (copy == NULL) {
7305
10
                xmlErrMemory(ctxt);
7306
10
                break;
7307
10
            }
7308
7309
7.13k
            if (ctxt->parseMode == XML_PARSE_READER) {
7310
                /* Needed for reader */
7311
0
                copy->extra = cur->extra;
7312
                /* Maybe needed for reader */
7313
0
                copy->_private = cur->_private;
7314
0
            }
7315
7316
7.13k
            copy->parent = ctxt->node;
7317
7.13k
            last = ctxt->node->last;
7318
7.13k
            if (last == NULL) {
7319
209
                ctxt->node->children = copy;
7320
6.92k
            } else {
7321
6.92k
                last->next = copy;
7322
6.92k
                copy->prev = last;
7323
6.92k
            }
7324
7.13k
            ctxt->node->last = copy;
7325
7326
7.13k
            cur = cur->next;
7327
7.13k
        }
7328
4.74k
    }
7329
9.73k
}
7330
7331
static void
7332
380k
xmlHandleUndeclaredEntity(xmlParserCtxtPtr ctxt, const xmlChar *name) {
7333
    /*
7334
     * [ WFC: Entity Declared ]
7335
     * In a document without any DTD, a document with only an
7336
     * internal DTD subset which contains no parameter entity
7337
     * references, or a document with "standalone='yes'", the
7338
     * Name given in the entity reference must match that in an
7339
     * entity declaration, except that well-formed documents
7340
     * need not declare any of the following entities: amp, lt,
7341
     * gt, apos, quot.
7342
     * The declaration of a parameter entity must precede any
7343
     * reference to it.
7344
     * Similarly, the declaration of a general entity must
7345
     * precede any reference to it which appears in a default
7346
     * value in an attribute-list declaration. Note that if
7347
     * entities are declared in the external subset or in
7348
     * external parameter entities, a non-validating processor
7349
     * is not obligated to read and process their declarations;
7350
     * for such documents, the rule that an entity must be
7351
     * declared is a well-formedness constraint only if
7352
     * standalone='yes'.
7353
     */
7354
380k
    if ((ctxt->standalone == 1) ||
7355
379k
        ((ctxt->hasExternalSubset == 0) &&
7356
377k
         (ctxt->hasPErefs == 0))) {
7357
272k
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY,
7358
272k
                          "Entity '%s' not defined\n", name);
7359
272k
#ifdef LIBXML_VALID_ENABLED
7360
272k
    } else if (ctxt->validate) {
7361
        /*
7362
         * [ VC: Entity Declared ]
7363
         * In a document with an external subset or external
7364
         * parameter entities with "standalone='no'", ...
7365
         * ... The declaration of a parameter entity must
7366
         * precede any reference to it...
7367
         */
7368
0
        xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY,
7369
0
                         "Entity '%s' not defined\n", name, NULL);
7370
0
#endif
7371
107k
    } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) ||
7372
8.18k
               ((ctxt->replaceEntities) &&
7373
106k
                ((ctxt->options & XML_PARSE_NO_XXE) == 0))) {
7374
        /*
7375
         * Also raise a non-fatal error
7376
         *
7377
         * - if the external subset is loaded and all entity declarations
7378
         *   should be available, or
7379
         * - entity substition was requested without restricting
7380
         *   external entity access.
7381
         */
7382
106k
        xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY,
7383
106k
                     "Entity '%s' not defined\n", name);
7384
106k
    } else {
7385
1.42k
        xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7386
1.42k
                      "Entity '%s' not defined\n", name, NULL);
7387
1.42k
    }
7388
7389
380k
    ctxt->valid = 0;
7390
380k
}
7391
7392
static xmlEntityPtr
7393
1.29M
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) {
7394
1.29M
    xmlEntityPtr ent = NULL;
7395
7396
    /*
7397
     * Predefined entities override any extra definition
7398
     */
7399
1.29M
    if ((ctxt->options & XML_PARSE_OLDSAX) == 0) {
7400
687k
        ent = xmlGetPredefinedEntity(name);
7401
687k
        if (ent != NULL)
7402
116k
            return(ent);
7403
687k
    }
7404
7405
    /*
7406
     * Ask first SAX for entity resolution, otherwise try the
7407
     * entities which may have stored in the parser context.
7408
     */
7409
1.17M
    if (ctxt->sax != NULL) {
7410
1.17M
  if (ctxt->sax->getEntity != NULL)
7411
1.17M
      ent = ctxt->sax->getEntity(ctxt->userData, name);
7412
1.17M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7413
7.50k
      (ctxt->options & XML_PARSE_OLDSAX))
7414
1.15k
      ent = xmlGetPredefinedEntity(name);
7415
1.17M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7416
7.50k
      (ctxt->userData==ctxt)) {
7417
7.50k
      ent = xmlSAX2GetEntity(ctxt, name);
7418
7.50k
  }
7419
1.17M
    }
7420
7421
1.17M
    if (ent == NULL) {
7422
294k
        xmlHandleUndeclaredEntity(ctxt, name);
7423
294k
    }
7424
7425
    /*
7426
     * [ WFC: Parsed Entity ]
7427
     * An entity reference must not contain the name of an
7428
     * unparsed entity
7429
     */
7430
879k
    else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) {
7431
386
  xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY,
7432
386
     "Entity reference to unparsed entity %s\n", name);
7433
386
        ent = NULL;
7434
386
    }
7435
7436
    /*
7437
     * [ WFC: No External Entity References ]
7438
     * Attribute values cannot contain direct or indirect
7439
     * entity references to external entities.
7440
     */
7441
879k
    else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) {
7442
17.2k
        if (inAttr) {
7443
1.70k
            xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL,
7444
1.70k
                 "Attribute references external entity '%s'\n", name);
7445
1.70k
            ent = NULL;
7446
1.70k
        }
7447
17.2k
    }
7448
7449
1.17M
    return(ent);
7450
1.29M
}
7451
7452
/**
7453
 * Parse an entity reference. Always consumes '&'.
7454
 *
7455
 *     [68] EntityRef ::= '&' Name ';'
7456
 *
7457
 * @param ctxt  an XML parser context
7458
 * @returns the name, or NULL in case of error.
7459
 */
7460
static const xmlChar *
7461
749k
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) {
7462
749k
    const xmlChar *name;
7463
7464
749k
    GROW;
7465
7466
749k
    if (RAW != '&')
7467
0
        return(NULL);
7468
749k
    NEXT;
7469
749k
    name = xmlParseName(ctxt);
7470
749k
    if (name == NULL) {
7471
145k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7472
145k
           "xmlParseEntityRef: no name\n");
7473
145k
        return(NULL);
7474
145k
    }
7475
603k
    if (RAW != ';') {
7476
93.2k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7477
93.2k
  return(NULL);
7478
93.2k
    }
7479
510k
    NEXT;
7480
7481
510k
    return(name);
7482
603k
}
7483
7484
/**
7485
 * @deprecated Internal function, don't use.
7486
 *
7487
 * @param ctxt  an XML parser context
7488
 * @returns the xmlEntity if found, or NULL otherwise.
7489
 */
7490
xmlEntity *
7491
0
xmlParseEntityRef(xmlParserCtxt *ctxt) {
7492
0
    const xmlChar *name;
7493
7494
0
    if (ctxt == NULL)
7495
0
        return(NULL);
7496
7497
0
    name = xmlParseEntityRefInternal(ctxt);
7498
0
    if (name == NULL)
7499
0
        return(NULL);
7500
7501
0
    return(xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 0));
7502
0
}
7503
7504
/**
7505
 * Parse ENTITY references declarations, but this version parses it from
7506
 * a string value.
7507
 *
7508
 *     [68] EntityRef ::= '&' Name ';'
7509
 *
7510
 * [ WFC: Entity Declared ]
7511
 * In a document without any DTD, a document with only an internal DTD
7512
 * subset which contains no parameter entity references, or a document
7513
 * with "standalone='yes'", the Name given in the entity reference
7514
 * must match that in an entity declaration, except that well-formed
7515
 * documents need not declare any of the following entities: amp, lt,
7516
 * gt, apos, quot.  The declaration of a parameter entity must precede
7517
 * any reference to it.  Similarly, the declaration of a general entity
7518
 * must precede any reference to it which appears in a default value in an
7519
 * attribute-list declaration. Note that if entities are declared in the
7520
 * external subset or in external parameter entities, a non-validating
7521
 * processor is not obligated to read and process their declarations;
7522
 * for such documents, the rule that an entity must be declared is a
7523
 * well-formedness constraint only if standalone='yes'.
7524
 *
7525
 * [ WFC: Parsed Entity ]
7526
 * An entity reference must not contain the name of an unparsed entity
7527
 *
7528
 * @param ctxt  an XML parser context
7529
 * @param str  a pointer to an index in the string
7530
 * @returns the xmlEntity if found, or NULL otherwise. The str pointer
7531
 * is updated to the current location in the string.
7532
 */
7533
static xmlChar *
7534
780k
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) {
7535
780k
    xmlChar *name;
7536
780k
    const xmlChar *ptr;
7537
780k
    xmlChar cur;
7538
7539
780k
    if ((str == NULL) || (*str == NULL))
7540
0
        return(NULL);
7541
780k
    ptr = *str;
7542
780k
    cur = *ptr;
7543
780k
    if (cur != '&')
7544
0
  return(NULL);
7545
7546
780k
    ptr++;
7547
780k
    name = xmlParseStringName(ctxt, &ptr);
7548
780k
    if (name == NULL) {
7549
24
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7550
24
           "xmlParseStringEntityRef: no name\n");
7551
24
  *str = ptr;
7552
24
  return(NULL);
7553
24
    }
7554
780k
    if (*ptr != ';') {
7555
11
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7556
11
        xmlFree(name);
7557
11
  *str = ptr;
7558
11
  return(NULL);
7559
11
    }
7560
780k
    ptr++;
7561
7562
780k
    *str = ptr;
7563
780k
    return(name);
7564
780k
}
7565
7566
/**
7567
 * Parse a parameter entity reference. Always consumes '%'.
7568
 *
7569
 * The entity content is handled directly by pushing it's content as
7570
 * a new input stream.
7571
 *
7572
 *     [69] PEReference ::= '%' Name ';'
7573
 *
7574
 * [ WFC: No Recursion ]
7575
 * A parsed entity must not contain a recursive
7576
 * reference to itself, either directly or indirectly.
7577
 *
7578
 * [ WFC: Entity Declared ]
7579
 * In a document without any DTD, a document with only an internal DTD
7580
 * subset which contains no parameter entity references, or a document
7581
 * with "standalone='yes'", ...  ... The declaration of a parameter
7582
 * entity must precede any reference to it...
7583
 *
7584
 * [ VC: Entity Declared ]
7585
 * In a document with an external subset or external parameter entities
7586
 * with "standalone='no'", ...  ... The declaration of a parameter entity
7587
 * must precede any reference to it...
7588
 *
7589
 * [ WFC: In DTD ]
7590
 * Parameter-entity references may only appear in the DTD.
7591
 * NOTE: misleading but this is handled.
7592
 *
7593
 * @param ctxt  an XML parser context
7594
 * @param markupDecl  whether the PERef starts a markup declaration
7595
 */
7596
static void
7597
219k
xmlParsePERefInternal(xmlParserCtxt *ctxt, int markupDecl) {
7598
219k
    const xmlChar *name;
7599
219k
    xmlEntityPtr entity = NULL;
7600
219k
    xmlParserInputPtr input;
7601
7602
219k
    if (RAW != '%')
7603
0
        return;
7604
219k
    NEXT;
7605
219k
    name = xmlParseName(ctxt);
7606
219k
    if (name == NULL) {
7607
10.9k
  xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n");
7608
10.9k
  return;
7609
10.9k
    }
7610
208k
    if (RAW != ';') {
7611
31.1k
  xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL);
7612
31.1k
        return;
7613
31.1k
    }
7614
7615
177k
    NEXT;
7616
7617
    /* Must be set before xmlHandleUndeclaredEntity */
7618
177k
    ctxt->hasPErefs = 1;
7619
7620
    /*
7621
     * Request the entity from SAX
7622
     */
7623
177k
    if ((ctxt->sax != NULL) &&
7624
177k
  (ctxt->sax->getParameterEntity != NULL))
7625
177k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7626
7627
177k
    if (entity == NULL) {
7628
82.8k
        xmlHandleUndeclaredEntity(ctxt, name);
7629
94.4k
    } else {
7630
  /*
7631
   * Internal checking in case the entity quest barfed
7632
   */
7633
94.4k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7634
81.1k
      (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) {
7635
0
      xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7636
0
      "Internal: %%%s; is not a parameter entity\n",
7637
0
        name, NULL);
7638
94.4k
  } else {
7639
94.4k
      if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) &&
7640
81.1k
                ((ctxt->options & XML_PARSE_NO_XXE) ||
7641
80.9k
     (((ctxt->loadsubset & ~XML_SKIP_IDS) == 0) &&
7642
7.96k
      (ctxt->replaceEntities == 0) &&
7643
205
      (ctxt->validate == 0))))
7644
407
    return;
7645
7646
94.0k
            if (entity->flags & XML_ENT_EXPANDING) {
7647
13
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
7648
13
                return;
7649
13
            }
7650
7651
94.0k
      input = xmlNewEntityInputStream(ctxt, entity);
7652
94.0k
      if (xmlCtxtPushInput(ctxt, input) < 0) {
7653
14.0k
                xmlFreeInputStream(input);
7654
14.0k
    return;
7655
14.0k
            }
7656
7657
79.9k
            entity->flags |= XML_ENT_EXPANDING;
7658
7659
79.9k
            if (markupDecl)
7660
71.7k
                input->flags |= XML_INPUT_MARKUP_DECL;
7661
7662
79.9k
            GROW;
7663
7664
79.9k
      if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
7665
66.7k
                xmlDetectEncoding(ctxt);
7666
7667
66.7k
                if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
7668
24.5k
                    (IS_BLANK_CH(NXT(5)))) {
7669
23.6k
                    xmlParseTextDecl(ctxt);
7670
23.6k
                }
7671
66.7k
            }
7672
79.9k
  }
7673
94.4k
    }
7674
177k
}
7675
7676
/**
7677
 * Parse a parameter entity reference.
7678
 *
7679
 * @deprecated Internal function, don't use.
7680
 *
7681
 * @param ctxt  an XML parser context
7682
 */
7683
void
7684
0
xmlParsePEReference(xmlParserCtxt *ctxt) {
7685
0
    xmlParsePERefInternal(ctxt, 0);
7686
0
}
7687
7688
/**
7689
 * Load the content of an entity.
7690
 *
7691
 * @param ctxt  an XML parser context
7692
 * @param entity  an unloaded system entity
7693
 * @returns 0 in case of success and -1 in case of failure
7694
 */
7695
static int
7696
14.2k
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) {
7697
14.2k
    xmlParserInputPtr oldinput, input = NULL;
7698
14.2k
    xmlParserInputPtr *oldinputTab;
7699
14.2k
    xmlChar *oldencoding;
7700
14.2k
    xmlChar *content = NULL;
7701
14.2k
    xmlResourceType rtype;
7702
14.2k
    size_t length, i;
7703
14.2k
    int oldinputNr, oldinputMax;
7704
14.2k
    int ret = -1;
7705
14.2k
    int res;
7706
7707
14.2k
    if ((ctxt == NULL) || (entity == NULL) ||
7708
14.2k
        ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) &&
7709
0
   (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) ||
7710
14.2k
  (entity->content != NULL)) {
7711
0
  xmlFatalErr(ctxt, XML_ERR_ARGUMENT,
7712
0
              "xmlLoadEntityContent parameter error");
7713
0
        return(-1);
7714
0
    }
7715
7716
14.2k
    if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY)
7717
14.2k
        rtype = XML_RESOURCE_PARAMETER_ENTITY;
7718
0
    else
7719
0
        rtype = XML_RESOURCE_GENERAL_ENTITY;
7720
7721
14.2k
    input = xmlLoadResource(ctxt, (char *) entity->URI,
7722
14.2k
                            (char *) entity->ExternalID, rtype);
7723
14.2k
    if (input == NULL)
7724
754
        return(-1);
7725
7726
13.4k
    oldinput = ctxt->input;
7727
13.4k
    oldinputNr = ctxt->inputNr;
7728
13.4k
    oldinputMax = ctxt->inputMax;
7729
13.4k
    oldinputTab = ctxt->inputTab;
7730
13.4k
    oldencoding = ctxt->encoding;
7731
7732
13.4k
    ctxt->input = NULL;
7733
13.4k
    ctxt->inputNr = 0;
7734
13.4k
    ctxt->inputMax = 1;
7735
13.4k
    ctxt->encoding = NULL;
7736
13.4k
    ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr));
7737
13.4k
    if (ctxt->inputTab == NULL) {
7738
2
        xmlErrMemory(ctxt);
7739
2
        xmlFreeInputStream(input);
7740
2
        goto error;
7741
2
    }
7742
7743
13.4k
    xmlBufResetInput(input->buf->buffer, input);
7744
7745
13.4k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
7746
5
        xmlFreeInputStream(input);
7747
5
        goto error;
7748
5
    }
7749
7750
13.4k
    xmlDetectEncoding(ctxt);
7751
7752
    /*
7753
     * Parse a possible text declaration first
7754
     */
7755
13.4k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
7756
6.32k
  xmlParseTextDecl(ctxt);
7757
        /*
7758
         * An XML-1.0 document can't reference an entity not XML-1.0
7759
         */
7760
6.32k
        if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
7761
5.82k
            (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
7762
476
            xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
7763
476
                           "Version mismatch between document and entity\n");
7764
476
        }
7765
6.32k
    }
7766
7767
13.4k
    length = input->cur - input->base;
7768
13.4k
    xmlBufShrink(input->buf->buffer, length);
7769
13.4k
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7770
7771
22.5k
    while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0)
7772
9.05k
        ;
7773
7774
13.4k
    xmlBufResetInput(input->buf->buffer, input);
7775
7776
13.4k
    if (res < 0) {
7777
1.23k
        xmlCtxtErrIO(ctxt, input->buf->error, NULL);
7778
1.23k
        goto error;
7779
1.23k
    }
7780
7781
12.2k
    length = xmlBufUse(input->buf->buffer);
7782
12.2k
    if (length > INT_MAX) {
7783
0
        xmlErrMemory(ctxt);
7784
0
        goto error;
7785
0
    }
7786
7787
12.2k
    content = xmlStrndup(xmlBufContent(input->buf->buffer), length);
7788
12.2k
    if (content == NULL) {
7789
2
        xmlErrMemory(ctxt);
7790
2
        goto error;
7791
2
    }
7792
7793
4.23M
    for (i = 0; i < length; ) {
7794
4.23M
        int clen = length - i;
7795
4.23M
        int c = xmlGetUTF8Char(content + i, &clen);
7796
7797
4.23M
        if ((c < 0) || (!IS_CHAR(c))) {
7798
12.1k
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
7799
12.1k
                              "xmlLoadEntityContent: invalid char value %d\n",
7800
12.1k
                              content[i]);
7801
12.1k
            goto error;
7802
12.1k
        }
7803
4.22M
        i += clen;
7804
4.22M
    }
7805
7806
67
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7807
67
    entity->content = content;
7808
67
    entity->length = length;
7809
67
    content = NULL;
7810
67
    ret = 0;
7811
7812
13.4k
error:
7813
26.9k
    while (ctxt->inputNr > 0)
7814
13.4k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
7815
13.4k
    xmlFree(ctxt->inputTab);
7816
13.4k
    xmlFree(ctxt->encoding);
7817
7818
13.4k
    ctxt->input = oldinput;
7819
13.4k
    ctxt->inputNr = oldinputNr;
7820
13.4k
    ctxt->inputMax = oldinputMax;
7821
13.4k
    ctxt->inputTab = oldinputTab;
7822
13.4k
    ctxt->encoding = oldencoding;
7823
7824
13.4k
    xmlFree(content);
7825
7826
13.4k
    return(ret);
7827
67
}
7828
7829
/**
7830
 * Parse PEReference declarations
7831
 *
7832
 *     [69] PEReference ::= '%' Name ';'
7833
 *
7834
 * [ WFC: No Recursion ]
7835
 * A parsed entity must not contain a recursive
7836
 * reference to itself, either directly or indirectly.
7837
 *
7838
 * [ WFC: Entity Declared ]
7839
 * In a document without any DTD, a document with only an internal DTD
7840
 * subset which contains no parameter entity references, or a document
7841
 * with "standalone='yes'", ...  ... The declaration of a parameter
7842
 * entity must precede any reference to it...
7843
 *
7844
 * [ VC: Entity Declared ]
7845
 * In a document with an external subset or external parameter entities
7846
 * with "standalone='no'", ...  ... The declaration of a parameter entity
7847
 * must precede any reference to it...
7848
 *
7849
 * [ WFC: In DTD ]
7850
 * Parameter-entity references may only appear in the DTD.
7851
 * NOTE: misleading but this is handled.
7852
 *
7853
 * @param ctxt  an XML parser context
7854
 * @param str  a pointer to an index in the string
7855
 * @returns the string of the entity content.
7856
 *         str is updated to the current value of the index
7857
 */
7858
static xmlEntityPtr
7859
32.2k
xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) {
7860
32.2k
    const xmlChar *ptr;
7861
32.2k
    xmlChar cur;
7862
32.2k
    xmlChar *name;
7863
32.2k
    xmlEntityPtr entity = NULL;
7864
7865
32.2k
    if ((str == NULL) || (*str == NULL)) return(NULL);
7866
32.2k
    ptr = *str;
7867
32.2k
    cur = *ptr;
7868
32.2k
    if (cur != '%')
7869
0
        return(NULL);
7870
32.2k
    ptr++;
7871
32.2k
    name = xmlParseStringName(ctxt, &ptr);
7872
32.2k
    if (name == NULL) {
7873
1.68k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7874
1.68k
           "xmlParseStringPEReference: no name\n");
7875
1.68k
  *str = ptr;
7876
1.68k
  return(NULL);
7877
1.68k
    }
7878
30.5k
    cur = *ptr;
7879
30.5k
    if (cur != ';') {
7880
2.71k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7881
2.71k
  xmlFree(name);
7882
2.71k
  *str = ptr;
7883
2.71k
  return(NULL);
7884
2.71k
    }
7885
27.8k
    ptr++;
7886
7887
    /* Must be set before xmlHandleUndeclaredEntity */
7888
27.8k
    ctxt->hasPErefs = 1;
7889
7890
    /*
7891
     * Request the entity from SAX
7892
     */
7893
27.8k
    if ((ctxt->sax != NULL) &&
7894
27.8k
  (ctxt->sax->getParameterEntity != NULL))
7895
27.8k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7896
7897
27.8k
    if (entity == NULL) {
7898
3.22k
        xmlHandleUndeclaredEntity(ctxt, name);
7899
24.6k
    } else {
7900
  /*
7901
   * Internal checking in case the entity quest barfed
7902
   */
7903
24.6k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7904
14.4k
      (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) {
7905
0
      xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7906
0
        "%%%s; is not a parameter entity\n",
7907
0
        name, NULL);
7908
0
  }
7909
24.6k
    }
7910
7911
27.8k
    xmlFree(name);
7912
27.8k
    *str = ptr;
7913
27.8k
    return(entity);
7914
30.5k
}
7915
7916
/**
7917
 * Parse a DOCTYPE declaration
7918
 *
7919
 * @deprecated Internal function, don't use.
7920
 *
7921
 *     [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S?
7922
 *                          ('[' (markupdecl | PEReference | S)* ']' S?)? '>'
7923
 *
7924
 * [ VC: Root Element Type ]
7925
 * The Name in the document type declaration must match the element
7926
 * type of the root element.
7927
 *
7928
 * @param ctxt  an XML parser context
7929
 */
7930
7931
void
7932
14.3k
xmlParseDocTypeDecl(xmlParserCtxt *ctxt) {
7933
14.3k
    const xmlChar *name = NULL;
7934
14.3k
    xmlChar *publicId = NULL;
7935
14.3k
    xmlChar *URI = NULL;
7936
7937
    /*
7938
     * We know that '<!DOCTYPE' has been detected.
7939
     */
7940
14.3k
    SKIP(9);
7941
7942
14.3k
    if (SKIP_BLANKS == 0) {
7943
3.71k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7944
3.71k
                       "Space required after 'DOCTYPE'\n");
7945
3.71k
    }
7946
7947
    /*
7948
     * Parse the DOCTYPE name.
7949
     */
7950
14.3k
    name = xmlParseName(ctxt);
7951
14.3k
    if (name == NULL) {
7952
2.46k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7953
2.46k
           "xmlParseDocTypeDecl : no DOCTYPE name !\n");
7954
2.46k
    }
7955
14.3k
    ctxt->intSubName = name;
7956
7957
14.3k
    SKIP_BLANKS;
7958
7959
    /*
7960
     * Check for public and system identifier (URI)
7961
     */
7962
14.3k
    URI = xmlParseExternalID(ctxt, &publicId, 1);
7963
7964
14.3k
    if ((URI != NULL) || (publicId != NULL)) {
7965
3.10k
        ctxt->hasExternalSubset = 1;
7966
3.10k
    }
7967
14.3k
    ctxt->extSubURI = URI;
7968
14.3k
    ctxt->extSubSystem = publicId;
7969
7970
14.3k
    SKIP_BLANKS;
7971
7972
    /*
7973
     * Create and update the internal subset.
7974
     */
7975
14.3k
    if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) &&
7976
14.3k
  (!ctxt->disableSAX))
7977
13.3k
  ctxt->sax->internalSubset(ctxt->userData, name, publicId, URI);
7978
7979
14.3k
    if ((RAW != '[') && (RAW != '>')) {
7980
312
  xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
7981
312
    }
7982
14.3k
}
7983
7984
/**
7985
 * Parse the internal subset declaration
7986
 *
7987
 *     [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>'
7988
 * @param ctxt  an XML parser context
7989
 */
7990
7991
static void
7992
11.3k
xmlParseInternalSubset(xmlParserCtxtPtr ctxt) {
7993
    /*
7994
     * Is there any DTD definition ?
7995
     */
7996
11.3k
    if (RAW == '[') {
7997
11.3k
        int oldInputNr = ctxt->inputNr;
7998
7999
11.3k
        NEXT;
8000
  /*
8001
   * Parse the succession of Markup declarations and
8002
   * PEReferences.
8003
   * Subsequence (markupdecl | PEReference | S)*
8004
   */
8005
11.3k
  SKIP_BLANKS;
8006
500k
        while (1) {
8007
500k
            if (PARSER_STOPPED(ctxt)) {
8008
601
                return;
8009
499k
            } else if (ctxt->input->cur >= ctxt->input->end) {
8010
70.6k
                if (ctxt->inputNr <= oldInputNr) {
8011
1.47k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8012
1.47k
                    return;
8013
1.47k
                }
8014
69.2k
                xmlPopPE(ctxt);
8015
429k
            } else if ((RAW == ']') && (ctxt->inputNr <= oldInputNr)) {
8016
5.85k
                NEXT;
8017
5.85k
                SKIP_BLANKS;
8018
5.85k
                break;
8019
423k
            } else if ((PARSER_EXTERNAL(ctxt)) &&
8020
343k
                       (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
8021
                /*
8022
                 * Conditional sections are allowed in external entities
8023
                 * included by PE References in the internal subset.
8024
                 */
8025
2.28k
                xmlParseConditionalSections(ctxt);
8026
421k
            } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
8027
223k
                xmlParseMarkupDecl(ctxt);
8028
223k
            } else if (RAW == '%') {
8029
193k
                xmlParsePERefInternal(ctxt, 1);
8030
193k
            } else {
8031
3.46k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8032
8033
4.20k
                while (ctxt->inputNr > oldInputNr)
8034
739
                    xmlPopPE(ctxt);
8035
3.46k
                return;
8036
3.46k
            }
8037
489k
            SKIP_BLANKS;
8038
489k
            SHRINK;
8039
489k
            GROW;
8040
489k
        }
8041
11.3k
    }
8042
8043
    /*
8044
     * We should be at the end of the DOCTYPE declaration.
8045
     */
8046
5.85k
    if (RAW != '>') {
8047
235
        xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
8048
235
        return;
8049
235
    }
8050
5.62k
    NEXT;
8051
5.62k
}
8052
8053
#ifdef LIBXML_SAX1_ENABLED
8054
/**
8055
 * Parse an attribute
8056
 *
8057
 * @deprecated Internal function, don't use.
8058
 *
8059
 *     [41] Attribute ::= Name Eq AttValue
8060
 *
8061
 * [ WFC: No External Entity References ]
8062
 * Attribute values cannot contain direct or indirect entity references
8063
 * to external entities.
8064
 *
8065
 * [ WFC: No < in Attribute Values ]
8066
 * The replacement text of any entity referred to directly or indirectly in
8067
 * an attribute value (other than "&lt;") must not contain a <.
8068
 *
8069
 * [ VC: Attribute Value Type ]
8070
 * The attribute must have been declared; the value must be of the type
8071
 * declared for it.
8072
 *
8073
 *     [25] Eq ::= S? '=' S?
8074
 *
8075
 * With namespace:
8076
 *
8077
 *     [NS 11] Attribute ::= QName Eq AttValue
8078
 *
8079
 * Also the case QName == xmlns:??? is handled independently as a namespace
8080
 * definition.
8081
 *
8082
 * @param ctxt  an XML parser context
8083
 * @param value  a xmlChar ** used to store the value of the attribute
8084
 * @returns the attribute name, and the value in *value.
8085
 */
8086
8087
const xmlChar *
8088
0
xmlParseAttribute(xmlParserCtxt *ctxt, xmlChar **value) {
8089
0
    const xmlChar *name;
8090
0
    xmlChar *val;
8091
8092
0
    *value = NULL;
8093
0
    GROW;
8094
0
    name = xmlParseName(ctxt);
8095
0
    if (name == NULL) {
8096
0
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8097
0
                 "error parsing attribute name\n");
8098
0
        return(NULL);
8099
0
    }
8100
8101
    /*
8102
     * read the value
8103
     */
8104
0
    SKIP_BLANKS;
8105
0
    if (RAW == '=') {
8106
0
        NEXT;
8107
0
  SKIP_BLANKS;
8108
0
  val = xmlParseAttValue(ctxt);
8109
0
    } else {
8110
0
  xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8111
0
         "Specification mandates value for attribute %s\n", name);
8112
0
  return(name);
8113
0
    }
8114
8115
    /*
8116
     * Check that xml:lang conforms to the specification
8117
     * No more registered as an error, just generate a warning now
8118
     * since this was deprecated in XML second edition
8119
     */
8120
0
    if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) {
8121
0
  if (!xmlCheckLanguageID(val)) {
8122
0
      xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8123
0
              "Malformed value for xml:lang : %s\n",
8124
0
        val, NULL);
8125
0
  }
8126
0
    }
8127
8128
    /*
8129
     * Check that xml:space conforms to the specification
8130
     */
8131
0
    if (xmlStrEqual(name, BAD_CAST "xml:space")) {
8132
0
  if (xmlStrEqual(val, BAD_CAST "default"))
8133
0
      *(ctxt->space) = 0;
8134
0
  else if (xmlStrEqual(val, BAD_CAST "preserve"))
8135
0
      *(ctxt->space) = 1;
8136
0
  else {
8137
0
    xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8138
0
"Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8139
0
                                 val, NULL);
8140
0
  }
8141
0
    }
8142
8143
0
    *value = val;
8144
0
    return(name);
8145
0
}
8146
8147
/**
8148
 * Parse a start tag. Always consumes '<'.
8149
 *
8150
 * @deprecated Internal function, don't use.
8151
 *
8152
 *     [40] STag ::= '<' Name (S Attribute)* S? '>'
8153
 *
8154
 * [ WFC: Unique Att Spec ]
8155
 * No attribute name may appear more than once in the same start-tag or
8156
 * empty-element tag.
8157
 *
8158
 *     [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
8159
 *
8160
 * [ WFC: Unique Att Spec ]
8161
 * No attribute name may appear more than once in the same start-tag or
8162
 * empty-element tag.
8163
 *
8164
 * With namespace:
8165
 *
8166
 *     [NS 8] STag ::= '<' QName (S Attribute)* S? '>'
8167
 *
8168
 *     [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>'
8169
 *
8170
 * @param ctxt  an XML parser context
8171
 * @returns the element name parsed
8172
 */
8173
8174
const xmlChar *
8175
0
xmlParseStartTag(xmlParserCtxt *ctxt) {
8176
0
    const xmlChar *name;
8177
0
    const xmlChar *attname;
8178
0
    xmlChar *attvalue;
8179
0
    const xmlChar **atts = ctxt->atts;
8180
0
    int nbatts = 0;
8181
0
    int maxatts = ctxt->maxatts;
8182
0
    int i;
8183
8184
0
    if (RAW != '<') return(NULL);
8185
0
    NEXT1;
8186
8187
0
    name = xmlParseName(ctxt);
8188
0
    if (name == NULL) {
8189
0
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8190
0
       "xmlParseStartTag: invalid element name\n");
8191
0
        return(NULL);
8192
0
    }
8193
8194
    /*
8195
     * Now parse the attributes, it ends up with the ending
8196
     *
8197
     * (S Attribute)* S?
8198
     */
8199
0
    SKIP_BLANKS;
8200
0
    GROW;
8201
8202
0
    while (((RAW != '>') &&
8203
0
     ((RAW != '/') || (NXT(1) != '>')) &&
8204
0
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8205
0
  attname = xmlParseAttribute(ctxt, &attvalue);
8206
0
        if (attname == NULL)
8207
0
      break;
8208
0
        if (attvalue != NULL) {
8209
      /*
8210
       * [ WFC: Unique Att Spec ]
8211
       * No attribute name may appear more than once in the same
8212
       * start-tag or empty-element tag.
8213
       */
8214
0
      for (i = 0; i < nbatts;i += 2) {
8215
0
          if (xmlStrEqual(atts[i], attname)) {
8216
0
        xmlErrAttributeDup(ctxt, NULL, attname);
8217
0
        goto failed;
8218
0
    }
8219
0
      }
8220
      /*
8221
       * Add the pair to atts
8222
       */
8223
0
      if (nbatts + 4 > maxatts) {
8224
0
          const xmlChar **n;
8225
0
                int newSize;
8226
8227
0
                newSize = xmlGrowCapacity(maxatts, sizeof(n[0]) * 2,
8228
0
                                          11, XML_MAX_ATTRS);
8229
0
                if (newSize < 0) {
8230
0
        xmlErrMemory(ctxt);
8231
0
        goto failed;
8232
0
    }
8233
0
#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
8234
0
                if (newSize < 2)
8235
0
                    newSize = 2;
8236
0
#endif
8237
0
          n = xmlRealloc(atts, newSize * sizeof(n[0]) * 2);
8238
0
    if (n == NULL) {
8239
0
        xmlErrMemory(ctxt);
8240
0
        goto failed;
8241
0
    }
8242
0
    atts = n;
8243
0
                maxatts = newSize * 2;
8244
0
    ctxt->atts = atts;
8245
0
    ctxt->maxatts = maxatts;
8246
0
      }
8247
8248
0
      atts[nbatts++] = attname;
8249
0
      atts[nbatts++] = attvalue;
8250
0
      atts[nbatts] = NULL;
8251
0
      atts[nbatts + 1] = NULL;
8252
8253
0
            attvalue = NULL;
8254
0
  }
8255
8256
0
failed:
8257
8258
0
        if (attvalue != NULL)
8259
0
            xmlFree(attvalue);
8260
8261
0
  GROW
8262
0
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
8263
0
      break;
8264
0
  if (SKIP_BLANKS == 0) {
8265
0
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
8266
0
         "attributes construct error\n");
8267
0
  }
8268
0
  SHRINK;
8269
0
        GROW;
8270
0
    }
8271
8272
    /*
8273
     * SAX: Start of Element !
8274
     */
8275
0
    if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) &&
8276
0
  (!ctxt->disableSAX)) {
8277
0
  if (nbatts > 0)
8278
0
      ctxt->sax->startElement(ctxt->userData, name, atts);
8279
0
  else
8280
0
      ctxt->sax->startElement(ctxt->userData, name, NULL);
8281
0
    }
8282
8283
0
    if (atts != NULL) {
8284
        /* Free only the content strings */
8285
0
        for (i = 1;i < nbatts;i+=2)
8286
0
      if (atts[i] != NULL)
8287
0
         xmlFree((xmlChar *) atts[i]);
8288
0
    }
8289
0
    return(name);
8290
0
}
8291
8292
/**
8293
 * Parse an end tag. Always consumes '</'.
8294
 *
8295
 *     [42] ETag ::= '</' Name S? '>'
8296
 *
8297
 * With namespace
8298
 *
8299
 *     [NS 9] ETag ::= '</' QName S? '>'
8300
 * @param ctxt  an XML parser context
8301
 * @param line  line of the start tag
8302
 */
8303
8304
static void
8305
0
xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) {
8306
0
    const xmlChar *name;
8307
8308
0
    GROW;
8309
0
    if ((RAW != '<') || (NXT(1) != '/')) {
8310
0
  xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED,
8311
0
           "xmlParseEndTag: '</' not found\n");
8312
0
  return;
8313
0
    }
8314
0
    SKIP(2);
8315
8316
0
    name = xmlParseNameAndCompare(ctxt,ctxt->name);
8317
8318
    /*
8319
     * We should definitely be at the ending "S? '>'" part
8320
     */
8321
0
    GROW;
8322
0
    SKIP_BLANKS;
8323
0
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
8324
0
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
8325
0
    } else
8326
0
  NEXT1;
8327
8328
    /*
8329
     * [ WFC: Element Type Match ]
8330
     * The Name in an element's end-tag must match the element type in the
8331
     * start-tag.
8332
     *
8333
     */
8334
0
    if (name != (xmlChar*)1) {
8335
0
        if (name == NULL) name = BAD_CAST "unparsable";
8336
0
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
8337
0
         "Opening and ending tag mismatch: %s line %d and %s\n",
8338
0
                    ctxt->name, line, name);
8339
0
    }
8340
8341
    /*
8342
     * SAX: End of Tag
8343
     */
8344
0
    if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
8345
0
  (!ctxt->disableSAX))
8346
0
        ctxt->sax->endElement(ctxt->userData, ctxt->name);
8347
8348
0
    namePop(ctxt);
8349
0
    spacePop(ctxt);
8350
0
}
8351
8352
/**
8353
 * Parse an end of tag
8354
 *
8355
 * @deprecated Internal function, don't use.
8356
 *
8357
 *     [42] ETag ::= '</' Name S? '>'
8358
 *
8359
 * With namespace
8360
 *
8361
 *     [NS 9] ETag ::= '</' QName S? '>'
8362
 * @param ctxt  an XML parser context
8363
 */
8364
8365
void
8366
0
xmlParseEndTag(xmlParserCtxt *ctxt) {
8367
0
    xmlParseEndTag1(ctxt, 0);
8368
0
}
8369
#endif /* LIBXML_SAX1_ENABLED */
8370
8371
/************************************************************************
8372
 *                  *
8373
 *          SAX 2 specific operations       *
8374
 *                  *
8375
 ************************************************************************/
8376
8377
/**
8378
 * Parse an XML Namespace QName
8379
 *
8380
 *     [6]  QName  ::= (Prefix ':')? LocalPart
8381
 *     [7]  Prefix  ::= NCName
8382
 *     [8]  LocalPart  ::= NCName
8383
 *
8384
 * @param ctxt  an XML parser context
8385
 * @param prefix  pointer to store the prefix part
8386
 * @returns the Name parsed or NULL
8387
 */
8388
8389
static xmlHashedString
8390
2.45M
xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) {
8391
2.45M
    xmlHashedString l, p;
8392
2.45M
    int start, isNCName = 0;
8393
8394
2.45M
    l.name = NULL;
8395
2.45M
    p.name = NULL;
8396
8397
2.45M
    GROW;
8398
2.45M
    start = CUR_PTR - BASE_PTR;
8399
8400
2.45M
    l = xmlParseNCName(ctxt);
8401
2.45M
    if (l.name != NULL) {
8402
1.52M
        isNCName = 1;
8403
1.52M
        if (CUR == ':') {
8404
313k
            NEXT;
8405
313k
            p = l;
8406
313k
            l = xmlParseNCName(ctxt);
8407
313k
        }
8408
1.52M
    }
8409
2.45M
    if ((l.name == NULL) || (CUR == ':')) {
8410
944k
        xmlChar *tmp;
8411
8412
944k
        l.name = NULL;
8413
944k
        p.name = NULL;
8414
944k
        if ((isNCName == 0) && (CUR != ':'))
8415
916k
            return(l);
8416
28.4k
        tmp = xmlParseNmtoken(ctxt);
8417
28.4k
        if (tmp != NULL)
8418
16.5k
            xmlFree(tmp);
8419
28.4k
        l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start,
8420
28.4k
                                CUR_PTR - (BASE_PTR + start));
8421
28.4k
        if (l.name == NULL) {
8422
1
            xmlErrMemory(ctxt);
8423
1
            return(l);
8424
1
        }
8425
28.4k
        xmlNsErr(ctxt, XML_NS_ERR_QNAME,
8426
28.4k
                 "Failed to parse QName '%s'\n", l.name, NULL, NULL);
8427
28.4k
    }
8428
8429
1.53M
    *prefix = p;
8430
1.53M
    return(l);
8431
2.45M
}
8432
8433
/**
8434
 * Parse an XML Namespace QName
8435
 *
8436
 *     [6]  QName  ::= (Prefix ':')? LocalPart
8437
 *     [7]  Prefix  ::= NCName
8438
 *     [8]  LocalPart  ::= NCName
8439
 *
8440
 * @param ctxt  an XML parser context
8441
 * @param prefix  pointer to store the prefix part
8442
 * @returns the Name parsed or NULL
8443
 */
8444
8445
static const xmlChar *
8446
4.76k
xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) {
8447
4.76k
    xmlHashedString n, p;
8448
8449
4.76k
    n = xmlParseQNameHashed(ctxt, &p);
8450
4.76k
    if (n.name == NULL)
8451
989
        return(NULL);
8452
3.77k
    *prefix = p.name;
8453
3.77k
    return(n.name);
8454
4.76k
}
8455
8456
/**
8457
 * Parse an XML name and compares for match
8458
 * (specialized for endtag parsing)
8459
 *
8460
 * @param ctxt  an XML parser context
8461
 * @param name  the localname
8462
 * @param prefix  the prefix, if any.
8463
 * @returns NULL for an illegal name, (xmlChar*) 1 for success
8464
 * and the name for mismatch
8465
 */
8466
8467
static const xmlChar *
8468
xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name,
8469
6.54k
                        xmlChar const *prefix) {
8470
6.54k
    const xmlChar *cmp;
8471
6.54k
    const xmlChar *in;
8472
6.54k
    const xmlChar *ret;
8473
6.54k
    const xmlChar *prefix2;
8474
8475
6.54k
    if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name));
8476
8477
6.54k
    GROW;
8478
6.54k
    in = ctxt->input->cur;
8479
8480
6.54k
    cmp = prefix;
8481
11.8k
    while (*in != 0 && *in == *cmp) {
8482
5.31k
  ++in;
8483
5.31k
  ++cmp;
8484
5.31k
    }
8485
6.54k
    if ((*cmp == 0) && (*in == ':')) {
8486
3.00k
        in++;
8487
3.00k
  cmp = name;
8488
8.14k
  while (*in != 0 && *in == *cmp) {
8489
5.13k
      ++in;
8490
5.13k
      ++cmp;
8491
5.13k
  }
8492
3.00k
  if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
8493
      /* success */
8494
1.78k
            ctxt->input->col += in - ctxt->input->cur;
8495
1.78k
      ctxt->input->cur = in;
8496
1.78k
      return((const xmlChar*) 1);
8497
1.78k
  }
8498
3.00k
    }
8499
    /*
8500
     * all strings coms from the dictionary, equality can be done directly
8501
     */
8502
4.76k
    ret = xmlParseQName (ctxt, &prefix2);
8503
4.76k
    if (ret == NULL)
8504
989
        return(NULL);
8505
3.77k
    if ((ret == name) && (prefix == prefix2))
8506
326
  return((const xmlChar*) 1);
8507
3.44k
    return ret;
8508
3.77k
}
8509
8510
/**
8511
 * Parse an attribute in the new SAX2 framework.
8512
 *
8513
 * @param ctxt  an XML parser context
8514
 * @param pref  the element prefix
8515
 * @param elem  the element name
8516
 * @param hprefix  resulting attribute prefix
8517
 * @param value  resulting value of the attribute
8518
 * @param len  resulting length of the attribute
8519
 * @param alloc  resulting indicator if the attribute was allocated
8520
 * @returns the attribute name, and the value in *value, .
8521
 */
8522
8523
static xmlHashedString
8524
xmlParseAttribute2(xmlParserCtxtPtr ctxt,
8525
                   const xmlChar * pref, const xmlChar * elem,
8526
                   xmlHashedString * hprefix, xmlChar ** value,
8527
                   int *len, int *alloc)
8528
1.08M
{
8529
1.08M
    xmlHashedString hname;
8530
1.08M
    const xmlChar *prefix, *name;
8531
1.08M
    xmlChar *val = NULL, *internal_val = NULL;
8532
1.08M
    int special = 0;
8533
1.08M
    int isNamespace;
8534
1.08M
    int flags;
8535
8536
1.08M
    *value = NULL;
8537
1.08M
    GROW;
8538
1.08M
    hname = xmlParseQNameHashed(ctxt, hprefix);
8539
1.08M
    if (hname.name == NULL) {
8540
744k
        xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8541
744k
                       "error parsing attribute name\n");
8542
744k
        return(hname);
8543
744k
    }
8544
341k
    name = hname.name;
8545
341k
    prefix = hprefix->name;
8546
8547
    /*
8548
     * get the type if needed
8549
     */
8550
341k
    if (ctxt->attsSpecial != NULL) {
8551
56.1k
        special = XML_PTR_TO_INT(xmlHashQLookup2(ctxt->attsSpecial, pref, elem,
8552
56.1k
                                              prefix, name));
8553
56.1k
    }
8554
8555
    /*
8556
     * read the value
8557
     */
8558
341k
    SKIP_BLANKS;
8559
341k
    if (RAW != '=') {
8560
43.1k
        xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8561
43.1k
                          "Specification mandates value for attribute %s\n",
8562
43.1k
                          name);
8563
43.1k
        goto error;
8564
43.1k
    }
8565
8566
8567
298k
    NEXT;
8568
298k
    SKIP_BLANKS;
8569
298k
    flags = 0;
8570
298k
    isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) ||
8571
281k
                   (prefix == ctxt->str_xmlns));
8572
298k
    val = xmlParseAttValueInternal(ctxt, len, &flags, special,
8573
298k
                                   isNamespace);
8574
298k
    if (val == NULL)
8575
5.73k
        goto error;
8576
8577
292k
    *alloc = (flags & XML_ATTVAL_ALLOC) != 0;
8578
8579
292k
#ifdef LIBXML_VALID_ENABLED
8580
292k
    if ((ctxt->validate) &&
8581
0
        (ctxt->standalone == 1) &&
8582
0
        (special & XML_SPECIAL_EXTERNAL) &&
8583
0
        (flags & XML_ATTVAL_NORM_CHANGE)) {
8584
0
        xmlValidityError(ctxt, XML_DTD_NOT_STANDALONE,
8585
0
                         "standalone: normalization of attribute %s on %s "
8586
0
                         "by external subset declaration\n",
8587
0
                         name, elem);
8588
0
    }
8589
292k
#endif
8590
8591
292k
    if (prefix == ctxt->str_xml) {
8592
        /*
8593
         * Check that xml:lang conforms to the specification
8594
         * No more registered as an error, just generate a warning now
8595
         * since this was deprecated in XML second edition
8596
         */
8597
35.0k
        if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) {
8598
24.3k
            internal_val = xmlStrndup(val, *len);
8599
24.3k
            if (internal_val == NULL)
8600
3
                goto mem_error;
8601
24.3k
            if (!xmlCheckLanguageID(internal_val)) {
8602
17.5k
                xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8603
17.5k
                              "Malformed value for xml:lang : %s\n",
8604
17.5k
                              internal_val, NULL);
8605
17.5k
            }
8606
24.3k
        }
8607
8608
        /*
8609
         * Check that xml:space conforms to the specification
8610
         */
8611
35.0k
        if (xmlStrEqual(name, BAD_CAST "space")) {
8612
1.52k
            internal_val = xmlStrndup(val, *len);
8613
1.52k
            if (internal_val == NULL)
8614
1
                goto mem_error;
8615
1.52k
            if (xmlStrEqual(internal_val, BAD_CAST "default"))
8616
862
                *(ctxt->space) = 0;
8617
665
            else if (xmlStrEqual(internal_val, BAD_CAST "preserve"))
8618
222
                *(ctxt->space) = 1;
8619
443
            else {
8620
443
                xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8621
443
                              "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8622
443
                              internal_val, NULL);
8623
443
            }
8624
1.52k
        }
8625
35.0k
        if (internal_val) {
8626
25.8k
            xmlFree(internal_val);
8627
25.8k
        }
8628
35.0k
    }
8629
8630
292k
    *value = val;
8631
292k
    return (hname);
8632
8633
4
mem_error:
8634
4
    xmlErrMemory(ctxt);
8635
48.8k
error:
8636
48.8k
    if ((val != NULL) && (*alloc != 0))
8637
1
        xmlFree(val);
8638
48.8k
    return(hname);
8639
4
}
8640
8641
/**
8642
 * Inserts a new attribute into the hash table.
8643
 *
8644
 * @param ctxt  parser context
8645
 * @param size  size of the hash table
8646
 * @param name  attribute name
8647
 * @param uri  namespace uri
8648
 * @param hashValue  combined hash value of name and uri
8649
 * @param aindex  attribute index (this is a multiple of 5)
8650
 * @returns INT_MAX if no existing attribute was found, the attribute
8651
 * index if an attribute was found, -1 if a memory allocation failed.
8652
 */
8653
static int
8654
xmlAttrHashInsert(xmlParserCtxtPtr ctxt, unsigned size, const xmlChar *name,
8655
166k
                  const xmlChar *uri, unsigned hashValue, int aindex) {
8656
166k
    xmlAttrHashBucket *table = ctxt->attrHash;
8657
166k
    xmlAttrHashBucket *bucket;
8658
166k
    unsigned hindex;
8659
8660
166k
    hindex = hashValue & (size - 1);
8661
166k
    bucket = &table[hindex];
8662
8663
190k
    while (bucket->index >= 0) {
8664
45.4k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8665
8666
45.4k
        if (name == atts[0]) {
8667
22.9k
            int nsIndex = XML_PTR_TO_INT(atts[2]);
8668
8669
22.9k
            if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) :
8670
22.9k
                (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) :
8671
8.88k
                (uri == ctxt->nsTab[nsIndex * 2 + 1]))
8672
21.4k
                return(bucket->index);
8673
22.9k
        }
8674
8675
23.9k
        hindex++;
8676
23.9k
        bucket++;
8677
23.9k
        if (hindex >= size) {
8678
6.01k
            hindex = 0;
8679
6.01k
            bucket = table;
8680
6.01k
        }
8681
23.9k
    }
8682
8683
145k
    bucket->index = aindex;
8684
8685
145k
    return(INT_MAX);
8686
166k
}
8687
8688
static int
8689
xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size,
8690
                       const xmlChar *name, const xmlChar *prefix,
8691
4.07k
                       unsigned hashValue, int aindex) {
8692
4.07k
    xmlAttrHashBucket *table = ctxt->attrHash;
8693
4.07k
    xmlAttrHashBucket *bucket;
8694
4.07k
    unsigned hindex;
8695
8696
4.07k
    hindex = hashValue & (size - 1);
8697
4.07k
    bucket = &table[hindex];
8698
8699
5.03k
    while (bucket->index >= 0) {
8700
2.49k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8701
8702
2.49k
        if ((name == atts[0]) && (prefix == atts[1]))
8703
1.53k
            return(bucket->index);
8704
8705
963
        hindex++;
8706
963
        bucket++;
8707
963
        if (hindex >= size) {
8708
194
            hindex = 0;
8709
194
            bucket = table;
8710
194
        }
8711
963
    }
8712
8713
2.53k
    bucket->index = aindex;
8714
8715
2.53k
    return(INT_MAX);
8716
4.07k
}
8717
/**
8718
 * Parse a start tag. Always consumes '<'.
8719
 *
8720
 * This routine is called when running SAX2 parsing
8721
 *
8722
 *     [40] STag ::= '<' Name (S Attribute)* S? '>'
8723
 *
8724
 * [ WFC: Unique Att Spec ]
8725
 * No attribute name may appear more than once in the same start-tag or
8726
 * empty-element tag.
8727
 *
8728
 *     [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
8729
 *
8730
 * [ WFC: Unique Att Spec ]
8731
 * No attribute name may appear more than once in the same start-tag or
8732
 * empty-element tag.
8733
 *
8734
 * With namespace:
8735
 *
8736
 *     [NS 8] STag ::= '<' QName (S Attribute)* S? '>'
8737
 *
8738
 *     [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>'
8739
 *
8740
 * @param ctxt  an XML parser context
8741
 * @param pref  resulting namespace prefix
8742
 * @param URI  resulting namespace URI
8743
 * @param nbNsPtr  resulting number of namespace declarations
8744
 * @returns the element name parsed
8745
 */
8746
8747
static const xmlChar *
8748
xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref,
8749
1.36M
                  const xmlChar **URI, int *nbNsPtr) {
8750
1.36M
    xmlHashedString hlocalname;
8751
1.36M
    xmlHashedString hprefix;
8752
1.36M
    xmlHashedString hattname;
8753
1.36M
    xmlHashedString haprefix;
8754
1.36M
    const xmlChar *localname;
8755
1.36M
    const xmlChar *prefix;
8756
1.36M
    const xmlChar *attname;
8757
1.36M
    const xmlChar *aprefix;
8758
1.36M
    const xmlChar *uri;
8759
1.36M
    xmlChar *attvalue = NULL;
8760
1.36M
    const xmlChar **atts = ctxt->atts;
8761
1.36M
    unsigned attrHashSize = 0;
8762
1.36M
    int maxatts = ctxt->maxatts;
8763
1.36M
    int nratts, nbatts, nbdef;
8764
1.36M
    int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts;
8765
1.36M
    int alloc = 0;
8766
1.36M
    int numNsErr = 0;
8767
1.36M
    int numDupErr = 0;
8768
8769
1.36M
    if (RAW != '<') return(NULL);
8770
1.36M
    NEXT1;
8771
8772
1.36M
    nbatts = 0;
8773
1.36M
    nratts = 0;
8774
1.36M
    nbdef = 0;
8775
1.36M
    nbNs = 0;
8776
1.36M
    nbTotalDef = 0;
8777
1.36M
    attval = 0;
8778
8779
1.36M
    if (xmlParserNsStartElement(ctxt->nsdb) < 0) {
8780
0
        xmlErrMemory(ctxt);
8781
0
        return(NULL);
8782
0
    }
8783
8784
1.36M
    hlocalname = xmlParseQNameHashed(ctxt, &hprefix);
8785
1.36M
    if (hlocalname.name == NULL) {
8786
170k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8787
170k
           "StartTag: invalid element name\n");
8788
170k
        return(NULL);
8789
170k
    }
8790
1.19M
    localname = hlocalname.name;
8791
1.19M
    prefix = hprefix.name;
8792
8793
    /*
8794
     * Now parse the attributes, it ends up with the ending
8795
     *
8796
     * (S Attribute)* S?
8797
     */
8798
1.19M
    SKIP_BLANKS;
8799
1.19M
    GROW;
8800
8801
    /*
8802
     * The ctxt->atts array will be ultimately passed to the SAX callback
8803
     * containing five xmlChar pointers for each attribute:
8804
     *
8805
     * [0] attribute name
8806
     * [1] attribute prefix
8807
     * [2] namespace URI
8808
     * [3] attribute value
8809
     * [4] end of attribute value
8810
     *
8811
     * To save memory, we reuse this array temporarily and store integers
8812
     * in these pointer variables.
8813
     *
8814
     * [0] attribute name
8815
     * [1] attribute prefix
8816
     * [2] hash value of attribute prefix, and later namespace index
8817
     * [3] for non-allocated values: ptrdiff_t offset into input buffer
8818
     * [4] for non-allocated values: ptrdiff_t offset into input buffer
8819
     *
8820
     * The ctxt->attallocs array contains an additional unsigned int for
8821
     * each attribute, containing the hash value of the attribute name
8822
     * and the alloc flag in bit 31.
8823
     */
8824
8825
1.34M
    while (((RAW != '>') &&
8826
1.12M
     ((RAW != '/') || (NXT(1) != '>')) &&
8827
1.09M
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8828
1.08M
  int len = -1;
8829
8830
1.08M
  hattname = xmlParseAttribute2(ctxt, prefix, localname,
8831
1.08M
                                          &haprefix, &attvalue, &len,
8832
1.08M
                                          &alloc);
8833
1.08M
        if (hattname.name == NULL)
8834
744k
      break;
8835
341k
        if (attvalue == NULL)
8836
48.8k
            goto next_attr;
8837
292k
        attname = hattname.name;
8838
292k
        aprefix = haprefix.name;
8839
292k
  if (len < 0) len = xmlStrlen(attvalue);
8840
8841
292k
        if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
8842
15.9k
            xmlHashedString huri;
8843
15.9k
            xmlURIPtr parsedUri;
8844
8845
15.9k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8846
15.9k
            uri = huri.name;
8847
15.9k
            if (uri == NULL) {
8848
1
                xmlErrMemory(ctxt);
8849
1
                goto next_attr;
8850
1
            }
8851
15.9k
            if (*uri != 0) {
8852
15.0k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8853
17
                    xmlErrMemory(ctxt);
8854
17
                    goto next_attr;
8855
17
                }
8856
15.0k
                if (parsedUri == NULL) {
8857
2.48k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8858
2.48k
                             "xmlns: '%s' is not a valid URI\n",
8859
2.48k
                                       uri, NULL, NULL);
8860
12.5k
                } else {
8861
12.5k
                    if (parsedUri->scheme == NULL) {
8862
10.8k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8863
10.8k
                                  "xmlns: URI %s is not absolute\n",
8864
10.8k
                                  uri, NULL, NULL);
8865
10.8k
                    }
8866
12.5k
                    xmlFreeURI(parsedUri);
8867
12.5k
                }
8868
15.0k
                if (uri == ctxt->str_xml_ns) {
8869
238
                    if (attname != ctxt->str_xml) {
8870
238
                        xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8871
238
                     "xml namespace URI cannot be the default namespace\n",
8872
238
                                 NULL, NULL, NULL);
8873
238
                    }
8874
238
                    goto next_attr;
8875
238
                }
8876
14.8k
                if ((len == 29) &&
8877
452
                    (xmlStrEqual(uri,
8878
452
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8879
194
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8880
194
                         "reuse of the xmlns namespace name is forbidden\n",
8881
194
                             NULL, NULL, NULL);
8882
194
                    goto next_attr;
8883
194
                }
8884
14.8k
            }
8885
8886
15.4k
            if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0)
8887
10.7k
                nbNs++;
8888
276k
        } else if (aprefix == ctxt->str_xmlns) {
8889
86.2k
            xmlHashedString huri;
8890
86.2k
            xmlURIPtr parsedUri;
8891
8892
86.2k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8893
86.2k
            uri = huri.name;
8894
86.2k
            if (uri == NULL) {
8895
1
                xmlErrMemory(ctxt);
8896
1
                goto next_attr;
8897
1
            }
8898
8899
86.2k
            if (attname == ctxt->str_xml) {
8900
682
                if (uri != ctxt->str_xml_ns) {
8901
488
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8902
488
                             "xml namespace prefix mapped to wrong URI\n",
8903
488
                             NULL, NULL, NULL);
8904
488
                }
8905
                /*
8906
                 * Do not keep a namespace definition node
8907
                 */
8908
682
                goto next_attr;
8909
682
            }
8910
85.5k
            if (uri == ctxt->str_xml_ns) {
8911
194
                if (attname != ctxt->str_xml) {
8912
194
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8913
194
                             "xml namespace URI mapped to wrong prefix\n",
8914
194
                             NULL, NULL, NULL);
8915
194
                }
8916
194
                goto next_attr;
8917
194
            }
8918
85.3k
            if (attname == ctxt->str_xmlns) {
8919
298
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8920
298
                         "redefinition of the xmlns prefix is forbidden\n",
8921
298
                         NULL, NULL, NULL);
8922
298
                goto next_attr;
8923
298
            }
8924
85.0k
            if ((len == 29) &&
8925
581
                (xmlStrEqual(uri,
8926
581
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8927
194
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8928
194
                         "reuse of the xmlns namespace name is forbidden\n",
8929
194
                         NULL, NULL, NULL);
8930
194
                goto next_attr;
8931
194
            }
8932
84.8k
            if ((uri == NULL) || (uri[0] == 0)) {
8933
275
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8934
275
                         "xmlns:%s: Empty XML namespace is not allowed\n",
8935
275
                              attname, NULL, NULL);
8936
275
                goto next_attr;
8937
84.5k
            } else {
8938
84.5k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8939
13
                    xmlErrMemory(ctxt);
8940
13
                    goto next_attr;
8941
13
                }
8942
84.5k
                if (parsedUri == NULL) {
8943
20.7k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8944
20.7k
                         "xmlns:%s: '%s' is not a valid URI\n",
8945
20.7k
                                       attname, uri, NULL);
8946
63.8k
                } else {
8947
63.8k
                    if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) {
8948
2.57k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8949
2.57k
                                  "xmlns:%s: URI %s is not absolute\n",
8950
2.57k
                                  attname, uri, NULL);
8951
2.57k
                    }
8952
63.8k
                    xmlFreeURI(parsedUri);
8953
63.8k
                }
8954
84.5k
            }
8955
8956
84.5k
            if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0)
8957
80.4k
                nbNs++;
8958
190k
        } else {
8959
            /*
8960
             * Populate attributes array, see above for repurposing
8961
             * of xmlChar pointers.
8962
             */
8963
190k
            if ((atts == NULL) || (nbatts + 5 > maxatts)) {
8964
18.8k
                int res = xmlCtxtGrowAttrs(ctxt);
8965
8966
18.8k
                maxatts = ctxt->maxatts;
8967
18.8k
                atts = ctxt->atts;
8968
8969
18.8k
                if (res < 0)
8970
14
                    goto next_attr;
8971
18.8k
            }
8972
190k
            ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) |
8973
190k
                                        ((unsigned) alloc << 31);
8974
190k
            atts[nbatts++] = attname;
8975
190k
            atts[nbatts++] = aprefix;
8976
190k
            atts[nbatts++] = XML_INT_TO_PTR(haprefix.hashValue);
8977
190k
            if (alloc) {
8978
40.3k
                atts[nbatts++] = attvalue;
8979
40.3k
                attvalue += len;
8980
40.3k
                atts[nbatts++] = attvalue;
8981
149k
            } else {
8982
                /*
8983
                 * attvalue points into the input buffer which can be
8984
                 * reallocated. Store differences to input->base instead.
8985
                 * The pointers will be reconstructed later.
8986
                 */
8987
149k
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8988
149k
                attvalue += len;
8989
149k
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8990
149k
            }
8991
            /*
8992
             * tag if some deallocation is needed
8993
             */
8994
190k
            if (alloc != 0) attval = 1;
8995
190k
            attvalue = NULL; /* moved into atts */
8996
190k
        }
8997
8998
341k
next_attr:
8999
341k
        if ((attvalue != NULL) && (alloc != 0)) {
9000
12.8k
            xmlFree(attvalue);
9001
12.8k
            attvalue = NULL;
9002
12.8k
        }
9003
9004
341k
  GROW
9005
341k
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
9006
73.8k
      break;
9007
267k
  if (SKIP_BLANKS == 0) {
9008
111k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
9009
111k
         "attributes construct error\n");
9010
111k
      break;
9011
111k
  }
9012
155k
        GROW;
9013
155k
    }
9014
9015
    /*
9016
     * Namespaces from default attributes
9017
     */
9018
1.19M
    if (ctxt->attsDefault != NULL) {
9019
168k
        xmlDefAttrsPtr defaults;
9020
9021
168k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9022
168k
  if (defaults != NULL) {
9023
355k
      for (i = 0; i < defaults->nbAttrs; i++) {
9024
241k
                xmlDefAttr *attr = &defaults->attrs[i];
9025
9026
241k
          attname = attr->name.name;
9027
241k
    aprefix = attr->prefix.name;
9028
9029
241k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
9030
3.19k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9031
9032
3.19k
                    if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0)
9033
2.52k
                        nbNs++;
9034
238k
    } else if (aprefix == ctxt->str_xmlns) {
9035
81.3k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9036
9037
81.3k
                    if (xmlParserNsPush(ctxt, &attr->name, &attr->value,
9038
81.3k
                                      NULL, 1) > 0)
9039
76.3k
                        nbNs++;
9040
156k
    } else {
9041
156k
                    if (nratts + nbTotalDef >= XML_MAX_ATTRS) {
9042
0
                        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
9043
0
                                    "Maximum number of attributes exceeded");
9044
0
                        break;
9045
0
                    }
9046
156k
                    nbTotalDef += 1;
9047
156k
                }
9048
241k
      }
9049
113k
  }
9050
168k
    }
9051
9052
    /*
9053
     * Resolve attribute namespaces
9054
     */
9055
1.38M
    for (i = 0; i < nbatts; i += 5) {
9056
190k
        attname = atts[i];
9057
190k
        aprefix = atts[i+1];
9058
9059
        /*
9060
  * The default namespace does not apply to attribute names.
9061
  */
9062
190k
  if (aprefix == NULL) {
9063
77.8k
            nsIndex = NS_INDEX_EMPTY;
9064
112k
        } else if (aprefix == ctxt->str_xml) {
9065
35.0k
            nsIndex = NS_INDEX_XML;
9066
77.2k
        } else {
9067
77.2k
            haprefix.name = aprefix;
9068
77.2k
            haprefix.hashValue = (size_t) atts[i+2];
9069
77.2k
            nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL);
9070
9071
77.2k
      if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) {
9072
53.3k
                xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9073
53.3k
        "Namespace prefix %s for %s on %s is not defined\n",
9074
53.3k
        aprefix, attname, localname);
9075
53.3k
                nsIndex = NS_INDEX_EMPTY;
9076
53.3k
            }
9077
77.2k
        }
9078
9079
190k
        atts[i+2] = XML_INT_TO_PTR(nsIndex);
9080
190k
    }
9081
9082
    /*
9083
     * Maximum number of attributes including default attributes.
9084
     */
9085
1.19M
    maxAtts = nratts + nbTotalDef;
9086
9087
    /*
9088
     * Verify that attribute names are unique.
9089
     */
9090
1.19M
    if (maxAtts > 1) {
9091
26.5k
        attrHashSize = 4;
9092
42.4k
        while (attrHashSize / 2 < (unsigned) maxAtts)
9093
15.9k
            attrHashSize *= 2;
9094
9095
26.5k
        if (attrHashSize > ctxt->attrHashMax) {
9096
2.25k
            xmlAttrHashBucket *tmp;
9097
9098
2.25k
            tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0]));
9099
2.25k
            if (tmp == NULL) {
9100
2
                xmlErrMemory(ctxt);
9101
2
                goto done;
9102
2
            }
9103
9104
2.25k
            ctxt->attrHash = tmp;
9105
2.25k
            ctxt->attrHashMax = attrHashSize;
9106
2.25k
        }
9107
9108
26.5k
        memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0]));
9109
9110
129k
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9111
102k
            const xmlChar *nsuri;
9112
102k
            unsigned hashValue, nameHashValue, uriHashValue;
9113
102k
            int res;
9114
9115
102k
            attname = atts[i];
9116
102k
            aprefix = atts[i+1];
9117
102k
            nsIndex = XML_PTR_TO_INT(atts[i+2]);
9118
            /* Hash values always have bit 31 set, see dict.c */
9119
102k
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9120
9121
102k
            if (nsIndex == NS_INDEX_EMPTY) {
9122
                /*
9123
                 * Prefix with empty namespace means an undeclared
9124
                 * prefix which was already reported above.
9125
                 */
9126
77.3k
                if (aprefix != NULL)
9127
47.2k
                    continue;
9128
30.1k
                nsuri = NULL;
9129
30.1k
                uriHashValue = URI_HASH_EMPTY;
9130
30.1k
            } else if (nsIndex == NS_INDEX_XML) {
9131
2.89k
                nsuri = ctxt->str_xml_ns;
9132
2.89k
                uriHashValue = URI_HASH_XML;
9133
22.6k
            } else {
9134
22.6k
                nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9135
22.6k
                uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9136
22.6k
            }
9137
9138
55.6k
            hashValue = xmlDictCombineHash(nameHashValue, uriHashValue);
9139
55.6k
            res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9140
55.6k
                                    hashValue, i);
9141
55.6k
            if (res < 0)
9142
0
                continue;
9143
9144
            /*
9145
             * [ WFC: Unique Att Spec ]
9146
             * No attribute name may appear more than once in the same
9147
             * start-tag or empty-element tag.
9148
             * As extended by the Namespace in XML REC.
9149
             */
9150
55.6k
            if (res < INT_MAX) {
9151
16.8k
                if (aprefix == atts[res+1]) {
9152
14.3k
                    xmlErrAttributeDup(ctxt, aprefix, attname);
9153
14.3k
                    numDupErr += 1;
9154
14.3k
                } else {
9155
2.50k
                    xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9156
2.50k
                             "Namespaced Attribute %s in '%s' redefined\n",
9157
2.50k
                             attname, nsuri, NULL);
9158
2.50k
                    numNsErr += 1;
9159
2.50k
                }
9160
16.8k
            }
9161
55.6k
        }
9162
26.5k
    }
9163
9164
    /*
9165
     * Default attributes
9166
     */
9167
1.19M
    if (ctxt->attsDefault != NULL) {
9168
168k
        xmlDefAttrsPtr defaults;
9169
9170
168k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9171
168k
  if (defaults != NULL) {
9172
355k
      for (i = 0; i < defaults->nbAttrs; i++) {
9173
241k
                xmlDefAttr *attr = &defaults->attrs[i];
9174
241k
                const xmlChar *nsuri = NULL;
9175
241k
                unsigned hashValue, uriHashValue = 0;
9176
241k
                int res;
9177
9178
241k
          attname = attr->name.name;
9179
241k
    aprefix = attr->prefix.name;
9180
9181
241k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL))
9182
3.19k
                    continue;
9183
238k
    if (aprefix == ctxt->str_xmlns)
9184
81.3k
                    continue;
9185
9186
156k
                if (aprefix == NULL) {
9187
144k
                    nsIndex = NS_INDEX_EMPTY;
9188
144k
                    nsuri = NULL;
9189
144k
                    uriHashValue = URI_HASH_EMPTY;
9190
144k
                } else if (aprefix == ctxt->str_xml) {
9191
4.82k
                    nsIndex = NS_INDEX_XML;
9192
4.82k
                    nsuri = ctxt->str_xml_ns;
9193
4.82k
                    uriHashValue = URI_HASH_XML;
9194
7.86k
                } else {
9195
7.86k
                    nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL);
9196
7.86k
                    if ((nsIndex == INT_MAX) ||
9197
4.48k
                        (nsIndex < ctxt->nsdb->minNsIndex)) {
9198
4.48k
                        xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9199
4.48k
                                 "Namespace prefix %s for %s on %s is not "
9200
4.48k
                                 "defined\n",
9201
4.48k
                                 aprefix, attname, localname);
9202
4.48k
                        nsIndex = NS_INDEX_EMPTY;
9203
4.48k
                        nsuri = NULL;
9204
4.48k
                        uriHashValue = URI_HASH_EMPTY;
9205
4.48k
                    } else {
9206
3.38k
                        nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9207
3.38k
                        uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9208
3.38k
                    }
9209
7.86k
                }
9210
9211
                /*
9212
                 * Check whether the attribute exists
9213
                 */
9214
156k
                if (maxAtts > 1) {
9215
110k
                    hashValue = xmlDictCombineHash(attr->name.hashValue,
9216
110k
                                                   uriHashValue);
9217
110k
                    res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9218
110k
                                            hashValue, nbatts);
9219
110k
                    if (res < 0)
9220
0
                        continue;
9221
110k
                    if (res < INT_MAX) {
9222
4.52k
                        if (aprefix == atts[res+1])
9223
3.67k
                            continue;
9224
858
                        xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9225
858
                                 "Namespaced Attribute %s in '%s' redefined\n",
9226
858
                                 attname, nsuri, NULL);
9227
858
                    }
9228
110k
                }
9229
9230
153k
                xmlParserEntityCheck(ctxt, attr->expandedSize);
9231
9232
153k
                if ((atts == NULL) || (nbatts + 5 > maxatts)) {
9233
1.03k
                    res = xmlCtxtGrowAttrs(ctxt);
9234
9235
1.03k
                    maxatts = ctxt->maxatts;
9236
1.03k
                    atts = ctxt->atts;
9237
9238
1.03k
                    if (res < 0) {
9239
3
                        localname = NULL;
9240
3
                        goto done;
9241
3
                    }
9242
1.03k
                }
9243
9244
153k
                atts[nbatts++] = attname;
9245
153k
                atts[nbatts++] = aprefix;
9246
153k
                atts[nbatts++] = XML_INT_TO_PTR(nsIndex);
9247
153k
                atts[nbatts++] = attr->value.name;
9248
153k
                atts[nbatts++] = attr->valueEnd;
9249
9250
153k
#ifdef LIBXML_VALID_ENABLED
9251
                /*
9252
                 * This should be moved to valid.c, but we don't keep track
9253
                 * whether an attribute was defaulted.
9254
                 */
9255
153k
                if ((ctxt->validate) &&
9256
0
                    (ctxt->standalone == 1) &&
9257
0
                    (attr->external != 0)) {
9258
0
                    xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED,
9259
0
                            "standalone: attribute %s on %s defaulted "
9260
0
                            "from external subset\n",
9261
0
                            attname, localname);
9262
0
                }
9263
153k
#endif
9264
153k
                nbdef++;
9265
153k
      }
9266
113k
  }
9267
168k
    }
9268
9269
    /*
9270
     * Using a single hash table for nsUri/localName pairs cannot
9271
     * detect duplicate QNames reliably. The following example will
9272
     * only result in two namespace errors.
9273
     *
9274
     * <doc xmlns:a="a" xmlns:b="a">
9275
     *   <elem a:a="" b:a="" b:a=""/>
9276
     * </doc>
9277
     *
9278
     * If we saw more than one namespace error but no duplicate QNames
9279
     * were found, we have to scan for duplicate QNames.
9280
     */
9281
1.19M
    if ((numDupErr == 0) && (numNsErr > 1)) {
9282
809
        memset(ctxt->attrHash, -1,
9283
809
               attrHashSize * sizeof(ctxt->attrHash[0]));
9284
9285
5.22k
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9286
4.41k
            unsigned hashValue, nameHashValue, prefixHashValue;
9287
4.41k
            int res;
9288
9289
4.41k
            aprefix = atts[i+1];
9290
4.41k
            if (aprefix == NULL)
9291
343
                continue;
9292
9293
4.07k
            attname = atts[i];
9294
            /* Hash values always have bit 31 set, see dict.c */
9295
4.07k
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9296
4.07k
            prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix);
9297
9298
4.07k
            hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue);
9299
4.07k
            res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname,
9300
4.07k
                                         aprefix, hashValue, i);
9301
4.07k
            if (res < INT_MAX)
9302
1.53k
                xmlErrAttributeDup(ctxt, aprefix, attname);
9303
4.07k
        }
9304
809
    }
9305
9306
    /*
9307
     * Reconstruct attribute pointers
9308
     */
9309
1.53M
    for (i = 0, j = 0; i < nbatts; i += 5, j++) {
9310
        /* namespace URI */
9311
343k
        nsIndex = XML_PTR_TO_INT(atts[i+2]);
9312
343k
        if (nsIndex == INT_MAX)
9313
276k
            atts[i+2] = NULL;
9314
66.8k
        else if (nsIndex == INT_MAX - 1)
9315
39.6k
            atts[i+2] = ctxt->str_xml_ns;
9316
27.1k
        else
9317
27.1k
            atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1];
9318
9319
343k
        if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) {
9320
149k
            atts[i+3] = BASE_PTR + XML_PTR_TO_INT(atts[i+3]);  /* value */
9321
149k
            atts[i+4] = BASE_PTR + XML_PTR_TO_INT(atts[i+4]);  /* valuend */
9322
149k
        }
9323
343k
    }
9324
9325
1.19M
    uri = xmlParserNsLookupUri(ctxt, &hprefix);
9326
1.19M
    if ((prefix != NULL) && (uri == NULL)) {
9327
24.2k
  xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9328
24.2k
           "Namespace prefix %s on %s is not defined\n",
9329
24.2k
     prefix, localname, NULL);
9330
24.2k
    }
9331
1.19M
    *pref = prefix;
9332
1.19M
    *URI = uri;
9333
9334
    /*
9335
     * SAX callback
9336
     */
9337
1.19M
    if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) &&
9338
1.19M
  (!ctxt->disableSAX)) {
9339
1.08M
  if (nbNs > 0)
9340
60.4k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9341
60.4k
                          nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs),
9342
60.4k
        nbatts / 5, nbdef, atts);
9343
1.02M
  else
9344
1.02M
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9345
1.02M
                          0, NULL, nbatts / 5, nbdef, atts);
9346
1.08M
    }
9347
9348
1.19M
done:
9349
    /*
9350
     * Free allocated attribute values
9351
     */
9352
1.19M
    if (attval != 0) {
9353
121k
  for (i = 0, j = 0; j < nratts; i += 5, j++)
9354
82.1k
      if (ctxt->attallocs[j] & 0x80000000)
9355
40.3k
          xmlFree((xmlChar *) atts[i+3]);
9356
39.0k
    }
9357
9358
1.19M
    *nbNsPtr = nbNs;
9359
1.19M
    return(localname);
9360
1.19M
}
9361
9362
/**
9363
 * Parse an end tag. Always consumes '</'.
9364
 *
9365
 *     [42] ETag ::= '</' Name S? '>'
9366
 *
9367
 * With namespace
9368
 *
9369
 *     [NS 9] ETag ::= '</' QName S? '>'
9370
 * @param ctxt  an XML parser context
9371
 * @param tag  the corresponding start tag
9372
 */
9373
9374
static void
9375
75.1k
xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) {
9376
75.1k
    const xmlChar *name;
9377
9378
75.1k
    GROW;
9379
75.1k
    if ((RAW != '<') || (NXT(1) != '/')) {
9380
575
  xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL);
9381
575
  return;
9382
575
    }
9383
74.6k
    SKIP(2);
9384
9385
74.6k
    if (tag->prefix == NULL)
9386
68.0k
        name = xmlParseNameAndCompare(ctxt, ctxt->name);
9387
6.54k
    else
9388
6.54k
        name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix);
9389
9390
    /*
9391
     * We should definitely be at the ending "S? '>'" part
9392
     */
9393
74.6k
    GROW;
9394
74.6k
    SKIP_BLANKS;
9395
74.6k
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
9396
12.8k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
9397
12.8k
    } else
9398
61.7k
  NEXT1;
9399
9400
    /*
9401
     * [ WFC: Element Type Match ]
9402
     * The Name in an element's end-tag must match the element type in the
9403
     * start-tag.
9404
     *
9405
     */
9406
74.6k
    if (name != (xmlChar*)1) {
9407
17.4k
        if (name == NULL) name = BAD_CAST "unparsable";
9408
17.4k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
9409
17.4k
         "Opening and ending tag mismatch: %s line %d and %s\n",
9410
17.4k
                    ctxt->name, tag->line, name);
9411
17.4k
    }
9412
9413
    /*
9414
     * SAX: End of Tag
9415
     */
9416
74.6k
    if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9417
74.6k
  (!ctxt->disableSAX))
9418
65.6k
  ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix,
9419
65.6k
                                tag->URI);
9420
9421
74.6k
    spacePop(ctxt);
9422
74.6k
    if (tag->nsNr != 0)
9423
3.40k
  xmlParserNsPop(ctxt, tag->nsNr);
9424
74.6k
}
9425
9426
/**
9427
 * Parse escaped pure raw content. Always consumes '<!['.
9428
 *
9429
 * @deprecated Internal function, don't use.
9430
 *
9431
 *     [18] CDSect ::= CDStart CData CDEnd
9432
 *
9433
 *     [19] CDStart ::= '<![CDATA['
9434
 *
9435
 *     [20] Data ::= (Char* - (Char* ']]>' Char*))
9436
 *
9437
 *     [21] CDEnd ::= ']]>'
9438
 * @param ctxt  an XML parser context
9439
 */
9440
void
9441
16.9k
xmlParseCDSect(xmlParserCtxt *ctxt) {
9442
16.9k
    xmlChar *buf = NULL;
9443
16.9k
    int len = 0;
9444
16.9k
    int size = XML_PARSER_BUFFER_SIZE;
9445
16.9k
    int r, rl;
9446
16.9k
    int s, sl;
9447
16.9k
    int cur, l;
9448
16.9k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9449
4.14k
                    XML_MAX_HUGE_LENGTH :
9450
16.9k
                    XML_MAX_TEXT_LENGTH;
9451
9452
16.9k
    if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '['))
9453
0
        return;
9454
16.9k
    SKIP(3);
9455
9456
16.9k
    if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '['))
9457
0
        return;
9458
16.9k
    SKIP(6);
9459
9460
16.9k
    r = xmlCurrentCharRecover(ctxt, &rl);
9461
16.9k
    if (!IS_CHAR(r)) {
9462
411
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9463
411
        goto out;
9464
411
    }
9465
16.5k
    NEXTL(rl);
9466
16.5k
    s = xmlCurrentCharRecover(ctxt, &sl);
9467
16.5k
    if (!IS_CHAR(s)) {
9468
750
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9469
750
        goto out;
9470
750
    }
9471
15.8k
    NEXTL(sl);
9472
15.8k
    cur = xmlCurrentCharRecover(ctxt, &l);
9473
15.8k
    buf = xmlMalloc(size);
9474
15.8k
    if (buf == NULL) {
9475
1
  xmlErrMemory(ctxt);
9476
1
        goto out;
9477
1
    }
9478
2.74M
    while (IS_CHAR(cur) &&
9479
2.74M
           ((r != ']') || (s != ']') || (cur != '>'))) {
9480
2.73M
  if (len + 5 >= size) {
9481
3.77k
      xmlChar *tmp;
9482
3.77k
            int newSize;
9483
9484
3.77k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9485
3.77k
            if (newSize < 0) {
9486
0
                xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9487
0
                               "CData section too big found\n");
9488
0
                goto out;
9489
0
            }
9490
3.77k
      tmp = xmlRealloc(buf, newSize);
9491
3.77k
      if (tmp == NULL) {
9492
1
    xmlErrMemory(ctxt);
9493
1
                goto out;
9494
1
      }
9495
3.77k
      buf = tmp;
9496
3.77k
      size = newSize;
9497
3.77k
  }
9498
2.73M
  COPY_BUF(buf, len, r);
9499
2.73M
  r = s;
9500
2.73M
  rl = sl;
9501
2.73M
  s = cur;
9502
2.73M
  sl = l;
9503
2.73M
  NEXTL(l);
9504
2.73M
  cur = xmlCurrentCharRecover(ctxt, &l);
9505
2.73M
    }
9506
15.7k
    buf[len] = 0;
9507
15.7k
    if (cur != '>') {
9508
2.97k
  xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9509
2.97k
                       "CData section not finished\n%.50s\n", buf);
9510
2.97k
        goto out;
9511
2.97k
    }
9512
12.8k
    NEXTL(l);
9513
9514
    /*
9515
     * OK the buffer is to be consumed as cdata.
9516
     */
9517
12.8k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) {
9518
11.6k
        if ((ctxt->sax->cdataBlock != NULL) &&
9519
11.6k
            ((ctxt->options & XML_PARSE_NOCDATA) == 0)) {
9520
9.75k
            ctxt->sax->cdataBlock(ctxt->userData, buf, len);
9521
9.75k
        } else if (ctxt->sax->characters != NULL) {
9522
1.94k
            ctxt->sax->characters(ctxt->userData, buf, len);
9523
1.94k
        }
9524
11.6k
    }
9525
9526
16.9k
out:
9527
16.9k
    xmlFree(buf);
9528
16.9k
}
9529
9530
/**
9531
 * Parse a content sequence. Stops at EOF or '</'. Leaves checking of
9532
 * unexpected EOF to the caller.
9533
 *
9534
 * @param ctxt  an XML parser context
9535
 */
9536
9537
static void
9538
23.4k
xmlParseContentInternal(xmlParserCtxtPtr ctxt) {
9539
23.4k
    int oldNameNr = ctxt->nameNr;
9540
23.4k
    int oldSpaceNr = ctxt->spaceNr;
9541
23.4k
    int oldNodeNr = ctxt->nodeNr;
9542
9543
23.4k
    GROW;
9544
4.27M
    while ((ctxt->input->cur < ctxt->input->end) &&
9545
4.25M
     (PARSER_STOPPED(ctxt) == 0)) {
9546
4.25M
  const xmlChar *cur = ctxt->input->cur;
9547
9548
  /*
9549
   * First case : a Processing Instruction.
9550
   */
9551
4.25M
  if ((*cur == '<') && (cur[1] == '?')) {
9552
11.6k
      xmlParsePI(ctxt);
9553
11.6k
  }
9554
9555
  /*
9556
   * Second case : a CDSection
9557
   */
9558
  /* 2.6.0 test was *cur not RAW */
9559
4.24M
  else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) {
9560
16.9k
      xmlParseCDSect(ctxt);
9561
16.9k
  }
9562
9563
  /*
9564
   * Third case :  a comment
9565
   */
9566
4.22M
  else if ((*cur == '<') && (NXT(1) == '!') &&
9567
59.9k
     (NXT(2) == '-') && (NXT(3) == '-')) {
9568
12.3k
      xmlParseComment(ctxt);
9569
12.3k
  }
9570
9571
  /*
9572
   * Fourth case :  a sub-element.
9573
   */
9574
4.21M
  else if (*cur == '<') {
9575
1.41M
            if (NXT(1) == '/') {
9576
74.5k
                if (ctxt->nameNr <= oldNameNr)
9577
1.04k
                    break;
9578
73.5k
          xmlParseElementEnd(ctxt);
9579
1.33M
            } else {
9580
1.33M
          xmlParseElementStart(ctxt);
9581
1.33M
            }
9582
1.41M
  }
9583
9584
  /*
9585
   * Fifth case : a reference. If if has not been resolved,
9586
   *    parsing returns it's Name, create the node
9587
   */
9588
9589
2.80M
  else if (*cur == '&') {
9590
159k
      xmlParseReference(ctxt);
9591
159k
  }
9592
9593
  /*
9594
   * Last case, text. Note that References are handled directly.
9595
   */
9596
2.64M
  else {
9597
2.64M
      xmlParseCharDataInternal(ctxt, 0);
9598
2.64M
  }
9599
9600
4.25M
  SHRINK;
9601
4.25M
  GROW;
9602
4.25M
    }
9603
9604
23.4k
    if ((ctxt->nameNr > oldNameNr) &&
9605
12.6k
        (ctxt->input->cur >= ctxt->input->end) &&
9606
12.3k
        (ctxt->wellFormed)) {
9607
156
        const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9608
156
        int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9609
156
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9610
156
                "Premature end of data in tag %s line %d\n",
9611
156
                name, line, NULL);
9612
156
    }
9613
9614
    /*
9615
     * Clean up in error case
9616
     */
9617
9618
196k
    while (ctxt->nodeNr > oldNodeNr)
9619
172k
        nodePop(ctxt);
9620
9621
224k
    while (ctxt->nameNr > oldNameNr) {
9622
201k
        xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1];
9623
9624
201k
        if (tag->nsNr != 0)
9625
40.1k
            xmlParserNsPop(ctxt, tag->nsNr);
9626
9627
201k
        namePop(ctxt);
9628
201k
    }
9629
9630
224k
    while (ctxt->spaceNr > oldSpaceNr)
9631
201k
        spacePop(ctxt);
9632
23.4k
}
9633
9634
/**
9635
 * Parse XML element content. This is useful if you're only interested
9636
 * in custom SAX callbacks. If you want a node list, use
9637
 * #xmlCtxtParseContent.
9638
 *
9639
 * @param ctxt  an XML parser context
9640
 */
9641
void
9642
0
xmlParseContent(xmlParserCtxt *ctxt) {
9643
0
    if ((ctxt == NULL) || (ctxt->input == NULL))
9644
0
        return;
9645
9646
0
    xmlCtxtInitializeLate(ctxt);
9647
9648
0
    xmlParseContentInternal(ctxt);
9649
9650
0
    xmlParserCheckEOF(ctxt, XML_ERR_NOT_WELL_BALANCED);
9651
0
}
9652
9653
/**
9654
 * Parse an XML element
9655
 *
9656
 * @deprecated Internal function, don't use.
9657
 *
9658
 *     [39] element ::= EmptyElemTag | STag content ETag
9659
 *
9660
 * [ WFC: Element Type Match ]
9661
 * The Name in an element's end-tag must match the element type in the
9662
 * start-tag.
9663
 *
9664
 * @param ctxt  an XML parser context
9665
 */
9666
9667
void
9668
24.7k
xmlParseElement(xmlParserCtxt *ctxt) {
9669
24.7k
    if (xmlParseElementStart(ctxt) != 0)
9670
6.01k
        return;
9671
9672
18.7k
    xmlParseContentInternal(ctxt);
9673
9674
18.7k
    if (ctxt->input->cur >= ctxt->input->end) {
9675
17.0k
        if (ctxt->wellFormed) {
9676
340
            const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9677
340
            int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9678
340
            xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9679
340
                    "Premature end of data in tag %s line %d\n",
9680
340
                    name, line, NULL);
9681
340
        }
9682
17.0k
        return;
9683
17.0k
    }
9684
9685
1.68k
    xmlParseElementEnd(ctxt);
9686
1.68k
}
9687
9688
/**
9689
 * Parse the start of an XML element. Returns -1 in case of error, 0 if an
9690
 * opening tag was parsed, 1 if an empty element was parsed.
9691
 *
9692
 * Always consumes '<'.
9693
 *
9694
 * @param ctxt  an XML parser context
9695
 */
9696
static int
9697
1.36M
xmlParseElementStart(xmlParserCtxtPtr ctxt) {
9698
1.36M
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
9699
1.36M
    const xmlChar *name;
9700
1.36M
    const xmlChar *prefix = NULL;
9701
1.36M
    const xmlChar *URI = NULL;
9702
1.36M
    xmlParserNodeInfo node_info;
9703
1.36M
    int line;
9704
1.36M
    xmlNodePtr cur;
9705
1.36M
    int nbNs = 0;
9706
9707
1.36M
    if (ctxt->nameNr > maxDepth) {
9708
2
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
9709
2
                "Excessive depth in document: %d use XML_PARSE_HUGE option\n",
9710
2
                ctxt->nameNr);
9711
2
  return(-1);
9712
2
    }
9713
9714
    /* Capture start position */
9715
1.36M
    if (ctxt->record_info) {
9716
0
        node_info.begin_pos = ctxt->input->consumed +
9717
0
                          (CUR_PTR - ctxt->input->base);
9718
0
  node_info.begin_line = ctxt->input->line;
9719
0
    }
9720
9721
1.36M
    if (ctxt->spaceNr == 0)
9722
23.3k
  spacePush(ctxt, -1);
9723
1.33M
    else if (*ctxt->space == -2)
9724
454k
  spacePush(ctxt, -1);
9725
883k
    else
9726
883k
  spacePush(ctxt, *ctxt->space);
9727
9728
1.36M
    line = ctxt->input->line;
9729
1.36M
#ifdef LIBXML_SAX1_ENABLED
9730
1.36M
    if (ctxt->sax2)
9731
1.36M
#endif /* LIBXML_SAX1_ENABLED */
9732
1.36M
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
9733
0
#ifdef LIBXML_SAX1_ENABLED
9734
0
    else
9735
0
  name = xmlParseStartTag(ctxt);
9736
1.36M
#endif /* LIBXML_SAX1_ENABLED */
9737
1.36M
    if (name == NULL) {
9738
170k
  spacePop(ctxt);
9739
170k
        return(-1);
9740
170k
    }
9741
1.19M
    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
9742
1.19M
    cur = ctxt->node;
9743
9744
1.19M
#ifdef LIBXML_VALID_ENABLED
9745
    /*
9746
     * [ VC: Root Element Type ]
9747
     * The Name in the document type declaration must match the element
9748
     * type of the root element.
9749
     */
9750
1.19M
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
9751
0
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
9752
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
9753
1.19M
#endif /* LIBXML_VALID_ENABLED */
9754
9755
    /*
9756
     * Check for an Empty Element.
9757
     */
9758
1.19M
    if ((RAW == '/') && (NXT(1) == '>')) {
9759
27.4k
        SKIP(2);
9760
27.4k
  if (ctxt->sax2) {
9761
27.4k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9762
27.4k
    (!ctxt->disableSAX))
9763
23.6k
    ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI);
9764
27.4k
#ifdef LIBXML_SAX1_ENABLED
9765
27.4k
  } else {
9766
0
      if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
9767
0
    (!ctxt->disableSAX))
9768
0
    ctxt->sax->endElement(ctxt->userData, name);
9769
0
#endif /* LIBXML_SAX1_ENABLED */
9770
0
  }
9771
27.4k
  namePop(ctxt);
9772
27.4k
  spacePop(ctxt);
9773
27.4k
  if (nbNs > 0)
9774
3.49k
      xmlParserNsPop(ctxt, nbNs);
9775
27.4k
  if (cur != NULL && ctxt->record_info) {
9776
0
            node_info.node = cur;
9777
0
            node_info.end_pos = ctxt->input->consumed +
9778
0
                                (CUR_PTR - ctxt->input->base);
9779
0
            node_info.end_line = ctxt->input->line;
9780
0
            xmlParserAddNodeInfo(ctxt, &node_info);
9781
0
  }
9782
27.4k
  return(1);
9783
27.4k
    }
9784
1.16M
    if (RAW == '>') {
9785
293k
        NEXT1;
9786
293k
        if (cur != NULL && ctxt->record_info) {
9787
0
            node_info.node = cur;
9788
0
            node_info.end_pos = 0;
9789
0
            node_info.end_line = 0;
9790
0
            xmlParserAddNodeInfo(ctxt, &node_info);
9791
0
        }
9792
869k
    } else {
9793
869k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED,
9794
869k
         "Couldn't find end of Start Tag %s line %d\n",
9795
869k
                    name, line, NULL);
9796
9797
  /*
9798
   * end of parsing of this node.
9799
   */
9800
869k
  nodePop(ctxt);
9801
869k
  namePop(ctxt);
9802
869k
  spacePop(ctxt);
9803
869k
  if (nbNs > 0)
9804
46.0k
      xmlParserNsPop(ctxt, nbNs);
9805
869k
  return(-1);
9806
869k
    }
9807
9808
293k
    return(0);
9809
1.16M
}
9810
9811
/**
9812
 * Parse the end of an XML element. Always consumes '</'.
9813
 *
9814
 * @param ctxt  an XML parser context
9815
 */
9816
static void
9817
75.1k
xmlParseElementEnd(xmlParserCtxtPtr ctxt) {
9818
75.1k
    xmlNodePtr cur = ctxt->node;
9819
9820
75.1k
    if (ctxt->nameNr <= 0) {
9821
13
        if ((RAW == '<') && (NXT(1) == '/'))
9822
2
            SKIP(2);
9823
13
        return;
9824
13
    }
9825
9826
    /*
9827
     * parse the end of tag: '</' should be here.
9828
     */
9829
75.1k
    if (ctxt->sax2) {
9830
75.1k
  xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
9831
75.1k
  namePop(ctxt);
9832
75.1k
    }
9833
0
#ifdef LIBXML_SAX1_ENABLED
9834
0
    else
9835
0
  xmlParseEndTag1(ctxt, 0);
9836
75.1k
#endif /* LIBXML_SAX1_ENABLED */
9837
9838
    /*
9839
     * Capture end position
9840
     */
9841
75.1k
    if (cur != NULL && ctxt->record_info) {
9842
0
        xmlParserNodeInfoPtr node_info;
9843
9844
0
        node_info = (xmlParserNodeInfoPtr) xmlParserFindNodeInfo(ctxt, cur);
9845
0
        if (node_info != NULL) {
9846
0
            node_info->end_pos = ctxt->input->consumed +
9847
0
                                 (CUR_PTR - ctxt->input->base);
9848
0
            node_info->end_line = ctxt->input->line;
9849
0
        }
9850
0
    }
9851
75.1k
}
9852
9853
/**
9854
 * Parse the XML version value.
9855
 *
9856
 * @deprecated Internal function, don't use.
9857
 *
9858
 *     [26] VersionNum ::= '1.' [0-9]+
9859
 *
9860
 * In practice allow [0-9].[0-9]+ at that level
9861
 *
9862
 * @param ctxt  an XML parser context
9863
 * @returns the string giving the XML version number, or NULL
9864
 */
9865
xmlChar *
9866
7.72k
xmlParseVersionNum(xmlParserCtxt *ctxt) {
9867
7.72k
    xmlChar *buf = NULL;
9868
7.72k
    int len = 0;
9869
7.72k
    int size = 10;
9870
7.72k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9871
2.13k
                    XML_MAX_TEXT_LENGTH :
9872
7.72k
                    XML_MAX_NAME_LENGTH;
9873
7.72k
    xmlChar cur;
9874
9875
7.72k
    buf = xmlMalloc(size);
9876
7.72k
    if (buf == NULL) {
9877
1
  xmlErrMemory(ctxt);
9878
1
  return(NULL);
9879
1
    }
9880
7.72k
    cur = CUR;
9881
7.72k
    if (!((cur >= '0') && (cur <= '9'))) {
9882
662
  xmlFree(buf);
9883
662
  return(NULL);
9884
662
    }
9885
7.06k
    buf[len++] = cur;
9886
7.06k
    NEXT;
9887
7.06k
    cur=CUR;
9888
7.06k
    if (cur != '.') {
9889
516
  xmlFree(buf);
9890
516
  return(NULL);
9891
516
    }
9892
6.54k
    buf[len++] = cur;
9893
6.54k
    NEXT;
9894
6.54k
    cur=CUR;
9895
22.8k
    while ((cur >= '0') && (cur <= '9')) {
9896
16.3k
  if (len + 1 >= size) {
9897
988
      xmlChar *tmp;
9898
988
            int newSize;
9899
9900
988
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9901
988
            if (newSize < 0) {
9902
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "VersionNum");
9903
0
                xmlFree(buf);
9904
0
                return(NULL);
9905
0
            }
9906
988
      tmp = xmlRealloc(buf, newSize);
9907
988
      if (tmp == NULL) {
9908
1
    xmlErrMemory(ctxt);
9909
1
          xmlFree(buf);
9910
1
    return(NULL);
9911
1
      }
9912
987
      buf = tmp;
9913
987
            size = newSize;
9914
987
  }
9915
16.3k
  buf[len++] = cur;
9916
16.3k
  NEXT;
9917
16.3k
  cur=CUR;
9918
16.3k
    }
9919
6.54k
    buf[len] = 0;
9920
6.54k
    return(buf);
9921
6.54k
}
9922
9923
/**
9924
 * Parse the XML version.
9925
 *
9926
 * @deprecated Internal function, don't use.
9927
 *
9928
 *     [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ")
9929
 *
9930
 *     [25] Eq ::= S? '=' S?
9931
 *
9932
 * @param ctxt  an XML parser context
9933
 * @returns the version string, e.g. "1.0"
9934
 */
9935
9936
xmlChar *
9937
33.0k
xmlParseVersionInfo(xmlParserCtxt *ctxt) {
9938
33.0k
    xmlChar *version = NULL;
9939
9940
33.0k
    if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) {
9941
8.87k
  SKIP(7);
9942
8.87k
  SKIP_BLANKS;
9943
8.87k
  if (RAW != '=') {
9944
676
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
9945
676
      return(NULL);
9946
676
        }
9947
8.19k
  NEXT;
9948
8.19k
  SKIP_BLANKS;
9949
8.19k
  if (RAW == '"') {
9950
2.36k
      NEXT;
9951
2.36k
      version = xmlParseVersionNum(ctxt);
9952
2.36k
      if (RAW != '"') {
9953
321
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9954
321
      } else
9955
2.03k
          NEXT;
9956
5.83k
  } else if (RAW == '\''){
9957
5.36k
      NEXT;
9958
5.36k
      version = xmlParseVersionNum(ctxt);
9959
5.36k
      if (RAW != '\'') {
9960
1.30k
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9961
1.30k
      } else
9962
4.05k
          NEXT;
9963
5.36k
  } else {
9964
471
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
9965
471
  }
9966
8.19k
    }
9967
32.4k
    return(version);
9968
33.0k
}
9969
9970
/**
9971
 * Parse the XML encoding name
9972
 *
9973
 * @deprecated Internal function, don't use.
9974
 *
9975
 *     [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')*
9976
 *
9977
 * @param ctxt  an XML parser context
9978
 * @returns the encoding name value or NULL
9979
 */
9980
xmlChar *
9981
6.55k
xmlParseEncName(xmlParserCtxt *ctxt) {
9982
6.55k
    xmlChar *buf = NULL;
9983
6.55k
    int len = 0;
9984
6.55k
    int size = 10;
9985
6.55k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9986
1.45k
                    XML_MAX_TEXT_LENGTH :
9987
6.55k
                    XML_MAX_NAME_LENGTH;
9988
6.55k
    xmlChar cur;
9989
9990
6.55k
    cur = CUR;
9991
6.55k
    if (((cur >= 'a') && (cur <= 'z')) ||
9992
6.08k
        ((cur >= 'A') && (cur <= 'Z'))) {
9993
6.08k
  buf = xmlMalloc(size);
9994
6.08k
  if (buf == NULL) {
9995
1
      xmlErrMemory(ctxt);
9996
1
      return(NULL);
9997
1
  }
9998
9999
6.08k
  buf[len++] = cur;
10000
6.08k
  NEXT;
10001
6.08k
  cur = CUR;
10002
131k
  while (((cur >= 'a') && (cur <= 'z')) ||
10003
44.6k
         ((cur >= 'A') && (cur <= 'Z')) ||
10004
29.6k
         ((cur >= '0') && (cur <= '9')) ||
10005
13.8k
         (cur == '.') || (cur == '_') ||
10006
125k
         (cur == '-')) {
10007
125k
      if (len + 1 >= size) {
10008
2.80k
          xmlChar *tmp;
10009
2.80k
                int newSize;
10010
10011
2.80k
                newSize = xmlGrowCapacity(size, 1, 1, maxLength);
10012
2.80k
                if (newSize < 0) {
10013
1
                    xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName");
10014
1
                    xmlFree(buf);
10015
1
                    return(NULL);
10016
1
                }
10017
2.80k
    tmp = xmlRealloc(buf, newSize);
10018
2.80k
    if (tmp == NULL) {
10019
1
        xmlErrMemory(ctxt);
10020
1
        xmlFree(buf);
10021
1
        return(NULL);
10022
1
    }
10023
2.80k
    buf = tmp;
10024
2.80k
                size = newSize;
10025
2.80k
      }
10026
125k
      buf[len++] = cur;
10027
125k
      NEXT;
10028
125k
      cur = CUR;
10029
125k
        }
10030
6.07k
  buf[len] = 0;
10031
6.07k
    } else {
10032
474
  xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL);
10033
474
    }
10034
6.55k
    return(buf);
10035
6.55k
}
10036
10037
/**
10038
 * Parse the XML encoding declaration
10039
 *
10040
 * @deprecated Internal function, don't use.
10041
 *
10042
 *     [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | 
10043
 *                           "'" EncName "'")
10044
 *
10045
 * this setups the conversion filters.
10046
 *
10047
 * @param ctxt  an XML parser context
10048
 * @returns the encoding value or NULL
10049
 */
10050
10051
const xmlChar *
10052
32.4k
xmlParseEncodingDecl(xmlParserCtxt *ctxt) {
10053
32.4k
    xmlChar *encoding = NULL;
10054
10055
32.4k
    SKIP_BLANKS;
10056
32.4k
    if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0)
10057
24.9k
        return(NULL);
10058
10059
7.47k
    SKIP(8);
10060
7.47k
    SKIP_BLANKS;
10061
7.47k
    if (RAW != '=') {
10062
492
        xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10063
492
        return(NULL);
10064
492
    }
10065
6.98k
    NEXT;
10066
6.98k
    SKIP_BLANKS;
10067
6.98k
    if (RAW == '"') {
10068
3.28k
        NEXT;
10069
3.28k
        encoding = xmlParseEncName(ctxt);
10070
3.28k
        if (RAW != '"') {
10071
730
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10072
730
            xmlFree(encoding);
10073
730
            return(NULL);
10074
730
        } else
10075
2.55k
            NEXT;
10076
3.69k
    } else if (RAW == '\''){
10077
3.26k
        NEXT;
10078
3.26k
        encoding = xmlParseEncName(ctxt);
10079
3.26k
        if (RAW != '\'') {
10080
608
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10081
608
            xmlFree(encoding);
10082
608
            return(NULL);
10083
608
        } else
10084
2.66k
            NEXT;
10085
3.26k
    } else {
10086
429
        xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10087
429
    }
10088
10089
5.64k
    if (encoding == NULL)
10090
429
        return(NULL);
10091
10092
5.21k
    xmlSetDeclaredEncoding(ctxt, encoding);
10093
10094
5.21k
    return(ctxt->encoding);
10095
5.64k
}
10096
10097
/**
10098
 * Parse the XML standalone declaration
10099
 *
10100
 * @deprecated Internal function, don't use.
10101
 *
10102
 *     [32] SDDecl ::= S 'standalone' Eq
10103
 *                     (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"'))
10104
 *
10105
 * [ VC: Standalone Document Declaration ]
10106
 * TODO The standalone document declaration must have the value "no"
10107
 * if any external markup declarations contain declarations of:
10108
 *  - attributes with default values, if elements to which these
10109
 *    attributes apply appear in the document without specifications
10110
 *    of values for these attributes, or
10111
 *  - entities (other than amp, lt, gt, apos, quot), if references
10112
 *    to those entities appear in the document, or
10113
 *  - attributes with values subject to normalization, where the
10114
 *    attribute appears in the document with a value which will change
10115
 *    as a result of normalization, or
10116
 *  - element types with element content, if white space occurs directly
10117
 *    within any instance of those types.
10118
 *
10119
 * @param ctxt  an XML parser context
10120
 * @returns
10121
 *   1 if standalone="yes"
10122
 *   0 if standalone="no"
10123
 *  -2 if standalone attribute is missing or invalid
10124
 *    (A standalone value of -2 means that the XML declaration was found,
10125
 *     but no value was specified for the standalone attribute).
10126
 */
10127
10128
int
10129
1.52k
xmlParseSDDecl(xmlParserCtxt *ctxt) {
10130
1.52k
    int standalone = -2;
10131
10132
1.52k
    SKIP_BLANKS;
10133
1.52k
    if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) {
10134
115
  SKIP(10);
10135
115
        SKIP_BLANKS;
10136
115
  if (RAW != '=') {
10137
4
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10138
4
      return(standalone);
10139
4
        }
10140
111
  NEXT;
10141
111
  SKIP_BLANKS;
10142
111
        if (RAW == '\''){
10143
34
      NEXT;
10144
34
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10145
16
          standalone = 0;
10146
16
                SKIP(2);
10147
18
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10148
3
                 (NXT(2) == 's')) {
10149
2
          standalone = 1;
10150
2
    SKIP(3);
10151
16
            } else {
10152
16
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10153
16
      }
10154
34
      if (RAW != '\'') {
10155
19
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10156
19
      } else
10157
15
          NEXT;
10158
77
  } else if (RAW == '"'){
10159
73
      NEXT;
10160
73
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10161
4
          standalone = 0;
10162
4
    SKIP(2);
10163
69
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10164
54
                 (NXT(2) == 's')) {
10165
53
          standalone = 1;
10166
53
                SKIP(3);
10167
53
            } else {
10168
16
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10169
16
      }
10170
73
      if (RAW != '"') {
10171
25
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10172
25
      } else
10173
48
          NEXT;
10174
73
  } else {
10175
4
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10176
4
        }
10177
111
    }
10178
1.52k
    return(standalone);
10179
1.52k
}
10180
10181
/**
10182
 * Parse an XML declaration header
10183
 *
10184
 * @deprecated Internal function, don't use.
10185
 *
10186
 *     [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>'
10187
 * @param ctxt  an XML parser context
10188
 */
10189
10190
void
10191
2.73k
xmlParseXMLDecl(xmlParserCtxt *ctxt) {
10192
2.73k
    xmlChar *version;
10193
10194
    /*
10195
     * This value for standalone indicates that the document has an
10196
     * XML declaration but it does not have a standalone attribute.
10197
     * It will be overwritten later if a standalone attribute is found.
10198
     */
10199
10200
2.73k
    ctxt->standalone = -2;
10201
10202
    /*
10203
     * We know that '<?xml' is here.
10204
     */
10205
2.73k
    SKIP(5);
10206
10207
2.73k
    if (!IS_BLANK_CH(RAW)) {
10208
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
10209
0
                 "Blank needed after '<?xml'\n");
10210
0
    }
10211
2.73k
    SKIP_BLANKS;
10212
10213
    /*
10214
     * We must have the VersionInfo here.
10215
     */
10216
2.73k
    version = xmlParseVersionInfo(ctxt);
10217
2.73k
    if (version == NULL) {
10218
1.09k
  xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL);
10219
1.64k
    } else {
10220
1.64k
  if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) {
10221
      /*
10222
       * Changed here for XML-1.0 5th edition
10223
       */
10224
504
      if (ctxt->options & XML_PARSE_OLD10) {
10225
136
    xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10226
136
                "Unsupported version '%s'\n",
10227
136
                version);
10228
368
      } else {
10229
368
          if ((version[0] == '1') && ((version[1] == '.'))) {
10230
137
        xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION,
10231
137
                      "Unsupported version '%s'\n",
10232
137
          version, NULL);
10233
231
    } else {
10234
231
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10235
231
              "Unsupported version '%s'\n",
10236
231
              version);
10237
231
    }
10238
368
      }
10239
504
  }
10240
1.64k
  if (ctxt->version != NULL)
10241
0
      xmlFree(ctxt->version);
10242
1.64k
  ctxt->version = version;
10243
1.64k
    }
10244
10245
    /*
10246
     * We may have the encoding declaration
10247
     */
10248
2.73k
    if (!IS_BLANK_CH(RAW)) {
10249
1.77k
        if ((RAW == '?') && (NXT(1) == '>')) {
10250
646
      SKIP(2);
10251
646
      return;
10252
646
  }
10253
1.12k
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10254
1.12k
    }
10255
2.09k
    xmlParseEncodingDecl(ctxt);
10256
10257
    /*
10258
     * We may have the standalone status.
10259
     */
10260
2.09k
    if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) {
10261
752
        if ((RAW == '?') && (NXT(1) == '>')) {
10262
564
      SKIP(2);
10263
564
      return;
10264
564
  }
10265
188
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10266
188
    }
10267
10268
    /*
10269
     * We can grow the input buffer freely at that point
10270
     */
10271
1.52k
    GROW;
10272
10273
1.52k
    SKIP_BLANKS;
10274
1.52k
    ctxt->standalone = xmlParseSDDecl(ctxt);
10275
10276
1.52k
    SKIP_BLANKS;
10277
1.52k
    if ((RAW == '?') && (NXT(1) == '>')) {
10278
65
        SKIP(2);
10279
1.46k
    } else if (RAW == '>') {
10280
        /* Deprecated old WD ... */
10281
141
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10282
141
  NEXT;
10283
1.32k
    } else {
10284
1.32k
        int c;
10285
10286
1.32k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10287
31.4k
        while ((PARSER_STOPPED(ctxt) == 0) &&
10288
31.3k
               ((c = CUR) != 0)) {
10289
31.0k
            NEXT;
10290
31.0k
            if (c == '>')
10291
927
                break;
10292
31.0k
        }
10293
1.32k
    }
10294
1.52k
}
10295
10296
/**
10297
 * @since 2.14.0
10298
 *
10299
 * @param ctxt  parser context
10300
 * @returns the version from the XML declaration.
10301
 */
10302
const xmlChar *
10303
0
xmlCtxtGetVersion(xmlParserCtxt *ctxt) {
10304
0
    if (ctxt == NULL)
10305
0
        return(NULL);
10306
10307
0
    return(ctxt->version);
10308
0
}
10309
10310
/**
10311
 * @since 2.14.0
10312
 *
10313
 * @param ctxt  parser context
10314
 * @returns the value from the standalone document declaration.
10315
 */
10316
int
10317
0
xmlCtxtGetStandalone(xmlParserCtxt *ctxt) {
10318
0
    if (ctxt == NULL)
10319
0
        return(0);
10320
10321
0
    return(ctxt->standalone);
10322
0
}
10323
10324
/**
10325
 * Parse an XML Misc* optional field.
10326
 *
10327
 * @deprecated Internal function, don't use.
10328
 *
10329
 *     [27] Misc ::= Comment | PI |  S
10330
 * @param ctxt  an XML parser context
10331
 */
10332
10333
void
10334
72.0k
xmlParseMisc(xmlParserCtxt *ctxt) {
10335
75.8k
    while (PARSER_STOPPED(ctxt) == 0) {
10336
72.7k
        SKIP_BLANKS;
10337
72.7k
        GROW;
10338
72.7k
        if ((RAW == '<') && (NXT(1) == '?')) {
10339
2.75k
      xmlParsePI(ctxt);
10340
69.9k
        } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) {
10341
998
      xmlParseComment(ctxt);
10342
68.9k
        } else {
10343
68.9k
            break;
10344
68.9k
        }
10345
72.7k
    }
10346
72.0k
}
10347
10348
static void
10349
32.9k
xmlFinishDocument(xmlParserCtxtPtr ctxt) {
10350
32.9k
    xmlDocPtr doc;
10351
10352
    /*
10353
     * SAX: end of the document processing.
10354
     */
10355
32.9k
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10356
32.9k
        ctxt->sax->endDocument(ctxt->userData);
10357
10358
    /*
10359
     * Remove locally kept entity definitions if the tree was not built
10360
     */
10361
32.9k
    doc = ctxt->myDoc;
10362
32.9k
    if ((doc != NULL) &&
10363
32.4k
        (xmlStrEqual(doc->version, SAX_COMPAT_MODE))) {
10364
98
        xmlFreeDoc(doc);
10365
98
        ctxt->myDoc = NULL;
10366
98
    }
10367
32.9k
}
10368
10369
/**
10370
 * Parse an XML document and invoke the SAX handlers. This is useful
10371
 * if you're only interested in custom SAX callbacks. If you want a
10372
 * document tree, use #xmlCtxtParseDocument.
10373
 *
10374
 * @param ctxt  an XML parser context
10375
 * @returns 0, -1 in case of error.
10376
 */
10377
10378
int
10379
33.1k
xmlParseDocument(xmlParserCtxt *ctxt) {
10380
33.1k
    if ((ctxt == NULL) || (ctxt->input == NULL))
10381
0
        return(-1);
10382
10383
33.1k
    GROW;
10384
10385
    /*
10386
     * SAX: detecting the level.
10387
     */
10388
33.1k
    xmlCtxtInitializeLate(ctxt);
10389
10390
33.1k
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10391
33.1k
        ctxt->sax->setDocumentLocator(ctxt->userData,
10392
33.1k
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10393
33.1k
    }
10394
10395
33.1k
    xmlDetectEncoding(ctxt);
10396
10397
33.1k
    if (CUR == 0) {
10398
193
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10399
193
  return(-1);
10400
193
    }
10401
10402
32.9k
    GROW;
10403
32.9k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
10404
10405
  /*
10406
   * Note that we will switch encoding on the fly.
10407
   */
10408
2.73k
  xmlParseXMLDecl(ctxt);
10409
2.73k
  SKIP_BLANKS;
10410
30.2k
    } else {
10411
30.2k
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10412
30.2k
        if (ctxt->version == NULL) {
10413
16
            xmlErrMemory(ctxt);
10414
16
            return(-1);
10415
16
        }
10416
30.2k
    }
10417
32.9k
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10418
32.3k
        ctxt->sax->startDocument(ctxt->userData);
10419
32.9k
    if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) &&
10420
32.3k
        (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) {
10421
0
  ctxt->myDoc->compression = ctxt->input->buf->compressed;
10422
0
    }
10423
10424
    /*
10425
     * The Misc part of the Prolog
10426
     */
10427
32.9k
    xmlParseMisc(ctxt);
10428
10429
    /*
10430
     * Then possibly doc type declaration(s) and more Misc
10431
     * (doctypedecl Misc*)?
10432
     */
10433
32.9k
    GROW;
10434
32.9k
    if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) {
10435
10436
14.3k
  ctxt->inSubset = 1;
10437
14.3k
  xmlParseDocTypeDecl(ctxt);
10438
14.3k
  if (RAW == '[') {
10439
11.3k
      xmlParseInternalSubset(ctxt);
10440
11.3k
  } else if (RAW == '>') {
10441
2.66k
            NEXT;
10442
2.66k
        }
10443
10444
  /*
10445
   * Create and update the external subset.
10446
   */
10447
14.3k
  ctxt->inSubset = 2;
10448
14.3k
  if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) &&
10449
14.3k
      (!ctxt->disableSAX))
10450
12.2k
      ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
10451
12.2k
                                ctxt->extSubSystem, ctxt->extSubURI);
10452
14.3k
  ctxt->inSubset = 0;
10453
10454
14.3k
        xmlCleanSpecialAttr(ctxt);
10455
10456
14.3k
  xmlParseMisc(ctxt);
10457
14.3k
    }
10458
10459
    /*
10460
     * Time to start parsing the tree itself
10461
     */
10462
32.9k
    GROW;
10463
32.9k
    if (RAW != '<') {
10464
8.15k
        if (ctxt->wellFormed)
10465
1.06k
            xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10466
1.06k
                           "Start tag expected, '<' not found\n");
10467
24.7k
    } else {
10468
24.7k
  xmlParseElement(ctxt);
10469
10470
  /*
10471
   * The Misc part at the end
10472
   */
10473
24.7k
  xmlParseMisc(ctxt);
10474
10475
24.7k
        xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
10476
24.7k
    }
10477
10478
32.9k
    ctxt->instate = XML_PARSER_EOF;
10479
32.9k
    xmlFinishDocument(ctxt);
10480
10481
32.9k
    if (! ctxt->wellFormed) {
10482
32.6k
  ctxt->valid = 0;
10483
32.6k
  return(-1);
10484
32.6k
    }
10485
10486
294
    return(0);
10487
32.9k
}
10488
10489
/**
10490
 * Parse a general parsed entity
10491
 * An external general parsed entity is well-formed if it matches the
10492
 * production labeled extParsedEnt.
10493
 *
10494
 * @deprecated Internal function, don't use.
10495
 *
10496
 *     [78] extParsedEnt ::= TextDecl? content
10497
 *
10498
 * @param ctxt  an XML parser context
10499
 * @returns 0, -1 in case of error. the parser context is augmented
10500
 *                as a result of the parsing.
10501
 */
10502
10503
int
10504
0
xmlParseExtParsedEnt(xmlParserCtxt *ctxt) {
10505
0
    if ((ctxt == NULL) || (ctxt->input == NULL))
10506
0
        return(-1);
10507
10508
0
    xmlCtxtInitializeLate(ctxt);
10509
10510
0
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10511
0
        ctxt->sax->setDocumentLocator(ctxt->userData,
10512
0
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10513
0
    }
10514
10515
0
    xmlDetectEncoding(ctxt);
10516
10517
0
    if (CUR == 0) {
10518
0
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10519
0
    }
10520
10521
    /*
10522
     * Check for the XMLDecl in the Prolog.
10523
     */
10524
0
    GROW;
10525
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
10526
10527
  /*
10528
   * Note that we will switch encoding on the fly.
10529
   */
10530
0
  xmlParseXMLDecl(ctxt);
10531
0
  SKIP_BLANKS;
10532
0
    } else {
10533
0
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10534
0
    }
10535
0
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10536
0
        ctxt->sax->startDocument(ctxt->userData);
10537
10538
    /*
10539
     * Doing validity checking on chunk doesn't make sense
10540
     */
10541
0
    ctxt->options &= ~XML_PARSE_DTDVALID;
10542
0
    ctxt->validate = 0;
10543
0
    ctxt->depth = 0;
10544
10545
0
    xmlParseContentInternal(ctxt);
10546
10547
0
    if (ctxt->input->cur < ctxt->input->end)
10548
0
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
10549
10550
    /*
10551
     * SAX: end of the document processing.
10552
     */
10553
0
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10554
0
        ctxt->sax->endDocument(ctxt->userData);
10555
10556
0
    if (! ctxt->wellFormed) return(-1);
10557
0
    return(0);
10558
0
}
10559
10560
#ifdef LIBXML_PUSH_ENABLED
10561
/************************************************************************
10562
 *                  *
10563
 *    Progressive parsing interfaces        *
10564
 *                  *
10565
 ************************************************************************/
10566
10567
/**
10568
 * Check whether the input buffer contains a character.
10569
 *
10570
 * @param ctxt  an XML parser context
10571
 * @param c  character
10572
 */
10573
static int
10574
0
xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) {
10575
0
    const xmlChar *cur;
10576
10577
0
    if (ctxt->checkIndex == 0) {
10578
0
        cur = ctxt->input->cur + 1;
10579
0
    } else {
10580
0
        cur = ctxt->input->cur + ctxt->checkIndex;
10581
0
    }
10582
10583
0
    if (memchr(cur, c, ctxt->input->end - cur) == NULL) {
10584
0
        size_t index = ctxt->input->end - ctxt->input->cur;
10585
10586
0
        if (index > LONG_MAX) {
10587
0
            ctxt->checkIndex = 0;
10588
0
            return(1);
10589
0
        }
10590
0
        ctxt->checkIndex = index;
10591
0
        return(0);
10592
0
    } else {
10593
0
        ctxt->checkIndex = 0;
10594
0
        return(1);
10595
0
    }
10596
0
}
10597
10598
/**
10599
 * Check whether the input buffer contains a string.
10600
 *
10601
 * @param ctxt  an XML parser context
10602
 * @param startDelta  delta to apply at the start
10603
 * @param str  string
10604
 * @param strLen  length of string
10605
 */
10606
static const xmlChar *
10607
xmlParseLookupString(xmlParserCtxtPtr ctxt, size_t startDelta,
10608
0
                     const char *str, size_t strLen) {
10609
0
    const xmlChar *cur, *term;
10610
10611
0
    if (ctxt->checkIndex == 0) {
10612
0
        cur = ctxt->input->cur + startDelta;
10613
0
    } else {
10614
0
        cur = ctxt->input->cur + ctxt->checkIndex;
10615
0
    }
10616
10617
0
    term = BAD_CAST strstr((const char *) cur, str);
10618
0
    if (term == NULL) {
10619
0
        const xmlChar *end = ctxt->input->end;
10620
0
        size_t index;
10621
10622
        /* Rescan (strLen - 1) characters. */
10623
0
        if ((size_t) (end - cur) < strLen)
10624
0
            end = cur;
10625
0
        else
10626
0
            end -= strLen - 1;
10627
0
        index = end - ctxt->input->cur;
10628
0
        if (index > LONG_MAX) {
10629
0
            ctxt->checkIndex = 0;
10630
0
            return(ctxt->input->end - strLen);
10631
0
        }
10632
0
        ctxt->checkIndex = index;
10633
0
    } else {
10634
0
        ctxt->checkIndex = 0;
10635
0
    }
10636
10637
0
    return(term);
10638
0
}
10639
10640
/**
10641
 * Check whether the input buffer contains terminated char data.
10642
 *
10643
 * @param ctxt  an XML parser context
10644
 */
10645
static int
10646
0
xmlParseLookupCharData(xmlParserCtxtPtr ctxt) {
10647
0
    const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex;
10648
0
    const xmlChar *end = ctxt->input->end;
10649
0
    size_t index;
10650
10651
0
    while (cur < end) {
10652
0
        if ((*cur == '<') || (*cur == '&')) {
10653
0
            ctxt->checkIndex = 0;
10654
0
            return(1);
10655
0
        }
10656
0
        cur++;
10657
0
    }
10658
10659
0
    index = cur - ctxt->input->cur;
10660
0
    if (index > LONG_MAX) {
10661
0
        ctxt->checkIndex = 0;
10662
0
        return(1);
10663
0
    }
10664
0
    ctxt->checkIndex = index;
10665
0
    return(0);
10666
0
}
10667
10668
/**
10669
 * Check whether there's enough data in the input buffer to finish parsing
10670
 * a start tag. This has to take quotes into account.
10671
 *
10672
 * @param ctxt  an XML parser context
10673
 */
10674
static int
10675
0
xmlParseLookupGt(xmlParserCtxtPtr ctxt) {
10676
0
    const xmlChar *cur;
10677
0
    const xmlChar *end = ctxt->input->end;
10678
0
    int state = ctxt->endCheckState;
10679
0
    size_t index;
10680
10681
0
    if (ctxt->checkIndex == 0)
10682
0
        cur = ctxt->input->cur + 1;
10683
0
    else
10684
0
        cur = ctxt->input->cur + ctxt->checkIndex;
10685
10686
0
    while (cur < end) {
10687
0
        if (state) {
10688
0
            if (*cur == state)
10689
0
                state = 0;
10690
0
        } else if (*cur == '\'' || *cur == '"') {
10691
0
            state = *cur;
10692
0
        } else if (*cur == '>') {
10693
0
            ctxt->checkIndex = 0;
10694
0
            ctxt->endCheckState = 0;
10695
0
            return(1);
10696
0
        }
10697
0
        cur++;
10698
0
    }
10699
10700
0
    index = cur - ctxt->input->cur;
10701
0
    if (index > LONG_MAX) {
10702
0
        ctxt->checkIndex = 0;
10703
0
        ctxt->endCheckState = 0;
10704
0
        return(1);
10705
0
    }
10706
0
    ctxt->checkIndex = index;
10707
0
    ctxt->endCheckState = state;
10708
0
    return(0);
10709
0
}
10710
10711
/**
10712
 * Check whether there's enough data in the input buffer to finish parsing
10713
 * the internal subset.
10714
 *
10715
 * @param ctxt  an XML parser context
10716
 */
10717
static int
10718
0
xmlParseLookupInternalSubset(xmlParserCtxtPtr ctxt) {
10719
    /*
10720
     * Sorry, but progressive parsing of the internal subset is not
10721
     * supported. We first check that the full content of the internal
10722
     * subset is available and parsing is launched only at that point.
10723
     * Internal subset ends with "']' S? '>'" in an unescaped section and
10724
     * not in a ']]>' sequence which are conditional sections.
10725
     */
10726
0
    const xmlChar *cur, *start;
10727
0
    const xmlChar *end = ctxt->input->end;
10728
0
    int state = ctxt->endCheckState;
10729
0
    size_t index;
10730
10731
0
    if (ctxt->checkIndex == 0) {
10732
0
        cur = ctxt->input->cur + 1;
10733
0
    } else {
10734
0
        cur = ctxt->input->cur + ctxt->checkIndex;
10735
0
    }
10736
0
    start = cur;
10737
10738
0
    while (cur < end) {
10739
0
        if (state == '-') {
10740
0
            if ((*cur == '-') &&
10741
0
                (cur[1] == '-') &&
10742
0
                (cur[2] == '>')) {
10743
0
                state = 0;
10744
0
                cur += 3;
10745
0
                start = cur;
10746
0
                continue;
10747
0
            }
10748
0
        }
10749
0
        else if (state == ']') {
10750
0
            if (*cur == '>') {
10751
0
                ctxt->checkIndex = 0;
10752
0
                ctxt->endCheckState = 0;
10753
0
                return(1);
10754
0
            }
10755
0
            if (IS_BLANK_CH(*cur)) {
10756
0
                state = ' ';
10757
0
            } else if (*cur != ']') {
10758
0
                state = 0;
10759
0
                start = cur;
10760
0
                continue;
10761
0
            }
10762
0
        }
10763
0
        else if (state == ' ') {
10764
0
            if (*cur == '>') {
10765
0
                ctxt->checkIndex = 0;
10766
0
                ctxt->endCheckState = 0;
10767
0
                return(1);
10768
0
            }
10769
0
            if (!IS_BLANK_CH(*cur)) {
10770
0
                state = 0;
10771
0
                start = cur;
10772
0
                continue;
10773
0
            }
10774
0
        }
10775
0
        else if (state != 0) {
10776
0
            if (*cur == state) {
10777
0
                state = 0;
10778
0
                start = cur + 1;
10779
0
            }
10780
0
        }
10781
0
        else if (*cur == '<') {
10782
0
            if ((cur[1] == '!') &&
10783
0
                (cur[2] == '-') &&
10784
0
                (cur[3] == '-')) {
10785
0
                state = '-';
10786
0
                cur += 4;
10787
                /* Don't treat <!--> as comment */
10788
0
                start = cur;
10789
0
                continue;
10790
0
            }
10791
0
        }
10792
0
        else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) {
10793
0
            state = *cur;
10794
0
        }
10795
10796
0
        cur++;
10797
0
    }
10798
10799
    /*
10800
     * Rescan the three last characters to detect "<!--" and "-->"
10801
     * split across chunks.
10802
     */
10803
0
    if ((state == 0) || (state == '-')) {
10804
0
        if (cur - start < 3)
10805
0
            cur = start;
10806
0
        else
10807
0
            cur -= 3;
10808
0
    }
10809
0
    index = cur - ctxt->input->cur;
10810
0
    if (index > LONG_MAX) {
10811
0
        ctxt->checkIndex = 0;
10812
0
        ctxt->endCheckState = 0;
10813
0
        return(1);
10814
0
    }
10815
0
    ctxt->checkIndex = index;
10816
0
    ctxt->endCheckState = state;
10817
0
    return(0);
10818
0
}
10819
10820
/**
10821
 * Try to progress on parsing
10822
 *
10823
 * @param ctxt  an XML parser context
10824
 * @param terminate  last chunk indicator
10825
 * @returns zero if no parsing was possible
10826
 */
10827
static int
10828
0
xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) {
10829
0
    int ret = 0;
10830
0
    size_t avail;
10831
0
    xmlChar cur, next;
10832
10833
0
    if (ctxt->input == NULL)
10834
0
        return(0);
10835
10836
0
    if ((ctxt->input != NULL) &&
10837
0
        (ctxt->input->cur - ctxt->input->base > 4096)) {
10838
0
        xmlParserShrink(ctxt);
10839
0
    }
10840
10841
0
    while (ctxt->disableSAX == 0) {
10842
0
        avail = ctxt->input->end - ctxt->input->cur;
10843
0
        if (avail < 1)
10844
0
      goto done;
10845
0
        switch (ctxt->instate) {
10846
0
            case XML_PARSER_EOF:
10847
          /*
10848
     * Document parsing is done !
10849
     */
10850
0
          goto done;
10851
0
            case XML_PARSER_START:
10852
                /*
10853
                 * Very first chars read from the document flow.
10854
                 */
10855
0
                if ((!terminate) && (avail < 4))
10856
0
                    goto done;
10857
10858
                /*
10859
                 * We need more bytes to detect EBCDIC code pages.
10860
                 * See xmlDetectEBCDIC.
10861
                 */
10862
0
                if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) &&
10863
0
                    (!terminate) && (avail < 200))
10864
0
                    goto done;
10865
10866
0
                xmlDetectEncoding(ctxt);
10867
0
                ctxt->instate = XML_PARSER_XML_DECL;
10868
0
    break;
10869
10870
0
            case XML_PARSER_XML_DECL:
10871
0
    if ((!terminate) && (avail < 2))
10872
0
        goto done;
10873
0
    cur = ctxt->input->cur[0];
10874
0
    next = ctxt->input->cur[1];
10875
0
          if ((cur == '<') && (next == '?')) {
10876
        /* PI or XML decl */
10877
0
        if ((!terminate) &&
10878
0
                        (!xmlParseLookupString(ctxt, 2, "?>", 2)))
10879
0
      goto done;
10880
0
        if ((ctxt->input->cur[2] == 'x') &&
10881
0
      (ctxt->input->cur[3] == 'm') &&
10882
0
      (ctxt->input->cur[4] == 'l') &&
10883
0
      (IS_BLANK_CH(ctxt->input->cur[5]))) {
10884
0
      ret += 5;
10885
0
      xmlParseXMLDecl(ctxt);
10886
0
        } else {
10887
0
      ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10888
0
                        if (ctxt->version == NULL) {
10889
0
                            xmlErrMemory(ctxt);
10890
0
                            break;
10891
0
                        }
10892
0
        }
10893
0
    } else {
10894
0
        ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10895
0
        if (ctxt->version == NULL) {
10896
0
            xmlErrMemory(ctxt);
10897
0
      break;
10898
0
        }
10899
0
    }
10900
0
                if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10901
0
                    ctxt->sax->setDocumentLocator(ctxt->userData,
10902
0
                            (xmlSAXLocator *) &xmlDefaultSAXLocator);
10903
0
                }
10904
0
                if ((ctxt->sax) && (ctxt->sax->startDocument) &&
10905
0
                    (!ctxt->disableSAX))
10906
0
                    ctxt->sax->startDocument(ctxt->userData);
10907
0
                ctxt->instate = XML_PARSER_MISC;
10908
0
    break;
10909
0
            case XML_PARSER_START_TAG: {
10910
0
          const xmlChar *name;
10911
0
    const xmlChar *prefix = NULL;
10912
0
    const xmlChar *URI = NULL;
10913
0
                int line = ctxt->input->line;
10914
0
    int nbNs = 0;
10915
10916
0
    if ((!terminate) && (avail < 2))
10917
0
        goto done;
10918
0
    cur = ctxt->input->cur[0];
10919
0
          if (cur != '<') {
10920
0
        xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10921
0
                                   "Start tag expected, '<' not found");
10922
0
                    ctxt->instate = XML_PARSER_EOF;
10923
0
                    xmlFinishDocument(ctxt);
10924
0
        goto done;
10925
0
    }
10926
0
    if ((!terminate) && (!xmlParseLookupGt(ctxt)))
10927
0
                    goto done;
10928
0
    if (ctxt->spaceNr == 0)
10929
0
        spacePush(ctxt, -1);
10930
0
    else if (*ctxt->space == -2)
10931
0
        spacePush(ctxt, -1);
10932
0
    else
10933
0
        spacePush(ctxt, *ctxt->space);
10934
0
#ifdef LIBXML_SAX1_ENABLED
10935
0
    if (ctxt->sax2)
10936
0
#endif /* LIBXML_SAX1_ENABLED */
10937
0
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
10938
0
#ifdef LIBXML_SAX1_ENABLED
10939
0
    else
10940
0
        name = xmlParseStartTag(ctxt);
10941
0
#endif /* LIBXML_SAX1_ENABLED */
10942
0
    if (name == NULL) {
10943
0
        spacePop(ctxt);
10944
0
                    ctxt->instate = XML_PARSER_EOF;
10945
0
                    xmlFinishDocument(ctxt);
10946
0
        goto done;
10947
0
    }
10948
0
#ifdef LIBXML_VALID_ENABLED
10949
    /*
10950
     * [ VC: Root Element Type ]
10951
     * The Name in the document type declaration must match
10952
     * the element type of the root element.
10953
     */
10954
0
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
10955
0
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
10956
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
10957
0
#endif /* LIBXML_VALID_ENABLED */
10958
10959
    /*
10960
     * Check for an Empty Element.
10961
     */
10962
0
    if ((RAW == '/') && (NXT(1) == '>')) {
10963
0
        SKIP(2);
10964
10965
0
        if (ctxt->sax2) {
10966
0
      if ((ctxt->sax != NULL) &&
10967
0
          (ctxt->sax->endElementNs != NULL) &&
10968
0
          (!ctxt->disableSAX))
10969
0
          ctxt->sax->endElementNs(ctxt->userData, name,
10970
0
                                  prefix, URI);
10971
0
      if (nbNs > 0)
10972
0
          xmlParserNsPop(ctxt, nbNs);
10973
0
#ifdef LIBXML_SAX1_ENABLED
10974
0
        } else {
10975
0
      if ((ctxt->sax != NULL) &&
10976
0
          (ctxt->sax->endElement != NULL) &&
10977
0
          (!ctxt->disableSAX))
10978
0
          ctxt->sax->endElement(ctxt->userData, name);
10979
0
#endif /* LIBXML_SAX1_ENABLED */
10980
0
        }
10981
0
        spacePop(ctxt);
10982
0
    } else if (RAW == '>') {
10983
0
        NEXT;
10984
0
                    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
10985
0
    } else {
10986
0
        xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED,
10987
0
           "Couldn't find end of Start Tag %s\n",
10988
0
           name);
10989
0
        nodePop(ctxt);
10990
0
        spacePop(ctxt);
10991
0
                    if (nbNs > 0)
10992
0
                        xmlParserNsPop(ctxt, nbNs);
10993
0
    }
10994
10995
0
                if (ctxt->nameNr == 0)
10996
0
                    ctxt->instate = XML_PARSER_EPILOG;
10997
0
                else
10998
0
                    ctxt->instate = XML_PARSER_CONTENT;
10999
0
                break;
11000
0
      }
11001
0
            case XML_PARSER_CONTENT: {
11002
0
    cur = ctxt->input->cur[0];
11003
11004
0
    if (cur == '<') {
11005
0
                    if ((!terminate) && (avail < 2))
11006
0
                        goto done;
11007
0
        next = ctxt->input->cur[1];
11008
11009
0
                    if (next == '/') {
11010
0
                        ctxt->instate = XML_PARSER_END_TAG;
11011
0
                        break;
11012
0
                    } else if (next == '?') {
11013
0
                        if ((!terminate) &&
11014
0
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11015
0
                            goto done;
11016
0
                        xmlParsePI(ctxt);
11017
0
                        ctxt->instate = XML_PARSER_CONTENT;
11018
0
                        break;
11019
0
                    } else if (next == '!') {
11020
0
                        if ((!terminate) && (avail < 3))
11021
0
                            goto done;
11022
0
                        next = ctxt->input->cur[2];
11023
11024
0
                        if (next == '-') {
11025
0
                            if ((!terminate) && (avail < 4))
11026
0
                                goto done;
11027
0
                            if (ctxt->input->cur[3] == '-') {
11028
0
                                if ((!terminate) &&
11029
0
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11030
0
                                    goto done;
11031
0
                                xmlParseComment(ctxt);
11032
0
                                ctxt->instate = XML_PARSER_CONTENT;
11033
0
                                break;
11034
0
                            }
11035
0
                        } else if (next == '[') {
11036
0
                            if ((!terminate) && (avail < 9))
11037
0
                                goto done;
11038
0
                            if ((ctxt->input->cur[2] == '[') &&
11039
0
                                (ctxt->input->cur[3] == 'C') &&
11040
0
                                (ctxt->input->cur[4] == 'D') &&
11041
0
                                (ctxt->input->cur[5] == 'A') &&
11042
0
                                (ctxt->input->cur[6] == 'T') &&
11043
0
                                (ctxt->input->cur[7] == 'A') &&
11044
0
                                (ctxt->input->cur[8] == '[')) {
11045
0
                                if ((!terminate) &&
11046
0
                                    (!xmlParseLookupString(ctxt, 9, "]]>", 3)))
11047
0
                                    goto done;
11048
0
                                ctxt->instate = XML_PARSER_CDATA_SECTION;
11049
0
                                xmlParseCDSect(ctxt);
11050
0
                                ctxt->instate = XML_PARSER_CONTENT;
11051
0
                                break;
11052
0
                            }
11053
0
                        }
11054
0
                    }
11055
0
    } else if (cur == '&') {
11056
0
        if ((!terminate) && (!xmlParseLookupChar(ctxt, ';')))
11057
0
      goto done;
11058
0
        xmlParseReference(ctxt);
11059
0
                    break;
11060
0
    } else {
11061
        /* TODO Avoid the extra copy, handle directly !!! */
11062
        /*
11063
         * Goal of the following test is:
11064
         *  - minimize calls to the SAX 'character' callback
11065
         *    when they are mergeable
11066
         *  - handle an problem for isBlank when we only parse
11067
         *    a sequence of blank chars and the next one is
11068
         *    not available to check against '<' presence.
11069
         *  - tries to homogenize the differences in SAX
11070
         *    callbacks between the push and pull versions
11071
         *    of the parser.
11072
         */
11073
0
        if (avail < XML_PARSER_BIG_BUFFER_SIZE) {
11074
0
      if ((!terminate) && (!xmlParseLookupCharData(ctxt)))
11075
0
          goto done;
11076
0
                    }
11077
0
                    ctxt->checkIndex = 0;
11078
0
        xmlParseCharDataInternal(ctxt, !terminate);
11079
0
                    break;
11080
0
    }
11081
11082
0
                ctxt->instate = XML_PARSER_START_TAG;
11083
0
    break;
11084
0
      }
11085
0
            case XML_PARSER_END_TAG:
11086
0
    if ((!terminate) && (!xmlParseLookupChar(ctxt, '>')))
11087
0
        goto done;
11088
0
    if (ctxt->sax2) {
11089
0
              xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
11090
0
        nameNsPop(ctxt);
11091
0
    }
11092
0
#ifdef LIBXML_SAX1_ENABLED
11093
0
      else
11094
0
        xmlParseEndTag1(ctxt, 0);
11095
0
#endif /* LIBXML_SAX1_ENABLED */
11096
0
    if (ctxt->nameNr == 0) {
11097
0
        ctxt->instate = XML_PARSER_EPILOG;
11098
0
    } else {
11099
0
        ctxt->instate = XML_PARSER_CONTENT;
11100
0
    }
11101
0
    break;
11102
0
            case XML_PARSER_MISC:
11103
0
            case XML_PARSER_PROLOG:
11104
0
            case XML_PARSER_EPILOG:
11105
0
    SKIP_BLANKS;
11106
0
                avail = ctxt->input->end - ctxt->input->cur;
11107
0
    if (avail < 1)
11108
0
        goto done;
11109
0
    if (ctxt->input->cur[0] == '<') {
11110
0
                    if ((!terminate) && (avail < 2))
11111
0
                        goto done;
11112
0
                    next = ctxt->input->cur[1];
11113
0
                    if (next == '?') {
11114
0
                        if ((!terminate) &&
11115
0
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11116
0
                            goto done;
11117
0
                        xmlParsePI(ctxt);
11118
0
                        break;
11119
0
                    } else if (next == '!') {
11120
0
                        if ((!terminate) && (avail < 3))
11121
0
                            goto done;
11122
11123
0
                        if (ctxt->input->cur[2] == '-') {
11124
0
                            if ((!terminate) && (avail < 4))
11125
0
                                goto done;
11126
0
                            if (ctxt->input->cur[3] == '-') {
11127
0
                                if ((!terminate) &&
11128
0
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11129
0
                                    goto done;
11130
0
                                xmlParseComment(ctxt);
11131
0
                                break;
11132
0
                            }
11133
0
                        } else if (ctxt->instate == XML_PARSER_MISC) {
11134
0
                            if ((!terminate) && (avail < 9))
11135
0
                                goto done;
11136
0
                            if ((ctxt->input->cur[2] == 'D') &&
11137
0
                                (ctxt->input->cur[3] == 'O') &&
11138
0
                                (ctxt->input->cur[4] == 'C') &&
11139
0
                                (ctxt->input->cur[5] == 'T') &&
11140
0
                                (ctxt->input->cur[6] == 'Y') &&
11141
0
                                (ctxt->input->cur[7] == 'P') &&
11142
0
                                (ctxt->input->cur[8] == 'E')) {
11143
0
                                if ((!terminate) && (!xmlParseLookupGt(ctxt)))
11144
0
                                    goto done;
11145
0
                                ctxt->inSubset = 1;
11146
0
                                xmlParseDocTypeDecl(ctxt);
11147
0
                                if (RAW == '[') {
11148
0
                                    ctxt->instate = XML_PARSER_DTD;
11149
0
                                } else {
11150
0
                                    if (RAW == '>')
11151
0
                                        NEXT;
11152
                                    /*
11153
                                     * Create and update the external subset.
11154
                                     */
11155
0
                                    ctxt->inSubset = 2;
11156
0
                                    if ((ctxt->sax != NULL) &&
11157
0
                                        (!ctxt->disableSAX) &&
11158
0
                                        (ctxt->sax->externalSubset != NULL))
11159
0
                                        ctxt->sax->externalSubset(
11160
0
                                                ctxt->userData,
11161
0
                                                ctxt->intSubName,
11162
0
                                                ctxt->extSubSystem,
11163
0
                                                ctxt->extSubURI);
11164
0
                                    ctxt->inSubset = 0;
11165
0
                                    xmlCleanSpecialAttr(ctxt);
11166
0
                                    ctxt->instate = XML_PARSER_PROLOG;
11167
0
                                }
11168
0
                                break;
11169
0
                            }
11170
0
                        }
11171
0
                    }
11172
0
                }
11173
11174
0
                if (ctxt->instate == XML_PARSER_EPILOG) {
11175
0
                    if (ctxt->errNo == XML_ERR_OK)
11176
0
                        xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL);
11177
0
        ctxt->instate = XML_PARSER_EOF;
11178
0
                    xmlFinishDocument(ctxt);
11179
0
                } else {
11180
0
        ctxt->instate = XML_PARSER_START_TAG;
11181
0
    }
11182
0
    break;
11183
0
            case XML_PARSER_DTD: {
11184
0
                if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt)))
11185
0
                    goto done;
11186
0
    xmlParseInternalSubset(ctxt);
11187
0
    ctxt->inSubset = 2;
11188
0
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
11189
0
        (ctxt->sax->externalSubset != NULL))
11190
0
        ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
11191
0
          ctxt->extSubSystem, ctxt->extSubURI);
11192
0
    ctxt->inSubset = 0;
11193
0
    xmlCleanSpecialAttr(ctxt);
11194
0
    ctxt->instate = XML_PARSER_PROLOG;
11195
0
                break;
11196
0
      }
11197
0
            default:
11198
0
                xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
11199
0
      "PP: internal error\n");
11200
0
    ctxt->instate = XML_PARSER_EOF;
11201
0
    break;
11202
0
  }
11203
0
    }
11204
0
done:
11205
0
    return(ret);
11206
0
}
11207
11208
/**
11209
 * Parse a chunk of memory in push parser mode.
11210
 *
11211
 * Assumes that the parser context was initialized with
11212
 * #xmlCreatePushParserCtxt.
11213
 *
11214
 * The last chunk, which will often be empty, must be marked with
11215
 * the `terminate` flag. With the default SAX callbacks, the resulting
11216
 * document will be available in ctxt->myDoc. This pointer will not
11217
 * be freed when calling #xmlFreeParserCtxt and must be freed by the
11218
 * caller. If the document isn't well-formed, it will still be returned
11219
 * in ctxt->myDoc.
11220
 *
11221
 * As an exception, #xmlCtxtResetPush will free the document in
11222
 * ctxt->myDoc. So ctxt->myDoc should be set to NULL after extracting
11223
 * the document.
11224
 *
11225
 * Since 2.14.0, #xmlCtxtGetDocument can be used to retrieve the
11226
 * result document.
11227
 *
11228
 * @param ctxt  an XML parser context
11229
 * @param chunk  chunk of memory
11230
 * @param size  size of chunk in bytes
11231
 * @param terminate  last chunk indicator
11232
 * @returns an xmlParserErrors code (0 on success).
11233
 */
11234
int
11235
xmlParseChunk(xmlParserCtxt *ctxt, const char *chunk, int size,
11236
0
              int terminate) {
11237
0
    size_t curBase;
11238
0
    size_t maxLength;
11239
0
    size_t pos;
11240
0
    int end_in_lf = 0;
11241
0
    int res;
11242
11243
0
    if ((ctxt == NULL) || (size < 0))
11244
0
        return(XML_ERR_ARGUMENT);
11245
0
    if ((chunk == NULL) && (size > 0))
11246
0
        return(XML_ERR_ARGUMENT);
11247
0
    if ((ctxt->input == NULL) || (ctxt->input->buf == NULL))
11248
0
        return(XML_ERR_ARGUMENT);
11249
0
    if (ctxt->disableSAX != 0)
11250
0
        return(ctxt->errNo);
11251
11252
0
    ctxt->input->flags |= XML_INPUT_PROGRESSIVE;
11253
0
    if (ctxt->instate == XML_PARSER_START)
11254
0
        xmlCtxtInitializeLate(ctxt);
11255
0
    if ((size > 0) && (chunk != NULL) && (!terminate) &&
11256
0
        (chunk[size - 1] == '\r')) {
11257
0
  end_in_lf = 1;
11258
0
  size--;
11259
0
    }
11260
11261
    /*
11262
     * Also push an empty chunk to make sure that the raw buffer
11263
     * will be flushed if there is an encoder.
11264
     */
11265
0
    pos = ctxt->input->cur - ctxt->input->base;
11266
0
    res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk);
11267
0
    xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11268
0
    if (res < 0) {
11269
0
        xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11270
0
        return(ctxt->errNo);
11271
0
    }
11272
11273
0
    xmlParseTryOrFinish(ctxt, terminate);
11274
11275
0
    curBase = ctxt->input->cur - ctxt->input->base;
11276
0
    maxLength = (ctxt->options & XML_PARSE_HUGE) ?
11277
0
                XML_MAX_HUGE_LENGTH :
11278
0
                XML_MAX_LOOKUP_LIMIT;
11279
0
    if (curBase > maxLength) {
11280
0
        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
11281
0
                    "Buffer size limit exceeded, try XML_PARSE_HUGE\n");
11282
0
    }
11283
11284
0
    if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0))
11285
0
        return(ctxt->errNo);
11286
11287
0
    if (end_in_lf == 1) {
11288
0
  pos = ctxt->input->cur - ctxt->input->base;
11289
0
  res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r");
11290
0
  xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11291
0
        if (res < 0) {
11292
0
            xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11293
0
            return(ctxt->errNo);
11294
0
        }
11295
0
    }
11296
0
    if (terminate) {
11297
  /*
11298
   * Check for termination
11299
   */
11300
0
        if ((ctxt->instate != XML_PARSER_EOF) &&
11301
0
            (ctxt->instate != XML_PARSER_EPILOG)) {
11302
0
            if (ctxt->nameNr > 0) {
11303
0
                const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
11304
0
                int line = ctxt->pushTab[ctxt->nameNr - 1].line;
11305
0
                xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
11306
0
                        "Premature end of data in tag %s line %d\n",
11307
0
                        name, line, NULL);
11308
0
            } else if (ctxt->instate == XML_PARSER_START) {
11309
0
                xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
11310
0
            } else {
11311
0
                xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
11312
0
                               "Start tag expected, '<' not found\n");
11313
0
            }
11314
0
        } else {
11315
0
            xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
11316
0
        }
11317
0
  if (ctxt->instate != XML_PARSER_EOF) {
11318
0
            ctxt->instate = XML_PARSER_EOF;
11319
0
            xmlFinishDocument(ctxt);
11320
0
  }
11321
0
    }
11322
0
    if (ctxt->wellFormed == 0)
11323
0
  return((xmlParserErrors) ctxt->errNo);
11324
0
    else
11325
0
        return(0);
11326
0
}
11327
11328
/************************************************************************
11329
 *                  *
11330
 *    I/O front end functions to the parser     *
11331
 *                  *
11332
 ************************************************************************/
11333
11334
/**
11335
 * Create a parser context for using the XML parser in push mode.
11336
 * See #xmlParseChunk.
11337
 *
11338
 * Passing an initial chunk is useless and deprecated.
11339
 *
11340
 * The push parser doesn't support recovery mode or the
11341
 * XML_PARSE_NOBLANKS option.
11342
 *
11343
 * `filename` is used as base URI to fetch external entities and for
11344
 * error reports.
11345
 *
11346
 * @param sax  a SAX handler (optional)
11347
 * @param user_data  user data for SAX callbacks (optional)
11348
 * @param chunk  initial chunk (optional, deprecated)
11349
 * @param size  size of initial chunk in bytes
11350
 * @param filename  file name or URI (optional)
11351
 * @returns the new parser context or NULL if a memory allocation
11352
 * failed.
11353
 */
11354
11355
xmlParserCtxt *
11356
xmlCreatePushParserCtxt(xmlSAXHandler *sax, void *user_data,
11357
0
                        const char *chunk, int size, const char *filename) {
11358
0
    xmlParserCtxtPtr ctxt;
11359
0
    xmlParserInputPtr input;
11360
11361
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11362
0
    if (ctxt == NULL)
11363
0
  return(NULL);
11364
11365
0
    ctxt->options &= ~XML_PARSE_NODICT;
11366
0
    ctxt->dictNames = 1;
11367
11368
0
    input = xmlNewPushInput(filename, chunk, size);
11369
0
    if (input == NULL) {
11370
0
  xmlFreeParserCtxt(ctxt);
11371
0
  return(NULL);
11372
0
    }
11373
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11374
0
        xmlFreeInputStream(input);
11375
0
        xmlFreeParserCtxt(ctxt);
11376
0
        return(NULL);
11377
0
    }
11378
11379
0
    return(ctxt);
11380
0
}
11381
#endif /* LIBXML_PUSH_ENABLED */
11382
11383
/**
11384
 * Blocks further parser processing
11385
 *
11386
 * @param ctxt  an XML parser context
11387
 */
11388
void
11389
0
xmlStopParser(xmlParserCtxt *ctxt) {
11390
0
    if (ctxt == NULL)
11391
0
        return;
11392
11393
    /* This stops the parser */
11394
0
    ctxt->disableSAX = 2;
11395
11396
    /*
11397
     * xmlStopParser is often called from error handlers,
11398
     * so we can't raise an error here to avoid infinite
11399
     * loops. Just make sure that an error condition is
11400
     * reported.
11401
     */
11402
0
    if (ctxt->errNo == XML_ERR_OK) {
11403
0
        ctxt->errNo = XML_ERR_USER_STOP;
11404
0
        ctxt->lastError.code = XML_ERR_USER_STOP;
11405
0
        ctxt->wellFormed = 0;
11406
0
    }
11407
0
}
11408
11409
/**
11410
 * Create a parser context for using the XML parser with an existing
11411
 * I/O stream
11412
 *
11413
 * @param sax  a SAX handler (optional)
11414
 * @param user_data  user data for SAX callbacks (optional)
11415
 * @param ioread  an I/O read function
11416
 * @param ioclose  an I/O close function (optional)
11417
 * @param ioctx  an I/O handler
11418
 * @param enc  the charset encoding if known (deprecated)
11419
 * @returns the new parser context or NULL
11420
 */
11421
xmlParserCtxt *
11422
xmlCreateIOParserCtxt(xmlSAXHandler *sax, void *user_data,
11423
                      xmlInputReadCallback ioread,
11424
                      xmlInputCloseCallback ioclose,
11425
0
                      void *ioctx, xmlCharEncoding enc) {
11426
0
    xmlParserCtxtPtr ctxt;
11427
0
    xmlParserInputPtr input;
11428
0
    const char *encoding;
11429
11430
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11431
0
    if (ctxt == NULL)
11432
0
  return(NULL);
11433
11434
0
    encoding = xmlGetCharEncodingName(enc);
11435
0
    input = xmlCtxtNewInputFromIO(ctxt, NULL, ioread, ioclose, ioctx,
11436
0
                                  encoding, 0);
11437
0
    if (input == NULL) {
11438
0
  xmlFreeParserCtxt(ctxt);
11439
0
        return (NULL);
11440
0
    }
11441
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11442
0
        xmlFreeInputStream(input);
11443
0
        xmlFreeParserCtxt(ctxt);
11444
0
        return(NULL);
11445
0
    }
11446
11447
0
    return(ctxt);
11448
0
}
11449
11450
#ifdef LIBXML_VALID_ENABLED
11451
/************************************************************************
11452
 *                  *
11453
 *    Front ends when parsing a DTD       *
11454
 *                  *
11455
 ************************************************************************/
11456
11457
/**
11458
 * Parse a DTD.
11459
 *
11460
 * Option XML_PARSE_DTDLOAD should be enabled in the parser context
11461
 * to make external entities work.
11462
 *
11463
 * @since 2.14.0
11464
 *
11465
 * @param ctxt  a parser context
11466
 * @param input  a parser input
11467
 * @param publicId  public ID of the DTD (optional)
11468
 * @param systemId  system ID of the DTD (optional)
11469
 * @returns the resulting xmlDtd or NULL in case of error.
11470
 * `input` will be freed by the function in any case.
11471
 */
11472
xmlDtd *
11473
xmlCtxtParseDtd(xmlParserCtxt *ctxt, xmlParserInput *input,
11474
0
                const xmlChar *publicId, const xmlChar *systemId) {
11475
0
    xmlDtdPtr ret = NULL;
11476
11477
0
    if ((ctxt == NULL) || (input == NULL)) {
11478
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
11479
0
        xmlFreeInputStream(input);
11480
0
        return(NULL);
11481
0
    }
11482
11483
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11484
0
        xmlFreeInputStream(input);
11485
0
        return(NULL);
11486
0
    }
11487
11488
0
    if (publicId == NULL)
11489
0
        publicId = BAD_CAST "none";
11490
0
    if (systemId == NULL)
11491
0
        systemId = BAD_CAST "none";
11492
11493
0
    ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0");
11494
0
    if (ctxt->myDoc == NULL) {
11495
0
        xmlErrMemory(ctxt);
11496
0
        goto error;
11497
0
    }
11498
0
    ctxt->myDoc->properties = XML_DOC_INTERNAL;
11499
0
    ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none",
11500
0
                                       publicId, systemId);
11501
0
    if (ctxt->myDoc->extSubset == NULL) {
11502
0
        xmlErrMemory(ctxt);
11503
0
        xmlFreeDoc(ctxt->myDoc);
11504
0
        goto error;
11505
0
    }
11506
11507
0
    xmlParseExternalSubset(ctxt, publicId, systemId);
11508
11509
0
    if (ctxt->wellFormed) {
11510
0
        ret = ctxt->myDoc->extSubset;
11511
0
        ctxt->myDoc->extSubset = NULL;
11512
0
        if (ret != NULL) {
11513
0
            xmlNodePtr tmp;
11514
11515
0
            ret->doc = NULL;
11516
0
            tmp = ret->children;
11517
0
            while (tmp != NULL) {
11518
0
                tmp->doc = NULL;
11519
0
                tmp = tmp->next;
11520
0
            }
11521
0
        }
11522
0
    } else {
11523
0
        ret = NULL;
11524
0
    }
11525
0
    xmlFreeDoc(ctxt->myDoc);
11526
0
    ctxt->myDoc = NULL;
11527
11528
0
error:
11529
0
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
11530
11531
0
    return(ret);
11532
0
}
11533
11534
/**
11535
 * Load and parse a DTD
11536
 *
11537
 * @deprecated Use #xmlCtxtParseDtd.
11538
 *
11539
 * @param sax  the SAX handler block or NULL
11540
 * @param input  an Input Buffer
11541
 * @param enc  the charset encoding if known
11542
 * @returns the resulting xmlDtd or NULL in case of error.
11543
 * `input` will be freed by the function in any case.
11544
 */
11545
11546
xmlDtd *
11547
xmlIOParseDTD(xmlSAXHandler *sax, xmlParserInputBuffer *input,
11548
0
        xmlCharEncoding enc) {
11549
0
    xmlDtdPtr ret = NULL;
11550
0
    xmlParserCtxtPtr ctxt;
11551
0
    xmlParserInputPtr pinput = NULL;
11552
11553
0
    if (input == NULL)
11554
0
  return(NULL);
11555
11556
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
11557
0
    if (ctxt == NULL) {
11558
0
        xmlFreeParserInputBuffer(input);
11559
0
  return(NULL);
11560
0
    }
11561
0
    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
11562
11563
    /*
11564
     * generate a parser input from the I/O handler
11565
     */
11566
11567
0
    pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE);
11568
0
    if (pinput == NULL) {
11569
0
        xmlFreeParserInputBuffer(input);
11570
0
  xmlFreeParserCtxt(ctxt);
11571
0
  return(NULL);
11572
0
    }
11573
11574
0
    if (enc != XML_CHAR_ENCODING_NONE) {
11575
0
        xmlSwitchEncoding(ctxt, enc);
11576
0
    }
11577
11578
0
    ret = xmlCtxtParseDtd(ctxt, pinput, NULL, NULL);
11579
11580
0
    xmlFreeParserCtxt(ctxt);
11581
0
    return(ret);
11582
0
}
11583
11584
/**
11585
 * Load and parse an external subset.
11586
 *
11587
 * @deprecated Use #xmlCtxtParseDtd.
11588
 *
11589
 * @param sax  the SAX handler block
11590
 * @param publicId  public identifier of the DTD (optional)
11591
 * @param systemId  system identifier (URL) of the DTD
11592
 * @returns the resulting xmlDtd or NULL in case of error.
11593
 */
11594
11595
xmlDtd *
11596
xmlSAXParseDTD(xmlSAXHandler *sax, const xmlChar *publicId,
11597
0
               const xmlChar *systemId) {
11598
0
    xmlDtdPtr ret = NULL;
11599
0
    xmlParserCtxtPtr ctxt;
11600
0
    xmlParserInputPtr input = NULL;
11601
0
    xmlChar* systemIdCanonic;
11602
11603
0
    if ((publicId == NULL) && (systemId == NULL)) return(NULL);
11604
11605
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
11606
0
    if (ctxt == NULL) {
11607
0
  return(NULL);
11608
0
    }
11609
0
    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
11610
11611
    /*
11612
     * Canonicalise the system ID
11613
     */
11614
0
    systemIdCanonic = xmlCanonicPath(systemId);
11615
0
    if ((systemId != NULL) && (systemIdCanonic == NULL)) {
11616
0
  xmlFreeParserCtxt(ctxt);
11617
0
  return(NULL);
11618
0
    }
11619
11620
    /*
11621
     * Ask the Entity resolver to load the damn thing
11622
     */
11623
11624
0
    if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL))
11625
0
  input = ctxt->sax->resolveEntity(ctxt->userData, publicId,
11626
0
                                   systemIdCanonic);
11627
0
    if (input == NULL) {
11628
0
  xmlFreeParserCtxt(ctxt);
11629
0
  if (systemIdCanonic != NULL)
11630
0
      xmlFree(systemIdCanonic);
11631
0
  return(NULL);
11632
0
    }
11633
11634
0
    if (input->filename == NULL)
11635
0
  input->filename = (char *) systemIdCanonic;
11636
0
    else
11637
0
  xmlFree(systemIdCanonic);
11638
11639
0
    ret = xmlCtxtParseDtd(ctxt, input, publicId, systemId);
11640
11641
0
    xmlFreeParserCtxt(ctxt);
11642
0
    return(ret);
11643
0
}
11644
11645
11646
/**
11647
 * Load and parse an external subset.
11648
 *
11649
 * @param publicId  public identifier of the DTD (optional)
11650
 * @param systemId  system identifier (URL) of the DTD
11651
 * @returns the resulting xmlDtd or NULL in case of error.
11652
 */
11653
11654
xmlDtd *
11655
0
xmlParseDTD(const xmlChar *publicId, const xmlChar *systemId) {
11656
0
    return(xmlSAXParseDTD(NULL, publicId, systemId));
11657
0
}
11658
#endif /* LIBXML_VALID_ENABLED */
11659
11660
/************************************************************************
11661
 *                  *
11662
 *    Front ends when parsing an Entity     *
11663
 *                  *
11664
 ************************************************************************/
11665
11666
static xmlNodePtr
11667
xmlCtxtParseContentInternal(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
11668
4.64k
                            int hasTextDecl, int buildTree) {
11669
4.64k
    xmlNodePtr root = NULL;
11670
4.64k
    xmlNodePtr list = NULL;
11671
4.64k
    xmlChar *rootName = BAD_CAST "#root";
11672
4.64k
    int result;
11673
11674
4.64k
    if (buildTree) {
11675
4.64k
        root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL);
11676
4.64k
        if (root == NULL) {
11677
2
            xmlErrMemory(ctxt);
11678
2
            goto error;
11679
2
        }
11680
4.64k
    }
11681
11682
4.64k
    if (xmlCtxtPushInput(ctxt, input) < 0)
11683
2
        goto error;
11684
11685
4.64k
    nameNsPush(ctxt, rootName, NULL, NULL, 0, 0);
11686
4.64k
    spacePush(ctxt, -1);
11687
11688
4.64k
    if (buildTree)
11689
4.64k
        nodePush(ctxt, root);
11690
11691
4.64k
    if (hasTextDecl) {
11692
3.10k
        xmlDetectEncoding(ctxt);
11693
11694
        /*
11695
         * Parse a possible text declaration first
11696
         */
11697
3.10k
        if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
11698
353
            (IS_BLANK_CH(NXT(5)))) {
11699
347
            xmlParseTextDecl(ctxt);
11700
            /*
11701
             * An XML-1.0 document can't reference an entity not XML-1.0
11702
             */
11703
347
            if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
11704
252
                (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
11705
1
                xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
11706
1
                               "Version mismatch between document and "
11707
1
                               "entity\n");
11708
1
            }
11709
347
        }
11710
3.10k
    }
11711
11712
4.64k
    xmlParseContentInternal(ctxt);
11713
11714
4.64k
    if (ctxt->input->cur < ctxt->input->end)
11715
312
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
11716
11717
4.64k
    if ((ctxt->wellFormed) ||
11718
4.01k
        ((ctxt->recovery) && (!xmlCtxtIsCatastrophicError(ctxt)))) {
11719
4.01k
        if (root != NULL) {
11720
4.01k
            xmlNodePtr cur;
11721
11722
            /*
11723
             * Unlink newly created node list.
11724
             */
11725
4.01k
            list = root->children;
11726
4.01k
            root->children = NULL;
11727
4.01k
            root->last = NULL;
11728
11.2k
            for (cur = list; cur != NULL; cur = cur->next)
11729
7.24k
                cur->parent = NULL;
11730
4.01k
        }
11731
4.01k
    }
11732
11733
    /*
11734
     * Read the rest of the stream in case of errors. We want
11735
     * to account for the whole entity size.
11736
     */
11737
4.87k
    do {
11738
4.87k
        ctxt->input->cur = ctxt->input->end;
11739
4.87k
        xmlParserShrink(ctxt);
11740
4.87k
        result = xmlParserGrow(ctxt);
11741
4.87k
    } while (result > 0);
11742
11743
4.64k
    if (buildTree)
11744
4.64k
        nodePop(ctxt);
11745
11746
4.64k
    namePop(ctxt);
11747
4.64k
    spacePop(ctxt);
11748
11749
4.64k
    xmlCtxtPopInput(ctxt);
11750
11751
4.64k
error:
11752
4.64k
    xmlFreeNode(root);
11753
11754
4.64k
    return(list);
11755
4.64k
}
11756
11757
static void
11758
7.24k
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
11759
7.24k
    xmlParserInputPtr input;
11760
7.24k
    xmlNodePtr list;
11761
7.24k
    unsigned long consumed;
11762
7.24k
    int isExternal;
11763
7.24k
    int buildTree;
11764
7.24k
    int oldMinNsIndex;
11765
7.24k
    int oldNodelen, oldNodemem;
11766
11767
7.24k
    isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY);
11768
7.24k
    buildTree = (ctxt->node != NULL);
11769
11770
    /*
11771
     * Recursion check
11772
     */
11773
7.24k
    if (ent->flags & XML_ENT_EXPANDING) {
11774
8
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
11775
8
        goto error;
11776
8
    }
11777
11778
    /*
11779
     * Load entity
11780
     */
11781
7.23k
    input = xmlNewEntityInputStream(ctxt, ent);
11782
7.23k
    if (input == NULL)
11783
2.59k
        goto error;
11784
11785
    /*
11786
     * When building a tree, we need to limit the scope of namespace
11787
     * declarations, so that entities don't reference xmlNs structs
11788
     * from the parent of a reference.
11789
     */
11790
4.64k
    oldMinNsIndex = ctxt->nsdb->minNsIndex;
11791
4.64k
    if (buildTree)
11792
4.64k
        ctxt->nsdb->minNsIndex = ctxt->nsNr;
11793
11794
4.64k
    oldNodelen = ctxt->nodelen;
11795
4.64k
    oldNodemem = ctxt->nodemem;
11796
4.64k
    ctxt->nodelen = 0;
11797
4.64k
    ctxt->nodemem = 0;
11798
11799
    /*
11800
     * Parse content
11801
     *
11802
     * This initiates a recursive call chain:
11803
     *
11804
     * - xmlCtxtParseContentInternal
11805
     * - xmlParseContentInternal
11806
     * - xmlParseReference
11807
     * - xmlCtxtParseEntity
11808
     *
11809
     * The nesting depth is limited by the maximum number of inputs,
11810
     * see xmlCtxtPushInput.
11811
     *
11812
     * It's possible to make this non-recursive (minNsIndex must be
11813
     * stored in the input struct) at the expense of code readability.
11814
     */
11815
11816
4.64k
    ent->flags |= XML_ENT_EXPANDING;
11817
11818
4.64k
    list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree);
11819
11820
4.64k
    ent->flags &= ~XML_ENT_EXPANDING;
11821
11822
4.64k
    ctxt->nsdb->minNsIndex = oldMinNsIndex;
11823
4.64k
    ctxt->nodelen = oldNodelen;
11824
4.64k
    ctxt->nodemem = oldNodemem;
11825
11826
    /*
11827
     * Entity size accounting
11828
     */
11829
4.64k
    consumed = input->consumed;
11830
4.64k
    xmlSaturatedAddSizeT(&consumed, input->end - input->base);
11831
11832
4.64k
    if ((ent->flags & XML_ENT_CHECKED) == 0)
11833
3.69k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
11834
11835
4.64k
    if ((ent->flags & XML_ENT_PARSED) == 0) {
11836
3.69k
        if (isExternal)
11837
2.79k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
11838
11839
3.69k
        ent->children = list;
11840
11841
10.9k
        while (list != NULL) {
11842
7.23k
            list->parent = (xmlNodePtr) ent;
11843
11844
            /*
11845
             * Downstream code like the nginx xslt module can set
11846
             * ctxt->myDoc->extSubset to a separate DTD, so the entity
11847
             * might have a different or a NULL document.
11848
             */
11849
7.23k
            if (list->doc != ent->doc)
11850
0
                xmlSetTreeDoc(list, ent->doc);
11851
11852
7.23k
            if (list->next == NULL)
11853
2.94k
                ent->last = list;
11854
7.23k
            list = list->next;
11855
7.23k
        }
11856
3.69k
    } else {
11857
947
        xmlFreeNodeList(list);
11858
947
    }
11859
11860
4.64k
    xmlFreeInputStream(input);
11861
11862
7.24k
error:
11863
7.24k
    ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED;
11864
7.24k
}
11865
11866
/**
11867
 * Parse an external general entity within an existing parsing context
11868
 * An external general parsed entity is well-formed if it matches the
11869
 * production labeled extParsedEnt.
11870
 *
11871
 *     [78] extParsedEnt ::= TextDecl? content
11872
 *
11873
 * @param ctxt  the existing parsing context
11874
 * @param URL  the URL for the entity to load
11875
 * @param ID  the System ID for the entity to load
11876
 * @param listOut  the return value for the set of parsed nodes
11877
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11878
 *    the parser error code otherwise
11879
 */
11880
11881
int
11882
xmlParseCtxtExternalEntity(xmlParserCtxt *ctxt, const xmlChar *URL,
11883
0
                           const xmlChar *ID, xmlNode **listOut) {
11884
0
    xmlParserInputPtr input;
11885
0
    xmlNodePtr list;
11886
11887
0
    if (listOut != NULL)
11888
0
        *listOut = NULL;
11889
11890
0
    if (ctxt == NULL)
11891
0
        return(XML_ERR_ARGUMENT);
11892
11893
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
11894
0
                            XML_RESOURCE_GENERAL_ENTITY);
11895
0
    if (input == NULL)
11896
0
        return(ctxt->errNo);
11897
11898
0
    xmlCtxtInitializeLate(ctxt);
11899
11900
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 1, 1);
11901
0
    if (listOut != NULL)
11902
0
        *listOut = list;
11903
0
    else
11904
0
        xmlFreeNodeList(list);
11905
11906
0
    xmlFreeInputStream(input);
11907
0
    return(ctxt->errNo);
11908
0
}
11909
11910
#ifdef LIBXML_SAX1_ENABLED
11911
/**
11912
 * Parse an external general entity
11913
 * An external general parsed entity is well-formed if it matches the
11914
 * production labeled extParsedEnt.
11915
 *
11916
 * This function uses deprecated global variables to set parser options
11917
 * which default to XML_PARSE_NODICT.
11918
 *
11919
 * @deprecated Use #xmlParseCtxtExternalEntity.
11920
 *
11921
 *     [78] extParsedEnt ::= TextDecl? content
11922
 *
11923
 * @param doc  the document the chunk pertains to
11924
 * @param sax  the SAX handler block (possibly NULL)
11925
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
11926
 * @param depth  Used for loop detection, use 0
11927
 * @param URL  the URL for the entity to load
11928
 * @param ID  the System ID for the entity to load
11929
 * @param list  the return value for the set of parsed nodes
11930
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11931
 *    the parser error code otherwise
11932
 */
11933
11934
int
11935
xmlParseExternalEntity(xmlDoc *doc, xmlSAXHandler *sax, void *user_data,
11936
0
    int depth, const xmlChar *URL, const xmlChar *ID, xmlNode **list) {
11937
0
    xmlParserCtxtPtr ctxt;
11938
0
    int ret;
11939
11940
0
    if (list != NULL)
11941
0
        *list = NULL;
11942
11943
0
    if (doc == NULL)
11944
0
        return(XML_ERR_ARGUMENT);
11945
11946
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11947
0
    if (ctxt == NULL)
11948
0
        return(XML_ERR_NO_MEMORY);
11949
11950
0
    ctxt->depth = depth;
11951
0
    ctxt->myDoc = doc;
11952
0
    ret = xmlParseCtxtExternalEntity(ctxt, URL, ID, list);
11953
11954
0
    xmlFreeParserCtxt(ctxt);
11955
0
    return(ret);
11956
0
}
11957
11958
/**
11959
 * Parse a well-balanced chunk of an XML document
11960
 * called by the parser
11961
 * The allowed sequence for the Well Balanced Chunk is the one defined by
11962
 * the content production in the XML grammar:
11963
 *
11964
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
11965
 *                       Comment)*
11966
 *
11967
 * This function uses deprecated global variables to set parser options
11968
 * which default to XML_PARSE_NODICT.
11969
 *
11970
 * @param doc  the document the chunk pertains to (must not be NULL)
11971
 * @param sax  the SAX handler block (possibly NULL)
11972
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
11973
 * @param depth  Used for loop detection, use 0
11974
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
11975
 * @param lst  the return value for the set of parsed nodes
11976
 * @returns 0 if the chunk is well balanced, -1 in case of args problem and
11977
 *    the parser error code otherwise
11978
 */
11979
11980
int
11981
xmlParseBalancedChunkMemory(xmlDoc *doc, xmlSAXHandler *sax,
11982
0
     void *user_data, int depth, const xmlChar *string, xmlNode **lst) {
11983
0
    return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data,
11984
0
                                                depth, string, lst, 0 );
11985
0
}
11986
#endif /* LIBXML_SAX1_ENABLED */
11987
11988
/**
11989
 * Parse a well-balanced chunk of XML matching the 'content' production.
11990
 *
11991
 * Namespaces in scope of `node` and entities of `node`'s document are
11992
 * recognized. When validating, the DTD of `node`'s document is used.
11993
 *
11994
 * Always consumes `input` even in error case.
11995
 *
11996
 * @since 2.14.0
11997
 *
11998
 * @param ctxt  parser context
11999
 * @param input  parser input
12000
 * @param node  target node or document
12001
 * @param hasTextDecl  whether to parse text declaration
12002
 * @returns a node list or NULL in case of error.
12003
 */
12004
xmlNode *
12005
xmlCtxtParseContent(xmlParserCtxt *ctxt, xmlParserInput *input,
12006
0
                    xmlNode *node, int hasTextDecl) {
12007
0
    xmlDocPtr doc;
12008
0
    xmlNodePtr cur, list = NULL;
12009
0
    int nsnr = 0;
12010
0
    xmlDictPtr oldDict;
12011
0
    int oldOptions, oldDictNames, oldLoadSubset;
12012
12013
0
    if ((ctxt == NULL) || (input == NULL) || (node == NULL)) {
12014
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12015
0
        goto exit;
12016
0
    }
12017
12018
0
    doc = node->doc;
12019
0
    if (doc == NULL) {
12020
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12021
0
        goto exit;
12022
0
    }
12023
12024
0
    switch (node->type) {
12025
0
        case XML_ELEMENT_NODE:
12026
0
        case XML_DOCUMENT_NODE:
12027
0
        case XML_HTML_DOCUMENT_NODE:
12028
0
            break;
12029
12030
0
        case XML_ATTRIBUTE_NODE:
12031
0
        case XML_TEXT_NODE:
12032
0
        case XML_CDATA_SECTION_NODE:
12033
0
        case XML_ENTITY_REF_NODE:
12034
0
        case XML_PI_NODE:
12035
0
        case XML_COMMENT_NODE:
12036
0
            for (cur = node->parent; cur != NULL; cur = node->parent) {
12037
0
                if ((cur->type == XML_ELEMENT_NODE) ||
12038
0
                    (cur->type == XML_DOCUMENT_NODE) ||
12039
0
                    (cur->type == XML_HTML_DOCUMENT_NODE)) {
12040
0
                    node = cur;
12041
0
                    break;
12042
0
                }
12043
0
            }
12044
0
            break;
12045
12046
0
        default:
12047
0
            xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12048
0
            goto exit;
12049
0
    }
12050
12051
0
    xmlCtxtReset(ctxt);
12052
12053
0
    oldDict = ctxt->dict;
12054
0
    oldOptions = ctxt->options;
12055
0
    oldDictNames = ctxt->dictNames;
12056
0
    oldLoadSubset = ctxt->loadsubset;
12057
12058
    /*
12059
     * Use input doc's dict if present, else assure XML_PARSE_NODICT is set.
12060
     */
12061
0
    if (doc->dict != NULL) {
12062
0
        ctxt->dict = doc->dict;
12063
0
    } else {
12064
0
        ctxt->options |= XML_PARSE_NODICT;
12065
0
        ctxt->dictNames = 0;
12066
0
    }
12067
12068
    /*
12069
     * Disable IDs
12070
     */
12071
0
    ctxt->loadsubset |= XML_SKIP_IDS;
12072
0
    ctxt->options |= XML_PARSE_SKIP_IDS;
12073
12074
0
    ctxt->myDoc = doc;
12075
12076
0
#ifdef LIBXML_HTML_ENABLED
12077
0
    if (ctxt->html) {
12078
        /*
12079
         * When parsing in context, it makes no sense to add implied
12080
         * elements like html/body/etc...
12081
         */
12082
0
        ctxt->options |= HTML_PARSE_NOIMPLIED;
12083
12084
0
        list = htmlCtxtParseContentInternal(ctxt, input);
12085
0
    } else
12086
0
#endif
12087
0
    {
12088
0
        xmlCtxtInitializeLate(ctxt);
12089
12090
        /*
12091
         * initialize the SAX2 namespaces stack
12092
         */
12093
0
        cur = node;
12094
0
        while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) {
12095
0
            xmlNsPtr ns = cur->nsDef;
12096
0
            xmlHashedString hprefix, huri;
12097
12098
0
            while (ns != NULL) {
12099
0
                hprefix = xmlDictLookupHashed(ctxt->dict, ns->prefix, -1);
12100
0
                huri = xmlDictLookupHashed(ctxt->dict, ns->href, -1);
12101
0
                if (xmlParserNsPush(ctxt, &hprefix, &huri, ns, 1) > 0)
12102
0
                    nsnr++;
12103
0
                ns = ns->next;
12104
0
            }
12105
0
            cur = cur->parent;
12106
0
        }
12107
12108
0
        list = xmlCtxtParseContentInternal(ctxt, input, hasTextDecl, 1);
12109
12110
0
        if (nsnr > 0)
12111
0
            xmlParserNsPop(ctxt, nsnr);
12112
0
    }
12113
12114
0
    ctxt->dict = oldDict;
12115
0
    ctxt->options = oldOptions;
12116
0
    ctxt->dictNames = oldDictNames;
12117
0
    ctxt->loadsubset = oldLoadSubset;
12118
0
    ctxt->myDoc = NULL;
12119
0
    ctxt->node = NULL;
12120
12121
0
exit:
12122
0
    xmlFreeInputStream(input);
12123
0
    return(list);
12124
0
}
12125
12126
/**
12127
 * Parse a well-balanced chunk of an XML document
12128
 * within the context (DTD, namespaces, etc ...) of the given node.
12129
 *
12130
 * The allowed sequence for the data is a Well Balanced Chunk defined by
12131
 * the content production in the XML grammar:
12132
 *
12133
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12134
 *                       Comment)*
12135
 *
12136
 * This function assumes the encoding of `node`'s document which is
12137
 * typically not what you want. A better alternative is
12138
 * #xmlCtxtParseContent.
12139
 *
12140
 * @param node  the context node
12141
 * @param data  the input string
12142
 * @param datalen  the input string length in bytes
12143
 * @param options  a combination of xmlParserOption
12144
 * @param listOut  the return value for the set of parsed nodes
12145
 * @returns XML_ERR_OK if the chunk is well balanced, and the parser
12146
 * error code otherwise
12147
 */
12148
xmlParserErrors
12149
xmlParseInNodeContext(xmlNode *node, const char *data, int datalen,
12150
0
                      int options, xmlNode **listOut) {
12151
0
    xmlParserCtxtPtr ctxt;
12152
0
    xmlParserInputPtr input;
12153
0
    xmlDocPtr doc;
12154
0
    xmlNodePtr list;
12155
0
    xmlParserErrors ret;
12156
12157
0
    if (listOut == NULL)
12158
0
        return(XML_ERR_INTERNAL_ERROR);
12159
0
    *listOut = NULL;
12160
12161
0
    if ((node == NULL) || (data == NULL) || (datalen < 0))
12162
0
        return(XML_ERR_INTERNAL_ERROR);
12163
12164
0
    doc = node->doc;
12165
0
    if (doc == NULL)
12166
0
        return(XML_ERR_INTERNAL_ERROR);
12167
12168
0
#ifdef LIBXML_HTML_ENABLED
12169
0
    if (doc->type == XML_HTML_DOCUMENT_NODE) {
12170
0
        ctxt = htmlNewParserCtxt();
12171
0
    }
12172
0
    else
12173
0
#endif
12174
0
        ctxt = xmlNewParserCtxt();
12175
12176
0
    if (ctxt == NULL)
12177
0
        return(XML_ERR_NO_MEMORY);
12178
12179
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, data, datalen,
12180
0
                                      (const char *) doc->encoding,
12181
0
                                      XML_INPUT_BUF_STATIC);
12182
0
    if (input == NULL) {
12183
0
        xmlFreeParserCtxt(ctxt);
12184
0
        return(XML_ERR_NO_MEMORY);
12185
0
    }
12186
12187
0
    xmlCtxtUseOptions(ctxt, options);
12188
12189
0
    list = xmlCtxtParseContent(ctxt, input, node, /* hasTextDecl */ 0);
12190
12191
0
    if (list == NULL) {
12192
0
        ret = ctxt->errNo;
12193
0
        if (ret == XML_ERR_ARGUMENT)
12194
0
            ret = XML_ERR_INTERNAL_ERROR;
12195
0
    } else {
12196
0
        ret = XML_ERR_OK;
12197
0
        *listOut = list;
12198
0
    }
12199
12200
0
    xmlFreeParserCtxt(ctxt);
12201
12202
0
    return(ret);
12203
0
}
12204
12205
#ifdef LIBXML_SAX1_ENABLED
12206
/**
12207
 * Parse a well-balanced chunk of an XML document
12208
 *
12209
 * The allowed sequence for the Well Balanced Chunk is the one defined by
12210
 * the content production in the XML grammar:
12211
 *
12212
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12213
 *                       Comment)*
12214
 *
12215
 * In case recover is set to 1, the nodelist will not be empty even if
12216
 * the parsed chunk is not well balanced, assuming the parsing succeeded to
12217
 * some extent.
12218
 *
12219
 * This function uses deprecated global variables to set parser options
12220
 * which default to XML_PARSE_NODICT.
12221
 *
12222
 * @param doc  the document the chunk pertains to (must not be NULL)
12223
 * @param sax  the SAX handler block (possibly NULL)
12224
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
12225
 * @param depth  Used for loop detection, use 0
12226
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
12227
 * @param listOut  the return value for the set of parsed nodes
12228
 * @param recover  return nodes even if the data is broken (use 0)
12229
 * @returns 0 if the chunk is well balanced, or thehe parser error code
12230
 * otherwise.
12231
 */
12232
int
12233
xmlParseBalancedChunkMemoryRecover(xmlDoc *doc, xmlSAXHandler *sax,
12234
     void *user_data, int depth, const xmlChar *string, xmlNode **listOut,
12235
0
     int recover) {
12236
0
    xmlParserCtxtPtr ctxt;
12237
0
    xmlParserInputPtr input;
12238
0
    xmlNodePtr list;
12239
0
    int ret;
12240
12241
0
    if (listOut != NULL)
12242
0
        *listOut = NULL;
12243
12244
0
    if (string == NULL)
12245
0
        return(XML_ERR_ARGUMENT);
12246
12247
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
12248
0
    if (ctxt == NULL)
12249
0
        return(XML_ERR_NO_MEMORY);
12250
12251
0
    xmlCtxtInitializeLate(ctxt);
12252
12253
0
    ctxt->depth = depth;
12254
0
    ctxt->myDoc = doc;
12255
0
    if (recover) {
12256
0
        ctxt->options |= XML_PARSE_RECOVER;
12257
0
        ctxt->recovery = 1;
12258
0
    }
12259
12260
0
    input = xmlNewStringInputStream(ctxt, string);
12261
0
    if (input == NULL) {
12262
0
        ret = ctxt->errNo;
12263
0
        goto error;
12264
0
    }
12265
12266
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 0, 1);
12267
0
    if (listOut != NULL)
12268
0
        *listOut = list;
12269
0
    else
12270
0
        xmlFreeNodeList(list);
12271
12272
0
    if (!ctxt->wellFormed)
12273
0
        ret = ctxt->errNo;
12274
0
    else
12275
0
        ret = XML_ERR_OK;
12276
12277
0
error:
12278
0
    xmlFreeInputStream(input);
12279
0
    xmlFreeParserCtxt(ctxt);
12280
0
    return(ret);
12281
0
}
12282
12283
/**
12284
 * Parse an XML external entity out of context and build a tree.
12285
 * It use the given SAX function block to handle the parsing callback.
12286
 * If sax is NULL, fallback to the default DOM tree building routines.
12287
 *
12288
 * @deprecated Don't use.
12289
 *
12290
 *     [78] extParsedEnt ::= TextDecl? content
12291
 *
12292
 * This correspond to a "Well Balanced" chunk
12293
 *
12294
 * This function uses deprecated global variables to set parser options
12295
 * which default to XML_PARSE_NODICT.
12296
 *
12297
 * @param sax  the SAX handler block
12298
 * @param filename  the filename
12299
 * @returns the resulting document tree
12300
 */
12301
12302
xmlDoc *
12303
0
xmlSAXParseEntity(xmlSAXHandler *sax, const char *filename) {
12304
0
    xmlDocPtr ret;
12305
0
    xmlParserCtxtPtr ctxt;
12306
12307
0
    ctxt = xmlCreateFileParserCtxt(filename);
12308
0
    if (ctxt == NULL) {
12309
0
  return(NULL);
12310
0
    }
12311
0
    if (sax != NULL) {
12312
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12313
0
            *ctxt->sax = *sax;
12314
0
        } else {
12315
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12316
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12317
0
        }
12318
0
        ctxt->userData = NULL;
12319
0
    }
12320
12321
0
    xmlParseExtParsedEnt(ctxt);
12322
12323
0
    if (ctxt->wellFormed) {
12324
0
  ret = ctxt->myDoc;
12325
0
    } else {
12326
0
        ret = NULL;
12327
0
        xmlFreeDoc(ctxt->myDoc);
12328
0
    }
12329
12330
0
    xmlFreeParserCtxt(ctxt);
12331
12332
0
    return(ret);
12333
0
}
12334
12335
/**
12336
 * Parse an XML external entity out of context and build a tree.
12337
 *
12338
 *     [78] extParsedEnt ::= TextDecl? content
12339
 *
12340
 * This correspond to a "Well Balanced" chunk
12341
 *
12342
 * This function uses deprecated global variables to set parser options
12343
 * which default to XML_PARSE_NODICT.
12344
 *
12345
 * @deprecated Don't use.
12346
 *
12347
 * @param filename  the filename
12348
 * @returns the resulting document tree
12349
 */
12350
12351
xmlDoc *
12352
0
xmlParseEntity(const char *filename) {
12353
0
    return(xmlSAXParseEntity(NULL, filename));
12354
0
}
12355
#endif /* LIBXML_SAX1_ENABLED */
12356
12357
/**
12358
 * Create a parser context for an external entity
12359
 * Automatic support for ZLIB/Compress compressed document is provided
12360
 * by default if found at compile-time.
12361
 *
12362
 * @deprecated Don't use.
12363
 *
12364
 * @param URL  the entity URL
12365
 * @param ID  the entity PUBLIC ID
12366
 * @param base  a possible base for the target URI
12367
 * @returns the new parser context or NULL
12368
 */
12369
xmlParserCtxt *
12370
xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID,
12371
0
                    const xmlChar *base) {
12372
0
    xmlParserCtxtPtr ctxt;
12373
0
    xmlParserInputPtr input;
12374
0
    xmlChar *uri = NULL;
12375
12376
0
    ctxt = xmlNewParserCtxt();
12377
0
    if (ctxt == NULL)
12378
0
  return(NULL);
12379
12380
0
    if (base != NULL) {
12381
0
        if (xmlBuildURISafe(URL, base, &uri) < 0)
12382
0
            goto error;
12383
0
        if (uri != NULL)
12384
0
            URL = uri;
12385
0
    }
12386
12387
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
12388
0
                            XML_RESOURCE_UNKNOWN);
12389
0
    if (input == NULL)
12390
0
        goto error;
12391
12392
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12393
0
        xmlFreeInputStream(input);
12394
0
        goto error;
12395
0
    }
12396
12397
0
    xmlFree(uri);
12398
0
    return(ctxt);
12399
12400
0
error:
12401
0
    xmlFree(uri);
12402
0
    xmlFreeParserCtxt(ctxt);
12403
0
    return(NULL);
12404
0
}
12405
12406
/************************************************************************
12407
 *                  *
12408
 *    Front ends when parsing from a file     *
12409
 *                  *
12410
 ************************************************************************/
12411
12412
/**
12413
 * Create a parser context for a file or URL content.
12414
 * Automatic support for ZLIB/Compress compressed document is provided
12415
 * by default if found at compile-time and for file accesses
12416
 *
12417
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12418
 *
12419
 * @param filename  the filename or URL
12420
 * @param options  a combination of xmlParserOption
12421
 * @returns the new parser context or NULL
12422
 */
12423
xmlParserCtxt *
12424
xmlCreateURLParserCtxt(const char *filename, int options)
12425
0
{
12426
0
    xmlParserCtxtPtr ctxt;
12427
0
    xmlParserInputPtr input;
12428
12429
0
    ctxt = xmlNewParserCtxt();
12430
0
    if (ctxt == NULL)
12431
0
  return(NULL);
12432
12433
0
    xmlCtxtUseOptions(ctxt, options);
12434
12435
0
    input = xmlLoadResource(ctxt, filename, NULL, XML_RESOURCE_MAIN_DOCUMENT);
12436
0
    if (input == NULL) {
12437
0
  xmlFreeParserCtxt(ctxt);
12438
0
  return(NULL);
12439
0
    }
12440
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12441
0
        xmlFreeInputStream(input);
12442
0
        xmlFreeParserCtxt(ctxt);
12443
0
        return(NULL);
12444
0
    }
12445
12446
0
    return(ctxt);
12447
0
}
12448
12449
/**
12450
 * Create a parser context for a file content.
12451
 * Automatic support for ZLIB/Compress compressed document is provided
12452
 * by default if found at compile-time.
12453
 *
12454
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12455
 *
12456
 * @param filename  the filename
12457
 * @returns the new parser context or NULL
12458
 */
12459
xmlParserCtxt *
12460
xmlCreateFileParserCtxt(const char *filename)
12461
0
{
12462
0
    return(xmlCreateURLParserCtxt(filename, 0));
12463
0
}
12464
12465
#ifdef LIBXML_SAX1_ENABLED
12466
/**
12467
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12468
 * compressed document is provided by default if found at compile-time.
12469
 * It use the given SAX function block to handle the parsing callback.
12470
 * If sax is NULL, fallback to the default DOM tree building routines.
12471
 *
12472
 * This function uses deprecated global variables to set parser options
12473
 * which default to XML_PARSE_NODICT.
12474
 *
12475
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12476
 *
12477
 * User data (void *) is stored within the parser context in the
12478
 * context's _private member, so it is available nearly everywhere in libxml
12479
 *
12480
 * @param sax  the SAX handler block
12481
 * @param filename  the filename
12482
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12483
 *             documents
12484
 * @param data  the userdata
12485
 * @returns the resulting document tree
12486
 */
12487
12488
xmlDoc *
12489
xmlSAXParseFileWithData(xmlSAXHandler *sax, const char *filename,
12490
0
                        int recovery, void *data) {
12491
0
    xmlDocPtr ret = NULL;
12492
0
    xmlParserCtxtPtr ctxt;
12493
0
    xmlParserInputPtr input;
12494
12495
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12496
0
    if (ctxt == NULL)
12497
0
  return(NULL);
12498
12499
0
    if (data != NULL)
12500
0
  ctxt->_private = data;
12501
12502
0
    if (recovery) {
12503
0
        ctxt->options |= XML_PARSE_RECOVER;
12504
0
        ctxt->recovery = 1;
12505
0
    }
12506
12507
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
12508
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, NULL, 0);
12509
0
    else
12510
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, NULL, 0);
12511
12512
0
    if (input != NULL)
12513
0
        ret = xmlCtxtParseDocument(ctxt, input);
12514
12515
0
    xmlFreeParserCtxt(ctxt);
12516
0
    return(ret);
12517
0
}
12518
12519
/**
12520
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12521
 * compressed document is provided by default if found at compile-time.
12522
 * It use the given SAX function block to handle the parsing callback.
12523
 * If sax is NULL, fallback to the default DOM tree building routines.
12524
 *
12525
 * This function uses deprecated global variables to set parser options
12526
 * which default to XML_PARSE_NODICT.
12527
 *
12528
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12529
 *
12530
 * @param sax  the SAX handler block
12531
 * @param filename  the filename
12532
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12533
 *             documents
12534
 * @returns the resulting document tree
12535
 */
12536
12537
xmlDoc *
12538
xmlSAXParseFile(xmlSAXHandler *sax, const char *filename,
12539
0
                          int recovery) {
12540
0
    return(xmlSAXParseFileWithData(sax,filename,recovery,NULL));
12541
0
}
12542
12543
/**
12544
 * Parse an XML in-memory document and build a tree.
12545
 * In the case the document is not Well Formed, a attempt to build a
12546
 * tree is tried anyway
12547
 *
12548
 * This function uses deprecated global variables to set parser options
12549
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12550
 *
12551
 * @deprecated Use #xmlReadDoc with XML_PARSE_RECOVER.
12552
 *
12553
 * @param cur  a pointer to an array of xmlChar
12554
 * @returns the resulting document tree or NULL in case of failure
12555
 */
12556
12557
xmlDoc *
12558
0
xmlRecoverDoc(const xmlChar *cur) {
12559
0
    return(xmlSAXParseDoc(NULL, cur, 1));
12560
0
}
12561
12562
/**
12563
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12564
 * compressed document is provided by default if found at compile-time.
12565
 *
12566
 * This function uses deprecated global variables to set parser options
12567
 * which default to XML_PARSE_NODICT.
12568
 *
12569
 * @deprecated Use #xmlReadFile.
12570
 *
12571
 * @param filename  the filename
12572
 * @returns the resulting document tree if the file was wellformed,
12573
 * NULL otherwise.
12574
 */
12575
12576
xmlDoc *
12577
0
xmlParseFile(const char *filename) {
12578
0
    return(xmlSAXParseFile(NULL, filename, 0));
12579
0
}
12580
12581
/**
12582
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12583
 * compressed document is provided by default if found at compile-time.
12584
 * In the case the document is not Well Formed, it attempts to build
12585
 * a tree anyway
12586
 *
12587
 * This function uses deprecated global variables to set parser options
12588
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12589
 *
12590
 * @deprecated Use #xmlReadFile with XML_PARSE_RECOVER.
12591
 *
12592
 * @param filename  the filename
12593
 * @returns the resulting document tree or NULL in case of failure
12594
 */
12595
12596
xmlDoc *
12597
0
xmlRecoverFile(const char *filename) {
12598
0
    return(xmlSAXParseFile(NULL, filename, 1));
12599
0
}
12600
12601
12602
/**
12603
 * Setup the parser context to parse a new buffer; Clears any prior
12604
 * contents from the parser context. The buffer parameter must not be
12605
 * NULL, but the filename parameter can be
12606
 *
12607
 * @deprecated Don't use.
12608
 *
12609
 * @param ctxt  an XML parser context
12610
 * @param buffer  a xmlChar * buffer
12611
 * @param filename  a file name
12612
 */
12613
void
12614
xmlSetupParserForBuffer(xmlParserCtxt *ctxt, const xmlChar* buffer,
12615
                             const char* filename)
12616
0
{
12617
0
    xmlParserInputPtr input;
12618
12619
0
    if ((ctxt == NULL) || (buffer == NULL))
12620
0
        return;
12621
12622
0
    xmlCtxtReset(ctxt);
12623
12624
0
    input = xmlCtxtNewInputFromString(ctxt, filename, (const char *) buffer,
12625
0
                                      NULL, 0);
12626
0
    if (input == NULL)
12627
0
        return;
12628
0
    if (xmlCtxtPushInput(ctxt, input) < 0)
12629
0
        xmlFreeInputStream(input);
12630
0
}
12631
12632
/**
12633
 * Parse an XML file and call the given SAX handler routines.
12634
 * Automatic support for ZLIB/Compress compressed document is provided
12635
 *
12636
 * This function uses deprecated global variables to set parser options
12637
 * which default to XML_PARSE_NODICT.
12638
 *
12639
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12640
 *
12641
 * @param sax  a SAX handler
12642
 * @param user_data  The user data returned on SAX callbacks
12643
 * @param filename  a file name
12644
 * @returns 0 in case of success or a error number otherwise
12645
 */
12646
int
12647
xmlSAXUserParseFile(xmlSAXHandler *sax, void *user_data,
12648
0
                    const char *filename) {
12649
0
    int ret = 0;
12650
0
    xmlParserCtxtPtr ctxt;
12651
12652
0
    ctxt = xmlCreateFileParserCtxt(filename);
12653
0
    if (ctxt == NULL) return -1;
12654
0
    if (sax != NULL) {
12655
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12656
0
            *ctxt->sax = *sax;
12657
0
        } else {
12658
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12659
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12660
0
        }
12661
0
  ctxt->userData = user_data;
12662
0
    }
12663
12664
0
    xmlParseDocument(ctxt);
12665
12666
0
    if (ctxt->wellFormed)
12667
0
  ret = 0;
12668
0
    else {
12669
0
        if (ctxt->errNo != 0)
12670
0
      ret = ctxt->errNo;
12671
0
  else
12672
0
      ret = -1;
12673
0
    }
12674
0
    if (ctxt->myDoc != NULL) {
12675
0
        xmlFreeDoc(ctxt->myDoc);
12676
0
  ctxt->myDoc = NULL;
12677
0
    }
12678
0
    xmlFreeParserCtxt(ctxt);
12679
12680
0
    return ret;
12681
0
}
12682
#endif /* LIBXML_SAX1_ENABLED */
12683
12684
/************************************************************************
12685
 *                  *
12686
 *    Front ends when parsing from memory     *
12687
 *                  *
12688
 ************************************************************************/
12689
12690
/**
12691
 * Create a parser context for an XML in-memory document. The input buffer
12692
 * must not contain a terminating null byte.
12693
 *
12694
 * @param buffer  a pointer to a char array
12695
 * @param size  the size of the array
12696
 * @returns the new parser context or NULL
12697
 */
12698
xmlParserCtxt *
12699
0
xmlCreateMemoryParserCtxt(const char *buffer, int size) {
12700
0
    xmlParserCtxtPtr ctxt;
12701
0
    xmlParserInputPtr input;
12702
12703
0
    if (size < 0)
12704
0
  return(NULL);
12705
12706
0
    ctxt = xmlNewParserCtxt();
12707
0
    if (ctxt == NULL)
12708
0
  return(NULL);
12709
12710
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, 0);
12711
0
    if (input == NULL) {
12712
0
  xmlFreeParserCtxt(ctxt);
12713
0
  return(NULL);
12714
0
    }
12715
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12716
0
        xmlFreeInputStream(input);
12717
0
        xmlFreeParserCtxt(ctxt);
12718
0
        return(NULL);
12719
0
    }
12720
12721
0
    return(ctxt);
12722
0
}
12723
12724
#ifdef LIBXML_SAX1_ENABLED
12725
/**
12726
 * Parse an XML in-memory block and use the given SAX function block
12727
 * to handle the parsing callback. If sax is NULL, fallback to the default
12728
 * DOM tree building routines.
12729
 *
12730
 * This function uses deprecated global variables to set parser options
12731
 * which default to XML_PARSE_NODICT.
12732
 *
12733
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12734
 *
12735
 * User data (void *) is stored within the parser context in the
12736
 * context's _private member, so it is available nearly everywhere in libxml
12737
 *
12738
 * @param sax  the SAX handler block
12739
 * @param buffer  an pointer to a char array
12740
 * @param size  the size of the array
12741
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12742
 *             documents
12743
 * @param data  the userdata
12744
 * @returns the resulting document tree
12745
 */
12746
12747
xmlDoc *
12748
xmlSAXParseMemoryWithData(xmlSAXHandler *sax, const char *buffer,
12749
0
                          int size, int recovery, void *data) {
12750
0
    xmlDocPtr ret = NULL;
12751
0
    xmlParserCtxtPtr ctxt;
12752
0
    xmlParserInputPtr input;
12753
12754
0
    if (size < 0)
12755
0
        return(NULL);
12756
12757
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12758
0
    if (ctxt == NULL)
12759
0
        return(NULL);
12760
12761
0
    if (data != NULL)
12762
0
  ctxt->_private=data;
12763
12764
0
    if (recovery) {
12765
0
        ctxt->options |= XML_PARSE_RECOVER;
12766
0
        ctxt->recovery = 1;
12767
0
    }
12768
12769
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL,
12770
0
                                      XML_INPUT_BUF_STATIC);
12771
12772
0
    if (input != NULL)
12773
0
        ret = xmlCtxtParseDocument(ctxt, input);
12774
12775
0
    xmlFreeParserCtxt(ctxt);
12776
0
    return(ret);
12777
0
}
12778
12779
/**
12780
 * Parse an XML in-memory block and use the given SAX function block
12781
 * to handle the parsing callback. If sax is NULL, fallback to the default
12782
 * DOM tree building routines.
12783
 *
12784
 * This function uses deprecated global variables to set parser options
12785
 * which default to XML_PARSE_NODICT.
12786
 *
12787
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12788
 *
12789
 * @param sax  the SAX handler block
12790
 * @param buffer  an pointer to a char array
12791
 * @param size  the size of the array
12792
 * @param recovery  work in recovery mode, i.e. tries to read not Well Formed
12793
 *             documents
12794
 * @returns the resulting document tree
12795
 */
12796
xmlDoc *
12797
xmlSAXParseMemory(xmlSAXHandler *sax, const char *buffer,
12798
0
            int size, int recovery) {
12799
0
    return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL);
12800
0
}
12801
12802
/**
12803
 * Parse an XML in-memory block and build a tree.
12804
 *
12805
 * This function uses deprecated global variables to set parser options
12806
 * which default to XML_PARSE_NODICT.
12807
 *
12808
 * @deprecated Use #xmlReadMemory.
12809
 *
12810
 * @param buffer  an pointer to a char array
12811
 * @param size  the size of the array
12812
 * @returns the resulting document tree
12813
 */
12814
12815
0
xmlDoc *xmlParseMemory(const char *buffer, int size) {
12816
0
   return(xmlSAXParseMemory(NULL, buffer, size, 0));
12817
0
}
12818
12819
/**
12820
 * Parse an XML in-memory block and build a tree.
12821
 * In the case the document is not Well Formed, an attempt to
12822
 * build a tree is tried anyway
12823
 *
12824
 * This function uses deprecated global variables to set parser options
12825
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12826
 *
12827
 * @deprecated Use #xmlReadMemory with XML_PARSE_RECOVER.
12828
 *
12829
 * @param buffer  an pointer to a char array
12830
 * @param size  the size of the array
12831
 * @returns the resulting document tree or NULL in case of error
12832
 */
12833
12834
0
xmlDoc *xmlRecoverMemory(const char *buffer, int size) {
12835
0
   return(xmlSAXParseMemory(NULL, buffer, size, 1));
12836
0
}
12837
12838
/**
12839
 * Parse an XML in-memory buffer and call the given SAX handler routines.
12840
 *
12841
 * This function uses deprecated global variables to set parser options
12842
 * which default to XML_PARSE_NODICT.
12843
 *
12844
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12845
 *
12846
 * @param sax  a SAX handler
12847
 * @param user_data  The user data returned on SAX callbacks
12848
 * @param buffer  an in-memory XML document input
12849
 * @param size  the length of the XML document in bytes
12850
 * @returns 0 in case of success or a error number otherwise
12851
 */
12852
int xmlSAXUserParseMemory(xmlSAXHandler *sax, void *user_data,
12853
0
        const char *buffer, int size) {
12854
0
    int ret = 0;
12855
0
    xmlParserCtxtPtr ctxt;
12856
12857
0
    ctxt = xmlCreateMemoryParserCtxt(buffer, size);
12858
0
    if (ctxt == NULL) return -1;
12859
0
    if (sax != NULL) {
12860
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12861
0
            *ctxt->sax = *sax;
12862
0
        } else {
12863
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12864
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12865
0
        }
12866
0
  ctxt->userData = user_data;
12867
0
    }
12868
12869
0
    xmlParseDocument(ctxt);
12870
12871
0
    if (ctxt->wellFormed)
12872
0
  ret = 0;
12873
0
    else {
12874
0
        if (ctxt->errNo != 0)
12875
0
      ret = ctxt->errNo;
12876
0
  else
12877
0
      ret = -1;
12878
0
    }
12879
0
    if (ctxt->myDoc != NULL) {
12880
0
        xmlFreeDoc(ctxt->myDoc);
12881
0
  ctxt->myDoc = NULL;
12882
0
    }
12883
0
    xmlFreeParserCtxt(ctxt);
12884
12885
0
    return ret;
12886
0
}
12887
#endif /* LIBXML_SAX1_ENABLED */
12888
12889
/**
12890
 * Creates a parser context for an XML in-memory document.
12891
 *
12892
 * @param str  a pointer to an array of xmlChar
12893
 * @returns the new parser context or NULL
12894
 */
12895
xmlParserCtxt *
12896
0
xmlCreateDocParserCtxt(const xmlChar *str) {
12897
0
    xmlParserCtxtPtr ctxt;
12898
0
    xmlParserInputPtr input;
12899
12900
0
    ctxt = xmlNewParserCtxt();
12901
0
    if (ctxt == NULL)
12902
0
  return(NULL);
12903
12904
0
    input = xmlCtxtNewInputFromString(ctxt, NULL, (const char *) str, NULL, 0);
12905
0
    if (input == NULL) {
12906
0
  xmlFreeParserCtxt(ctxt);
12907
0
  return(NULL);
12908
0
    }
12909
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12910
0
        xmlFreeInputStream(input);
12911
0
        xmlFreeParserCtxt(ctxt);
12912
0
        return(NULL);
12913
0
    }
12914
12915
0
    return(ctxt);
12916
0
}
12917
12918
#ifdef LIBXML_SAX1_ENABLED
12919
/**
12920
 * Parse an XML in-memory document and build a tree.
12921
 * It use the given SAX function block to handle the parsing callback.
12922
 * If sax is NULL, fallback to the default DOM tree building routines.
12923
 *
12924
 * This function uses deprecated global variables to set parser options
12925
 * which default to XML_PARSE_NODICT.
12926
 *
12927
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadDoc.
12928
 *
12929
 * @param sax  the SAX handler block
12930
 * @param cur  a pointer to an array of xmlChar
12931
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12932
 *             documents
12933
 * @returns the resulting document tree
12934
 */
12935
12936
xmlDoc *
12937
0
xmlSAXParseDoc(xmlSAXHandler *sax, const xmlChar *cur, int recovery) {
12938
0
    xmlDocPtr ret;
12939
0
    xmlParserCtxtPtr ctxt;
12940
0
    xmlSAXHandlerPtr oldsax = NULL;
12941
12942
0
    if (cur == NULL) return(NULL);
12943
12944
12945
0
    ctxt = xmlCreateDocParserCtxt(cur);
12946
0
    if (ctxt == NULL) return(NULL);
12947
0
    if (sax != NULL) {
12948
0
        oldsax = ctxt->sax;
12949
0
        ctxt->sax = sax;
12950
0
        ctxt->userData = NULL;
12951
0
    }
12952
12953
0
    xmlParseDocument(ctxt);
12954
0
    if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc;
12955
0
    else {
12956
0
       ret = NULL;
12957
0
       xmlFreeDoc(ctxt->myDoc);
12958
0
       ctxt->myDoc = NULL;
12959
0
    }
12960
0
    if (sax != NULL)
12961
0
  ctxt->sax = oldsax;
12962
0
    xmlFreeParserCtxt(ctxt);
12963
12964
0
    return(ret);
12965
0
}
12966
12967
/**
12968
 * Parse an XML in-memory document and build a tree.
12969
 *
12970
 * This function uses deprecated global variables to set parser options
12971
 * which default to XML_PARSE_NODICT.
12972
 *
12973
 * @deprecated Use #xmlReadDoc.
12974
 *
12975
 * @param cur  a pointer to an array of xmlChar
12976
 * @returns the resulting document tree
12977
 */
12978
12979
xmlDoc *
12980
0
xmlParseDoc(const xmlChar *cur) {
12981
0
    return(xmlSAXParseDoc(NULL, cur, 0));
12982
0
}
12983
#endif /* LIBXML_SAX1_ENABLED */
12984
12985
/************************************************************************
12986
 *                  *
12987
 *  New set (2.6.0) of simpler and more flexible APIs   *
12988
 *                  *
12989
 ************************************************************************/
12990
12991
/**
12992
 * Reset a parser context
12993
 *
12994
 * @param ctxt  an XML parser context
12995
 */
12996
void
12997
xmlCtxtReset(xmlParserCtxt *ctxt)
12998
31.4k
{
12999
31.4k
    xmlParserInputPtr input;
13000
13001
31.4k
    if (ctxt == NULL)
13002
0
        return;
13003
13004
31.4k
    while ((input = xmlCtxtPopInput(ctxt)) != NULL) { /* Non consuming */
13005
0
        xmlFreeInputStream(input);
13006
0
    }
13007
31.4k
    ctxt->inputNr = 0;
13008
31.4k
    ctxt->input = NULL;
13009
13010
31.4k
    ctxt->spaceNr = 0;
13011
31.4k
    if (ctxt->spaceTab != NULL) {
13012
31.4k
  ctxt->spaceTab[0] = -1;
13013
31.4k
  ctxt->space = &ctxt->spaceTab[0];
13014
31.4k
    } else {
13015
0
        ctxt->space = NULL;
13016
0
    }
13017
13018
13019
31.4k
    ctxt->nodeNr = 0;
13020
31.4k
    ctxt->node = NULL;
13021
13022
31.4k
    ctxt->nameNr = 0;
13023
31.4k
    ctxt->name = NULL;
13024
13025
31.4k
    ctxt->nsNr = 0;
13026
31.4k
    xmlParserNsReset(ctxt->nsdb);
13027
13028
31.4k
    if (ctxt->version != NULL) {
13029
0
        xmlFree(ctxt->version);
13030
0
        ctxt->version = NULL;
13031
0
    }
13032
31.4k
    if (ctxt->encoding != NULL) {
13033
0
        xmlFree(ctxt->encoding);
13034
0
        ctxt->encoding = NULL;
13035
0
    }
13036
31.4k
    if (ctxt->extSubURI != NULL) {
13037
0
        xmlFree(ctxt->extSubURI);
13038
0
        ctxt->extSubURI = NULL;
13039
0
    }
13040
31.4k
    if (ctxt->extSubSystem != NULL) {
13041
0
        xmlFree(ctxt->extSubSystem);
13042
0
        ctxt->extSubSystem = NULL;
13043
0
    }
13044
31.4k
    if (ctxt->directory != NULL) {
13045
0
        xmlFree(ctxt->directory);
13046
0
        ctxt->directory = NULL;
13047
0
    }
13048
13049
31.4k
    if (ctxt->myDoc != NULL)
13050
0
        xmlFreeDoc(ctxt->myDoc);
13051
31.4k
    ctxt->myDoc = NULL;
13052
13053
31.4k
    ctxt->standalone = -1;
13054
31.4k
    ctxt->hasExternalSubset = 0;
13055
31.4k
    ctxt->hasPErefs = 0;
13056
31.4k
    ctxt->html = ctxt->html ? 1 : 0;
13057
31.4k
    ctxt->instate = XML_PARSER_START;
13058
13059
31.4k
    ctxt->wellFormed = 1;
13060
31.4k
    ctxt->nsWellFormed = 1;
13061
31.4k
    ctxt->disableSAX = 0;
13062
31.4k
    ctxt->valid = 1;
13063
31.4k
    ctxt->record_info = 0;
13064
31.4k
    ctxt->checkIndex = 0;
13065
31.4k
    ctxt->endCheckState = 0;
13066
31.4k
    ctxt->inSubset = 0;
13067
31.4k
    ctxt->errNo = XML_ERR_OK;
13068
31.4k
    ctxt->depth = 0;
13069
31.4k
    ctxt->catalogs = NULL;
13070
31.4k
    ctxt->sizeentities = 0;
13071
31.4k
    ctxt->sizeentcopy = 0;
13072
31.4k
    xmlInitNodeInfoSeq(&ctxt->node_seq);
13073
13074
31.4k
    if (ctxt->attsDefault != NULL) {
13075
0
        xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator);
13076
0
        ctxt->attsDefault = NULL;
13077
0
    }
13078
31.4k
    if (ctxt->attsSpecial != NULL) {
13079
0
        xmlHashFree(ctxt->attsSpecial, NULL);
13080
0
        ctxt->attsSpecial = NULL;
13081
0
    }
13082
13083
31.4k
#ifdef LIBXML_CATALOG_ENABLED
13084
31.4k
    if (ctxt->catalogs != NULL)
13085
0
  xmlCatalogFreeLocal(ctxt->catalogs);
13086
31.4k
#endif
13087
31.4k
    ctxt->nbErrors = 0;
13088
31.4k
    ctxt->nbWarnings = 0;
13089
31.4k
    if (ctxt->lastError.code != XML_ERR_OK)
13090
0
        xmlResetError(&ctxt->lastError);
13091
31.4k
}
13092
13093
/**
13094
 * Reset a push parser context
13095
 *
13096
 * @param ctxt  an XML parser context
13097
 * @param chunk  a pointer to an array of chars
13098
 * @param size  number of chars in the array
13099
 * @param filename  an optional file name or URI
13100
 * @param encoding  the document encoding, or NULL
13101
 * @returns 0 in case of success and 1 in case of error
13102
 */
13103
int
13104
xmlCtxtResetPush(xmlParserCtxt *ctxt, const char *chunk,
13105
                 int size, const char *filename, const char *encoding)
13106
0
{
13107
0
    xmlParserInputPtr input;
13108
13109
0
    if (ctxt == NULL)
13110
0
        return(1);
13111
13112
0
    xmlCtxtReset(ctxt);
13113
13114
0
    input = xmlNewPushInput(filename, chunk, size);
13115
0
    if (input == NULL)
13116
0
        return(1);
13117
13118
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13119
0
        xmlFreeInputStream(input);
13120
0
        return(1);
13121
0
    }
13122
13123
0
    if (encoding != NULL)
13124
0
        xmlSwitchEncodingName(ctxt, encoding);
13125
13126
0
    return(0);
13127
0
}
13128
13129
static int
13130
xmlCtxtSetOptionsInternal(xmlParserCtxtPtr ctxt, int options, int keepMask)
13131
35.9k
{
13132
35.9k
    int allMask;
13133
13134
35.9k
    if (ctxt == NULL)
13135
0
        return(-1);
13136
13137
    /*
13138
     * XInclude options aren't handled by the parser.
13139
     *
13140
     * XML_PARSE_XINCLUDE
13141
     * XML_PARSE_NOXINCNODE
13142
     * XML_PARSE_NOBASEFIX
13143
     */
13144
35.9k
    allMask = XML_PARSE_RECOVER |
13145
35.9k
              XML_PARSE_NOENT |
13146
35.9k
              XML_PARSE_DTDLOAD |
13147
35.9k
              XML_PARSE_DTDATTR |
13148
35.9k
              XML_PARSE_DTDVALID |
13149
35.9k
              XML_PARSE_NOERROR |
13150
35.9k
              XML_PARSE_NOWARNING |
13151
35.9k
              XML_PARSE_PEDANTIC |
13152
35.9k
              XML_PARSE_NOBLANKS |
13153
35.9k
#ifdef LIBXML_SAX1_ENABLED
13154
35.9k
              XML_PARSE_SAX1 |
13155
35.9k
#endif
13156
35.9k
              XML_PARSE_NONET |
13157
35.9k
              XML_PARSE_NODICT |
13158
35.9k
              XML_PARSE_NSCLEAN |
13159
35.9k
              XML_PARSE_NOCDATA |
13160
35.9k
              XML_PARSE_COMPACT |
13161
35.9k
              XML_PARSE_OLD10 |
13162
35.9k
              XML_PARSE_HUGE |
13163
35.9k
              XML_PARSE_OLDSAX |
13164
35.9k
              XML_PARSE_IGNORE_ENC |
13165
35.9k
              XML_PARSE_BIG_LINES |
13166
35.9k
              XML_PARSE_NO_XXE |
13167
35.9k
              XML_PARSE_UNZIP |
13168
35.9k
              XML_PARSE_NO_SYS_CATALOG |
13169
35.9k
              XML_PARSE_CATALOG_PI;
13170
13171
35.9k
    ctxt->options = (ctxt->options & keepMask) | (options & allMask);
13172
13173
    /*
13174
     * For some options, struct members are historically the source
13175
     * of truth. The values are initalized from global variables and
13176
     * old code could also modify them directly. Several older API
13177
     * functions that don't take an options argument rely on these
13178
     * deprecated mechanisms.
13179
     *
13180
     * Once public access to struct members and the globals are
13181
     * disabled, we can use the options bitmask as source of
13182
     * truth, making all these struct members obsolete.
13183
     *
13184
     * The XML_DETECT_IDS flags is misnamed. It simply enables
13185
     * loading of the external subset.
13186
     */
13187
35.9k
    ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0;
13188
35.9k
    ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0;
13189
35.9k
    ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0;
13190
35.9k
    ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0;
13191
35.9k
    ctxt->loadsubset |= (options & XML_PARSE_SKIP_IDS) ? XML_SKIP_IDS : 0;
13192
35.9k
    ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0;
13193
35.9k
    ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0;
13194
35.9k
    ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1;
13195
35.9k
    ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1;
13196
13197
35.9k
    return(options & ~allMask);
13198
35.9k
}
13199
13200
/**
13201
 * Applies the options to the parser context. Unset options are
13202
 * cleared.
13203
 *
13204
 * @since 2.13.0
13205
 *
13206
 * With older versions, you can use #xmlCtxtUseOptions.
13207
 *
13208
 * @param ctxt  an XML parser context
13209
 * @param options  a bitmask of xmlParserOption values
13210
 * @returns 0 in case of success, the set of unknown or unimplemented options
13211
 *         in case of error.
13212
 */
13213
int
13214
xmlCtxtSetOptions(xmlParserCtxt *ctxt, int options)
13215
0
{
13216
0
#ifdef LIBXML_HTML_ENABLED
13217
0
    if ((ctxt != NULL) && (ctxt->html))
13218
0
        return(htmlCtxtSetOptions(ctxt, options));
13219
0
#endif
13220
13221
0
    return(xmlCtxtSetOptionsInternal(ctxt, options, 0));
13222
0
}
13223
13224
/**
13225
 * Get the current options of the parser context.
13226
 *
13227
 * @since 2.14.0
13228
 *
13229
 * @param ctxt  an XML parser context
13230
 * @returns the current options set in the parser context, or -1 if ctxt is NULL.
13231
 */
13232
int
13233
xmlCtxtGetOptions(xmlParserCtxt *ctxt)
13234
0
{
13235
0
    if (ctxt == NULL)
13236
0
        return(-1);
13237
13238
0
    return(ctxt->options);
13239
0
}
13240
13241
/**
13242
 * Applies the options to the parser context. The following options
13243
 * are never cleared and can only be enabled:
13244
 *
13245
 * - XML_PARSE_NOERROR
13246
 * - XML_PARSE_NOWARNING
13247
 * - XML_PARSE_NONET
13248
 * - XML_PARSE_NSCLEAN
13249
 * - XML_PARSE_NOCDATA
13250
 * - XML_PARSE_COMPACT
13251
 * - XML_PARSE_OLD10
13252
 * - XML_PARSE_HUGE
13253
 * - XML_PARSE_OLDSAX
13254
 * - XML_PARSE_IGNORE_ENC
13255
 * - XML_PARSE_BIG_LINES
13256
 *
13257
 * @deprecated Use #xmlCtxtSetOptions.
13258
 *
13259
 * @param ctxt  an XML parser context
13260
 * @param options  a combination of xmlParserOption
13261
 * @returns 0 in case of success, the set of unknown or unimplemented options
13262
 *         in case of error.
13263
 */
13264
int
13265
xmlCtxtUseOptions(xmlParserCtxt *ctxt, int options)
13266
35.9k
{
13267
35.9k
    int keepMask;
13268
13269
35.9k
#ifdef LIBXML_HTML_ENABLED
13270
35.9k
    if ((ctxt != NULL) && (ctxt->html))
13271
0
        return(htmlCtxtUseOptions(ctxt, options));
13272
35.9k
#endif
13273
13274
    /*
13275
     * For historic reasons, some options can only be enabled.
13276
     */
13277
35.9k
    keepMask = XML_PARSE_NOERROR |
13278
35.9k
               XML_PARSE_NOWARNING |
13279
35.9k
               XML_PARSE_NONET |
13280
35.9k
               XML_PARSE_NSCLEAN |
13281
35.9k
               XML_PARSE_NOCDATA |
13282
35.9k
               XML_PARSE_COMPACT |
13283
35.9k
               XML_PARSE_OLD10 |
13284
35.9k
               XML_PARSE_HUGE |
13285
35.9k
               XML_PARSE_OLDSAX |
13286
35.9k
               XML_PARSE_IGNORE_ENC |
13287
35.9k
               XML_PARSE_BIG_LINES;
13288
13289
35.9k
    return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask));
13290
35.9k
}
13291
13292
/**
13293
 * To protect against exponential entity expansion ("billion laughs"), the
13294
 * size of serialized output is (roughly) limited to the input size
13295
 * multiplied by this factor. The default value is 5.
13296
 *
13297
 * When working with documents making heavy use of entity expansion, it can
13298
 * be necessary to increase the value. For security reasons, this should only
13299
 * be considered when processing trusted input.
13300
 *
13301
 * @param ctxt  an XML parser context
13302
 * @param maxAmpl  maximum amplification factor
13303
 */
13304
void
13305
xmlCtxtSetMaxAmplification(xmlParserCtxt *ctxt, unsigned maxAmpl)
13306
0
{
13307
0
    if (ctxt == NULL)
13308
0
        return;
13309
0
    ctxt->maxAmpl = maxAmpl;
13310
0
}
13311
13312
/**
13313
 * Parse an XML document and return the resulting document tree.
13314
 * Takes ownership of the input object.
13315
 *
13316
 * @since 2.13.0
13317
 *
13318
 * @param ctxt  an XML parser context
13319
 * @param input  parser input
13320
 * @returns the resulting document tree or NULL
13321
 */
13322
xmlDoc *
13323
xmlCtxtParseDocument(xmlParserCtxt *ctxt, xmlParserInput *input)
13324
31.4k
{
13325
31.4k
    xmlDocPtr ret = NULL;
13326
13327
31.4k
    if ((ctxt == NULL) || (input == NULL)) {
13328
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
13329
0
        xmlFreeInputStream(input);
13330
0
        return(NULL);
13331
0
    }
13332
13333
    /* assert(ctxt->inputNr == 0); */
13334
31.4k
    while (ctxt->inputNr > 0)
13335
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13336
13337
31.4k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13338
2
        xmlFreeInputStream(input);
13339
2
        return(NULL);
13340
2
    }
13341
13342
31.4k
    xmlParseDocument(ctxt);
13343
13344
31.4k
    ret = xmlCtxtGetDocument(ctxt);
13345
13346
    /* assert(ctxt->inputNr == 1); */
13347
63.3k
    while (ctxt->inputNr > 0)
13348
31.9k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13349
13350
31.4k
    return(ret);
13351
31.4k
}
13352
13353
/**
13354
 * Convenience function to parse an XML document from a
13355
 * zero-terminated string.
13356
 *
13357
 * See #xmlCtxtReadDoc for details.
13358
 *
13359
 * @param cur  a pointer to a zero terminated string
13360
 * @param URL  base URL (optional)
13361
 * @param encoding  the document encoding (optional)
13362
 * @param options  a combination of xmlParserOption
13363
 * @returns the resulting document tree
13364
 */
13365
xmlDoc *
13366
xmlReadDoc(const xmlChar *cur, const char *URL, const char *encoding,
13367
           int options)
13368
0
{
13369
0
    xmlParserCtxtPtr ctxt;
13370
0
    xmlParserInputPtr input;
13371
0
    xmlDocPtr doc = NULL;
13372
13373
0
    ctxt = xmlNewParserCtxt();
13374
0
    if (ctxt == NULL)
13375
0
        return(NULL);
13376
13377
0
    xmlCtxtUseOptions(ctxt, options);
13378
13379
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) cur, encoding,
13380
0
                                      XML_INPUT_BUF_STATIC);
13381
13382
0
    if (input != NULL)
13383
0
        doc = xmlCtxtParseDocument(ctxt, input);
13384
13385
0
    xmlFreeParserCtxt(ctxt);
13386
0
    return(doc);
13387
0
}
13388
13389
/**
13390
 * Convenience function to parse an XML file from the filesystem
13391
 * or a global, user-defined resource loader.
13392
 *
13393
 * If a "-" filename is passed, the function will read from stdin.
13394
 * This feature is potentially insecure and might be removed from
13395
 * later versions.
13396
 *
13397
 * See #xmlCtxtReadFile for details.
13398
 *
13399
 * @param filename  a file or URL
13400
 * @param encoding  the document encoding (optional)
13401
 * @param options  a combination of xmlParserOption
13402
 * @returns the resulting document tree
13403
 */
13404
xmlDoc *
13405
xmlReadFile(const char *filename, const char *encoding, int options)
13406
0
{
13407
0
    xmlParserCtxtPtr ctxt;
13408
0
    xmlParserInputPtr input;
13409
0
    xmlDocPtr doc = NULL;
13410
13411
0
    ctxt = xmlNewParserCtxt();
13412
0
    if (ctxt == NULL)
13413
0
        return(NULL);
13414
13415
0
    xmlCtxtUseOptions(ctxt, options);
13416
13417
    /*
13418
     * Backward compatibility for users of command line utilities like
13419
     * xmlstarlet expecting "-" to mean stdin. This is dangerous and
13420
     * should be removed at some point.
13421
     */
13422
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
13423
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO,
13424
0
                                      encoding, 0);
13425
0
    else
13426
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13427
13428
0
    if (input != NULL)
13429
0
        doc = xmlCtxtParseDocument(ctxt, input);
13430
13431
0
    xmlFreeParserCtxt(ctxt);
13432
0
    return(doc);
13433
0
}
13434
13435
/**
13436
 * Parse an XML in-memory document and build a tree. The input buffer must
13437
 * not contain a terminating null byte.
13438
 *
13439
 * See #xmlCtxtReadMemory for details.
13440
 *
13441
 * @param buffer  a pointer to a char array
13442
 * @param size  the size of the array
13443
 * @param url  base URL (optional)
13444
 * @param encoding  the document encoding (optional)
13445
 * @param options  a combination of xmlParserOption
13446
 * @returns the resulting document tree
13447
 */
13448
xmlDoc *
13449
xmlReadMemory(const char *buffer, int size, const char *url,
13450
              const char *encoding, int options)
13451
0
{
13452
0
    xmlParserCtxtPtr ctxt;
13453
0
    xmlParserInputPtr input;
13454
0
    xmlDocPtr doc = NULL;
13455
13456
0
    if (size < 0)
13457
0
  return(NULL);
13458
13459
0
    ctxt = xmlNewParserCtxt();
13460
0
    if (ctxt == NULL)
13461
0
        return(NULL);
13462
13463
0
    xmlCtxtUseOptions(ctxt, options);
13464
13465
0
    input = xmlCtxtNewInputFromMemory(ctxt, url, buffer, size, encoding,
13466
0
                                      XML_INPUT_BUF_STATIC);
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 from a file descriptor and build a tree.
13477
 *
13478
 * See #xmlCtxtReadFd for details.
13479
 *
13480
 * NOTE that the file descriptor will not be closed when the
13481
 * context is freed or reset.
13482
 *
13483
 * @param fd  an open file descriptor
13484
 * @param URL  base URL (optional)
13485
 * @param encoding  the document encoding (optional)
13486
 * @param options  a combination of xmlParserOption
13487
 * @returns the resulting document tree
13488
 */
13489
xmlDoc *
13490
xmlReadFd(int fd, const char *URL, const char *encoding, int options)
13491
0
{
13492
0
    xmlParserCtxtPtr ctxt;
13493
0
    xmlParserInputPtr input;
13494
0
    xmlDocPtr doc = NULL;
13495
13496
0
    ctxt = xmlNewParserCtxt();
13497
0
    if (ctxt == NULL)
13498
0
        return(NULL);
13499
13500
0
    xmlCtxtUseOptions(ctxt, options);
13501
13502
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
13503
13504
0
    if (input != NULL)
13505
0
        doc = xmlCtxtParseDocument(ctxt, input);
13506
13507
0
    xmlFreeParserCtxt(ctxt);
13508
0
    return(doc);
13509
0
}
13510
13511
/**
13512
 * Parse an XML document from I/O functions and context and build a tree.
13513
 *
13514
 * See #xmlCtxtReadIO for details.
13515
 *
13516
 * @param ioread  an I/O read function
13517
 * @param ioclose  an I/O close function (optional)
13518
 * @param ioctx  an I/O handler
13519
 * @param URL  base URL (optional)
13520
 * @param encoding  the document encoding (optional)
13521
 * @param options  a combination of xmlParserOption
13522
 * @returns the resulting document tree
13523
 */
13524
xmlDoc *
13525
xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose,
13526
          void *ioctx, const char *URL, const char *encoding, int options)
13527
0
{
13528
0
    xmlParserCtxtPtr ctxt;
13529
0
    xmlParserInputPtr input;
13530
0
    xmlDocPtr doc = NULL;
13531
13532
0
    ctxt = xmlNewParserCtxt();
13533
0
    if (ctxt == NULL)
13534
0
        return(NULL);
13535
13536
0
    xmlCtxtUseOptions(ctxt, options);
13537
13538
0
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
13539
0
                                  encoding, 0);
13540
13541
0
    if (input != NULL)
13542
0
        doc = xmlCtxtParseDocument(ctxt, input);
13543
13544
0
    xmlFreeParserCtxt(ctxt);
13545
0
    return(doc);
13546
0
}
13547
13548
/**
13549
 * Parse an XML in-memory document and build a tree.
13550
 *
13551
 * `URL` is used as base to resolve external entities and for error
13552
 * reporting.
13553
 *
13554
 * @param ctxt  an XML parser context
13555
 * @param str  a pointer to a zero terminated string
13556
 * @param URL  base URL (optional)
13557
 * @param encoding  the document encoding (optional)
13558
 * @param options  a combination of xmlParserOption
13559
 * @returns the resulting document tree
13560
 */
13561
xmlDoc *
13562
xmlCtxtReadDoc(xmlParserCtxt *ctxt, const xmlChar *str,
13563
               const char *URL, const char *encoding, int options)
13564
0
{
13565
0
    xmlParserInputPtr input;
13566
13567
0
    if (ctxt == NULL)
13568
0
        return(NULL);
13569
13570
0
    xmlCtxtReset(ctxt);
13571
0
    xmlCtxtUseOptions(ctxt, options);
13572
13573
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) str, encoding,
13574
0
                                      XML_INPUT_BUF_STATIC);
13575
0
    if (input == NULL)
13576
0
        return(NULL);
13577
13578
0
    return(xmlCtxtParseDocument(ctxt, input));
13579
0
}
13580
13581
/**
13582
 * Parse an XML file from the filesystem or a global, user-defined
13583
 * resource loader.
13584
 *
13585
 * @param ctxt  an XML parser context
13586
 * @param filename  a file or URL
13587
 * @param encoding  the document encoding (optional)
13588
 * @param options  a combination of xmlParserOption
13589
 * @returns the resulting document tree
13590
 */
13591
xmlDoc *
13592
xmlCtxtReadFile(xmlParserCtxt *ctxt, const char *filename,
13593
                const char *encoding, int options)
13594
0
{
13595
0
    xmlParserInputPtr input;
13596
13597
0
    if (ctxt == NULL)
13598
0
        return(NULL);
13599
13600
0
    xmlCtxtReset(ctxt);
13601
0
    xmlCtxtUseOptions(ctxt, options);
13602
13603
0
    input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13604
0
    if (input == NULL)
13605
0
        return(NULL);
13606
13607
0
    return(xmlCtxtParseDocument(ctxt, input));
13608
0
}
13609
13610
/**
13611
 * Parse an XML in-memory document and build a tree. The input buffer must
13612
 * not contain a terminating null byte.
13613
 *
13614
 * `URL` is used as base to resolve external entities and for error
13615
 * reporting.
13616
 *
13617
 * @param ctxt  an XML parser context
13618
 * @param buffer  a pointer to a char array
13619
 * @param size  the size of the array
13620
 * @param URL  base URL (optional)
13621
 * @param encoding  the document encoding (optional)
13622
 * @param options  a combination of xmlParserOption
13623
 * @returns the resulting document tree
13624
 */
13625
xmlDoc *
13626
xmlCtxtReadMemory(xmlParserCtxt *ctxt, const char *buffer, int size,
13627
                  const char *URL, const char *encoding, int options)
13628
31.4k
{
13629
31.4k
    xmlParserInputPtr input;
13630
13631
31.4k
    if ((ctxt == NULL) || (size < 0))
13632
0
        return(NULL);
13633
13634
31.4k
    xmlCtxtReset(ctxt);
13635
31.4k
    xmlCtxtUseOptions(ctxt, options);
13636
13637
31.4k
    input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding,
13638
31.4k
                                      XML_INPUT_BUF_STATIC);
13639
31.4k
    if (input == NULL)
13640
15
        return(NULL);
13641
13642
31.4k
    return(xmlCtxtParseDocument(ctxt, input));
13643
31.4k
}
13644
13645
/**
13646
 * Parse an XML document from a file descriptor and build a tree.
13647
 *
13648
 * NOTE that the file descriptor will not be closed when the
13649
 * context is freed or reset.
13650
 *
13651
 * `URL` is used as base to resolve external entities and for error
13652
 * reporting.
13653
 *
13654
 * @param ctxt  an XML parser context
13655
 * @param fd  an open file descriptor
13656
 * @param URL  base URL (optional)
13657
 * @param encoding  the document encoding (optional)
13658
 * @param options  a combination of xmlParserOption
13659
 * @returns the resulting document tree
13660
 */
13661
xmlDoc *
13662
xmlCtxtReadFd(xmlParserCtxt *ctxt, int fd,
13663
              const char *URL, const char *encoding, int options)
13664
0
{
13665
0
    xmlParserInputPtr input;
13666
13667
0
    if (ctxt == NULL)
13668
0
        return(NULL);
13669
13670
0
    xmlCtxtReset(ctxt);
13671
0
    xmlCtxtUseOptions(ctxt, options);
13672
13673
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
13674
0
    if (input == NULL)
13675
0
        return(NULL);
13676
13677
0
    return(xmlCtxtParseDocument(ctxt, input));
13678
0
}
13679
13680
/**
13681
 * Parse an XML document from I/O functions and source and build a tree.
13682
 * This reuses the existing `ctxt` parser context
13683
 *
13684
 * `URL` is used as base to resolve external entities and for error
13685
 * reporting.
13686
 *
13687
 * @param ctxt  an XML parser context
13688
 * @param ioread  an I/O read function
13689
 * @param ioclose  an I/O close function
13690
 * @param ioctx  an I/O handler
13691
 * @param URL  the base URL to use for the document
13692
 * @param encoding  the document encoding, or NULL
13693
 * @param options  a combination of xmlParserOption
13694
 * @returns the resulting document tree
13695
 */
13696
xmlDoc *
13697
xmlCtxtReadIO(xmlParserCtxt *ctxt, xmlInputReadCallback ioread,
13698
              xmlInputCloseCallback ioclose, void *ioctx,
13699
        const char *URL,
13700
              const char *encoding, int options)
13701
0
{
13702
0
    xmlParserInputPtr input;
13703
13704
0
    if (ctxt == NULL)
13705
0
        return(NULL);
13706
13707
0
    xmlCtxtReset(ctxt);
13708
0
    xmlCtxtUseOptions(ctxt, options);
13709
13710
0
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
13711
0
                                  encoding, 0);
13712
0
    if (input == NULL)
13713
0
        return(NULL);
13714
13715
0
    return(xmlCtxtParseDocument(ctxt, input));
13716
0
}
13717