Coverage Report

Created: 2026-04-09 11:41

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/work/workdir/UnpackedTarball/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
30.4M
#define NS_INDEX_EMPTY  INT_MAX
81
5.28M
#define NS_INDEX_XML    (INT_MAX - 1)
82
11.0M
#define URI_HASH_EMPTY  0xD943A04E
83
48.8k
#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
971k
#define XML_MAX_ATTRS 100000000 /* 100 million */
94
95
21.9M
#define XML_SPECIAL_EXTERNAL    (1 << 20)
96
21.9M
#define XML_SPECIAL_TYPE_MASK   (XML_SPECIAL_EXTERNAL - 1)
97
98
22.1M
#define XML_ATTVAL_ALLOC        (1 << 0)
99
39.2k
#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
738k
#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
738k
#define XML_ENT_FIXED_COST 20
170
171
74.9M
#define XML_PARSER_BIG_BUFFER_SIZE 300
172
216k
#define XML_PARSER_BUFFER_SIZE 100
173
42.4k
#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
0
xmlErrMemory(xmlParserCtxtPtr ctxt) {
225
0
    xmlCtxtErrMemory(ctxt);
226
0
}
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
223k
{
239
223k
    if (prefix == NULL)
240
174k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
241
174k
                   XML_ERR_FATAL, localname, NULL, NULL, 0,
242
174k
                   "Attribute %s redefined\n", localname);
243
48.8k
    else
244
48.8k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
245
48.8k
                   XML_ERR_FATAL, prefix, localname, NULL, 0,
246
48.8k
                   "Attribute %s:%s redefined\n", prefix, localname);
247
223k
}
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
1.83M
{
260
1.83M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
261
1.83M
               NULL, NULL, NULL, 0, "%s", msg);
262
1.83M
}
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
27.9k
{
277
27.9k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING,
278
27.9k
               str1, str2, NULL, 0, msg, str1, str2);
279
27.9k
}
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
14.4k
{
314
14.4k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
315
14.4k
               NULL, NULL, NULL, val, msg, val);
316
14.4k
}
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
75.1k
{
333
75.1k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
334
75.1k
               str1, str2, NULL, val, msg, str1, val, str2);
335
75.1k
}
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
128k
{
349
128k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
350
128k
               val, NULL, NULL, 0, msg, val);
351
128k
}
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
10.5k
{
365
10.5k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR,
366
10.5k
               val, NULL, NULL, 0, msg, val);
367
10.5k
}
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
1.79M
{
385
1.79M
    ctxt->nsWellFormed = 0;
386
387
1.79M
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR,
388
1.79M
               info1, info2, info3, 0, msg, info1, info2, info3);
389
1.79M
}
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
70.2k
{
407
70.2k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING,
408
70.2k
               info1, info2, info3, 0, msg, info1, info2, info3);
409
70.2k
}
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
738k
{
436
738k
    unsigned long consumed;
437
738k
    unsigned long *expandedSize;
438
738k
    xmlParserInputPtr input = ctxt->input;
439
738k
    xmlEntityPtr entity = input->entity;
440
441
738k
    if ((entity) && (entity->flags & XML_ENT_CHECKED))
442
0
        return(0);
443
444
    /*
445
     * Compute total consumed bytes so far, including input streams of
446
     * external entities.
447
     */
448
738k
    consumed = input->consumed;
449
738k
    xmlSaturatedAddSizeT(&consumed, input->cur - input->base);
450
738k
    xmlSaturatedAdd(&consumed, ctxt->sizeentities);
451
452
738k
    if (entity)
453
0
        expandedSize = &entity->expandedSize;
454
738k
    else
455
738k
        expandedSize = &ctxt->sizeentcopy;
456
457
    /*
458
     * Add extra cost and some fixed cost.
459
     */
460
738k
    xmlSaturatedAdd(expandedSize, extra);
461
738k
    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
738k
    if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) &&
469
168
        ((*expandedSize >= ULONG_MAX) ||
470
168
         (*expandedSize / ctxt->maxAmpl > consumed))) {
471
168
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
472
168
                       "Maximum entity amplification factor exceeded, see "
473
168
                       "xmlCtxtSetMaxAmplification.\n");
474
168
        return(1);
475
168
    }
476
477
737k
    return(0);
478
738k
}
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
#ifdef LIBXML_ICONV_ENABLED
589
            return(1);
590
#else
591
0
            return(0);
592
0
#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
0
#ifdef LIBXML_DEBUG_ENABLED
641
0
            return(1);
642
#else
643
            return(0);
644
#endif
645
0
        case XML_WITH_DEBUG_MEM:
646
0
            return(0);
647
0
        case XML_WITH_ZLIB:
648
#ifdef LIBXML_ZLIB_ENABLED
649
            return(1);
650
#else
651
0
            return(0);
652
0
#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
22.0M
xmlSBufInit(xmlSBuf *buf, unsigned max) {
683
22.0M
    buf->mem = NULL;
684
22.0M
    buf->size = 0;
685
22.0M
    buf->cap = 0;
686
22.0M
    buf->max = max;
687
22.0M
    buf->code = XML_ERR_OK;
688
22.0M
}
689
690
static int
691
283k
xmlSBufGrow(xmlSBuf *buf, unsigned len) {
692
283k
    xmlChar *mem;
693
283k
    unsigned cap;
694
695
283k
    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
283k
    cap = (buf->size + len) * 2;
702
283k
    if (cap < 240)
703
236k
        cap = 240;
704
705
283k
    mem = xmlRealloc(buf->mem, cap);
706
283k
    if (mem == NULL) {
707
0
        buf->code = XML_ERR_NO_MEMORY;
708
0
        return(-1);
709
0
    }
710
711
283k
    buf->mem = mem;
712
283k
    buf->cap = cap;
713
714
283k
    return(0);
715
283k
}
716
717
static void
718
5.52M
xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) {
719
5.52M
    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
5.52M
    if (buf->cap - buf->size <= len) {
726
280k
        if (xmlSBufGrow(buf, len) < 0)
727
0
            return;
728
280k
    }
729
730
5.52M
    if (len > 0)
731
5.52M
        memcpy(buf->mem + buf->size, str, len);
732
5.52M
    buf->size += len;
733
5.52M
}
734
735
static void
736
4.40M
xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) {
737
4.40M
    xmlSBufAddString(buf, (const xmlChar *) str, len);
738
4.40M
}
739
740
static void
741
69.7k
xmlSBufAddChar(xmlSBuf *buf, int c) {
742
69.7k
    xmlChar *end;
743
744
69.7k
    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
69.7k
    if (buf->cap - buf->size <= 4) {
751
3.32k
        if (xmlSBufGrow(buf, 4) < 0)
752
0
            return;
753
3.32k
    }
754
755
69.7k
    end = buf->mem + buf->size;
756
757
69.7k
    if (c < 0x80) {
758
18.5k
        *end = (xmlChar) c;
759
18.5k
        buf->size += 1;
760
51.2k
    } else {
761
51.2k
        buf->size += xmlCopyCharMultiByte(end, c);
762
51.2k
    }
763
69.7k
}
764
765
static void
766
3.60M
xmlSBufAddReplChar(xmlSBuf *buf) {
767
3.60M
    xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3);
768
3.60M
}
769
770
static void
771
0
xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
772
0
    if (buf->code == XML_ERR_NO_MEMORY)
773
0
        xmlCtxtErrMemory(ctxt);
774
0
    else
775
0
        xmlFatalErr(ctxt, buf->code, errMsg);
776
0
}
777
778
static xmlChar *
779
xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt,
780
272k
              const char *errMsg) {
781
272k
    if (buf->mem == NULL) {
782
27.3k
        buf->mem = xmlMalloc(1);
783
27.3k
        if (buf->mem == NULL) {
784
0
            buf->code = XML_ERR_NO_MEMORY;
785
27.3k
        } else {
786
27.3k
            buf->mem[0] = 0;
787
27.3k
        }
788
245k
    } else {
789
245k
        buf->mem[buf->size] = 0;
790
245k
    }
791
792
272k
    if (buf->code == XML_ERR_OK) {
793
272k
        if (sizeOut != NULL)
794
191k
            *sizeOut = buf->size;
795
272k
        return(buf->mem);
796
272k
    }
797
798
0
    xmlSBufReportError(buf, ctxt, errMsg);
799
800
0
    xmlFree(buf->mem);
801
802
0
    if (sizeOut != NULL)
803
0
        *sizeOut = 0;
804
0
    return(NULL);
805
272k
}
806
807
static void
808
21.7M
xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
809
21.7M
    if (buf->code != XML_ERR_OK)
810
0
        xmlSBufReportError(buf, ctxt, errMsg);
811
812
21.7M
    xmlFree(buf->mem);
813
21.7M
}
814
815
static int
816
xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str,
817
6.10M
                    const char *errMsg) {
818
6.10M
    int c = str[0];
819
6.10M
    int c1 = str[1];
820
821
6.10M
    if ((c1 & 0xC0) != 0x80)
822
1.62M
        goto encoding_error;
823
824
4.47M
    if (c < 0xE0) {
825
        /* 2-byte sequence */
826
1.24M
        if (c < 0xC2)
827
462k
            goto encoding_error;
828
829
787k
        return(2);
830
3.22M
    } else {
831
3.22M
        int c2 = str[2];
832
833
3.22M
        if ((c2 & 0xC0) != 0x80)
834
21.4k
            goto encoding_error;
835
836
3.20M
        if (c < 0xF0) {
837
            /* 3-byte sequence */
838
3.10M
            if (c == 0xE0) {
839
                /* overlong */
840
160k
                if (c1 < 0xA0)
841
1.54k
                    goto encoding_error;
842
2.94M
            } else if (c == 0xED) {
843
                /* surrogate */
844
12.6k
                if (c1 >= 0xA0)
845
7.00k
                    goto encoding_error;
846
2.93M
            } else if (c == 0xEF) {
847
                /* U+FFFE and U+FFFF are invalid Chars */
848
1.49M
                if ((c1 == 0xBF) && (c2 >= 0xBE))
849
26.4k
                    xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg);
850
1.49M
            }
851
852
3.09M
            return(3);
853
3.10M
        } else {
854
            /* 4-byte sequence */
855
105k
            if ((str[3] & 0xC0) != 0x80)
856
5.32k
                goto encoding_error;
857
99.7k
            if (c == 0xF0) {
858
                /* overlong */
859
3.36k
                if (c1 < 0x90)
860
1.09k
                    goto encoding_error;
861
96.3k
            } else if (c >= 0xF4) {
862
                /* greater than 0x10FFFF */
863
6.69k
                if ((c > 0xF4) || (c1 >= 0x90))
864
3.25k
                    goto encoding_error;
865
6.69k
            }
866
867
95.3k
            return(4);
868
99.7k
        }
869
3.20M
    }
870
871
2.13M
encoding_error:
872
    /* Only report the first error */
873
2.13M
    if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) {
874
5.06k
        xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL);
875
5.06k
        ctxt->input->flags |= XML_INPUT_ENCODING_ERROR;
876
5.06k
    }
877
878
2.13M
    return(0);
879
4.47M
}
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
218k
xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) {
897
218k
    xmlSAXHandlerPtr sax;
898
899
    /* Avoid unused variable warning if features are disabled. */
900
218k
    (void) sax;
901
902
    /*
903
     * Changing the SAX struct directly is still widespread practice
904
     * in internal and external code.
905
     */
906
218k
    if (ctxt == NULL) return;
907
218k
    sax = ctxt->sax;
908
218k
#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
218k
    if (((ctxt->options & XML_PARSE_SAX1) == 0) &&
914
218k
        (sax) &&
915
218k
        (sax->initialized == XML_SAX2_MAGIC) &&
916
218k
        ((sax->startElementNs != NULL) ||
917
0
         (sax->endElementNs != NULL) ||
918
0
         ((sax->startElement == NULL) && (sax->endElement == NULL))))
919
218k
        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
218k
    ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3);
929
218k
    ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5);
930
218k
    ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36);
931
218k
    if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) ||
932
218k
    (ctxt->str_xml_ns == NULL)) {
933
0
        xmlErrMemory(ctxt);
934
0
    }
935
936
218k
    xmlDictSetLimit(ctxt->dict,
937
218k
                    (ctxt->options & XML_PARSE_HUGE) ?
938
213k
                        0 :
939
218k
                        XML_MAX_DICTIONARY_LIMIT);
940
941
218k
#ifdef LIBXML_VALID_ENABLED
942
218k
    if (ctxt->validate)
943
0
        ctxt->vctxt.flags |= XML_VCTXT_VALIDATE;
944
218k
    else
945
218k
        ctxt->vctxt.flags &= ~XML_VCTXT_VALIDATE;
946
218k
#endif /* LIBXML_VALID_ENABLED */
947
218k
}
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
34.5k
{
989
34.5k
    if ((src == NULL) || (dst == NULL))
990
0
        return(NULL);
991
992
49.9k
    while (*src == 0x20) src++;
993
780k
    while (*src != 0) {
994
746k
  if (*src == 0x20) {
995
131k
      while (*src == 0x20) src++;
996
35.1k
      if (*src != 0)
997
29.4k
    *dst++ = 0x20;
998
710k
  } else {
999
710k
      *dst++ = *src++;
1000
710k
  }
1001
746k
    }
1002
34.5k
    *dst = 0;
1003
34.5k
    if (dst == src)
1004
21.6k
       return(NULL);
1005
12.9k
    return(dst);
1006
34.5k
}
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
37.0k
               const xmlChar *value) {
1021
37.0k
    xmlDefAttrsPtr defaults;
1022
37.0k
    xmlDefAttr *attr;
1023
37.0k
    int len, expandedSize;
1024
37.0k
    xmlHashedString name;
1025
37.0k
    xmlHashedString prefix;
1026
37.0k
    xmlHashedString hvalue;
1027
37.0k
    const xmlChar *localname;
1028
1029
    /*
1030
     * Allows to detect attribute redefinitions
1031
     */
1032
37.0k
    if (ctxt->attsSpecial != NULL) {
1033
34.2k
        if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL)
1034
10.0k
      return;
1035
34.2k
    }
1036
1037
27.0k
    if (ctxt->attsDefault == NULL) {
1038
2.86k
        ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict);
1039
2.86k
  if (ctxt->attsDefault == NULL)
1040
0
      goto mem_error;
1041
2.86k
    }
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
27.0k
    localname = xmlSplitQName3(fullname, &len);
1048
27.0k
    if (localname == NULL) {
1049
24.4k
        name = xmlDictLookupHashed(ctxt->dict, fullname, -1);
1050
24.4k
  prefix.name = NULL;
1051
24.4k
    } else {
1052
2.62k
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1053
2.62k
  prefix = xmlDictLookupHashed(ctxt->dict, fullname, len);
1054
2.62k
        if (prefix.name == NULL)
1055
0
            goto mem_error;
1056
2.62k
    }
1057
27.0k
    if (name.name == NULL)
1058
0
        goto mem_error;
1059
1060
    /*
1061
     * make sure there is some storage
1062
     */
1063
27.0k
    defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name);
1064
27.0k
    if ((defaults == NULL) ||
1065
20.8k
        (defaults->nbAttrs >= defaults->maxAttrs)) {
1066
9.30k
        xmlDefAttrsPtr temp;
1067
9.30k
        int newSize;
1068
1069
9.30k
        if (defaults == NULL) {
1070
6.17k
            newSize = 4;
1071
6.17k
        } else {
1072
3.13k
            if ((defaults->maxAttrs >= XML_MAX_ATTRS) ||
1073
3.13k
                ((size_t) defaults->maxAttrs >
1074
3.13k
                     SIZE_MAX / 2 / sizeof(temp[0]) - sizeof(*defaults)))
1075
0
                goto mem_error;
1076
1077
3.13k
            if (defaults->maxAttrs > XML_MAX_ATTRS / 2)
1078
0
                newSize = XML_MAX_ATTRS;
1079
3.13k
            else
1080
3.13k
                newSize = defaults->maxAttrs * 2;
1081
3.13k
        }
1082
9.30k
        temp = xmlRealloc(defaults,
1083
9.30k
                          sizeof(*defaults) + newSize * sizeof(xmlDefAttr));
1084
9.30k
  if (temp == NULL)
1085
0
      goto mem_error;
1086
9.30k
        if (defaults == NULL)
1087
6.17k
            temp->nbAttrs = 0;
1088
9.30k
  temp->maxAttrs = newSize;
1089
9.30k
        defaults = temp;
1090
9.30k
  if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name,
1091
9.30k
                          defaults, NULL) < 0) {
1092
0
      xmlFree(defaults);
1093
0
      goto mem_error;
1094
0
  }
1095
9.30k
    }
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
27.0k
    localname = xmlSplitQName3(fullattr, &len);
1102
27.0k
    if (localname == NULL) {
1103
12.9k
        name = xmlDictLookupHashed(ctxt->dict, fullattr, -1);
1104
12.9k
  prefix.name = NULL;
1105
14.0k
    } else {
1106
14.0k
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1107
14.0k
  prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len);
1108
14.0k
        if (prefix.name == NULL)
1109
0
            goto mem_error;
1110
14.0k
    }
1111
27.0k
    if (name.name == NULL)
1112
0
        goto mem_error;
1113
1114
    /* intern the string and precompute the end */
1115
27.0k
    len = strlen((const char *) value);
1116
27.0k
    hvalue = xmlDictLookupHashed(ctxt->dict, value, len);
1117
27.0k
    if (hvalue.name == NULL)
1118
0
        goto mem_error;
1119
1120
27.0k
    expandedSize = strlen((const char *) name.name);
1121
27.0k
    if (prefix.name != NULL)
1122
14.0k
        expandedSize += strlen((const char *) prefix.name);
1123
27.0k
    expandedSize += len;
1124
1125
27.0k
    attr = &defaults->attrs[defaults->nbAttrs++];
1126
27.0k
    attr->name = name;
1127
27.0k
    attr->prefix = prefix;
1128
27.0k
    attr->value = hvalue;
1129
27.0k
    attr->valueEnd = hvalue.name + len;
1130
27.0k
    attr->external = PARSER_EXTERNAL(ctxt);
1131
27.0k
    attr->expandedSize = expandedSize;
1132
1133
27.0k
    return;
1134
1135
0
mem_error:
1136
0
    xmlErrMemory(ctxt);
1137
0
}
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
43.0k
{
1153
43.0k
    if (ctxt->attsSpecial == NULL) {
1154
3.18k
        ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict);
1155
3.18k
  if (ctxt->attsSpecial == NULL)
1156
0
      goto mem_error;
1157
3.18k
    }
1158
1159
43.0k
    if (PARSER_EXTERNAL(ctxt))
1160
0
        type |= XML_SPECIAL_EXTERNAL;
1161
1162
43.0k
    if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr,
1163
43.0k
                    XML_INT_TO_PTR(type)) < 0)
1164
0
        goto mem_error;
1165
43.0k
    return;
1166
1167
43.0k
mem_error:
1168
0
    xmlErrMemory(ctxt);
1169
0
}
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
30.1k
                            const xmlChar *unused ATTRIBUTE_UNUSED) {
1178
30.1k
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data;
1179
1180
30.1k
    if (XML_PTR_TO_INT(payload) == XML_ATTRIBUTE_CDATA) {
1181
2.32k
        xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL);
1182
2.32k
    }
1183
30.1k
}
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
7.50k
{
1195
7.50k
    if (ctxt->attsSpecial == NULL)
1196
4.32k
        return;
1197
1198
3.18k
    xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt);
1199
1200
3.18k
    if (xmlHashSize(ctxt->attsSpecial) == 0) {
1201
213
        xmlHashFree(ctxt->attsSpecial, NULL);
1202
213
        ctxt->attsSpecial = NULL;
1203
213
    }
1204
3.18k
}
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
0
{
1268
0
    const xmlChar *cur = lang, *nxt;
1269
1270
0
    if (cur == NULL)
1271
0
        return (0);
1272
0
    if (((cur[0] == 'i') && (cur[1] == '-')) ||
1273
0
        ((cur[0] == 'I') && (cur[1] == '-')) ||
1274
0
        ((cur[0] == 'x') && (cur[1] == '-')) ||
1275
0
        ((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
0
        cur += 2;
1282
0
        while (((cur[0] >= 'A') && (cur[0] <= 'Z')) ||
1283
0
               ((cur[0] >= 'a') && (cur[0] <= 'z')))
1284
0
            cur++;
1285
0
        return(cur[0] == 0);
1286
0
    }
1287
0
    nxt = cur;
1288
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1289
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1290
0
           nxt++;
1291
0
    if (nxt - cur >= 4) {
1292
        /*
1293
         * Reserved
1294
         */
1295
0
        if ((nxt - cur > 8) || (nxt[0] != 0))
1296
0
            return(0);
1297
0
        return(1);
1298
0
    }
1299
0
    if (nxt - cur < 2)
1300
0
        return(0);
1301
    /* we got an ISO 639 code */
1302
0
    if (nxt[0] == 0)
1303
0
        return(1);
1304
0
    if (nxt[0] != '-')
1305
0
        return(0);
1306
1307
0
    nxt++;
1308
0
    cur = nxt;
1309
    /* now we can have extlang or script or region or variant */
1310
0
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1311
0
        goto region_m49;
1312
1313
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1314
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1315
0
           nxt++;
1316
0
    if (nxt - cur == 4)
1317
0
        goto script;
1318
0
    if (nxt - cur == 2)
1319
0
        goto region;
1320
0
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1321
0
        goto variant;
1322
0
    if (nxt - cur != 3)
1323
0
        return(0);
1324
    /* we parsed an extlang */
1325
0
    if (nxt[0] == 0)
1326
0
        return(1);
1327
0
    if (nxt[0] != '-')
1328
0
        return(0);
1329
1330
0
    nxt++;
1331
0
    cur = nxt;
1332
    /* now we can have script or region or variant */
1333
0
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1334
0
        goto region_m49;
1335
1336
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1337
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1338
0
           nxt++;
1339
0
    if (nxt - cur == 2)
1340
0
        goto region;
1341
0
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1342
0
        goto variant;
1343
0
    if (nxt - cur != 4)
1344
0
        return(0);
1345
    /* we parsed a script */
1346
0
script:
1347
0
    if (nxt[0] == 0)
1348
0
        return(1);
1349
0
    if (nxt[0] != '-')
1350
0
        return(0);
1351
1352
0
    nxt++;
1353
0
    cur = nxt;
1354
    /* now we can have region or variant */
1355
0
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1356
0
        goto region_m49;
1357
1358
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1359
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1360
0
           nxt++;
1361
1362
0
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1363
0
        goto variant;
1364
0
    if (nxt - cur != 2)
1365
0
        return(0);
1366
    /* we parsed a region */
1367
0
region:
1368
0
    if (nxt[0] == 0)
1369
0
        return(1);
1370
0
    if (nxt[0] != '-')
1371
0
        return(0);
1372
1373
0
    nxt++;
1374
0
    cur = nxt;
1375
    /* now we can just have a variant */
1376
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1377
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1378
0
           nxt++;
1379
1380
0
    if ((nxt - cur < 5) || (nxt - cur > 8))
1381
0
        return(0);
1382
1383
    /* we parsed a variant */
1384
0
variant:
1385
0
    if (nxt[0] == 0)
1386
0
        return(1);
1387
0
    if (nxt[0] != '-')
1388
0
        return(0);
1389
    /* extensions and private use subtags not checked */
1390
0
    return (1);
1391
1392
0
region_m49:
1393
0
    if (((nxt[1] >= '0') && (nxt[1] <= '9')) &&
1394
0
        ((nxt[2] >= '0') && (nxt[2] <= '9'))) {
1395
0
        nxt += 3;
1396
0
        goto region;
1397
0
    }
1398
0
    return(0);
1399
0
}
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
218k
xmlParserNsCreate(void) {
1417
218k
    xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb));
1418
1419
218k
    if (nsdb == NULL)
1420
0
        return(NULL);
1421
218k
    memset(nsdb, 0, sizeof(*nsdb));
1422
218k
    nsdb->defaultNsIndex = INT_MAX;
1423
1424
218k
    return(nsdb);
1425
218k
}
1426
1427
/**
1428
 * Free a namespace database.
1429
 *
1430
 * @param nsdb  namespace database
1431
 */
1432
void
1433
218k
xmlParserNsFree(xmlParserNsData *nsdb) {
1434
218k
    if (nsdb == NULL)
1435
0
        return;
1436
1437
218k
    xmlFree(nsdb->extra);
1438
218k
    xmlFree(nsdb->hash);
1439
218k
    xmlFree(nsdb);
1440
218k
}
1441
1442
/**
1443
 * Reset a namespace database.
1444
 *
1445
 * @param nsdb  namespace database
1446
 */
1447
static void
1448
5.29k
xmlParserNsReset(xmlParserNsData *nsdb) {
1449
5.29k
    if (nsdb == NULL)
1450
0
        return;
1451
1452
5.29k
    nsdb->hashElems = 0;
1453
5.29k
    nsdb->elementId = 0;
1454
5.29k
    nsdb->defaultNsIndex = INT_MAX;
1455
1456
5.29k
    if (nsdb->hash)
1457
0
        memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0]));
1458
5.29k
}
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
33.6M
xmlParserNsStartElement(xmlParserNsData *nsdb) {
1468
33.6M
    if (nsdb->elementId == UINT_MAX)
1469
0
        return(-1);
1470
33.6M
    nsdb->elementId++;
1471
1472
33.6M
    return(0);
1473
33.6M
}
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
43.5M
                  xmlParserNsBucket **bucketPtr) {
1489
43.5M
    xmlParserNsBucket *bucket, *tombstone;
1490
43.5M
    unsigned index, hashValue;
1491
1492
43.5M
    if (prefix->name == NULL)
1493
21.4M
        return(ctxt->nsdb->defaultNsIndex);
1494
1495
22.0M
    if (ctxt->nsdb->hashSize == 0)
1496
469k
        return(INT_MAX);
1497
1498
21.5M
    hashValue = prefix->hashValue;
1499
21.5M
    index = hashValue & (ctxt->nsdb->hashSize - 1);
1500
21.5M
    bucket = &ctxt->nsdb->hash[index];
1501
21.5M
    tombstone = NULL;
1502
1503
22.8M
    while (bucket->hashValue) {
1504
21.3M
        if (bucket->index == INT_MAX) {
1505
245k
            if (tombstone == NULL)
1506
237k
                tombstone = bucket;
1507
21.0M
        } else if (bucket->hashValue == hashValue) {
1508
20.0M
            if (ctxt->nsTab[bucket->index * 2] == prefix->name) {
1509
20.0M
                if (bucketPtr != NULL)
1510
914k
                    *bucketPtr = bucket;
1511
20.0M
                return(bucket->index);
1512
20.0M
            }
1513
20.0M
        }
1514
1515
1.29M
        index++;
1516
1.29M
        bucket++;
1517
1.29M
        if (index == ctxt->nsdb->hashSize) {
1518
15.0k
            index = 0;
1519
15.0k
            bucket = ctxt->nsdb->hash;
1520
15.0k
        }
1521
1.29M
    }
1522
1523
1.53M
    if (bucketPtr != NULL)
1524
827k
        *bucketPtr = tombstone ? tombstone : bucket;
1525
1.53M
    return(INT_MAX);
1526
21.5M
}
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
33.5M
xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) {
1537
33.5M
    const xmlChar *ret;
1538
33.5M
    int nsIndex;
1539
1540
33.5M
    if (prefix->name == ctxt->str_xml)
1541
38.1k
        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
33.5M
    nsIndex = xmlParserNsLookup(ctxt, prefix, NULL);
1548
33.5M
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1549
17.3M
        return(NULL);
1550
1551
16.2M
    ret = ctxt->nsTab[nsIndex * 2 + 1];
1552
16.2M
    if (ret[0] == 0)
1553
29.0k
        ret = NULL;
1554
16.2M
    return(ret);
1555
33.5M
}
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
801k
xmlParserNsLookupSax(xmlParserCtxt *ctxt, const xmlChar *prefix) {
1567
801k
    xmlHashedString hprefix;
1568
801k
    int nsIndex;
1569
1570
801k
    if (prefix == ctxt->str_xml)
1571
0
        return(NULL);
1572
1573
801k
    hprefix.name = prefix;
1574
801k
    if (prefix != NULL)
1575
799k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1576
1.45k
    else
1577
1.45k
        hprefix.hashValue = 0;
1578
801k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1579
801k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1580
0
        return(NULL);
1581
1582
801k
    return(ctxt->nsdb->extra[nsIndex].saxData);
1583
801k
}
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
11.1k
                     void *saxData) {
1598
11.1k
    xmlHashedString hprefix;
1599
11.1k
    int nsIndex;
1600
1601
11.1k
    if (prefix == ctxt->str_xml)
1602
0
        return(-1);
1603
1604
11.1k
    hprefix.name = prefix;
1605
11.1k
    if (prefix != NULL)
1606
10.3k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1607
810
    else
1608
810
        hprefix.hashValue = 0;
1609
11.1k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1610
11.1k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1611
0
        return(-1);
1612
1613
11.1k
    ctxt->nsdb->extra[nsIndex].saxData = saxData;
1614
11.1k
    return(0);
1615
11.1k
}
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
524k
xmlParserNsGrow(xmlParserCtxtPtr ctxt) {
1625
524k
    const xmlChar **table;
1626
524k
    xmlParserNsExtra *extra;
1627
524k
    int newSize;
1628
1629
524k
    newSize = xmlGrowCapacity(ctxt->nsMax,
1630
524k
                              sizeof(table[0]) + sizeof(extra[0]),
1631
524k
                              16, XML_MAX_ITEMS);
1632
524k
    if (newSize < 0)
1633
0
        goto error;
1634
1635
524k
    table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0]));
1636
524k
    if (table == NULL)
1637
0
        goto error;
1638
524k
    ctxt->nsTab = table;
1639
1640
524k
    extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0]));
1641
524k
    if (extra == NULL)
1642
0
        goto error;
1643
524k
    ctxt->nsdb->extra = extra;
1644
1645
524k
    ctxt->nsMax = newSize;
1646
524k
    return(0);
1647
1648
0
error:
1649
0
    xmlErrMemory(ctxt);
1650
0
    return(-1);
1651
524k
}
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
1.52M
                const xmlHashedString *uri, void *saxData, int defAttr) {
1667
1.52M
    xmlParserNsBucket *bucket = NULL;
1668
1.52M
    xmlParserNsExtra *extra;
1669
1.52M
    const xmlChar **ns;
1670
1.52M
    unsigned hashValue, nsIndex, oldIndex;
1671
1672
1.52M
    if ((prefix != NULL) && (prefix->name == ctxt->str_xml))
1673
45
        return(0);
1674
1675
1.52M
    if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) {
1676
0
        xmlErrMemory(ctxt);
1677
0
        return(-1);
1678
0
    }
1679
1680
    /*
1681
     * Default namespace and 'xml' namespace
1682
     */
1683
1.52M
    if ((prefix == NULL) || (prefix->name == NULL)) {
1684
356k
        oldIndex = ctxt->nsdb->defaultNsIndex;
1685
1686
356k
        if (oldIndex != INT_MAX) {
1687
213k
            extra = &ctxt->nsdb->extra[oldIndex];
1688
1689
213k
            if (extra->elementId == ctxt->nsdb->elementId) {
1690
126k
                if (defAttr == 0)
1691
123k
                    xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns");
1692
126k
                return(0);
1693
126k
            }
1694
1695
86.7k
            if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1696
0
                (uri->name == ctxt->nsTab[oldIndex * 2 + 1]))
1697
0
                return(0);
1698
86.7k
        }
1699
1700
229k
        ctxt->nsdb->defaultNsIndex = ctxt->nsNr;
1701
229k
        goto populate_entry;
1702
356k
    }
1703
1704
    /*
1705
     * Hash table lookup
1706
     */
1707
1.16M
    oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket);
1708
1.16M
    if (oldIndex != INT_MAX) {
1709
223k
        extra = &ctxt->nsdb->extra[oldIndex];
1710
1711
        /*
1712
         * Check for duplicate definitions on the same element.
1713
         */
1714
223k
        if (extra->elementId == ctxt->nsdb->elementId) {
1715
46.1k
            if (defAttr == 0)
1716
46.1k
                xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name);
1717
46.1k
            return(0);
1718
46.1k
        }
1719
1720
177k
        if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1721
0
            (uri->name == ctxt->nsTab[bucket->index * 2 + 1]))
1722
0
            return(0);
1723
1724
177k
        bucket->index = ctxt->nsNr;
1725
177k
        goto populate_entry;
1726
177k
    }
1727
1728
    /*
1729
     * Insert new bucket
1730
     */
1731
1732
945k
    hashValue = prefix->hashValue;
1733
1734
    /*
1735
     * Grow hash table, 50% fill factor
1736
     */
1737
945k
    if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) {
1738
155k
        xmlParserNsBucket *newHash;
1739
155k
        unsigned newSize, i, index;
1740
1741
155k
        if (ctxt->nsdb->hashSize > UINT_MAX / 2) {
1742
0
            xmlErrMemory(ctxt);
1743
0
            return(-1);
1744
0
        }
1745
155k
        newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16;
1746
155k
        newHash = xmlMalloc(newSize * sizeof(newHash[0]));
1747
155k
        if (newHash == NULL) {
1748
0
            xmlErrMemory(ctxt);
1749
0
            return(-1);
1750
0
        }
1751
155k
        memset(newHash, 0, newSize * sizeof(newHash[0]));
1752
1753
1.65M
        for (i = 0; i < ctxt->nsdb->hashSize; i++) {
1754
1.49M
            unsigned hv = ctxt->nsdb->hash[i].hashValue;
1755
1.49M
            unsigned newIndex;
1756
1757
1.49M
            if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX))
1758
1.08M
                continue;
1759
416k
            newIndex = hv & (newSize - 1);
1760
1761
429k
            while (newHash[newIndex].hashValue != 0) {
1762
13.8k
                newIndex++;
1763
13.8k
                if (newIndex == newSize)
1764
94
                    newIndex = 0;
1765
13.8k
            }
1766
1767
416k
            newHash[newIndex] = ctxt->nsdb->hash[i];
1768
416k
        }
1769
1770
155k
        xmlFree(ctxt->nsdb->hash);
1771
155k
        ctxt->nsdb->hash = newHash;
1772
155k
        ctxt->nsdb->hashSize = newSize;
1773
1774
        /*
1775
         * Relookup
1776
         */
1777
155k
        index = hashValue & (newSize - 1);
1778
1779
157k
        while (newHash[index].hashValue != 0) {
1780
2.66k
            index++;
1781
2.66k
            if (index == newSize)
1782
44
                index = 0;
1783
2.66k
        }
1784
1785
155k
        bucket = &newHash[index];
1786
155k
    }
1787
1788
945k
    bucket->hashValue = hashValue;
1789
945k
    bucket->index = ctxt->nsNr;
1790
945k
    ctxt->nsdb->hashElems++;
1791
945k
    oldIndex = INT_MAX;
1792
1793
1.35M
populate_entry:
1794
1.35M
    nsIndex = ctxt->nsNr;
1795
1796
1.35M
    ns = &ctxt->nsTab[nsIndex * 2];
1797
1.35M
    ns[0] = prefix ? prefix->name : NULL;
1798
1.35M
    ns[1] = uri->name;
1799
1800
1.35M
    extra = &ctxt->nsdb->extra[nsIndex];
1801
1.35M
    extra->saxData = saxData;
1802
1.35M
    extra->prefixHashValue = prefix ? prefix->hashValue : 0;
1803
1.35M
    extra->uriHashValue = uri->hashValue;
1804
1.35M
    extra->elementId = ctxt->nsdb->elementId;
1805
1.35M
    extra->oldIndex = oldIndex;
1806
1807
1.35M
    ctxt->nsNr++;
1808
1809
1.35M
    return(1);
1810
945k
}
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
311k
{
1822
311k
    int i;
1823
1824
    /* assert(nr <= ctxt->nsNr); */
1825
1826
1.15M
    for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) {
1827
848k
        const xmlChar *prefix = ctxt->nsTab[i * 2];
1828
848k
        xmlParserNsExtra *extra = &ctxt->nsdb->extra[i];
1829
1830
848k
        if (prefix == NULL) {
1831
157k
            ctxt->nsdb->defaultNsIndex = extra->oldIndex;
1832
690k
        } else {
1833
690k
            xmlHashedString hprefix;
1834
690k
            xmlParserNsBucket *bucket = NULL;
1835
1836
690k
            hprefix.name = prefix;
1837
690k
            hprefix.hashValue = extra->prefixHashValue;
1838
690k
            xmlParserNsLookup(ctxt, &hprefix, &bucket);
1839
            /* assert(bucket && bucket->hashValue); */
1840
690k
            bucket->index = extra->oldIndex;
1841
690k
        }
1842
848k
    }
1843
1844
311k
    ctxt->nsNr -= nr;
1845
311k
    return(nr);
1846
311k
}
1847
1848
static int
1849
570k
xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt) {
1850
570k
    const xmlChar **atts;
1851
570k
    unsigned *attallocs;
1852
570k
    int newSize;
1853
1854
570k
    newSize = xmlGrowCapacity(ctxt->maxatts / 5,
1855
570k
                              sizeof(atts[0]) * 5 + sizeof(attallocs[0]),
1856
570k
                              10, XML_MAX_ATTRS);
1857
570k
    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
570k
    atts = xmlRealloc(ctxt->atts, newSize * sizeof(atts[0]) * 5);
1864
570k
    if (atts == NULL)
1865
0
        goto mem_error;
1866
570k
    ctxt->atts = atts;
1867
1868
570k
    attallocs = xmlRealloc(ctxt->attallocs,
1869
570k
                           newSize * sizeof(attallocs[0]));
1870
570k
    if (attallocs == NULL)
1871
0
        goto mem_error;
1872
570k
    ctxt->attallocs = attallocs;
1873
1874
570k
    ctxt->maxatts = newSize * 5;
1875
1876
570k
    return(0);
1877
1878
0
mem_error:
1879
0
    xmlErrMemory(ctxt);
1880
0
    return(-1);
1881
570k
}
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
218k
{
1893
218k
    char *directory = NULL;
1894
218k
    int maxDepth;
1895
1896
218k
    if ((ctxt == NULL) || (value == NULL))
1897
0
        return(-1);
1898
1899
218k
    maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
1900
1901
218k
    if (ctxt->inputNr >= ctxt->inputMax) {
1902
0
        xmlParserInputPtr *tmp;
1903
0
        int newSize;
1904
1905
0
        newSize = xmlGrowCapacity(ctxt->inputMax, sizeof(tmp[0]),
1906
0
                                  5, maxDepth);
1907
0
        if (newSize < 0) {
1908
0
            xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
1909
0
                           "Maximum entity nesting depth exceeded");
1910
0
            return(-1);
1911
0
        }
1912
0
        tmp = xmlRealloc(ctxt->inputTab, newSize * sizeof(tmp[0]));
1913
0
        if (tmp == NULL) {
1914
0
            xmlErrMemory(ctxt);
1915
0
            return(-1);
1916
0
        }
1917
0
        ctxt->inputTab = tmp;
1918
0
        ctxt->inputMax = newSize;
1919
0
    }
1920
1921
218k
    if ((ctxt->inputNr == 0) && (value->filename != NULL)) {
1922
0
        directory = xmlParserGetDirectory(value->filename);
1923
0
        if (directory == NULL) {
1924
0
            xmlErrMemory(ctxt);
1925
0
            return(-1);
1926
0
        }
1927
0
    }
1928
1929
218k
    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
218k
    ctxt->inputTab[ctxt->inputNr] = value;
1935
218k
    ctxt->input = value;
1936
1937
218k
    if (ctxt->inputNr == 0) {
1938
218k
        xmlFree(ctxt->directory);
1939
218k
        ctxt->directory = directory;
1940
218k
    }
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
218k
    value->id = ctxt->input_id++;
1948
1949
218k
    return(ctxt->inputNr++);
1950
218k
}
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
662k
{
1961
662k
    xmlParserInputPtr ret;
1962
1963
662k
    if (ctxt == NULL)
1964
0
        return(NULL);
1965
662k
    if (ctxt->inputNr <= 0)
1966
443k
        return (NULL);
1967
218k
    ctxt->inputNr--;
1968
218k
    if (ctxt->inputNr > 0)
1969
0
        ctxt->input = ctxt->inputTab[ctxt->inputNr - 1];
1970
218k
    else
1971
218k
        ctxt->input = NULL;
1972
218k
    ret = ctxt->inputTab[ctxt->inputNr];
1973
218k
    ctxt->inputTab[ctxt->inputNr] = NULL;
1974
218k
    return (ret);
1975
662k
}
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
799k
{
1989
799k
    if (ctxt == NULL)
1990
0
        return(0);
1991
1992
799k
    if (ctxt->nodeNr >= ctxt->nodeMax) {
1993
20.0k
        int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
1994
20.0k
        xmlNodePtr *tmp;
1995
20.0k
        int newSize;
1996
1997
20.0k
        newSize = xmlGrowCapacity(ctxt->nodeMax, sizeof(tmp[0]),
1998
20.0k
                                  10, maxDepth);
1999
20.0k
        if (newSize < 0) {
2000
0
            xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
2001
0
                    "Excessive depth in document: %d,"
2002
0
                    " use XML_PARSE_HUGE option\n",
2003
0
                    ctxt->nodeNr);
2004
0
            return(-1);
2005
0
        }
2006
2007
20.0k
  tmp = xmlRealloc(ctxt->nodeTab, newSize * sizeof(tmp[0]));
2008
20.0k
        if (tmp == NULL) {
2009
0
            xmlErrMemory(ctxt);
2010
0
            return (-1);
2011
0
        }
2012
20.0k
        ctxt->nodeTab = tmp;
2013
20.0k
  ctxt->nodeMax = newSize;
2014
20.0k
    }
2015
2016
799k
    ctxt->nodeTab[ctxt->nodeNr] = value;
2017
799k
    ctxt->node = value;
2018
799k
    return (ctxt->nodeNr++);
2019
799k
}
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
899k
{
2032
899k
    xmlNodePtr ret;
2033
2034
899k
    if (ctxt == NULL) return(NULL);
2035
899k
    if (ctxt->nodeNr <= 0)
2036
99.6k
        return (NULL);
2037
799k
    ctxt->nodeNr--;
2038
799k
    if (ctxt->nodeNr > 0)
2039
794k
        ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1];
2040
5.19k
    else
2041
5.19k
        ctxt->node = NULL;
2042
799k
    ret = ctxt->nodeTab[ctxt->nodeNr];
2043
799k
    ctxt->nodeTab[ctxt->nodeNr] = NULL;
2044
799k
    return (ret);
2045
899k
}
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
25.2M
{
2062
25.2M
    xmlStartTag *tag;
2063
2064
25.2M
    if (ctxt->nameNr >= ctxt->nameMax) {
2065
775k
        const xmlChar **tmp;
2066
775k
        xmlStartTag *tmp2;
2067
775k
        int newSize;
2068
2069
775k
        newSize = xmlGrowCapacity(ctxt->nameMax,
2070
775k
                                  sizeof(tmp[0]) + sizeof(tmp2[0]),
2071
775k
                                  10, XML_MAX_ITEMS);
2072
775k
        if (newSize < 0)
2073
0
            goto mem_error;
2074
2075
775k
        tmp = xmlRealloc(ctxt->nameTab, newSize * sizeof(tmp[0]));
2076
775k
        if (tmp == NULL)
2077
0
      goto mem_error;
2078
775k
  ctxt->nameTab = tmp;
2079
2080
775k
        tmp2 = xmlRealloc(ctxt->pushTab, newSize * sizeof(tmp2[0]));
2081
775k
        if (tmp2 == NULL)
2082
0
      goto mem_error;
2083
775k
  ctxt->pushTab = tmp2;
2084
2085
775k
        ctxt->nameMax = newSize;
2086
24.5M
    } else if (ctxt->pushTab == NULL) {
2087
199k
        ctxt->pushTab = xmlMalloc(ctxt->nameMax * sizeof(ctxt->pushTab[0]));
2088
199k
        if (ctxt->pushTab == NULL)
2089
0
            goto mem_error;
2090
199k
    }
2091
25.2M
    ctxt->nameTab[ctxt->nameNr] = value;
2092
25.2M
    ctxt->name = value;
2093
25.2M
    tag = &ctxt->pushTab[ctxt->nameNr];
2094
25.2M
    tag->prefix = prefix;
2095
25.2M
    tag->URI = URI;
2096
25.2M
    tag->line = line;
2097
25.2M
    tag->nsNr = nsNr;
2098
25.2M
    return (ctxt->nameNr++);
2099
0
mem_error:
2100
0
    xmlErrMemory(ctxt);
2101
0
    return (-1);
2102
25.2M
}
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
7.71M
{
2113
7.71M
    const xmlChar *ret;
2114
2115
7.71M
    if (ctxt->nameNr <= 0)
2116
0
        return (NULL);
2117
7.71M
    ctxt->nameNr--;
2118
7.71M
    if (ctxt->nameNr > 0)
2119
7.58M
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2120
123k
    else
2121
123k
        ctxt->name = NULL;
2122
7.71M
    ret = ctxt->nameTab[ctxt->nameNr];
2123
7.71M
    ctxt->nameTab[ctxt->nameNr] = NULL;
2124
7.71M
    return (ret);
2125
7.71M
}
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
890k
{
2139
890k
    const xmlChar *ret;
2140
2141
890k
    if ((ctxt == NULL) || (ctxt->nameNr <= 0))
2142
0
        return (NULL);
2143
890k
    ctxt->nameNr--;
2144
890k
    if (ctxt->nameNr > 0)
2145
885k
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2146
5.04k
    else
2147
5.04k
        ctxt->name = NULL;
2148
890k
    ret = ctxt->nameTab[ctxt->nameNr];
2149
890k
    ctxt->nameTab[ctxt->nameNr] = NULL;
2150
890k
    return (ret);
2151
890k
}
2152
2153
33.6M
static int spacePush(xmlParserCtxtPtr ctxt, int val) {
2154
33.6M
    if (ctxt->spaceNr >= ctxt->spaceMax) {
2155
967k
        int *tmp;
2156
967k
        int newSize;
2157
2158
967k
        newSize = xmlGrowCapacity(ctxt->spaceMax, sizeof(tmp[0]),
2159
967k
                                  10, XML_MAX_ITEMS);
2160
967k
        if (newSize < 0) {
2161
0
      xmlErrMemory(ctxt);
2162
0
      return(-1);
2163
0
        }
2164
2165
967k
        tmp = xmlRealloc(ctxt->spaceTab, newSize * sizeof(tmp[0]));
2166
967k
        if (tmp == NULL) {
2167
0
      xmlErrMemory(ctxt);
2168
0
      return(-1);
2169
0
  }
2170
967k
  ctxt->spaceTab = tmp;
2171
2172
967k
        ctxt->spaceMax = newSize;
2173
967k
    }
2174
33.6M
    ctxt->spaceTab[ctxt->spaceNr] = val;
2175
33.6M
    ctxt->space = &ctxt->spaceTab[ctxt->spaceNr];
2176
33.6M
    return(ctxt->spaceNr++);
2177
33.6M
}
2178
2179
16.9M
static int spacePop(xmlParserCtxtPtr ctxt) {
2180
16.9M
    int ret;
2181
16.9M
    if (ctxt->spaceNr <= 0) return(0);
2182
16.9M
    ctxt->spaceNr--;
2183
16.9M
    if (ctxt->spaceNr > 0)
2184
16.9M
  ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1];
2185
5.01k
    else
2186
5.01k
        ctxt->space = &ctxt->spaceTab[0];
2187
16.9M
    ret = ctxt->spaceTab[ctxt->spaceNr];
2188
16.9M
    ctxt->spaceTab[ctxt->spaceNr] = -1;
2189
16.9M
    return(ret);
2190
16.9M
}
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
247M
#define RAW (*ctxt->input->cur)
2225
386M
#define CUR (*ctxt->input->cur)
2226
30.9M
#define NXT(val) ctxt->input->cur[(val)]
2227
586M
#define CUR_PTR ctxt->input->cur
2228
100M
#define BASE_PTR ctxt->input->base
2229
2230
#define CMP4( s, c1, c2, c3, c4 ) \
2231
5.10M
  ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \
2232
2.66M
    ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 )
2233
#define CMP5( s, c1, c2, c3, c4, c5 ) \
2234
4.77M
  ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 )
2235
#define CMP6( s, c1, c2, c3, c4, c5, c6 ) \
2236
4.48M
  ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 )
2237
#define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \
2238
4.09M
  ( 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
3.68M
  ( 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
1.83M
  ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \
2243
1.83M
    ((unsigned char *) s)[ 8 ] == c9 )
2244
#define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \
2245
97.4k
  ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \
2246
97.4k
    ((unsigned char *) s)[ 9 ] == c10 )
2247
2248
18.3M
#define SKIP(val) do {             \
2249
18.3M
    ctxt->input->cur += (val),ctxt->input->col+=(val);      \
2250
18.3M
    if (*ctxt->input->cur == 0)           \
2251
18.3M
        xmlParserGrow(ctxt);           \
2252
18.3M
  } 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
2.91M
    if (!PARSER_PROGRESSIVE(ctxt)) \
2268
2.91M
  xmlParserShrink(ctxt);
2269
2270
#define GROW \
2271
208M
    if ((!PARSER_PROGRESSIVE(ctxt)) && \
2272
208M
        (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \
2273
462k
  xmlParserGrow(ctxt);
2274
2275
99.7M
#define SKIP_BLANKS xmlSkipBlankChars(ctxt)
2276
2277
378k
#define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt)
2278
2279
72.1M
#define NEXT xmlNextChar(ctxt)
2280
2281
42.0M
#define NEXT1 {               \
2282
42.0M
  ctxt->input->col++;           \
2283
42.0M
  ctxt->input->cur++;           \
2284
42.0M
  if (*ctxt->input->cur == 0)         \
2285
42.0M
      xmlParserGrow(ctxt);           \
2286
42.0M
    }
2287
2288
424M
#define NEXTL(l) do {             \
2289
424M
    if (*(ctxt->input->cur) == '\n') {         \
2290
1.47M
  ctxt->input->line++; ctxt->input->col = 1;      \
2291
422M
    } else ctxt->input->col++;           \
2292
424M
    ctxt->input->cur += l;        \
2293
424M
  } while (0)
2294
2295
#define COPY_BUF(b, i, v)           \
2296
122M
    if (v < 0x80) b[i++] = v;           \
2297
122M
    else i += xmlCopyCharMultiByte(&b[i],v)
2298
2299
static int
2300
118M
xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) {
2301
118M
    int c = xmlCurrentChar(ctxt, len);
2302
2303
118M
    if (c == XML_INVALID_CHAR)
2304
552k
        c = 0xFFFD; /* replacement character */
2305
2306
118M
    return(c);
2307
118M
}
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
100M
xmlSkipBlankChars(xmlParserCtxt *ctxt) {
2319
100M
    const xmlChar *cur;
2320
100M
    int res = 0;
2321
2322
100M
    cur = ctxt->input->cur;
2323
100M
    while (IS_BLANK_CH(*cur)) {
2324
30.3M
        if (*cur == '\n') {
2325
829k
            ctxt->input->line++; ctxt->input->col = 1;
2326
29.5M
        } else {
2327
29.5M
            ctxt->input->col++;
2328
29.5M
        }
2329
30.3M
        cur++;
2330
30.3M
        if (res < INT_MAX)
2331
30.3M
            res++;
2332
30.3M
        if (*cur == 0) {
2333
10.1k
            ctxt->input->cur = cur;
2334
10.1k
            xmlParserGrow(ctxt);
2335
10.1k
            cur = ctxt->input->cur;
2336
10.1k
        }
2337
30.3M
    }
2338
100M
    ctxt->input->cur = cur;
2339
2340
100M
    if (res > 4)
2341
403k
        GROW;
2342
2343
100M
    return(res);
2344
100M
}
2345
2346
static void
2347
0
xmlPopPE(xmlParserCtxtPtr ctxt) {
2348
0
    unsigned long consumed;
2349
0
    xmlEntityPtr ent;
2350
2351
0
    ent = ctxt->input->entity;
2352
2353
0
    ent->flags &= ~XML_ENT_EXPANDING;
2354
2355
0
    if ((ent->flags & XML_ENT_CHECKED) == 0) {
2356
0
        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
0
        do {
2363
0
            ctxt->input->cur = ctxt->input->end;
2364
0
            xmlParserShrink(ctxt);
2365
0
            result = xmlParserGrow(ctxt);
2366
0
        } while (result > 0);
2367
2368
0
        consumed = ctxt->input->consumed;
2369
0
        xmlSaturatedAddSizeT(&consumed,
2370
0
                             ctxt->input->end - ctxt->input->base);
2371
2372
0
        xmlSaturatedAdd(&ent->expandedSize, consumed);
2373
2374
        /*
2375
         * Add to sizeentities when parsing an external entity
2376
         * for the first time.
2377
         */
2378
0
        if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
2379
0
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
2380
0
        }
2381
2382
0
        ent->flags |= XML_ENT_CHECKED;
2383
0
    }
2384
2385
0
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
2386
2387
0
    xmlParserEntityCheck(ctxt, ent->expandedSize);
2388
2389
0
    GROW;
2390
0
}
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
378k
xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) {
2401
378k
    int res = 0;
2402
378k
    int inParam;
2403
378k
    int expandParam;
2404
2405
378k
    inParam = PARSER_IN_PE(ctxt);
2406
378k
    expandParam = PARSER_EXTERNAL(ctxt);
2407
2408
378k
    if (!inParam && !expandParam)
2409
378k
        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
0
    while (PARSER_STOPPED(ctxt) == 0) {
2416
0
        if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */
2417
0
            NEXT;
2418
0
        } else if (CUR == '%') {
2419
0
            if ((expandParam == 0) ||
2420
0
                (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0))
2421
0
                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
0
            xmlParsePERefInternal(ctxt, 0);
2430
2431
0
            inParam = PARSER_IN_PE(ctxt);
2432
0
            expandParam = PARSER_EXTERNAL(ctxt);
2433
0
        } else if (CUR == 0) {
2434
0
            if (inParam == 0)
2435
0
                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
0
            if (ctxt->input->flags & XML_INPUT_MARKUP_DECL)
2443
0
                break;
2444
2445
0
            xmlPopPE(ctxt);
2446
2447
0
            inParam = PARSER_IN_PE(ctxt);
2448
0
            expandParam = PARSER_EXTERNAL(ctxt);
2449
0
        } else {
2450
0
            break;
2451
0
        }
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
0
        if (res < INT_MAX)
2461
0
            res++;
2462
0
    }
2463
2464
0
    return(res);
2465
378k
}
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
147k
xmlParseCharRef(xmlParserCtxt *ctxt) {
2530
147k
    int val = 0;
2531
147k
    int count = 0;
2532
2533
    /*
2534
     * Using RAW/CUR/NEXT is okay since we are working on ASCII range here
2535
     */
2536
147k
    if ((RAW == '&') && (NXT(1) == '#') &&
2537
147k
        (NXT(2) == 'x')) {
2538
88.8k
  SKIP(3);
2539
88.8k
  GROW;
2540
427k
  while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) {
2541
338k
      if (count++ > 20) {
2542
344
    count = 0;
2543
344
    GROW;
2544
344
      }
2545
338k
      if ((RAW >= '0') && (RAW <= '9'))
2546
150k
          val = val * 16 + (CUR - '0');
2547
188k
      else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20))
2548
150k
          val = val * 16 + (CUR - 'a') + 10;
2549
37.9k
      else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20))
2550
37.5k
          val = val * 16 + (CUR - 'A') + 10;
2551
395
      else {
2552
395
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2553
395
    val = 0;
2554
395
    break;
2555
395
      }
2556
338k
      if (val > 0x110000)
2557
4.24k
          val = 0x110000;
2558
2559
338k
      NEXT;
2560
338k
      count++;
2561
338k
  }
2562
88.8k
  if (RAW == ';') {
2563
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2564
88.4k
      ctxt->input->col++;
2565
88.4k
      ctxt->input->cur++;
2566
88.4k
  }
2567
88.8k
    } else if  ((RAW == '&') && (NXT(1) == '#')) {
2568
58.8k
  SKIP(2);
2569
58.8k
  GROW;
2570
196k
  while (RAW != ';') { /* loop blocked by count */
2571
138k
      if (count++ > 20) {
2572
389
    count = 0;
2573
389
    GROW;
2574
389
      }
2575
138k
      if ((RAW >= '0') && (RAW <= '9'))
2576
137k
          val = val * 10 + (CUR - '0');
2577
497
      else {
2578
497
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2579
497
    val = 0;
2580
497
    break;
2581
497
      }
2582
137k
      if (val > 0x110000)
2583
4.54k
          val = 0x110000;
2584
2585
137k
      NEXT;
2586
137k
      count++;
2587
137k
  }
2588
58.8k
  if (RAW == ';') {
2589
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2590
58.3k
      ctxt->input->col++;
2591
58.3k
      ctxt->input->cur++;
2592
58.3k
  }
2593
58.8k
    } 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
147k
    if (val >= 0x110000) {
2605
356
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2606
356
                "xmlParseCharRef: character reference out of bounds\n",
2607
356
          val);
2608
356
        val = 0xFFFD;
2609
147k
    } else if (!IS_CHAR(val)) {
2610
3.20k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2611
3.20k
                          "xmlParseCharRef: invalid xmlChar value %d\n",
2612
3.20k
                    val);
2613
3.20k
    }
2614
147k
    return(val);
2615
147k
}
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
46.1k
xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) {
2635
46.1k
    const xmlChar *ptr;
2636
46.1k
    xmlChar cur;
2637
46.1k
    int val = 0;
2638
2639
46.1k
    if ((str == NULL) || (*str == NULL)) return(0);
2640
46.1k
    ptr = *str;
2641
46.1k
    cur = *ptr;
2642
46.1k
    if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) {
2643
18.7k
  ptr += 3;
2644
18.7k
  cur = *ptr;
2645
73.5k
  while (cur != ';') { /* Non input consuming loop */
2646
59.4k
      if ((cur >= '0') && (cur <= '9'))
2647
29.5k
          val = val * 16 + (cur - '0');
2648
29.9k
      else if ((cur >= 'a') && (cur <= 'f'))
2649
16.3k
          val = val * 16 + (cur - 'a') + 10;
2650
13.5k
      else if ((cur >= 'A') && (cur <= 'F'))
2651
8.86k
          val = val * 16 + (cur - 'A') + 10;
2652
4.69k
      else {
2653
4.69k
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2654
4.69k
    val = 0;
2655
4.69k
    break;
2656
4.69k
      }
2657
54.7k
      if (val > 0x110000)
2658
3.75k
          val = 0x110000;
2659
2660
54.7k
      ptr++;
2661
54.7k
      cur = *ptr;
2662
54.7k
  }
2663
18.7k
  if (cur == ';')
2664
14.0k
      ptr++;
2665
27.3k
    } else if  ((cur == '&') && (ptr[1] == '#')){
2666
27.3k
  ptr += 2;
2667
27.3k
  cur = *ptr;
2668
155k
  while (cur != ';') { /* Non input consuming loops */
2669
130k
      if ((cur >= '0') && (cur <= '9'))
2670
127k
          val = val * 10 + (cur - '0');
2671
2.12k
      else {
2672
2.12k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2673
2.12k
    val = 0;
2674
2.12k
    break;
2675
2.12k
      }
2676
127k
      if (val > 0x110000)
2677
1.45k
          val = 0x110000;
2678
2679
127k
      ptr++;
2680
127k
      cur = *ptr;
2681
127k
  }
2682
27.3k
  if (cur == ';')
2683
25.2k
      ptr++;
2684
27.3k
    } else {
2685
0
  xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2686
0
  return(0);
2687
0
    }
2688
46.1k
    *str = ptr;
2689
2690
    /*
2691
     * [ WFC: Legal Character ]
2692
     * Characters referred to using character references must match the
2693
     * production for Char.
2694
     */
2695
46.1k
    if (val >= 0x110000) {
2696
215
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2697
215
                "xmlParseStringCharRef: character reference out of bounds\n",
2698
215
                val);
2699
45.9k
    } else if (IS_CHAR(val)) {
2700
38.0k
        return(val);
2701
38.0k
    } else {
2702
7.84k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2703
7.84k
        "xmlParseStringCharRef: invalid xmlChar value %d\n",
2704
7.84k
        val);
2705
7.84k
    }
2706
8.06k
    return(0);
2707
46.1k
}
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
11.6M
                     int blank_chars) {
2816
11.6M
    int i;
2817
11.6M
    xmlNodePtr lastChild;
2818
2819
    /*
2820
     * Check for xml:space value.
2821
     */
2822
11.6M
    if ((ctxt->space == NULL) || (*(ctxt->space) == 1) ||
2823
11.5M
        (*(ctxt->space) == -2))
2824
4.63M
  return(0);
2825
2826
    /*
2827
     * Check that the string is made of blanks
2828
     */
2829
7.01M
    if (blank_chars == 0) {
2830
7.37M
  for (i = 0;i < len;i++)
2831
7.24M
      if (!(IS_BLANK_CH(str[i]))) return(0);
2832
6.17M
    }
2833
2834
    /*
2835
     * Look if the element is mixed content in the DTD if available
2836
     */
2837
969k
    if (ctxt->node == NULL) return(0);
2838
18.4E
    if (ctxt->myDoc != NULL) {
2839
0
        xmlElementPtr elemDecl = NULL;
2840
0
        xmlDocPtr doc = ctxt->myDoc;
2841
0
        const xmlChar *prefix = NULL;
2842
2843
0
        if (ctxt->node->ns)
2844
0
            prefix = ctxt->node->ns->prefix;
2845
0
        if (doc->intSubset != NULL)
2846
0
            elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name,
2847
0
                                      prefix);
2848
0
        if ((elemDecl == NULL) && (doc->extSubset != NULL))
2849
0
            elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name,
2850
0
                                      prefix);
2851
0
        if (elemDecl != NULL) {
2852
0
            if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)
2853
0
                return(1);
2854
0
            if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) ||
2855
0
                (elemDecl->etype == XML_ELEMENT_TYPE_MIXED))
2856
0
                return(0);
2857
0
        }
2858
0
    }
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
18.4E
    if ((RAW != '<') && (RAW != 0xD)) return(0);
2868
18.4E
    if ((ctxt->node->children == NULL) &&
2869
0
  (RAW == '<') && (NXT(1) == '/')) return(0);
2870
2871
18.4E
    lastChild = xmlGetLastChild(ctxt->node);
2872
18.4E
    if (lastChild == NULL) {
2873
0
        if ((ctxt->node->type != XML_ELEMENT_NODE) &&
2874
0
            (ctxt->node->content != NULL)) return(0);
2875
18.4E
    } else if (xmlNodeIsText(lastChild))
2876
0
        return(0);
2877
18.4E
    else if ((ctxt->node->children != NULL) &&
2878
0
             (xmlNodeIsText(ctxt->node->children)))
2879
0
        return(0);
2880
18.4E
    return(1);
2881
18.4E
}
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
1.88M
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
1.88M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2953
1.87M
        (((c >= 'a') && (c <= 'z')) ||
2954
979k
         ((c >= 'A') && (c <= 'Z')) ||
2955
692k
         (c == '_') || (c == ':') ||
2956
486k
         ((c >= 0xC0) && (c <= 0xD6)) ||
2957
482k
         ((c >= 0xD8) && (c <= 0xF6)) ||
2958
471k
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2959
460k
         ((c >= 0x370) && (c <= 0x37D)) ||
2960
458k
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2961
448k
         ((c >= 0x200C) && (c <= 0x200D)) ||
2962
445k
         ((c >= 0x2070) && (c <= 0x218F)) ||
2963
433k
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2964
429k
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2965
413k
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2966
412k
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2967
281k
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2968
1.59M
        return(1);
2969
287k
    return(0);
2970
1.88M
}
2971
2972
static int
2973
36.3M
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
36.3M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2979
35.9M
        (((c >= 'a') && (c <= 'z')) ||
2980
20.4M
         ((c >= 'A') && (c <= 'Z')) ||
2981
15.8M
         ((c >= '0') && (c <= '9')) || /* !start */
2982
13.9M
         (c == '_') || (c == ':') ||
2983
13.0M
         (c == '-') || (c == '.') || (c == 0xB7) || /* !start */
2984
12.5M
         ((c >= 0xC0) && (c <= 0xD6)) ||
2985
12.4M
         ((c >= 0xD8) && (c <= 0xF6)) ||
2986
12.4M
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2987
12.1M
         ((c >= 0x300) && (c <= 0x36F)) || /* !start */
2988
12.1M
         ((c >= 0x370) && (c <= 0x37D)) ||
2989
12.1M
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2990
12.0M
         ((c >= 0x200C) && (c <= 0x200D)) ||
2991
12.0M
         ((c >= 0x203F) && (c <= 0x2040)) || /* !start */
2992
12.0M
         ((c >= 0x2070) && (c <= 0x218F)) ||
2993
12.0M
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2994
12.0M
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2995
11.8M
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2996
11.8M
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2997
322k
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2998
35.6M
         return(1);
2999
726k
    return(0);
3000
36.3M
}
3001
3002
static int
3003
0
xmlIsNameStartCharOld(int c) {
3004
0
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3005
0
        ((IS_LETTER(c) || (c == '_') || (c == ':'))))
3006
0
        return(1);
3007
0
    return(0);
3008
0
}
3009
3010
static int
3011
0
xmlIsNameCharOld(int c) {
3012
0
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3013
0
        ((IS_LETTER(c)) || (IS_DIGIT(c)) ||
3014
0
         (c == '.') || (c == '-') ||
3015
0
         (c == '_') || (c == ':') ||
3016
0
         (IS_COMBINING(c)) ||
3017
0
         (IS_EXTENDER(c))))
3018
0
        return(1);
3019
0
    return(0);
3020
0
}
3021
3022
static int
3023
1.88M
xmlIsNameStartChar(int c, int old10) {
3024
1.88M
    if (!old10)
3025
1.88M
        return(xmlIsNameStartCharNew(c));
3026
0
    else
3027
0
        return(xmlIsNameStartCharOld(c));
3028
1.88M
}
3029
3030
static int
3031
36.3M
xmlIsNameChar(int c, int old10) {
3032
36.3M
    if (!old10)
3033
36.3M
        return(xmlIsNameCharNew(c));
3034
0
    else
3035
0
        return(xmlIsNameCharOld(c));
3036
36.3M
}
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
66.9k
xmlScanName(const xmlChar *ptr, size_t maxSize, int flags) {
3052
66.9k
    int stop = flags & XML_SCAN_NC ? ':' : 0;
3053
66.9k
    int old10 = flags & XML_SCAN_OLD10 ? 1 : 0;
3054
3055
552k
    while (1) {
3056
552k
        int c, len;
3057
3058
552k
        c = *ptr;
3059
552k
        if (c < 0x80) {
3060
310k
            if (c == stop)
3061
1.63k
                break;
3062
309k
            len = 1;
3063
309k
        } else {
3064
242k
            len = 4;
3065
242k
            c = xmlGetUTF8Char(ptr, &len);
3066
242k
            if (c < 0)
3067
5.93k
                break;
3068
242k
        }
3069
3070
545k
        if (flags & XML_SCAN_NMTOKEN ?
3071
482k
                !xmlIsNameChar(c, old10) :
3072
545k
                !xmlIsNameStartChar(c, old10))
3073
59.3k
            break;
3074
3075
486k
        if ((size_t) len > maxSize)
3076
0
            return(NULL);
3077
486k
        ptr += len;
3078
486k
        maxSize -= len;
3079
486k
        flags |= XML_SCAN_NMTOKEN;
3080
486k
    }
3081
3082
66.9k
    return(ptr);
3083
66.9k
}
3084
3085
static const xmlChar *
3086
289k
xmlParseNameComplex(xmlParserCtxtPtr ctxt) {
3087
289k
    const xmlChar *ret;
3088
289k
    int len = 0, l;
3089
289k
    int c;
3090
289k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3091
288k
                    XML_MAX_TEXT_LENGTH :
3092
289k
                    XML_MAX_NAME_LENGTH;
3093
289k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3094
3095
    /*
3096
     * Handler for more complex cases
3097
     */
3098
289k
    c = xmlCurrentChar(ctxt, &l);
3099
289k
    if (!xmlIsNameStartChar(c, old10))
3100
168k
        return(NULL);
3101
121k
    len += l;
3102
121k
    NEXTL(l);
3103
121k
    c = xmlCurrentChar(ctxt, &l);
3104
917k
    while (xmlIsNameChar(c, old10)) {
3105
796k
        if (len <= INT_MAX - l)
3106
796k
            len += l;
3107
796k
        NEXTL(l);
3108
796k
        c = xmlCurrentChar(ctxt, &l);
3109
796k
    }
3110
121k
    if (len > maxLength) {
3111
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3112
0
        return(NULL);
3113
0
    }
3114
121k
    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
121k
    if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r'))
3125
783
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len);
3126
120k
    else
3127
120k
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len);
3128
121k
    if (ret == NULL)
3129
0
        xmlErrMemory(ctxt);
3130
121k
    return(ret);
3131
121k
}
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
1.02M
xmlParseName(xmlParserCtxt *ctxt) {
3151
1.02M
    const xmlChar *in;
3152
1.02M
    const xmlChar *ret;
3153
1.02M
    size_t count = 0;
3154
1.02M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3155
1.02M
                       XML_MAX_TEXT_LENGTH :
3156
1.02M
                       XML_MAX_NAME_LENGTH;
3157
3158
1.02M
    GROW;
3159
3160
    /*
3161
     * Accelerator for simple ASCII names
3162
     */
3163
1.02M
    in = ctxt->input->cur;
3164
1.02M
    if (((*in >= 0x61) && (*in <= 0x7A)) ||
3165
320k
  ((*in >= 0x41) && (*in <= 0x5A)) ||
3166
790k
  (*in == '_') || (*in == ':')) {
3167
790k
  in++;
3168
3.85M
  while (((*in >= 0x61) && (*in <= 0x7A)) ||
3169
1.56M
         ((*in >= 0x41) && (*in <= 0x5A)) ||
3170
1.27M
         ((*in >= 0x30) && (*in <= 0x39)) ||
3171
1.01M
         (*in == '_') || (*in == '-') ||
3172
865k
         (*in == ':') || (*in == '.'))
3173
3.06M
      in++;
3174
790k
  if ((*in > 0) && (*in < 0x80)) {
3175
739k
      count = in - ctxt->input->cur;
3176
739k
            if (count > maxLength) {
3177
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3178
0
                return(NULL);
3179
0
            }
3180
739k
      ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count);
3181
739k
      ctxt->input->cur = in;
3182
739k
      ctxt->input->col += count;
3183
739k
      if (ret == NULL)
3184
0
          xmlErrMemory(ctxt);
3185
739k
      return(ret);
3186
739k
  }
3187
790k
    }
3188
    /* accelerator for special cases */
3189
289k
    return(xmlParseNameComplex(ctxt));
3190
1.02M
}
3191
3192
static xmlHashedString
3193
1.61M
xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) {
3194
1.61M
    xmlHashedString ret;
3195
1.61M
    int len = 0, l;
3196
1.61M
    int c;
3197
1.61M
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3198
1.56M
                    XML_MAX_TEXT_LENGTH :
3199
1.61M
                    XML_MAX_NAME_LENGTH;
3200
1.61M
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3201
1.61M
    size_t startPosition = 0;
3202
3203
1.61M
    ret.name = NULL;
3204
1.61M
    ret.hashValue = 0;
3205
3206
    /*
3207
     * Handler for more complex cases
3208
     */
3209
1.61M
    startPosition = CUR_PTR - BASE_PTR;
3210
1.61M
    c = xmlCurrentChar(ctxt, &l);
3211
1.61M
    if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */
3212
1.52M
  (!xmlIsNameStartChar(c, old10) || (c == ':'))) {
3213
392k
  return(ret);
3214
392k
    }
3215
3216
30.9M
    while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */
3217
29.8M
     (xmlIsNameChar(c, old10) && (c != ':'))) {
3218
29.7M
        if (len <= INT_MAX - l)
3219
29.7M
      len += l;
3220
29.7M
  NEXTL(l);
3221
29.7M
  c = xmlCurrentChar(ctxt, &l);
3222
29.7M
    }
3223
1.22M
    if (len > maxLength) {
3224
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3225
0
        return(ret);
3226
0
    }
3227
1.22M
    ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len);
3228
1.22M
    if (ret.name == NULL)
3229
0
        xmlErrMemory(ctxt);
3230
1.22M
    return(ret);
3231
1.22M
}
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
76.1M
xmlParseNCName(xmlParserCtxtPtr ctxt) {
3247
76.1M
    const xmlChar *in, *e;
3248
76.1M
    xmlHashedString ret;
3249
76.1M
    size_t count = 0;
3250
76.1M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3251
73.3M
                       XML_MAX_TEXT_LENGTH :
3252
76.1M
                       XML_MAX_NAME_LENGTH;
3253
3254
76.1M
    ret.name = NULL;
3255
3256
    /*
3257
     * Accelerator for simple ASCII names
3258
     */
3259
76.1M
    in = ctxt->input->cur;
3260
76.1M
    e = ctxt->input->end;
3261
76.1M
    if ((((*in >= 0x61) && (*in <= 0x7A)) ||
3262
5.67M
   ((*in >= 0x41) && (*in <= 0x5A)) ||
3263
75.6M
   (*in == '_')) && (in < e)) {
3264
75.6M
  in++;
3265
427M
  while ((((*in >= 0x61) && (*in <= 0x7A)) ||
3266
140M
          ((*in >= 0x41) && (*in <= 0x5A)) ||
3267
95.3M
          ((*in >= 0x30) && (*in <= 0x39)) ||
3268
83.2M
          (*in == '_') || (*in == '-') ||
3269
352M
          (*in == '.')) && (in < e))
3270
352M
      in++;
3271
75.6M
  if (in >= e)
3272
17.7k
      goto complex;
3273
75.6M
  if ((*in > 0) && (*in < 0x80)) {
3274
74.5M
      count = in - ctxt->input->cur;
3275
74.5M
            if (count > maxLength) {
3276
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3277
0
                return(ret);
3278
0
            }
3279
74.5M
      ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count);
3280
74.5M
      ctxt->input->cur = in;
3281
74.5M
      ctxt->input->col += count;
3282
74.5M
      if (ret.name == NULL) {
3283
0
          xmlErrMemory(ctxt);
3284
0
      }
3285
74.5M
      return(ret);
3286
74.5M
  }
3287
75.6M
    }
3288
1.61M
complex:
3289
1.61M
    return(xmlParseNCNameComplex(ctxt));
3290
76.1M
}
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
2.63M
xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) {
3304
2.63M
    register const xmlChar *cmp = other;
3305
2.63M
    register const xmlChar *in;
3306
2.63M
    const xmlChar *ret;
3307
3308
2.63M
    GROW;
3309
3310
2.63M
    in = ctxt->input->cur;
3311
8.61M
    while (*in != 0 && *in == *cmp) {
3312
5.97M
  ++in;
3313
5.97M
  ++cmp;
3314
5.97M
    }
3315
2.63M
    if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
3316
  /* success */
3317
2.63M
  ctxt->input->col += in - ctxt->input->cur;
3318
2.63M
  ctxt->input->cur = in;
3319
2.63M
  return (const xmlChar*) 1;
3320
2.63M
    }
3321
    /* failure (or end of input buffer), check with full function */
3322
5.37k
    ret = xmlParseName (ctxt);
3323
    /* strings coming from the dictionary direct compare possible */
3324
5.37k
    if (ret == other) {
3325
58
  return (const xmlChar*) 1;
3326
58
    }
3327
5.31k
    return ret;
3328
5.37k
}
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
58.0k
xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) {
3341
58.0k
    xmlChar *ret;
3342
58.0k
    const xmlChar *cur = *str;
3343
58.0k
    int flags = 0;
3344
58.0k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3345
58.0k
                    XML_MAX_TEXT_LENGTH :
3346
58.0k
                    XML_MAX_NAME_LENGTH;
3347
3348
58.0k
    if (ctxt->options & XML_PARSE_OLD10)
3349
0
        flags |= XML_SCAN_OLD10;
3350
3351
58.0k
    cur = xmlScanName(*str, maxLength, flags);
3352
58.0k
    if (cur == NULL) {
3353
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3354
0
        return(NULL);
3355
0
    }
3356
58.0k
    if (cur == *str)
3357
11.9k
        return(NULL);
3358
3359
46.1k
    ret = xmlStrndup(*str, cur - *str);
3360
46.1k
    if (ret == NULL)
3361
0
        xmlErrMemory(ctxt);
3362
46.1k
    *str = cur;
3363
46.1k
    return(ret);
3364
58.0k
}
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
507k
xmlParseNmtoken(xmlParserCtxt *ctxt) {
3381
507k
    xmlChar buf[XML_MAX_NAMELEN + 5];
3382
507k
    xmlChar *ret;
3383
507k
    int len = 0, l;
3384
507k
    int c;
3385
507k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3386
501k
                    XML_MAX_TEXT_LENGTH :
3387
507k
                    XML_MAX_NAME_LENGTH;
3388
507k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3389
3390
507k
    c = xmlCurrentChar(ctxt, &l);
3391
3392
4.56M
    while (xmlIsNameChar(c, old10)) {
3393
4.06M
  COPY_BUF(buf, len, c);
3394
4.06M
  NEXTL(l);
3395
4.06M
  c = xmlCurrentChar(ctxt, &l);
3396
4.06M
  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
4.25k
      xmlChar *buffer;
3402
4.25k
      int max = len * 2;
3403
3404
4.25k
      buffer = xmlMalloc(max);
3405
4.25k
      if (buffer == NULL) {
3406
0
          xmlErrMemory(ctxt);
3407
0
    return(NULL);
3408
0
      }
3409
4.25k
      memcpy(buffer, buf, len);
3410
488k
      while (xmlIsNameChar(c, old10)) {
3411
484k
    if (len + 10 > max) {
3412
1.59k
        xmlChar *tmp;
3413
1.59k
                    int newSize;
3414
3415
1.59k
                    newSize = xmlGrowCapacity(max, 1, 1, maxLength);
3416
1.59k
                    if (newSize < 0) {
3417
0
                        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3418
0
                        xmlFree(buffer);
3419
0
                        return(NULL);
3420
0
                    }
3421
1.59k
        tmp = xmlRealloc(buffer, newSize);
3422
1.59k
        if (tmp == NULL) {
3423
0
      xmlErrMemory(ctxt);
3424
0
      xmlFree(buffer);
3425
0
      return(NULL);
3426
0
        }
3427
1.59k
        buffer = tmp;
3428
1.59k
                    max = newSize;
3429
1.59k
    }
3430
484k
    COPY_BUF(buffer, len, c);
3431
484k
    NEXTL(l);
3432
484k
    c = xmlCurrentChar(ctxt, &l);
3433
484k
      }
3434
4.25k
      buffer[len] = 0;
3435
4.25k
      return(buffer);
3436
4.25k
  }
3437
4.06M
    }
3438
503k
    if (len == 0)
3439
79.8k
        return(NULL);
3440
423k
    if (len > maxLength) {
3441
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3442
0
        return(NULL);
3443
0
    }
3444
423k
    ret = xmlStrndup(buf, len);
3445
423k
    if (ret == NULL)
3446
0
        xmlErrMemory(ctxt);
3447
423k
    return(ret);
3448
423k
}
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
43.2k
                          const xmlChar *str, int length, int depth) {
3462
43.2k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3463
43.2k
    const xmlChar *end, *chunk;
3464
43.2k
    int c, l;
3465
3466
43.2k
    if (str == NULL)
3467
0
        return;
3468
3469
43.2k
    depth += 1;
3470
43.2k
    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
43.2k
    end = str + length;
3477
43.2k
    chunk = str;
3478
3479
2.45M
    while ((str < end) && (!PARSER_STOPPED(ctxt))) {
3480
2.43M
        c = *str;
3481
3482
2.43M
        if (c >= 0x80) {
3483
414k
            l = xmlUTF8MultibyteLen(ctxt, str,
3484
414k
                    "invalid character in entity value\n");
3485
414k
            if (l == 0) {
3486
210k
                if (chunk < str)
3487
25.6k
                    xmlSBufAddString(buf, chunk, str - chunk);
3488
210k
                xmlSBufAddReplChar(buf);
3489
210k
                str += 1;
3490
210k
                chunk = str;
3491
210k
            } else {
3492
204k
                str += l;
3493
204k
            }
3494
2.02M
        } else if (c == '&') {
3495
100k
            if (str[1] == '#') {
3496
46.1k
                if (chunk < str)
3497
29.0k
                    xmlSBufAddString(buf, chunk, str - chunk);
3498
3499
46.1k
                c = xmlParseStringCharRef(ctxt, &str);
3500
46.1k
                if (c == 0)
3501
8.06k
                    return;
3502
3503
38.0k
                xmlSBufAddChar(buf, c);
3504
3505
38.0k
                chunk = str;
3506
54.4k
            } else {
3507
54.4k
                xmlChar *name;
3508
3509
                /*
3510
                 * General entity references are checked for
3511
                 * syntactic validity.
3512
                 */
3513
54.4k
                str++;
3514
54.4k
                name = xmlParseStringName(ctxt, &str);
3515
3516
54.4k
                if ((name == NULL) || (*str++ != ';')) {
3517
19.4k
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR,
3518
19.4k
                            "EntityValue: '&' forbidden except for entities "
3519
19.4k
                            "references\n");
3520
19.4k
                    xmlFree(name);
3521
19.4k
                    return;
3522
19.4k
                }
3523
3524
35.0k
                xmlFree(name);
3525
35.0k
            }
3526
1.92M
        } else if (c == '%') {
3527
3.62k
            xmlEntityPtr ent;
3528
3529
3.62k
            if (chunk < str)
3530
3.23k
                xmlSBufAddString(buf, chunk, str - chunk);
3531
3532
3.62k
            ent = xmlParseStringPEReference(ctxt, &str);
3533
3.62k
            if (ent == NULL)
3534
3.62k
                return;
3535
3536
0
            if (!PARSER_EXTERNAL(ctxt)) {
3537
0
                xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL);
3538
0
                return;
3539
0
            }
3540
3541
0
            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
0
                if (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
3549
0
                    ((ctxt->replaceEntities) ||
3550
0
                     (ctxt->validate))) {
3551
0
                    xmlLoadEntityContent(ctxt, ent);
3552
0
                } else {
3553
0
                    xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING,
3554
0
                                  "not validating will not read content for "
3555
0
                                  "PE entity %s\n", ent->name, NULL);
3556
0
                }
3557
0
            }
3558
3559
            /*
3560
             * TODO: Skip if ent->content is still NULL.
3561
             */
3562
3563
0
            if (xmlParserEntityCheck(ctxt, ent->length))
3564
0
                return;
3565
3566
0
            if (ent->flags & XML_ENT_EXPANDING) {
3567
0
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3568
0
                return;
3569
0
            }
3570
3571
0
            ent->flags |= XML_ENT_EXPANDING;
3572
0
            xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length,
3573
0
                                      depth);
3574
0
            ent->flags &= ~XML_ENT_EXPANDING;
3575
3576
0
            chunk = str;
3577
1.92M
        } else {
3578
            /* Normal ASCII char */
3579
1.92M
            if (!IS_BYTE_CHAR(c)) {
3580
240k
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3581
240k
                        "invalid character in entity value\n");
3582
240k
                if (chunk < str)
3583
11.2k
                    xmlSBufAddString(buf, chunk, str - chunk);
3584
240k
                xmlSBufAddReplChar(buf);
3585
240k
                str += 1;
3586
240k
                chunk = str;
3587
1.67M
            } else {
3588
1.67M
                str += 1;
3589
1.67M
            }
3590
1.92M
        }
3591
2.43M
    }
3592
3593
12.0k
    if (chunk < str)
3594
8.34k
        xmlSBufAddString(buf, chunk, str - chunk);
3595
12.0k
}
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
43.6k
xmlParseEntityValue(xmlParserCtxt *ctxt, xmlChar **orig) {
3611
43.6k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3612
43.6k
                         XML_MAX_HUGE_LENGTH :
3613
43.6k
                         XML_MAX_TEXT_LENGTH;
3614
43.6k
    xmlSBuf buf;
3615
43.6k
    const xmlChar *start;
3616
43.6k
    int quote, length;
3617
3618
43.6k
    xmlSBufInit(&buf, maxLength);
3619
3620
43.6k
    GROW;
3621
3622
43.6k
    quote = CUR;
3623
43.6k
    if ((quote != '"') && (quote != '\'')) {
3624
0
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
3625
0
  return(NULL);
3626
0
    }
3627
43.6k
    CUR_PTR++;
3628
3629
43.6k
    length = 0;
3630
3631
    /*
3632
     * Copy raw content of the entity into a buffer
3633
     */
3634
6.82M
    while (1) {
3635
6.82M
        int c;
3636
3637
6.82M
        if (PARSER_STOPPED(ctxt))
3638
0
            goto error;
3639
3640
6.82M
        if (CUR_PTR >= ctxt->input->end) {
3641
373
            xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL);
3642
373
            goto error;
3643
373
        }
3644
3645
6.82M
        c = CUR;
3646
3647
6.82M
        if (c == 0) {
3648
81
            xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3649
81
                    "invalid character in entity value\n");
3650
81
            goto error;
3651
81
        }
3652
6.82M
        if (c == quote)
3653
43.2k
            break;
3654
6.78M
        NEXTL(1);
3655
6.78M
        length += 1;
3656
3657
        /*
3658
         * TODO: Check growth threshold
3659
         */
3660
6.78M
        if (ctxt->input->end - CUR_PTR < 10)
3661
4.36k
            GROW;
3662
6.78M
    }
3663
3664
43.2k
    start = CUR_PTR - length;
3665
3666
43.2k
    if (orig != NULL) {
3667
43.2k
        *orig = xmlStrndup(start, length);
3668
43.2k
        if (*orig == NULL)
3669
0
            xmlErrMemory(ctxt);
3670
43.2k
    }
3671
3672
43.2k
    xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr);
3673
3674
43.2k
    NEXTL(1);
3675
3676
43.2k
    return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long"));
3677
3678
454
error:
3679
454
    xmlSBufCleanup(&buf, ctxt, "entity length too long");
3680
454
    return(NULL);
3681
43.6k
}
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
0
xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) {
3693
0
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3694
0
    const xmlChar *str;
3695
0
    unsigned long expandedSize = pent->length;
3696
0
    int c, flags;
3697
3698
0
    depth += 1;
3699
0
    if (depth > maxDepth) {
3700
0
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3701
0
                       "Maximum entity nesting depth exceeded");
3702
0
  return;
3703
0
    }
3704
3705
0
    if (pent->flags & XML_ENT_EXPANDING) {
3706
0
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3707
0
        return;
3708
0
    }
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
0
    if (ctxt->inSubset == 0)
3716
0
        flags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
3717
0
    else
3718
0
        flags = XML_ENT_VALIDATED;
3719
3720
0
    str = pent->content;
3721
0
    if (str == NULL)
3722
0
        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
0
    while (!PARSER_STOPPED(ctxt)) {
3731
0
        c = *str;
3732
3733
0
  if (c != '&') {
3734
0
            if (c == 0)
3735
0
                break;
3736
3737
0
            if (c == '<')
3738
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3739
0
                        "'<' in entity '%s' is not allowed in attributes "
3740
0
                        "values\n", pent->name);
3741
3742
0
            str += 1;
3743
0
        } else if (str[1] == '#') {
3744
0
            int val;
3745
3746
0
      val = xmlParseStringCharRef(ctxt, &str);
3747
0
      if (val == 0) {
3748
0
                pent->content[0] = 0;
3749
0
                break;
3750
0
            }
3751
0
  } else {
3752
0
            xmlChar *name;
3753
0
            xmlEntityPtr ent;
3754
3755
0
      name = xmlParseStringEntityRef(ctxt, &str);
3756
0
      if (name == NULL) {
3757
0
                pent->content[0] = 0;
3758
0
                break;
3759
0
            }
3760
3761
0
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3762
0
            xmlFree(name);
3763
3764
0
            if ((ent != NULL) &&
3765
0
                (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) {
3766
0
                if ((ent->flags & flags) != flags) {
3767
0
                    pent->flags |= XML_ENT_EXPANDING;
3768
0
                    xmlCheckEntityInAttValue(ctxt, ent, depth);
3769
0
                    pent->flags &= ~XML_ENT_EXPANDING;
3770
0
                }
3771
3772
0
                xmlSaturatedAdd(&expandedSize, ent->expandedSize);
3773
0
                xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST);
3774
0
            }
3775
0
        }
3776
0
    }
3777
3778
0
done:
3779
0
    if (ctxt->inSubset == 0)
3780
0
        pent->expandedSize = expandedSize;
3781
3782
0
    pent->flags |= flags;
3783
0
}
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
0
                          int *inSpace, int depth, int check) {
3803
0
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3804
0
    int c, chunkSize;
3805
0
    int normChange = 0;
3806
3807
0
    if (str == NULL)
3808
0
        return(0);
3809
3810
0
    depth += 1;
3811
0
    if (depth > maxDepth) {
3812
0
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3813
0
                       "Maximum entity nesting depth exceeded");
3814
0
  return(0);
3815
0
    }
3816
3817
0
    if (pent != NULL) {
3818
0
        if (pent->flags & XML_ENT_EXPANDING) {
3819
0
            xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3820
0
            return(0);
3821
0
        }
3822
3823
0
        if (check) {
3824
0
            if (xmlParserEntityCheck(ctxt, pent->length))
3825
0
                return(0);
3826
0
        }
3827
0
    }
3828
3829
0
    chunkSize = 0;
3830
3831
    /*
3832
     * Note that entity values are already validated. No special
3833
     * handling for multi-byte characters is needed.
3834
     */
3835
0
    while (!PARSER_STOPPED(ctxt)) {
3836
0
        c = *str;
3837
3838
0
  if (c != '&') {
3839
0
            if (c == 0)
3840
0
                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
0
            if ((pent != NULL) && (c == '<')) {
3848
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3849
0
                        "'<' in entity '%s' is not allowed in attributes "
3850
0
                        "values\n", pent->name);
3851
0
                break;
3852
0
            }
3853
3854
0
            if (c <= 0x20) {
3855
0
                if ((normalize) && (*inSpace)) {
3856
                    /* Skip char */
3857
0
                    if (chunkSize > 0) {
3858
0
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3859
0
                        chunkSize = 0;
3860
0
                    }
3861
0
                    normChange = 1;
3862
0
                } else if (c < 0x20) {
3863
0
                    if (chunkSize > 0) {
3864
0
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3865
0
                        chunkSize = 0;
3866
0
                    }
3867
3868
0
                    xmlSBufAddCString(buf, " ", 1);
3869
0
                } else {
3870
0
                    chunkSize += 1;
3871
0
                }
3872
3873
0
                *inSpace = 1;
3874
0
            } else {
3875
0
                chunkSize += 1;
3876
0
                *inSpace = 0;
3877
0
            }
3878
3879
0
            str += 1;
3880
0
        } else if (str[1] == '#') {
3881
0
            int val;
3882
3883
0
            if (chunkSize > 0) {
3884
0
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3885
0
                chunkSize = 0;
3886
0
            }
3887
3888
0
      val = xmlParseStringCharRef(ctxt, &str);
3889
0
      if (val == 0) {
3890
0
                if (pent != NULL)
3891
0
                    pent->content[0] = 0;
3892
0
                break;
3893
0
            }
3894
3895
0
            if (val == ' ') {
3896
0
                if ((normalize) && (*inSpace))
3897
0
                    normChange = 1;
3898
0
                else
3899
0
                    xmlSBufAddCString(buf, " ", 1);
3900
0
                *inSpace = 1;
3901
0
            } else {
3902
0
                xmlSBufAddChar(buf, val);
3903
0
                *inSpace = 0;
3904
0
            }
3905
0
  } else {
3906
0
            xmlChar *name;
3907
0
            xmlEntityPtr ent;
3908
3909
0
            if (chunkSize > 0) {
3910
0
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3911
0
                chunkSize = 0;
3912
0
            }
3913
3914
0
      name = xmlParseStringEntityRef(ctxt, &str);
3915
0
            if (name == NULL) {
3916
0
                if (pent != NULL)
3917
0
                    pent->content[0] = 0;
3918
0
                break;
3919
0
            }
3920
3921
0
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3922
0
            xmlFree(name);
3923
3924
0
      if ((ent != NULL) &&
3925
0
    (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
3926
0
    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
0
                xmlSBufAddString(buf, ent->content, ent->length);
3933
3934
0
                *inSpace = 0;
3935
0
      } else if ((ent != NULL) && (ent->content != NULL)) {
3936
0
                if (pent != NULL)
3937
0
                    pent->flags |= XML_ENT_EXPANDING;
3938
0
    normChange |= xmlExpandEntityInAttValue(ctxt, buf,
3939
0
                        ent->content, ent, normalize, inSpace, depth, check);
3940
0
                if (pent != NULL)
3941
0
                    pent->flags &= ~XML_ENT_EXPANDING;
3942
0
      }
3943
0
        }
3944
0
    }
3945
3946
0
    if (chunkSize > 0)
3947
0
        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3948
3949
0
    return(normChange);
3950
0
}
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
21.9M
                         int special, int isNamespace) {
4020
21.9M
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4021
20.9M
                         XML_MAX_HUGE_LENGTH :
4022
21.9M
                         XML_MAX_TEXT_LENGTH;
4023
21.9M
    xmlSBuf buf;
4024
21.9M
    xmlChar *ret;
4025
21.9M
    int c, l, quote, entFlags, chunkSize;
4026
21.9M
    int inSpace = 1;
4027
21.9M
    int replaceEntities;
4028
21.9M
    int normalize = (special & XML_SPECIAL_TYPE_MASK) > XML_ATTRIBUTE_CDATA;
4029
21.9M
    int attvalFlags = 0;
4030
4031
    /* Always expand namespace URIs */
4032
21.9M
    replaceEntities = (ctxt->replaceEntities) || (isNamespace);
4033
4034
21.9M
    xmlSBufInit(&buf, maxLength);
4035
4036
21.9M
    GROW;
4037
4038
21.9M
    quote = CUR;
4039
21.9M
    if ((quote != '"') && (quote != '\'')) {
4040
3.60k
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
4041
3.60k
  return(NULL);
4042
3.60k
    }
4043
21.9M
    NEXTL(1);
4044
4045
21.9M
    if (ctxt->inSubset == 0)
4046
21.9M
        entFlags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
4047
38.1k
    else
4048
38.1k
        entFlags = XML_ENT_VALIDATED;
4049
4050
21.9M
    inSpace = 1;
4051
21.9M
    chunkSize = 0;
4052
4053
243M
    while (1) {
4054
243M
        if (PARSER_STOPPED(ctxt))
4055
0
            goto error;
4056
4057
243M
        if (CUR_PTR >= ctxt->input->end) {
4058
5.26k
            xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED,
4059
5.26k
                           "AttValue: ' expected\n");
4060
5.26k
            goto error;
4061
5.26k
        }
4062
4063
        /*
4064
         * TODO: Check growth threshold
4065
         */
4066
243M
        if (ctxt->input->end - CUR_PTR < 10)
4067
54.5k
            GROW;
4068
4069
243M
        c = CUR;
4070
4071
243M
        if (c >= 0x80) {
4072
5.69M
            l = xmlUTF8MultibyteLen(ctxt, CUR_PTR,
4073
5.69M
                    "invalid character in attribute value\n");
4074
5.69M
            if (l == 0) {
4075
1.92M
                if (chunkSize > 0) {
4076
217k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4077
217k
                    chunkSize = 0;
4078
217k
                }
4079
1.92M
                xmlSBufAddReplChar(&buf);
4080
1.92M
                NEXTL(1);
4081
3.77M
            } else {
4082
3.77M
                chunkSize += l;
4083
3.77M
                NEXTL(l);
4084
3.77M
            }
4085
4086
5.69M
            inSpace = 0;
4087
237M
        } else if (c != '&') {
4088
237M
            if (c > 0x20) {
4089
233M
                if (c == quote)
4090
21.9M
                    break;
4091
4092
211M
                if (c == '<')
4093
187k
                    xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL);
4094
4095
211M
                chunkSize += 1;
4096
211M
                inSpace = 0;
4097
211M
            } else if (!IS_BYTE_CHAR(c)) {
4098
1.23M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
4099
1.23M
                        "invalid character in attribute value\n");
4100
1.23M
                if (chunkSize > 0) {
4101
69.2k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4102
69.2k
                    chunkSize = 0;
4103
69.2k
                }
4104
1.23M
                xmlSBufAddReplChar(&buf);
4105
1.23M
                inSpace = 0;
4106
2.62M
            } else {
4107
                /* Whitespace */
4108
2.62M
                if ((normalize) && (inSpace)) {
4109
                    /* Skip char */
4110
28.3k
                    if (chunkSize > 0) {
4111
2.11k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4112
2.11k
                        chunkSize = 0;
4113
2.11k
                    }
4114
28.3k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4115
2.59M
                } else if (c < 0x20) {
4116
                    /* Convert to space */
4117
762k
                    if (chunkSize > 0) {
4118
167k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4119
167k
                        chunkSize = 0;
4120
167k
                    }
4121
4122
762k
                    xmlSBufAddCString(&buf, " ", 1);
4123
1.83M
                } else {
4124
1.83M
                    chunkSize += 1;
4125
1.83M
                }
4126
4127
2.62M
                inSpace = 1;
4128
4129
2.62M
                if ((c == 0xD) && (NXT(1) == 0xA))
4130
22.8k
                    CUR_PTR++;
4131
2.62M
            }
4132
4133
215M
            NEXTL(1);
4134
215M
        } else if (NXT(1) == '#') {
4135
76.0k
            int val;
4136
4137
76.0k
            if (chunkSize > 0) {
4138
35.2k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4139
35.2k
                chunkSize = 0;
4140
35.2k
            }
4141
4142
76.0k
            val = xmlParseCharRef(ctxt);
4143
76.0k
            if (val == 0)
4144
1.00k
                goto error;
4145
4146
75.0k
            if ((val == '&') && (!replaceEntities)) {
4147
                /*
4148
                 * The reparsing will be done in xmlNodeParseContent()
4149
                 * called from SAX2.c
4150
                 */
4151
0
                xmlSBufAddCString(&buf, "&#38;", 5);
4152
0
                inSpace = 0;
4153
75.0k
            } else if (val == ' ') {
4154
43.3k
                if ((normalize) && (inSpace))
4155
7.31k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4156
36.0k
                else
4157
36.0k
                    xmlSBufAddCString(&buf, " ", 1);
4158
43.3k
                inSpace = 1;
4159
43.3k
            } else {
4160
31.6k
                xmlSBufAddChar(&buf, val);
4161
31.6k
                inSpace = 0;
4162
31.6k
            }
4163
461k
        } else {
4164
461k
            const xmlChar *name;
4165
461k
            xmlEntityPtr ent;
4166
4167
461k
            if (chunkSize > 0) {
4168
239k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4169
239k
                chunkSize = 0;
4170
239k
            }
4171
4172
461k
            name = xmlParseEntityRefInternal(ctxt);
4173
461k
            if (name == NULL) {
4174
                /*
4175
                 * Probably a literal '&' which wasn't escaped.
4176
                 * TODO: Handle gracefully in recovery mode.
4177
                 */
4178
284k
                continue;
4179
284k
            }
4180
4181
177k
            ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1);
4182
177k
            if (ent == NULL)
4183
21.1k
                continue;
4184
4185
156k
            if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) {
4186
155k
                if ((ent->content[0] == '&') && (!replaceEntities))
4187
0
                    xmlSBufAddCString(&buf, "&#38;", 5);
4188
155k
                else
4189
155k
                    xmlSBufAddString(&buf, ent->content, ent->length);
4190
155k
                inSpace = 0;
4191
155k
            } else if (replaceEntities) {
4192
0
                if (xmlExpandEntityInAttValue(ctxt, &buf,
4193
0
                        ent->content, ent, normalize, &inSpace, ctxt->inputNr,
4194
0
                        /* check */ 1) > 0)
4195
0
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4196
62
            } else {
4197
62
                if ((ent->flags & entFlags) != entFlags)
4198
0
                    xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr);
4199
4200
62
                if (xmlParserEntityCheck(ctxt, ent->expandedSize)) {
4201
0
                    ent->content[0] = 0;
4202
0
                    goto error;
4203
0
                }
4204
4205
                /*
4206
                 * Just output the reference
4207
                 */
4208
62
                xmlSBufAddCString(&buf, "&", 1);
4209
62
                xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name));
4210
62
                xmlSBufAddCString(&buf, ";", 1);
4211
4212
62
                inSpace = 0;
4213
62
            }
4214
156k
  }
4215
243M
    }
4216
4217
21.9M
    if ((buf.mem == NULL) && (outFlags != NULL)) {
4218
21.7M
        ret = (xmlChar *) CUR_PTR - chunkSize;
4219
4220
21.7M
        if (attlen != NULL)
4221
21.7M
            *attlen = chunkSize;
4222
21.7M
        if ((normalize) && (inSpace) && (chunkSize > 0)) {
4223
243
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4224
243
            *attlen -= 1;
4225
243
        }
4226
4227
        /* Report potential error */
4228
21.7M
        xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4229
21.7M
    } else {
4230
231k
        if (chunkSize > 0)
4231
159k
            xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4232
4233
231k
        if ((normalize) && (inSpace) && (buf.size > 0)) {
4234
3.38k
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4235
3.38k
            buf.size--;
4236
3.38k
        }
4237
4238
231k
        ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long");
4239
231k
        attvalFlags |= XML_ATTVAL_ALLOC;
4240
4241
231k
        if (ret != NULL) {
4242
229k
            if (attlen != NULL)
4243
191k
                *attlen = buf.size;
4244
229k
        }
4245
231k
    }
4246
4247
21.9M
    if (outFlags != NULL)
4248
21.9M
        *outFlags = attvalFlags;
4249
4250
21.9M
    NEXTL(1);
4251
4252
21.9M
    return(ret);
4253
4254
6.26k
error:
4255
6.26k
    xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4256
6.26k
    return(NULL);
4257
21.9M
}
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
38.3k
xmlParseAttValue(xmlParserCtxt *ctxt) {
4297
38.3k
    if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL);
4298
38.3k
    return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0));
4299
38.3k
}
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
8.17k
xmlParseSystemLiteral(xmlParserCtxt *ctxt) {
4314
8.17k
    xmlChar *buf = NULL;
4315
8.17k
    int len = 0;
4316
8.17k
    int size = XML_PARSER_BUFFER_SIZE;
4317
8.17k
    int cur, l;
4318
8.17k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4319
8.17k
                    XML_MAX_TEXT_LENGTH :
4320
8.17k
                    XML_MAX_NAME_LENGTH;
4321
8.17k
    xmlChar stop;
4322
4323
8.17k
    if (RAW == '"') {
4324
5.99k
        NEXT;
4325
5.99k
  stop = '"';
4326
5.99k
    } else if (RAW == '\'') {
4327
1.56k
        NEXT;
4328
1.56k
  stop = '\'';
4329
1.56k
    } else {
4330
617
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4331
617
  return(NULL);
4332
617
    }
4333
4334
7.56k
    buf = xmlMalloc(size);
4335
7.56k
    if (buf == NULL) {
4336
0
        xmlErrMemory(ctxt);
4337
0
  return(NULL);
4338
0
    }
4339
7.56k
    cur = xmlCurrentCharRecover(ctxt, &l);
4340
423k
    while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */
4341
416k
  if (len + 5 >= size) {
4342
2.89k
      xmlChar *tmp;
4343
2.89k
            int newSize;
4344
4345
2.89k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4346
2.89k
            if (newSize < 0) {
4347
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral");
4348
0
                xmlFree(buf);
4349
0
                return(NULL);
4350
0
            }
4351
2.89k
      tmp = xmlRealloc(buf, newSize);
4352
2.89k
      if (tmp == NULL) {
4353
0
          xmlFree(buf);
4354
0
    xmlErrMemory(ctxt);
4355
0
    return(NULL);
4356
0
      }
4357
2.89k
      buf = tmp;
4358
2.89k
            size = newSize;
4359
2.89k
  }
4360
416k
  COPY_BUF(buf, len, cur);
4361
416k
  NEXTL(l);
4362
416k
  cur = xmlCurrentCharRecover(ctxt, &l);
4363
416k
    }
4364
7.56k
    buf[len] = 0;
4365
7.56k
    if (!IS_CHAR(cur)) {
4366
175
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4367
7.38k
    } else {
4368
7.38k
  NEXT;
4369
7.38k
    }
4370
7.56k
    return(buf);
4371
7.56k
}
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
3.95k
xmlParsePubidLiteral(xmlParserCtxt *ctxt) {
4386
3.95k
    xmlChar *buf = NULL;
4387
3.95k
    int len = 0;
4388
3.95k
    int size = XML_PARSER_BUFFER_SIZE;
4389
3.95k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4390
3.95k
                    XML_MAX_TEXT_LENGTH :
4391
3.95k
                    XML_MAX_NAME_LENGTH;
4392
3.95k
    xmlChar cur;
4393
3.95k
    xmlChar stop;
4394
4395
3.95k
    if (RAW == '"') {
4396
1.33k
        NEXT;
4397
1.33k
  stop = '"';
4398
2.62k
    } else if (RAW == '\'') {
4399
2.22k
        NEXT;
4400
2.22k
  stop = '\'';
4401
2.22k
    } else {
4402
407
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4403
407
  return(NULL);
4404
407
    }
4405
3.55k
    buf = xmlMalloc(size);
4406
3.55k
    if (buf == NULL) {
4407
0
  xmlErrMemory(ctxt);
4408
0
  return(NULL);
4409
0
    }
4410
3.55k
    cur = CUR;
4411
76.4k
    while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) &&
4412
72.9k
           (PARSER_STOPPED(ctxt) == 0)) { /* checked */
4413
72.9k
  if (len + 1 >= size) {
4414
114
      xmlChar *tmp;
4415
114
            int newSize;
4416
4417
114
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4418
114
            if (newSize < 0) {
4419
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID");
4420
0
                xmlFree(buf);
4421
0
                return(NULL);
4422
0
            }
4423
114
      tmp = xmlRealloc(buf, newSize);
4424
114
      if (tmp == NULL) {
4425
0
    xmlErrMemory(ctxt);
4426
0
    xmlFree(buf);
4427
0
    return(NULL);
4428
0
      }
4429
114
      buf = tmp;
4430
114
            size = newSize;
4431
114
  }
4432
72.9k
  buf[len++] = cur;
4433
72.9k
  NEXT;
4434
72.9k
  cur = CUR;
4435
72.9k
    }
4436
3.55k
    buf[len] = 0;
4437
3.55k
    if (cur != stop) {
4438
138
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4439
3.41k
    } else {
4440
3.41k
  NEXTL(1);
4441
3.41k
    }
4442
3.55k
    return(buf);
4443
3.55k
}
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
12.0M
              int isBlank) {
4488
12.0M
    int checkBlanks;
4489
4490
12.0M
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
4491
67.2k
        return;
4492
4493
11.9M
    checkBlanks = (!ctxt->keepBlanks) ||
4494
11.9M
                  (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
11.9M
    if ((checkBlanks) &&
4502
11.6M
        (areBlanks(ctxt, buf, size, isBlank))) {
4503
0
        if ((ctxt->sax->ignorableWhitespace != NULL) &&
4504
0
            (ctxt->keepBlanks))
4505
0
            ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size);
4506
11.9M
    } else {
4507
11.9M
        if (ctxt->sax->characters != NULL)
4508
11.9M
            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
11.9M
        if ((checkBlanks) && (*ctxt->space == -1))
4515
7.01M
            *ctxt->space = -2;
4516
11.9M
    }
4517
11.9M
}
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
11.2M
xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) {
4534
11.2M
    const xmlChar *in;
4535
11.2M
    int line = ctxt->input->line;
4536
11.2M
    int col = ctxt->input->col;
4537
11.2M
    int ccol;
4538
11.2M
    int terminate = 0;
4539
4540
11.2M
    GROW;
4541
    /*
4542
     * Accelerated common case where input don't need to be
4543
     * modified before passing it to the handler.
4544
     */
4545
11.2M
    in = ctxt->input->cur;
4546
11.6M
    do {
4547
15.0M
get_more_space:
4548
35.1M
        while (*in == 0x20) { in++; ctxt->input->col++; }
4549
15.0M
        if (*in == 0xA) {
4550
3.63M
            do {
4551
3.63M
                ctxt->input->line++; ctxt->input->col = 1;
4552
3.63M
                in++;
4553
3.63M
            } while (*in == 0xA);
4554
3.43M
            goto get_more_space;
4555
3.43M
        }
4556
11.6M
        if (*in == '<') {
4557
6.97M
            while (in > ctxt->input->cur) {
4558
3.48M
                const xmlChar *tmp = ctxt->input->cur;
4559
3.48M
                size_t nbchar = in - tmp;
4560
4561
3.48M
                if (nbchar > XML_MAX_ITEMS)
4562
0
                    nbchar = XML_MAX_ITEMS;
4563
3.48M
                ctxt->input->cur += nbchar;
4564
4565
3.48M
                xmlCharacters(ctxt, tmp, nbchar, 1);
4566
3.48M
            }
4567
3.48M
            return;
4568
3.48M
        }
4569
4570
9.02M
get_more:
4571
9.02M
        ccol = ctxt->input->col;
4572
117M
        while (test_char_data[*in]) {
4573
108M
            in++;
4574
108M
            ccol++;
4575
108M
        }
4576
9.02M
        ctxt->input->col = ccol;
4577
9.02M
        if (*in == 0xA) {
4578
861k
            do {
4579
861k
                ctxt->input->line++; ctxt->input->col = 1;
4580
861k
                in++;
4581
861k
            } while (*in == 0xA);
4582
710k
            goto get_more;
4583
710k
        }
4584
8.31M
        if (*in == ']') {
4585
201k
            size_t avail = ctxt->input->end - in;
4586
4587
201k
            if (partial && avail < 2) {
4588
0
                terminate = 1;
4589
0
                goto invoke_callback;
4590
0
            }
4591
201k
            if (in[1] == ']') {
4592
34.3k
                if (partial && avail < 3) {
4593
1
                    terminate = 1;
4594
1
                    goto invoke_callback;
4595
1
                }
4596
34.3k
                if (in[2] == '>')
4597
25
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4598
34.3k
            }
4599
4600
201k
            in++;
4601
201k
            ctxt->input->col++;
4602
201k
            goto get_more;
4603
201k
        }
4604
4605
8.11M
invoke_callback:
4606
15.5M
        while (in > ctxt->input->cur) {
4607
7.43M
            const xmlChar *tmp = ctxt->input->cur;
4608
7.43M
            size_t nbchar = in - tmp;
4609
4610
7.43M
            if (nbchar > XML_MAX_ITEMS)
4611
0
                nbchar = XML_MAX_ITEMS;
4612
7.43M
            ctxt->input->cur += nbchar;
4613
4614
7.43M
            xmlCharacters(ctxt, tmp, nbchar, 0);
4615
4616
7.43M
            line = ctxt->input->line;
4617
7.43M
            col = ctxt->input->col;
4618
7.43M
        }
4619
8.11M
        ctxt->input->cur = in;
4620
8.11M
        if (*in == 0xD) {
4621
433k
            in++;
4622
433k
            if (*in == 0xA) {
4623
405k
                ctxt->input->cur = in;
4624
405k
                in++;
4625
405k
                ctxt->input->line++; ctxt->input->col = 1;
4626
405k
                continue; /* while */
4627
405k
            }
4628
28.3k
            in--;
4629
28.3k
        }
4630
7.70M
        if (*in == '<') {
4631
6.63M
            return;
4632
6.63M
        }
4633
1.07M
        if (*in == '&') {
4634
197k
            return;
4635
197k
        }
4636
879k
        if (terminate) {
4637
1
            return;
4638
1
        }
4639
879k
        SHRINK;
4640
879k
        GROW;
4641
879k
        in = ctxt->input->cur;
4642
1.28M
    } while (((*in >= 0x20) && (*in <= 0x7F)) ||
4643
1.14M
             (*in == 0x09) || (*in == 0x0a));
4644
900k
    ctxt->input->line = line;
4645
900k
    ctxt->input->col = col;
4646
900k
    xmlParseCharDataComplex(ctxt, partial);
4647
900k
}
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
900k
xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) {
4661
900k
    xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5];
4662
900k
    int nbchar = 0;
4663
900k
    int cur, l;
4664
4665
900k
    cur = xmlCurrentCharRecover(ctxt, &l);
4666
64.9M
    while ((cur != '<') && /* checked */
4667
64.1M
           (cur != '&') &&
4668
64.0M
     (IS_CHAR(cur))) {
4669
64.0M
        if (cur == ']') {
4670
87.6k
            size_t avail = ctxt->input->end - ctxt->input->cur;
4671
4672
87.6k
            if (partial && avail < 2)
4673
1
                break;
4674
87.6k
            if (NXT(1) == ']') {
4675
64.5k
                if (partial && avail < 3)
4676
3
                    break;
4677
64.5k
                if (NXT(2) == '>')
4678
580
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4679
64.5k
            }
4680
87.6k
        }
4681
4682
64.0M
  COPY_BUF(buf, nbchar, cur);
4683
  /* move current position before possible calling of ctxt->sax->characters */
4684
64.0M
  NEXTL(l);
4685
64.0M
  if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) {
4686
185k
      buf[nbchar] = 0;
4687
4688
185k
            xmlCharacters(ctxt, buf, nbchar, 0);
4689
185k
      nbchar = 0;
4690
185k
            SHRINK;
4691
185k
  }
4692
64.0M
  cur = xmlCurrentCharRecover(ctxt, &l);
4693
64.0M
    }
4694
900k
    if (nbchar != 0) {
4695
892k
        buf[nbchar] = 0;
4696
4697
892k
        xmlCharacters(ctxt, buf, nbchar, 0);
4698
892k
    }
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
900k
    if (ctxt->input->cur < ctxt->input->end) {
4707
892k
        if ((cur == 0) && (CUR != 0)) {
4708
234
            if (partial == 0) {
4709
228
                xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4710
228
                        "Incomplete UTF-8 sequence starting with %02X\n", CUR);
4711
228
                NEXTL(1);
4712
228
            }
4713
891k
        } else if ((cur != '<') && (cur != '&') && (cur != ']')) {
4714
            /* Generate the error and skip the offending character */
4715
2.35k
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4716
2.35k
                              "PCDATA invalid Char value %d\n", cur);
4717
2.35k
            NEXTL(l);
4718
2.35k
        }
4719
892k
    }
4720
900k
}
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
17.8k
xmlParseExternalID(xmlParserCtxt *ctxt, xmlChar **publicId, int strict) {
4756
17.8k
    xmlChar *URI = NULL;
4757
4758
17.8k
    *publicId = NULL;
4759
17.8k
    if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) {
4760
5.81k
        SKIP(6);
4761
5.81k
  if (SKIP_BLANKS == 0) {
4762
665
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4763
665
                     "Space required after 'SYSTEM'\n");
4764
665
  }
4765
5.81k
  URI = xmlParseSystemLiteral(ctxt);
4766
5.81k
  if (URI == NULL) {
4767
199
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4768
199
        }
4769
12.0k
    } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) {
4770
3.95k
        SKIP(6);
4771
3.95k
  if (SKIP_BLANKS == 0) {
4772
25
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4773
25
        "Space required after 'PUBLIC'\n");
4774
25
  }
4775
3.95k
  *publicId = xmlParsePubidLiteral(ctxt);
4776
3.95k
  if (*publicId == NULL) {
4777
407
      xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL);
4778
407
  }
4779
3.95k
  if (strict) {
4780
      /*
4781
       * We don't handle [83] so "S SystemLiteral" is required.
4782
       */
4783
1.88k
      if (SKIP_BLANKS == 0) {
4784
435
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4785
435
      "Space required after the Public Identifier\n");
4786
435
      }
4787
2.07k
  } 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
2.07k
      if (SKIP_BLANKS == 0) return(NULL);
4795
1.13k
      if ((CUR != '\'') && (CUR != '"')) return(NULL);
4796
1.13k
  }
4797
2.35k
  URI = xmlParseSystemLiteral(ctxt);
4798
2.35k
  if (URI == NULL) {
4799
418
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4800
418
        }
4801
2.35k
    }
4802
16.2k
    return(URI);
4803
17.8k
}
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
33.6k
                       size_t len, size_t size) {
4820
33.6k
    int q, ql;
4821
33.6k
    int r, rl;
4822
33.6k
    int cur, l;
4823
33.6k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4824
33.6k
                    XML_MAX_HUGE_LENGTH :
4825
33.6k
                    XML_MAX_TEXT_LENGTH;
4826
4827
33.6k
    if (buf == NULL) {
4828
9.54k
        len = 0;
4829
9.54k
  size = XML_PARSER_BUFFER_SIZE;
4830
9.54k
  buf = xmlMalloc(size);
4831
9.54k
  if (buf == NULL) {
4832
0
      xmlErrMemory(ctxt);
4833
0
      return;
4834
0
  }
4835
9.54k
    }
4836
33.6k
    q = xmlCurrentCharRecover(ctxt, &ql);
4837
33.6k
    if (q == 0)
4838
279
        goto not_terminated;
4839
33.3k
    if (!IS_CHAR(q)) {
4840
50
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4841
50
                          "xmlParseComment: invalid xmlChar value %d\n",
4842
50
                    q);
4843
50
  xmlFree (buf);
4844
50
  return;
4845
50
    }
4846
33.3k
    NEXTL(ql);
4847
33.3k
    r = xmlCurrentCharRecover(ctxt, &rl);
4848
33.3k
    if (r == 0)
4849
31
        goto not_terminated;
4850
33.2k
    if (!IS_CHAR(r)) {
4851
32
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4852
32
                          "xmlParseComment: invalid xmlChar value %d\n",
4853
32
                    r);
4854
32
  xmlFree (buf);
4855
32
  return;
4856
32
    }
4857
33.2k
    NEXTL(rl);
4858
33.2k
    cur = xmlCurrentCharRecover(ctxt, &l);
4859
33.2k
    if (cur == 0)
4860
29
        goto not_terminated;
4861
2.19M
    while (IS_CHAR(cur) && /* checked */
4862
2.19M
           ((cur != '>') ||
4863
2.16M
      (r != '-') || (q != '-'))) {
4864
2.16M
  if ((r == '-') && (q == '-')) {
4865
18.9k
      xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL);
4866
18.9k
  }
4867
2.16M
  if (len + 5 >= size) {
4868
13.7k
      xmlChar *tmp;
4869
13.7k
            int newSize;
4870
4871
13.7k
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4872
13.7k
            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
13.7k
      tmp = xmlRealloc(buf, newSize);
4879
13.7k
      if (tmp == NULL) {
4880
0
    xmlErrMemory(ctxt);
4881
0
    xmlFree(buf);
4882
0
    return;
4883
0
      }
4884
13.7k
      buf = tmp;
4885
13.7k
            size = newSize;
4886
13.7k
  }
4887
2.16M
  COPY_BUF(buf, len, q);
4888
4889
2.16M
  q = r;
4890
2.16M
  ql = rl;
4891
2.16M
  r = cur;
4892
2.16M
  rl = l;
4893
4894
2.16M
  NEXTL(l);
4895
2.16M
  cur = xmlCurrentCharRecover(ctxt, &l);
4896
4897
2.16M
    }
4898
33.2k
    buf[len] = 0;
4899
33.2k
    if (cur == 0) {
4900
494
  xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4901
494
                       "Comment not terminated \n<!--%.50s\n", buf);
4902
32.7k
    } else if (!IS_CHAR(cur)) {
4903
105
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4904
105
                          "xmlParseComment: invalid xmlChar value %d\n",
4905
105
                    cur);
4906
32.6k
    } else {
4907
32.6k
        NEXT;
4908
32.6k
  if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
4909
0
      (!ctxt->disableSAX))
4910
0
      ctxt->sax->comment(ctxt->userData, buf);
4911
32.6k
    }
4912
33.2k
    xmlFree(buf);
4913
33.2k
    return;
4914
339
not_terminated:
4915
339
    xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4916
339
       "Comment not terminated\n", NULL);
4917
339
    xmlFree(buf);
4918
339
}
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
75.8k
xmlParseComment(xmlParserCtxt *ctxt) {
4933
75.8k
    xmlChar *buf = NULL;
4934
75.8k
    size_t size = XML_PARSER_BUFFER_SIZE;
4935
75.8k
    size_t len = 0;
4936
75.8k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4937
75.8k
                       XML_MAX_HUGE_LENGTH :
4938
75.8k
                       XML_MAX_TEXT_LENGTH;
4939
75.8k
    const xmlChar *in;
4940
75.8k
    size_t nbchar = 0;
4941
75.8k
    int ccol;
4942
4943
    /*
4944
     * Check that there is a comment right here.
4945
     */
4946
75.8k
    if ((RAW != '<') || (NXT(1) != '!'))
4947
0
        return;
4948
75.8k
    SKIP(2);
4949
75.8k
    if ((RAW != '-') || (NXT(1) != '-'))
4950
17
        return;
4951
75.8k
    SKIP(2);
4952
75.8k
    GROW;
4953
4954
    /*
4955
     * Accelerated common case where input don't need to be
4956
     * modified before passing it to the handler.
4957
     */
4958
75.8k
    in = ctxt->input->cur;
4959
75.8k
    do {
4960
75.8k
  if (*in == 0xA) {
4961
9.21k
      do {
4962
9.21k
    ctxt->input->line++; ctxt->input->col = 1;
4963
9.21k
    in++;
4964
9.21k
      } while (*in == 0xA);
4965
3.95k
  }
4966
152k
get_more:
4967
152k
        ccol = ctxt->input->col;
4968
1.78M
  while (((*in > '-') && (*in <= 0x7F)) ||
4969
420k
         ((*in >= 0x20) && (*in < '-')) ||
4970
1.63M
         (*in == 0x09)) {
4971
1.63M
        in++;
4972
1.63M
        ccol++;
4973
1.63M
  }
4974
152k
  ctxt->input->col = ccol;
4975
152k
  if (*in == 0xA) {
4976
26.3k
      do {
4977
26.3k
    ctxt->input->line++; ctxt->input->col = 1;
4978
26.3k
    in++;
4979
26.3k
      } while (*in == 0xA);
4980
16.3k
      goto get_more;
4981
16.3k
  }
4982
135k
  nbchar = in - ctxt->input->cur;
4983
  /*
4984
   * save current set of data
4985
   */
4986
135k
  if (nbchar > 0) {
4987
109k
            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
109k
            if (buf == NULL) {
4994
60.4k
                if ((*in == '-') && (in[1] == '-'))
4995
33.1k
                    size = nbchar + 1;
4996
27.3k
                else
4997
27.3k
                    size = XML_PARSER_BUFFER_SIZE + nbchar;
4998
60.4k
                buf = xmlMalloc(size);
4999
60.4k
                if (buf == NULL) {
5000
0
                    xmlErrMemory(ctxt);
5001
0
                    return;
5002
0
                }
5003
60.4k
                len = 0;
5004
60.4k
            } else if (len + nbchar + 1 >= size) {
5005
3.08k
                xmlChar *new_buf;
5006
3.08k
                size += len + nbchar + XML_PARSER_BUFFER_SIZE;
5007
3.08k
                new_buf = xmlRealloc(buf, size);
5008
3.08k
                if (new_buf == NULL) {
5009
0
                    xmlErrMemory(ctxt);
5010
0
                    xmlFree(buf);
5011
0
                    return;
5012
0
                }
5013
3.08k
                buf = new_buf;
5014
3.08k
            }
5015
109k
            memcpy(&buf[len], ctxt->input->cur, nbchar);
5016
109k
            len += nbchar;
5017
109k
            buf[len] = 0;
5018
109k
  }
5019
135k
  ctxt->input->cur = in;
5020
135k
  if (*in == 0xA) {
5021
0
      in++;
5022
0
      ctxt->input->line++; ctxt->input->col = 1;
5023
0
  }
5024
135k
  if (*in == 0xD) {
5025
12.1k
      in++;
5026
12.1k
      if (*in == 0xA) {
5027
6.03k
    ctxt->input->cur = in;
5028
6.03k
    in++;
5029
6.03k
    ctxt->input->line++; ctxt->input->col = 1;
5030
6.03k
    goto get_more;
5031
6.03k
      }
5032
6.14k
      in--;
5033
6.14k
  }
5034
129k
  SHRINK;
5035
129k
  GROW;
5036
129k
  in = ctxt->input->cur;
5037
129k
  if (*in == '-') {
5038
96.0k
      if (in[1] == '-') {
5039
67.8k
          if (in[2] == '>') {
5040
42.1k
        SKIP(3);
5041
42.1k
        if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
5042
0
            (!ctxt->disableSAX)) {
5043
0
      if (buf != NULL)
5044
0
          ctxt->sax->comment(ctxt->userData, buf);
5045
0
      else
5046
0
          ctxt->sax->comment(ctxt->userData, BAD_CAST "");
5047
0
        }
5048
42.1k
        if (buf != NULL)
5049
36.3k
            xmlFree(buf);
5050
42.1k
        return;
5051
42.1k
    }
5052
25.6k
    if (buf != NULL) {
5053
22.6k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5054
22.6k
                          "Double hyphen within comment: "
5055
22.6k
                                      "<!--%.50s\n",
5056
22.6k
              buf);
5057
22.6k
    } else
5058
3.05k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5059
3.05k
                          "Double hyphen within comment\n", NULL);
5060
25.6k
    in++;
5061
25.6k
    ctxt->input->col++;
5062
25.6k
      }
5063
53.9k
      in++;
5064
53.9k
      ctxt->input->col++;
5065
53.9k
      goto get_more;
5066
96.0k
  }
5067
129k
    } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a));
5068
33.6k
    xmlParseCommentComplex(ctxt, buf, len, size);
5069
33.6k
}
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
75.8k
xmlParsePITarget(xmlParserCtxt *ctxt) {
5085
75.8k
    const xmlChar *name;
5086
5087
75.8k
    name = xmlParseName(ctxt);
5088
75.8k
    if ((name != NULL) &&
5089
73.7k
        ((name[0] == 'x') || (name[0] == 'X')) &&
5090
38.9k
        ((name[1] == 'm') || (name[1] == 'M')) &&
5091
33.5k
        ((name[2] == 'l') || (name[2] == 'L'))) {
5092
24.9k
  int i;
5093
24.9k
  if ((name[0] == 'x') && (name[1] == 'm') &&
5094
18.8k
      (name[2] == 'l') && (name[3] == 0)) {
5095
1.82k
      xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5096
1.82k
     "XML declaration allowed only at the start of the document\n");
5097
1.82k
      return(name);
5098
23.0k
  } else if (name[3] == 0) {
5099
1.74k
      xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL);
5100
1.74k
      return(name);
5101
1.74k
  }
5102
62.3k
  for (i = 0;;i++) {
5103
62.3k
      if (xmlW3CPIs[i] == NULL) break;
5104
41.9k
      if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i]))
5105
925
          return(name);
5106
41.9k
  }
5107
20.4k
  xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5108
20.4k
          "xmlParsePITarget: invalid name prefix 'xml'\n",
5109
20.4k
          NULL, NULL);
5110
20.4k
    }
5111
71.3k
    if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) {
5112
6.55k
  xmlNsErr(ctxt, XML_NS_ERR_COLON,
5113
6.55k
     "colons are forbidden from PI names '%s'\n", name, NULL, NULL);
5114
6.55k
    }
5115
71.3k
    return(name);
5116
75.8k
}
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
75.8k
xmlParsePI(xmlParserCtxt *ctxt) {
5199
75.8k
    xmlChar *buf = NULL;
5200
75.8k
    size_t len = 0;
5201
75.8k
    size_t size = XML_PARSER_BUFFER_SIZE;
5202
75.8k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
5203
74.9k
                       XML_MAX_HUGE_LENGTH :
5204
75.8k
                       XML_MAX_TEXT_LENGTH;
5205
75.8k
    int cur, l;
5206
75.8k
    const xmlChar *target;
5207
5208
75.8k
    if ((RAW == '<') && (NXT(1) == '?')) {
5209
  /*
5210
   * this is a Processing Instruction.
5211
   */
5212
75.8k
  SKIP(2);
5213
5214
  /*
5215
   * Parse the target name and check for special support like
5216
   * namespace.
5217
   */
5218
75.8k
        target = xmlParsePITarget(ctxt);
5219
75.8k
  if (target != NULL) {
5220
73.7k
      if ((RAW == '?') && (NXT(1) == '>')) {
5221
12.9k
    SKIP(2);
5222
5223
    /*
5224
     * SAX: PI detected.
5225
     */
5226
12.9k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5227
6.36k
        (ctxt->sax->processingInstruction != NULL))
5228
6.36k
        ctxt->sax->processingInstruction(ctxt->userData,
5229
6.36k
                                         target, NULL);
5230
12.9k
    return;
5231
12.9k
      }
5232
60.7k
      buf = xmlMalloc(size);
5233
60.7k
      if (buf == NULL) {
5234
0
    xmlErrMemory(ctxt);
5235
0
    return;
5236
0
      }
5237
60.7k
      if (SKIP_BLANKS == 0) {
5238
10.5k
    xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED,
5239
10.5k
        "ParsePI: PI %s space expected\n", target);
5240
10.5k
      }
5241
60.7k
      cur = xmlCurrentCharRecover(ctxt, &l);
5242
5.83M
      while (IS_CHAR(cur) && /* checked */
5243
5.83M
       ((cur != '?') || (NXT(1) != '>'))) {
5244
5.77M
    if (len + 5 >= size) {
5245
35.7k
        xmlChar *tmp;
5246
35.7k
                    int newSize;
5247
5248
35.7k
                    newSize = xmlGrowCapacity(size, 1, 1, maxLength);
5249
35.7k
                    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
35.7k
        tmp = xmlRealloc(buf, newSize);
5256
35.7k
        if (tmp == NULL) {
5257
0
      xmlErrMemory(ctxt);
5258
0
      xmlFree(buf);
5259
0
      return;
5260
0
        }
5261
35.7k
        buf = tmp;
5262
35.7k
                    size = newSize;
5263
35.7k
    }
5264
5.77M
    COPY_BUF(buf, len, cur);
5265
5.77M
    NEXTL(l);
5266
5.77M
    cur = xmlCurrentCharRecover(ctxt, &l);
5267
5.77M
      }
5268
60.7k
      buf[len] = 0;
5269
60.7k
      if (cur != '?') {
5270
1.82k
    xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5271
1.82k
          "ParsePI: PI %s never end ...\n", target);
5272
58.9k
      } else {
5273
58.9k
    SKIP(2);
5274
5275
58.9k
#ifdef LIBXML_CATALOG_ENABLED
5276
58.9k
    if ((ctxt->inSubset == 0) &&
5277
46.4k
        (xmlStrEqual(target, XML_CATALOG_PI))) {
5278
12.1k
        xmlCatalogAllow allow = xmlCatalogGetDefaults();
5279
5280
12.1k
        if ((ctxt->options & XML_PARSE_CATALOG_PI) &&
5281
0
                        ((allow == XML_CATA_ALLOW_DOCUMENT) ||
5282
0
       (allow == XML_CATA_ALLOW_ALL)))
5283
0
      xmlParseCatalogPI(ctxt, buf);
5284
12.1k
    }
5285
58.9k
#endif
5286
5287
    /*
5288
     * SAX: PI detected.
5289
     */
5290
58.9k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5291
47.2k
        (ctxt->sax->processingInstruction != NULL))
5292
47.2k
        ctxt->sax->processingInstruction(ctxt->userData,
5293
47.2k
                                         target, buf);
5294
58.9k
      }
5295
60.7k
      xmlFree(buf);
5296
60.7k
  } else {
5297
2.13k
      xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL);
5298
2.13k
  }
5299
75.8k
    }
5300
75.8k
}
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
3.13k
xmlParseNotationDecl(xmlParserCtxt *ctxt) {
5323
3.13k
    const xmlChar *name;
5324
3.13k
    xmlChar *Pubid;
5325
3.13k
    xmlChar *Systemid;
5326
5327
3.13k
    if ((CUR != '<') || (NXT(1) != '!'))
5328
0
        return;
5329
3.13k
    SKIP(2);
5330
5331
3.13k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5332
3.06k
#ifdef LIBXML_VALID_ENABLED
5333
3.06k
  int oldInputNr = ctxt->inputNr;
5334
3.06k
#endif
5335
5336
3.06k
  SKIP(8);
5337
3.06k
  if (SKIP_BLANKS_PE == 0) {
5338
30
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5339
30
         "Space required after '<!NOTATION'\n");
5340
30
      return;
5341
30
  }
5342
5343
3.03k
        name = xmlParseName(ctxt);
5344
3.03k
  if (name == NULL) {
5345
51
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5346
51
      return;
5347
51
  }
5348
2.98k
  if (xmlStrchr(name, ':') != NULL) {
5349
6
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5350
6
         "colons are forbidden from notation names '%s'\n",
5351
6
         name, NULL, NULL);
5352
6
  }
5353
2.98k
  if (SKIP_BLANKS_PE == 0) {
5354
25
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5355
25
         "Space required after the NOTATION name'\n");
5356
25
      return;
5357
25
  }
5358
5359
  /*
5360
   * Parse the IDs.
5361
   */
5362
2.95k
  Systemid = xmlParseExternalID(ctxt, &Pubid, 0);
5363
2.95k
  SKIP_BLANKS_PE;
5364
5365
2.95k
  if (RAW == '>') {
5366
2.63k
#ifdef LIBXML_VALID_ENABLED
5367
2.63k
      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
2.63k
#endif
5374
2.63k
      NEXT;
5375
2.63k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5376
2.06k
    (ctxt->sax->notationDecl != NULL))
5377
0
    ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid);
5378
2.63k
  } else {
5379
321
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5380
321
  }
5381
2.95k
  if (Systemid != NULL) xmlFree(Systemid);
5382
2.95k
  if (Pubid != NULL) xmlFree(Pubid);
5383
2.95k
    }
5384
3.13k
}
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
51.3k
xmlParseEntityDecl(xmlParserCtxt *ctxt) {
5411
51.3k
    const xmlChar *name = NULL;
5412
51.3k
    xmlChar *value = NULL;
5413
51.3k
    xmlChar *URI = NULL, *literal = NULL;
5414
51.3k
    const xmlChar *ndata = NULL;
5415
51.3k
    int isParameter = 0;
5416
51.3k
    xmlChar *orig = NULL;
5417
5418
51.3k
    if ((CUR != '<') || (NXT(1) != '!'))
5419
0
        return;
5420
51.3k
    SKIP(2);
5421
5422
    /* GROW; done in the caller */
5423
51.3k
    if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5424
51.2k
#ifdef LIBXML_VALID_ENABLED
5425
51.2k
  int oldInputNr = ctxt->inputNr;
5426
51.2k
#endif
5427
5428
51.2k
  SKIP(6);
5429
51.2k
  if (SKIP_BLANKS_PE == 0) {
5430
12.8k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5431
12.8k
         "Space required after '<!ENTITY'\n");
5432
12.8k
  }
5433
5434
51.2k
  if (RAW == '%') {
5435
11.1k
      NEXT;
5436
11.1k
      if (SKIP_BLANKS_PE == 0) {
5437
6.26k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5438
6.26k
             "Space required after '%%'\n");
5439
6.26k
      }
5440
11.1k
      isParameter = 1;
5441
11.1k
  }
5442
5443
51.2k
        name = xmlParseName(ctxt);
5444
51.2k
  if (name == NULL) {
5445
204
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5446
204
                     "xmlParseEntityDecl: no name\n");
5447
204
            return;
5448
204
  }
5449
51.0k
  if (xmlStrchr(name, ':') != NULL) {
5450
1.71k
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5451
1.71k
         "colons are forbidden from entities names '%s'\n",
5452
1.71k
         name, NULL, NULL);
5453
1.71k
  }
5454
51.0k
  if (SKIP_BLANKS_PE == 0) {
5455
21.0k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5456
21.0k
         "Space required after the entity name\n");
5457
21.0k
  }
5458
5459
  /*
5460
   * handle the various case of definitions...
5461
   */
5462
51.0k
  if (isParameter) {
5463
11.1k
      if ((RAW == '"') || (RAW == '\'')) {
5464
10.1k
          value = xmlParseEntityValue(ctxt, &orig);
5465
10.1k
    if (value) {
5466
10.1k
        if ((ctxt->sax != NULL) &&
5467
10.1k
      (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5468
0
      ctxt->sax->entityDecl(ctxt->userData, name,
5469
0
                        XML_INTERNAL_PARAMETER_ENTITY,
5470
0
            NULL, NULL, value);
5471
10.1k
    }
5472
10.1k
      } else {
5473
970
          URI = xmlParseExternalID(ctxt, &literal, 1);
5474
970
    if ((URI == NULL) && (literal == NULL)) {
5475
140
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5476
140
    }
5477
970
    if (URI) {
5478
753
                    if (xmlStrchr(URI, '#')) {
5479
6
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5480
747
                    } else {
5481
747
                        if ((ctxt->sax != NULL) &&
5482
747
                            (!ctxt->disableSAX) &&
5483
428
                            (ctxt->sax->entityDecl != NULL))
5484
0
                            ctxt->sax->entityDecl(ctxt->userData, name,
5485
0
                                        XML_EXTERNAL_PARAMETER_ENTITY,
5486
0
                                        literal, URI, NULL);
5487
747
                    }
5488
753
    }
5489
970
      }
5490
39.8k
  } else {
5491
39.8k
      if ((RAW == '"') || (RAW == '\'')) {
5492
33.4k
          value = xmlParseEntityValue(ctxt, &orig);
5493
33.4k
    if ((ctxt->sax != NULL) &&
5494
33.4k
        (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5495
0
        ctxt->sax->entityDecl(ctxt->userData, name,
5496
0
        XML_INTERNAL_GENERAL_ENTITY,
5497
0
        NULL, NULL, value);
5498
    /*
5499
     * For expat compatibility in SAX mode.
5500
     */
5501
33.4k
    if ((ctxt->myDoc == NULL) ||
5502
33.4k
        (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) {
5503
33.4k
        if (ctxt->myDoc == NULL) {
5504
1.29k
      ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5505
1.29k
      if (ctxt->myDoc == NULL) {
5506
0
          xmlErrMemory(ctxt);
5507
0
          goto done;
5508
0
      }
5509
1.29k
      ctxt->myDoc->properties = XML_DOC_INTERNAL;
5510
1.29k
        }
5511
33.4k
        if (ctxt->myDoc->intSubset == NULL) {
5512
1.29k
      ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5513
1.29k
              BAD_CAST "fake", NULL, NULL);
5514
1.29k
                        if (ctxt->myDoc->intSubset == NULL) {
5515
0
                            xmlErrMemory(ctxt);
5516
0
                            goto done;
5517
0
                        }
5518
1.29k
                    }
5519
5520
33.4k
        xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY,
5521
33.4k
                    NULL, NULL, value);
5522
33.4k
    }
5523
33.4k
      } else {
5524
6.36k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5525
6.36k
    if ((URI == NULL) && (literal == NULL)) {
5526
1.04k
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5527
1.04k
    }
5528
6.36k
    if (URI) {
5529
5.29k
                    if (xmlStrchr(URI, '#')) {
5530
562
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5531
562
                    }
5532
5.29k
    }
5533
6.36k
    if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) {
5534
1.26k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5535
1.26k
           "Space required before 'NDATA'\n");
5536
1.26k
    }
5537
6.36k
    if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) {
5538
2.60k
        SKIP(5);
5539
2.60k
        if (SKIP_BLANKS_PE == 0) {
5540
72
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5541
72
               "Space required after 'NDATA'\n");
5542
72
        }
5543
2.60k
        ndata = xmlParseName(ctxt);
5544
2.60k
        if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5545
897
            (ctxt->sax->unparsedEntityDecl != NULL))
5546
0
      ctxt->sax->unparsedEntityDecl(ctxt->userData, name,
5547
0
            literal, URI, ndata);
5548
3.76k
    } else {
5549
3.76k
        if ((ctxt->sax != NULL) &&
5550
3.76k
            (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5551
0
      ctxt->sax->entityDecl(ctxt->userData, name,
5552
0
            XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5553
0
            literal, URI, NULL);
5554
        /*
5555
         * For expat compatibility in SAX mode.
5556
         * assuming the entity replacement was asked for
5557
         */
5558
3.76k
        if ((ctxt->replaceEntities != 0) &&
5559
3.76k
      ((ctxt->myDoc == NULL) ||
5560
3.76k
      (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) {
5561
3.76k
      if (ctxt->myDoc == NULL) {
5562
289
          ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5563
289
          if (ctxt->myDoc == NULL) {
5564
0
              xmlErrMemory(ctxt);
5565
0
        goto done;
5566
0
          }
5567
289
          ctxt->myDoc->properties = XML_DOC_INTERNAL;
5568
289
      }
5569
5570
3.76k
      if (ctxt->myDoc->intSubset == NULL) {
5571
289
          ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5572
289
            BAD_CAST "fake", NULL, NULL);
5573
289
                            if (ctxt->myDoc->intSubset == NULL) {
5574
0
                                xmlErrMemory(ctxt);
5575
0
                                goto done;
5576
0
                            }
5577
289
                        }
5578
3.76k
      xmlSAX2EntityDecl(ctxt, name,
5579
3.76k
                  XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5580
3.76k
                  literal, URI, NULL);
5581
3.76k
        }
5582
3.76k
    }
5583
6.36k
      }
5584
39.8k
  }
5585
51.0k
  SKIP_BLANKS_PE;
5586
51.0k
  if (RAW != '>') {
5587
1.21k
      xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED,
5588
1.21k
              "xmlParseEntityDecl: entity %s not terminated\n", name);
5589
49.7k
  } else {
5590
49.7k
#ifdef LIBXML_VALID_ENABLED
5591
49.7k
      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
49.7k
#endif
5598
49.7k
      NEXT;
5599
49.7k
  }
5600
51.0k
  if (orig != NULL) {
5601
      /*
5602
       * Ugly mechanism to save the raw entity value.
5603
       */
5604
43.2k
      xmlEntityPtr cur = NULL;
5605
5606
43.2k
      if (isParameter) {
5607
10.1k
          if ((ctxt->sax != NULL) &&
5608
10.1k
        (ctxt->sax->getParameterEntity != NULL))
5609
0
        cur = ctxt->sax->getParameterEntity(ctxt->userData, name);
5610
33.0k
      } else {
5611
33.0k
          if ((ctxt->sax != NULL) &&
5612
33.0k
        (ctxt->sax->getEntity != NULL))
5613
33.0k
        cur = ctxt->sax->getEntity(ctxt->userData, name);
5614
33.0k
    if ((cur == NULL) && (ctxt->userData==ctxt)) {
5615
0
        cur = xmlSAX2GetEntity(ctxt, name);
5616
0
    }
5617
33.0k
      }
5618
43.2k
            if ((cur != NULL) && (cur->orig == NULL)) {
5619
0
    cur->orig = orig;
5620
0
                orig = NULL;
5621
0
      }
5622
43.2k
  }
5623
5624
51.0k
done:
5625
51.0k
  if (value != NULL) xmlFree(value);
5626
51.0k
  if (URI != NULL) xmlFree(URI);
5627
51.0k
  if (literal != NULL) xmlFree(literal);
5628
51.0k
        if (orig != NULL) xmlFree(orig);
5629
51.0k
    }
5630
51.3k
}
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
44.3k
xmlParseDefaultDecl(xmlParserCtxt *ctxt, xmlChar **value) {
5663
44.3k
    int val;
5664
44.3k
    xmlChar *ret;
5665
5666
44.3k
    *value = NULL;
5667
44.3k
    if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) {
5668
2.10k
  SKIP(9);
5669
2.10k
  return(XML_ATTRIBUTE_REQUIRED);
5670
2.10k
    }
5671
42.2k
    if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) {
5672
3.91k
  SKIP(8);
5673
3.91k
  return(XML_ATTRIBUTE_IMPLIED);
5674
3.91k
    }
5675
38.3k
    val = XML_ATTRIBUTE_NONE;
5676
38.3k
    if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) {
5677
1.32k
  SKIP(6);
5678
1.32k
  val = XML_ATTRIBUTE_FIXED;
5679
1.32k
  if (SKIP_BLANKS_PE == 0) {
5680
394
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5681
394
         "Space required after '#FIXED'\n");
5682
394
  }
5683
1.32k
    }
5684
38.3k
    ret = xmlParseAttValue(ctxt);
5685
38.3k
    if (ret == NULL) {
5686
685
  xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo,
5687
685
           "Attribute default value declaration error\n");
5688
685
    } else
5689
37.6k
        *value = ret;
5690
38.3k
    return(val);
5691
42.2k
}
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
730
xmlParseNotationType(xmlParserCtxt *ctxt) {
5712
730
    const xmlChar *name;
5713
730
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5714
5715
730
    if (RAW != '(') {
5716
9
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5717
9
  return(NULL);
5718
9
    }
5719
1.48k
    do {
5720
1.48k
        NEXT;
5721
1.48k
  SKIP_BLANKS_PE;
5722
1.48k
        name = xmlParseName(ctxt);
5723
1.48k
  if (name == NULL) {
5724
25
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5725
25
         "Name expected in NOTATION declaration\n");
5726
25
            xmlFreeEnumeration(ret);
5727
25
      return(NULL);
5728
25
  }
5729
1.46k
        tmp = NULL;
5730
1.46k
#ifdef LIBXML_VALID_ENABLED
5731
1.46k
        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
1.46k
#endif /* LIBXML_VALID_ENABLED */
5746
1.46k
  if (tmp == NULL) {
5747
1.46k
      cur = xmlCreateEnumeration(name);
5748
1.46k
      if (cur == NULL) {
5749
0
                xmlErrMemory(ctxt);
5750
0
                xmlFreeEnumeration(ret);
5751
0
                return(NULL);
5752
0
            }
5753
1.46k
      if (last == NULL) ret = last = cur;
5754
753
      else {
5755
753
    last->next = cur;
5756
753
    last = cur;
5757
753
      }
5758
1.46k
  }
5759
1.46k
  SKIP_BLANKS_PE;
5760
1.46k
    } while (RAW == '|');
5761
696
    if (RAW != ')') {
5762
74
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5763
74
        xmlFreeEnumeration(ret);
5764
74
  return(NULL);
5765
74
    }
5766
622
    NEXT;
5767
622
    return(ret);
5768
696
}
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
6.61k
xmlParseEnumerationType(xmlParserCtxt *ctxt) {
5787
6.61k
    xmlChar *name;
5788
6.61k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5789
5790
6.61k
    if (RAW != '(') {
5791
247
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL);
5792
247
  return(NULL);
5793
247
    }
5794
8.22k
    do {
5795
8.22k
        NEXT;
5796
8.22k
  SKIP_BLANKS_PE;
5797
8.22k
        name = xmlParseNmtoken(ctxt);
5798
8.22k
  if (name == NULL) {
5799
51
      xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL);
5800
51
      return(ret);
5801
51
  }
5802
8.16k
        tmp = NULL;
5803
8.16k
#ifdef LIBXML_VALID_ENABLED
5804
8.16k
        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
8.16k
#endif /* LIBXML_VALID_ENABLED */
5819
8.16k
  if (tmp == NULL) {
5820
8.16k
      cur = xmlCreateEnumeration(name);
5821
8.16k
      if (!xmlDictOwns(ctxt->dict, name))
5822
8.16k
    xmlFree(name);
5823
8.16k
      if (cur == NULL) {
5824
0
                xmlErrMemory(ctxt);
5825
0
                xmlFreeEnumeration(ret);
5826
0
                return(NULL);
5827
0
            }
5828
8.16k
      if (last == NULL) ret = last = cur;
5829
1.80k
      else {
5830
1.80k
    last->next = cur;
5831
1.80k
    last = cur;
5832
1.80k
      }
5833
8.16k
  }
5834
8.16k
  SKIP_BLANKS_PE;
5835
8.16k
    } while (RAW == '|');
5836
6.31k
    if (RAW != ')') {
5837
74
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL);
5838
74
  return(ret);
5839
74
    }
5840
6.24k
    NEXT;
5841
6.24k
    return(ret);
5842
6.31k
}
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
7.35k
xmlParseEnumeratedType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5860
7.35k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5861
738
  SKIP(8);
5862
738
  if (SKIP_BLANKS_PE == 0) {
5863
8
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5864
8
         "Space required after 'NOTATION'\n");
5865
8
      return(0);
5866
8
  }
5867
730
  *tree = xmlParseNotationType(ctxt);
5868
730
  if (*tree == NULL) return(0);
5869
622
  return(XML_ATTRIBUTE_NOTATION);
5870
730
    }
5871
6.61k
    *tree = xmlParseEnumerationType(ctxt);
5872
6.61k
    if (*tree == NULL) return(0);
5873
6.36k
    return(XML_ATTRIBUTE_ENUMERATION);
5874
6.61k
}
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
44.9k
xmlParseAttributeType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5923
44.9k
    if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) {
5924
3.58k
  SKIP(5);
5925
3.58k
  return(XML_ATTRIBUTE_CDATA);
5926
41.3k
     } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) {
5927
557
  SKIP(6);
5928
557
  return(XML_ATTRIBUTE_IDREFS);
5929
40.8k
     } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) {
5930
557
  SKIP(5);
5931
557
  return(XML_ATTRIBUTE_IDREF);
5932
40.2k
     } else if ((RAW == 'I') && (NXT(1) == 'D')) {
5933
26.2k
        SKIP(2);
5934
26.2k
  return(XML_ATTRIBUTE_ID);
5935
26.2k
     } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5936
520
  SKIP(6);
5937
520
  return(XML_ATTRIBUTE_ENTITY);
5938
13.4k
     } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) {
5939
5.09k
  SKIP(8);
5940
5.09k
  return(XML_ATTRIBUTE_ENTITIES);
5941
8.37k
     } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) {
5942
307
  SKIP(8);
5943
307
  return(XML_ATTRIBUTE_NMTOKENS);
5944
8.06k
     } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) {
5945
715
  SKIP(7);
5946
715
  return(XML_ATTRIBUTE_NMTOKEN);
5947
715
     }
5948
7.35k
     return(xmlParseEnumeratedType(ctxt, tree));
5949
44.9k
}
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
12.5k
xmlParseAttributeListDecl(xmlParserCtxt *ctxt) {
5963
12.5k
    const xmlChar *elemName;
5964
12.5k
    const xmlChar *attrName;
5965
12.5k
    xmlEnumerationPtr tree;
5966
5967
12.5k
    if ((CUR != '<') || (NXT(1) != '!'))
5968
0
        return;
5969
12.5k
    SKIP(2);
5970
5971
12.5k
    if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) {
5972
12.5k
#ifdef LIBXML_VALID_ENABLED
5973
12.5k
  int oldInputNr = ctxt->inputNr;
5974
12.5k
#endif
5975
5976
12.5k
  SKIP(7);
5977
12.5k
  if (SKIP_BLANKS_PE == 0) {
5978
584
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5979
584
                     "Space required after '<!ATTLIST'\n");
5980
584
  }
5981
12.5k
        elemName = xmlParseName(ctxt);
5982
12.5k
  if (elemName == NULL) {
5983
47
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5984
47
         "ATTLIST: no name for Element\n");
5985
47
      return;
5986
47
  }
5987
12.4k
  SKIP_BLANKS_PE;
5988
12.4k
  GROW;
5989
55.5k
  while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) {
5990
45.5k
      int type;
5991
45.5k
      int def;
5992
45.5k
      xmlChar *defaultValue = NULL;
5993
5994
45.5k
      GROW;
5995
45.5k
            tree = NULL;
5996
45.5k
      attrName = xmlParseName(ctxt);
5997
45.5k
      if (attrName == NULL) {
5998
192
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5999
192
             "ATTLIST: no name for Attribute\n");
6000
192
    break;
6001
192
      }
6002
45.3k
      GROW;
6003
45.3k
      if (SKIP_BLANKS_PE == 0) {
6004
385
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6005
385
            "Space required after the attribute name\n");
6006
385
    break;
6007
385
      }
6008
6009
44.9k
      type = xmlParseAttributeType(ctxt, &tree);
6010
44.9k
      if (type <= 0) {
6011
370
          break;
6012
370
      }
6013
6014
44.5k
      GROW;
6015
44.5k
      if (SKIP_BLANKS_PE == 0) {
6016
197
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6017
197
             "Space required after the attribute type\n");
6018
197
          if (tree != NULL)
6019
133
        xmlFreeEnumeration(tree);
6020
197
    break;
6021
197
      }
6022
6023
44.3k
      def = xmlParseDefaultDecl(ctxt, &defaultValue);
6024
44.3k
      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
44.3k
      if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL))
6032
34.5k
          xmlAttrNormalizeSpace(defaultValue, defaultValue);
6033
6034
44.3k
      GROW;
6035
44.3k
            if (RAW != '>') {
6036
37.2k
    if (SKIP_BLANKS_PE == 0) {
6037
1.27k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6038
1.27k
      "Space required after the attribute default value\n");
6039
1.27k
        if (defaultValue != NULL)
6040
585
      xmlFree(defaultValue);
6041
1.27k
        if (tree != NULL)
6042
197
      xmlFreeEnumeration(tree);
6043
1.27k
        break;
6044
1.27k
    }
6045
37.2k
      }
6046
43.0k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6047
36.1k
    (ctxt->sax->attributeDecl != NULL))
6048
0
    ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName,
6049
0
                          type, def, defaultValue, tree);
6050
43.0k
      else if (tree != NULL)
6051
6.65k
    xmlFreeEnumeration(tree);
6052
6053
43.0k
      if ((ctxt->sax2) && (defaultValue != NULL) &&
6054
37.0k
          (def != XML_ATTRIBUTE_IMPLIED) &&
6055
37.0k
    (def != XML_ATTRIBUTE_REQUIRED)) {
6056
37.0k
    xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue);
6057
37.0k
      }
6058
43.0k
      if (ctxt->sax2) {
6059
43.0k
    xmlAddSpecialAttr(ctxt, elemName, attrName, type);
6060
43.0k
      }
6061
43.0k
      if (defaultValue != NULL)
6062
37.0k
          xmlFree(defaultValue);
6063
43.0k
      GROW;
6064
43.0k
  }
6065
12.4k
  if (RAW == '>') {
6066
10.1k
#ifdef LIBXML_VALID_ENABLED
6067
10.1k
      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
10.1k
#endif
6074
10.1k
      NEXT;
6075
10.1k
  }
6076
12.4k
    }
6077
12.5k
}
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
328k
xmlSkipBlankCharsPEBalanced(xmlParserCtxt *ctxt, int openInputNr) {
6088
328k
    SKIP_BLANKS;
6089
328k
    GROW;
6090
6091
328k
    (void) openInputNr;
6092
6093
328k
    if (!PARSER_EXTERNAL(ctxt) && !PARSER_IN_PE(ctxt))
6094
328k
        return;
6095
6096
0
    while (!PARSER_STOPPED(ctxt)) {
6097
0
        if (ctxt->input->cur >= ctxt->input->end) {
6098
0
#ifdef LIBXML_VALID_ENABLED
6099
0
            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
0
#endif
6106
0
            if (PARSER_IN_PE(ctxt))
6107
0
                xmlPopPE(ctxt);
6108
0
            else
6109
0
                break;
6110
0
        } else if (RAW == '%') {
6111
0
            xmlParsePERefInternal(ctxt, 0);
6112
0
        } else {
6113
0
            break;
6114
0
        }
6115
6116
0
        SKIP_BLANKS;
6117
0
        GROW;
6118
0
    }
6119
0
}
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
1.86k
xmlParseElementMixedContentDecl(xmlParserCtxt *ctxt, int openInputNr) {
6143
1.86k
    xmlElementContentPtr ret = NULL, cur = NULL, n;
6144
1.86k
    const xmlChar *elem = NULL;
6145
6146
1.86k
    GROW;
6147
1.86k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6148
1.86k
  SKIP(7);
6149
1.86k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6150
1.86k
  if (RAW == ')') {
6151
958
#ifdef LIBXML_VALID_ENABLED
6152
958
      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
958
#endif
6159
958
      NEXT;
6160
958
      ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6161
958
      if (ret == NULL)
6162
0
                goto mem_error;
6163
958
      if (RAW == '*') {
6164
405
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6165
405
    NEXT;
6166
405
      }
6167
958
      return(ret);
6168
958
  }
6169
911
  if ((RAW == '(') || (RAW == '|')) {
6170
900
      ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6171
900
      if (ret == NULL)
6172
0
                goto mem_error;
6173
900
  }
6174
2.63k
  while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) {
6175
1.77k
      NEXT;
6176
1.77k
            n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6177
1.77k
            if (n == NULL)
6178
0
                goto mem_error;
6179
1.77k
      if (elem == NULL) {
6180
897
    n->c1 = cur;
6181
897
    if (cur != NULL)
6182
897
        cur->parent = n;
6183
897
    ret = cur = n;
6184
897
      } else {
6185
877
          cur->c2 = n;
6186
877
    n->parent = cur;
6187
877
    n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6188
877
                if (n->c1 == NULL)
6189
0
                    goto mem_error;
6190
877
    n->c1->parent = n;
6191
877
    cur = n;
6192
877
      }
6193
1.77k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6194
1.77k
      elem = xmlParseName(ctxt);
6195
1.77k
      if (elem == NULL) {
6196
49
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6197
49
      "xmlParseElementMixedContentDecl : Name expected\n");
6198
49
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6199
49
    return(NULL);
6200
49
      }
6201
1.72k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6202
1.72k
  }
6203
862
  if ((RAW == ')') && (NXT(1) == '*')) {
6204
800
      if (elem != NULL) {
6205
800
    cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem,
6206
800
                                   XML_ELEMENT_CONTENT_ELEMENT);
6207
800
    if (cur->c2 == NULL)
6208
0
                    goto mem_error;
6209
800
    cur->c2->parent = cur;
6210
800
            }
6211
800
            if (ret != NULL)
6212
800
                ret->ocur = XML_ELEMENT_CONTENT_MULT;
6213
800
#ifdef LIBXML_VALID_ENABLED
6214
800
      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
800
#endif
6221
800
      SKIP(2);
6222
800
  } else {
6223
62
      xmlFreeDocElementContent(ctxt->myDoc, ret);
6224
62
      xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL);
6225
62
      return(NULL);
6226
62
  }
6227
6228
862
    } else {
6229
0
  xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL);
6230
0
    }
6231
800
    return(ret);
6232
6233
0
mem_error:
6234
0
    xmlErrMemory(ctxt);
6235
0
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6236
0
    return(NULL);
6237
1.86k
}
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
122k
                                       int depth) {
6272
122k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
6273
122k
    xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL;
6274
122k
    const xmlChar *elem;
6275
122k
    xmlChar type = 0;
6276
6277
122k
    if (depth > maxDepth) {
6278
5
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
6279
5
                "xmlParseElementChildrenContentDecl : depth %d too deep, "
6280
5
                "use XML_PARSE_HUGE\n", depth);
6281
5
  return(NULL);
6282
5
    }
6283
122k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6284
122k
    if (RAW == '(') {
6285
107k
        int newInputNr = ctxt->inputNr;
6286
6287
        /* Recurse on first child */
6288
107k
  NEXT;
6289
107k
        cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6290
107k
                                                           depth + 1);
6291
107k
        if (cur == NULL)
6292
84.0k
            return(NULL);
6293
107k
    } else {
6294
15.7k
  elem = xmlParseName(ctxt);
6295
15.7k
  if (elem == NULL) {
6296
173
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6297
173
      return(NULL);
6298
173
  }
6299
15.5k
        cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6300
15.5k
  if (cur == NULL) {
6301
0
      xmlErrMemory(ctxt);
6302
0
      return(NULL);
6303
0
  }
6304
15.5k
  GROW;
6305
15.5k
  if (RAW == '?') {
6306
1.83k
      cur->ocur = XML_ELEMENT_CONTENT_OPT;
6307
1.83k
      NEXT;
6308
13.7k
  } else if (RAW == '*') {
6309
1.06k
      cur->ocur = XML_ELEMENT_CONTENT_MULT;
6310
1.06k
      NEXT;
6311
12.6k
  } else if (RAW == '+') {
6312
2.23k
      cur->ocur = XML_ELEMENT_CONTENT_PLUS;
6313
2.23k
      NEXT;
6314
10.4k
  } else {
6315
10.4k
      cur->ocur = XML_ELEMENT_CONTENT_ONCE;
6316
10.4k
  }
6317
15.5k
  GROW;
6318
15.5k
    }
6319
113k
    while (!PARSER_STOPPED(ctxt)) {
6320
113k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6321
113k
        if (RAW == ')')
6322
31.8k
            break;
6323
        /*
6324
   * Each loop we parse one separator and one element.
6325
   */
6326
81.1k
        if (RAW == ',') {
6327
61.4k
      if (type == 0) type = CUR;
6328
6329
      /*
6330
       * Detect "Name | Name , Name" error
6331
       */
6332
56.6k
      else if (type != CUR) {
6333
4
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6334
4
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6335
4
                      type);
6336
4
    if ((last != NULL) && (last != ret))
6337
4
        xmlFreeDocElementContent(ctxt->myDoc, last);
6338
4
    if (ret != NULL)
6339
4
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6340
4
    return(NULL);
6341
4
      }
6342
61.4k
      NEXT;
6343
6344
61.4k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ);
6345
61.4k
      if (op == NULL) {
6346
0
                xmlErrMemory(ctxt);
6347
0
    if ((last != NULL) && (last != ret))
6348
0
        xmlFreeDocElementContent(ctxt->myDoc, last);
6349
0
          xmlFreeDocElementContent(ctxt->myDoc, ret);
6350
0
    return(NULL);
6351
0
      }
6352
61.4k
      if (last == NULL) {
6353
4.76k
    op->c1 = ret;
6354
4.76k
    if (ret != NULL)
6355
4.76k
        ret->parent = op;
6356
4.76k
    ret = cur = op;
6357
56.6k
      } else {
6358
56.6k
          cur->c2 = op;
6359
56.6k
    if (op != NULL)
6360
56.6k
        op->parent = cur;
6361
56.6k
    op->c1 = last;
6362
56.6k
    if (last != NULL)
6363
56.6k
        last->parent = op;
6364
56.6k
    cur =op;
6365
56.6k
    last = NULL;
6366
56.6k
      }
6367
61.4k
  } else if (RAW == '|') {
6368
19.4k
      if (type == 0) type = CUR;
6369
6370
      /*
6371
       * Detect "Name , Name | Name" error
6372
       */
6373
11.1k
      else if (type != CUR) {
6374
3
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6375
3
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6376
3
          type);
6377
3
    if ((last != NULL) && (last != ret))
6378
3
        xmlFreeDocElementContent(ctxt->myDoc, last);
6379
3
    if (ret != NULL)
6380
3
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6381
3
    return(NULL);
6382
3
      }
6383
19.4k
      NEXT;
6384
6385
19.4k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6386
19.4k
      if (op == NULL) {
6387
0
                xmlErrMemory(ctxt);
6388
0
    if ((last != NULL) && (last != ret))
6389
0
        xmlFreeDocElementContent(ctxt->myDoc, last);
6390
0
    if (ret != NULL)
6391
0
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6392
0
    return(NULL);
6393
0
      }
6394
19.4k
      if (last == NULL) {
6395
8.32k
    op->c1 = ret;
6396
8.32k
    if (ret != NULL)
6397
8.32k
        ret->parent = op;
6398
8.32k
    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
19.4k
  } else {
6410
255
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL);
6411
255
      if ((last != NULL) && (last != ret))
6412
112
          xmlFreeDocElementContent(ctxt->myDoc, last);
6413
255
      if (ret != NULL)
6414
255
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6415
255
      return(NULL);
6416
255
  }
6417
80.8k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6418
80.8k
        if (RAW == '(') {
6419
11.6k
            int newInputNr = ctxt->inputNr;
6420
6421
      /* Recurse on second child */
6422
11.6k
      NEXT;
6423
11.6k
      last = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6424
11.6k
                                                          depth + 1);
6425
11.6k
            if (last == NULL) {
6426
6.44k
    if (ret != NULL)
6427
6.44k
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6428
6.44k
    return(NULL);
6429
6.44k
            }
6430
69.2k
  } else {
6431
69.2k
      elem = xmlParseName(ctxt);
6432
69.2k
      if (elem == NULL) {
6433
78
    xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6434
78
    if (ret != NULL)
6435
78
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6436
78
    return(NULL);
6437
78
      }
6438
69.1k
      last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6439
69.1k
      if (last == NULL) {
6440
0
                xmlErrMemory(ctxt);
6441
0
    if (ret != NULL)
6442
0
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6443
0
    return(NULL);
6444
0
      }
6445
69.1k
      if (RAW == '?') {
6446
614
    last->ocur = XML_ELEMENT_CONTENT_OPT;
6447
614
    NEXT;
6448
68.5k
      } else if (RAW == '*') {
6449
466
    last->ocur = XML_ELEMENT_CONTENT_MULT;
6450
466
    NEXT;
6451
68.0k
      } else if (RAW == '+') {
6452
5.54k
    last->ocur = XML_ELEMENT_CONTENT_PLUS;
6453
5.54k
    NEXT;
6454
62.5k
      } else {
6455
62.5k
    last->ocur = XML_ELEMENT_CONTENT_ONCE;
6456
62.5k
      }
6457
69.1k
  }
6458
80.8k
    }
6459
31.8k
    if ((cur != NULL) && (last != NULL)) {
6460
6.45k
        cur->c2 = last;
6461
6.45k
  if (last != NULL)
6462
6.45k
      last->parent = cur;
6463
6.45k
    }
6464
31.8k
#ifdef LIBXML_VALID_ENABLED
6465
31.8k
    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
31.8k
#endif
6472
31.8k
    NEXT;
6473
31.8k
    if (RAW == '?') {
6474
3.24k
  if (ret != NULL) {
6475
3.24k
      if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) ||
6476
3.08k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6477
838
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6478
2.40k
      else
6479
2.40k
          ret->ocur = XML_ELEMENT_CONTENT_OPT;
6480
3.24k
  }
6481
3.24k
  NEXT;
6482
28.6k
    } else if (RAW == '*') {
6483
3.61k
  if (ret != NULL) {
6484
3.61k
      ret->ocur = XML_ELEMENT_CONTENT_MULT;
6485
3.61k
      cur = ret;
6486
      /*
6487
       * Some normalization:
6488
       * (a | b* | c?)* == (a | b | c)*
6489
       */
6490
7.29k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6491
3.67k
    if ((cur->c1 != NULL) &&
6492
3.67k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6493
3.65k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT)))
6494
1.24k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6495
3.67k
    if ((cur->c2 != NULL) &&
6496
3.67k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6497
3.66k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT)))
6498
137
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6499
3.67k
    cur = cur->c2;
6500
3.67k
      }
6501
3.61k
  }
6502
3.61k
  NEXT;
6503
25.0k
    } else if (RAW == '+') {
6504
7.91k
  if (ret != NULL) {
6505
7.91k
      int found = 0;
6506
6507
7.91k
      if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) ||
6508
5.87k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6509
3.78k
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6510
4.12k
      else
6511
4.12k
          ret->ocur = XML_ELEMENT_CONTENT_PLUS;
6512
      /*
6513
       * Some normalization:
6514
       * (a | b*)+ == (a | b)*
6515
       * (a | b?)+ == (a | b)*
6516
       */
6517
11.9k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6518
4.03k
    if ((cur->c1 != NULL) &&
6519
4.03k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6520
3.90k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) {
6521
856
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6522
856
        found = 1;
6523
856
    }
6524
4.03k
    if ((cur->c2 != NULL) &&
6525
4.03k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6526
3.65k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) {
6527
570
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6528
570
        found = 1;
6529
570
    }
6530
4.03k
    cur = cur->c2;
6531
4.03k
      }
6532
7.91k
      if (found)
6533
1.27k
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6534
7.91k
  }
6535
7.91k
  NEXT;
6536
7.91k
    }
6537
31.8k
    return(ret);
6538
38.6k
}
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
5.98k
                           xmlElementContent **result) {
6593
6594
5.98k
    xmlElementContentPtr tree = NULL;
6595
5.98k
    int openInputNr = ctxt->inputNr;
6596
5.98k
    int res;
6597
6598
5.98k
    *result = NULL;
6599
6600
5.98k
    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
5.98k
    NEXT;
6606
5.98k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6607
5.98k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6608
1.86k
        tree = xmlParseElementMixedContentDecl(ctxt, openInputNr);
6609
1.86k
  res = XML_ELEMENT_TYPE_MIXED;
6610
4.11k
    } else {
6611
4.11k
        tree = xmlParseElementChildrenContentDeclPriv(ctxt, openInputNr, 1);
6612
4.11k
  res = XML_ELEMENT_TYPE_ELEMENT;
6613
4.11k
    }
6614
5.98k
    if (tree == NULL)
6615
629
        return(-1);
6616
5.35k
    SKIP_BLANKS_PE;
6617
5.35k
    *result = tree;
6618
5.35k
    return(res);
6619
5.98k
}
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
7.21k
xmlParseElementDecl(xmlParserCtxt *ctxt) {
6636
7.21k
    const xmlChar *name;
6637
7.21k
    int ret = -1;
6638
7.21k
    xmlElementContentPtr content  = NULL;
6639
6640
7.21k
    if ((CUR != '<') || (NXT(1) != '!'))
6641
0
        return(ret);
6642
7.21k
    SKIP(2);
6643
6644
    /* GROW; done in the caller */
6645
7.21k
    if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) {
6646
7.14k
#ifdef LIBXML_VALID_ENABLED
6647
7.14k
  int oldInputNr = ctxt->inputNr;
6648
7.14k
#endif
6649
6650
7.14k
  SKIP(7);
6651
7.14k
  if (SKIP_BLANKS_PE == 0) {
6652
25
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6653
25
               "Space required after 'ELEMENT'\n");
6654
25
      return(-1);
6655
25
  }
6656
7.11k
        name = xmlParseName(ctxt);
6657
7.11k
  if (name == NULL) {
6658
48
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6659
48
         "xmlParseElementDecl: no name for Element\n");
6660
48
      return(-1);
6661
48
  }
6662
7.06k
  if (SKIP_BLANKS_PE == 0) {
6663
101
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6664
101
         "Space required after the element name\n");
6665
101
  }
6666
7.06k
  if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) {
6667
379
      SKIP(5);
6668
      /*
6669
       * Element must always be empty.
6670
       */
6671
379
      ret = XML_ELEMENT_TYPE_EMPTY;
6672
6.68k
  } else if ((RAW == 'A') && (NXT(1) == 'N') &&
6673
537
             (NXT(2) == 'Y')) {
6674
533
      SKIP(3);
6675
      /*
6676
       * Element is a generic container.
6677
       */
6678
533
      ret = XML_ELEMENT_TYPE_ANY;
6679
6.15k
  } else if (RAW == '(') {
6680
5.98k
      ret = xmlParseElementContentDecl(ctxt, name, &content);
6681
5.98k
            if (ret <= 0)
6682
629
                return(-1);
6683
5.98k
  } else {
6684
      /*
6685
       * [ WFC: PEs in Internal Subset ] error handling.
6686
       */
6687
170
            xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6688
170
                  "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n");
6689
170
      return(-1);
6690
170
  }
6691
6692
6.26k
  SKIP_BLANKS_PE;
6693
6694
6.26k
  if (RAW != '>') {
6695
110
      xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
6696
110
      if (content != NULL) {
6697
64
    xmlFreeDocElementContent(ctxt->myDoc, content);
6698
64
      }
6699
6.15k
  } else {
6700
6.15k
#ifdef LIBXML_VALID_ENABLED
6701
6.15k
      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
6.15k
#endif
6708
6709
6.15k
      NEXT;
6710
6.15k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6711
4.05k
    (ctxt->sax->elementDecl != NULL)) {
6712
0
    if (content != NULL)
6713
0
        content->parent = NULL;
6714
0
          ctxt->sax->elementDecl(ctxt->userData, name, ret,
6715
0
                           content);
6716
0
    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
0
        xmlFreeDocElementContent(ctxt->myDoc, content);
6724
0
    }
6725
6.15k
      } else if (content != NULL) {
6726
5.29k
    xmlFreeDocElementContent(ctxt->myDoc, content);
6727
5.29k
      }
6728
6.15k
  }
6729
6.26k
    }
6730
6.34k
    return(ret);
6731
7.21k
}
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
0
xmlParseConditionalSections(xmlParserCtxtPtr ctxt) {
6747
0
    size_t depth = 0;
6748
0
    int isFreshPE = 0;
6749
0
    int oldInputNr = ctxt->inputNr;
6750
0
    int declInputNr = ctxt->inputNr;
6751
6752
0
    while (!PARSER_STOPPED(ctxt)) {
6753
0
        if (ctxt->input->cur >= ctxt->input->end) {
6754
0
            if (ctxt->inputNr <= oldInputNr) {
6755
0
                xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6756
0
                return;
6757
0
            }
6758
6759
0
            xmlPopPE(ctxt);
6760
0
            declInputNr = ctxt->inputNr;
6761
0
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6762
0
            SKIP(3);
6763
0
            SKIP_BLANKS_PE;
6764
6765
0
            isFreshPE = 0;
6766
6767
0
            if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) {
6768
0
                SKIP(7);
6769
0
                SKIP_BLANKS_PE;
6770
0
                if (RAW != '[') {
6771
0
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6772
0
                    return;
6773
0
                }
6774
0
#ifdef LIBXML_VALID_ENABLED
6775
0
                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
0
#endif
6782
0
                NEXT;
6783
6784
0
                depth++;
6785
0
            } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) {
6786
0
                size_t ignoreDepth = 0;
6787
6788
0
                SKIP(6);
6789
0
                SKIP_BLANKS_PE;
6790
0
                if (RAW != '[') {
6791
0
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6792
0
                    return;
6793
0
                }
6794
0
#ifdef LIBXML_VALID_ENABLED
6795
0
                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
0
#endif
6802
0
                NEXT;
6803
6804
0
                while (PARSER_STOPPED(ctxt) == 0) {
6805
0
                    if (RAW == 0) {
6806
0
                        xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL);
6807
0
                        return;
6808
0
                    }
6809
0
                    if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6810
0
                        SKIP(3);
6811
0
                        ignoreDepth++;
6812
                        /* Check for integer overflow */
6813
0
                        if (ignoreDepth == 0) {
6814
0
                            xmlErrMemory(ctxt);
6815
0
                            return;
6816
0
                        }
6817
0
                    } else if ((RAW == ']') && (NXT(1) == ']') &&
6818
0
                               (NXT(2) == '>')) {
6819
0
                        SKIP(3);
6820
0
                        if (ignoreDepth == 0)
6821
0
                            break;
6822
0
                        ignoreDepth--;
6823
0
                    } else {
6824
0
                        NEXT;
6825
0
                    }
6826
0
                }
6827
6828
0
#ifdef LIBXML_VALID_ENABLED
6829
0
                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
0
#endif
6836
0
            } else {
6837
0
                xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL);
6838
0
                return;
6839
0
            }
6840
0
        } else if ((depth > 0) &&
6841
0
                   (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) {
6842
0
            if (isFreshPE) {
6843
0
                xmlFatalErrMsg(ctxt, XML_ERR_CONDSEC_INVALID,
6844
0
                               "Parameter entity must match "
6845
0
                               "extSubsetDecl\n");
6846
0
                return;
6847
0
            }
6848
6849
0
            depth--;
6850
0
#ifdef LIBXML_VALID_ENABLED
6851
0
            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
0
#endif
6858
0
            SKIP(3);
6859
0
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
6860
0
            isFreshPE = 0;
6861
0
            xmlParseMarkupDecl(ctxt);
6862
0
        } else if (RAW == '%') {
6863
0
            xmlParsePERefInternal(ctxt, 1);
6864
0
            if (ctxt->inputNr > declInputNr) {
6865
0
                isFreshPE = 1;
6866
0
                declInputNr = ctxt->inputNr;
6867
0
            }
6868
0
        } else {
6869
0
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6870
0
            return;
6871
0
        }
6872
6873
0
        if (depth == 0)
6874
0
            break;
6875
6876
0
        SKIP_BLANKS;
6877
0
        SHRINK;
6878
0
        GROW;
6879
0
    }
6880
0
}
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
118k
xmlParseMarkupDecl(xmlParserCtxt *ctxt) {
6907
118k
    GROW;
6908
118k
    if (CUR == '<') {
6909
118k
        if (NXT(1) == '!') {
6910
96.9k
      switch (NXT(2)) {
6911
58.6k
          case 'E':
6912
58.6k
        if (NXT(3) == 'L')
6913
7.21k
      xmlParseElementDecl(ctxt);
6914
51.3k
        else if (NXT(3) == 'N')
6915
51.3k
      xmlParseEntityDecl(ctxt);
6916
44
                    else
6917
44
                        SKIP(2);
6918
58.6k
        break;
6919
12.5k
          case 'A':
6920
12.5k
        xmlParseAttributeListDecl(ctxt);
6921
12.5k
        break;
6922
3.13k
          case 'N':
6923
3.13k
        xmlParseNotationDecl(ctxt);
6924
3.13k
        break;
6925
22.3k
          case '-':
6926
22.3k
        xmlParseComment(ctxt);
6927
22.3k
        break;
6928
192
    default:
6929
192
                    xmlFatalErr(ctxt,
6930
192
                                ctxt->inSubset == 2 ?
6931
0
                                    XML_ERR_EXT_SUBSET_NOT_FINISHED :
6932
192
                                    XML_ERR_INT_SUBSET_NOT_FINISHED,
6933
192
                                NULL);
6934
192
                    SKIP(2);
6935
192
        break;
6936
96.9k
      }
6937
96.9k
  } else if (NXT(1) == '?') {
6938
21.4k
      xmlParsePI(ctxt);
6939
21.4k
  }
6940
118k
    }
6941
118k
}
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
0
xmlParseTextDecl(xmlParserCtxt *ctxt) {
6954
0
    xmlChar *version;
6955
6956
    /*
6957
     * We know that '<?xml' is here.
6958
     */
6959
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
6960
0
  SKIP(5);
6961
0
    } else {
6962
0
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL);
6963
0
  return;
6964
0
    }
6965
6966
0
    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
0
    version = xmlParseVersionInfo(ctxt);
6975
0
    if (version == NULL) {
6976
0
  version = xmlCharStrdup(XML_DEFAULT_VERSION);
6977
0
        if (version == NULL) {
6978
0
            xmlErrMemory(ctxt);
6979
0
            return;
6980
0
        }
6981
0
    } else {
6982
0
  if (SKIP_BLANKS == 0) {
6983
0
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6984
0
               "Space needed here\n");
6985
0
  }
6986
0
    }
6987
0
    ctxt->input->version = version;
6988
6989
    /*
6990
     * We must have the encoding declaration
6991
     */
6992
0
    xmlParseEncodingDecl(ctxt);
6993
6994
0
    SKIP_BLANKS;
6995
0
    if ((RAW == '?') && (NXT(1) == '>')) {
6996
0
        SKIP(2);
6997
0
    } else if (RAW == '>') {
6998
        /* Deprecated old WD ... */
6999
0
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7000
0
  NEXT;
7001
0
    } else {
7002
0
        int c;
7003
7004
0
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7005
0
        while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) {
7006
0
            NEXT;
7007
0
            if (c == '>')
7008
0
                break;
7009
0
        }
7010
0
    }
7011
0
}
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
0
                       const xmlChar *systemId) {
7029
0
    int oldInputNr;
7030
7031
0
    xmlCtxtInitializeLate(ctxt);
7032
7033
0
    xmlDetectEncoding(ctxt);
7034
7035
0
    if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) {
7036
0
  xmlParseTextDecl(ctxt);
7037
0
    }
7038
0
    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
0
    if ((ctxt->myDoc->intSubset == NULL) &&
7047
0
        (xmlCreateIntSubset(ctxt->myDoc, NULL, publicId, systemId) == NULL)) {
7048
0
        xmlErrMemory(ctxt);
7049
0
    }
7050
7051
0
    ctxt->inSubset = 2;
7052
0
    oldInputNr = ctxt->inputNr;
7053
7054
0
    SKIP_BLANKS;
7055
0
    while (!PARSER_STOPPED(ctxt)) {
7056
0
        if (ctxt->input->cur >= ctxt->input->end) {
7057
0
            if (ctxt->inputNr <= oldInputNr) {
7058
0
                xmlParserCheckEOF(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED);
7059
0
                break;
7060
0
            }
7061
7062
0
            xmlPopPE(ctxt);
7063
0
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
7064
0
            xmlParseConditionalSections(ctxt);
7065
0
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
7066
0
            xmlParseMarkupDecl(ctxt);
7067
0
        } else if (RAW == '%') {
7068
0
            xmlParsePERefInternal(ctxt, 1);
7069
0
        } else {
7070
0
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
7071
7072
0
            while (ctxt->inputNr > oldInputNr)
7073
0
                xmlPopPE(ctxt);
7074
0
            break;
7075
0
        }
7076
0
        SKIP_BLANKS;
7077
0
        SHRINK;
7078
0
        GROW;
7079
0
    }
7080
0
}
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
328k
xmlParseReference(xmlParserCtxt *ctxt) {
7097
328k
    xmlEntityPtr ent = NULL;
7098
328k
    const xmlChar *name;
7099
328k
    xmlChar *val;
7100
7101
328k
    if (RAW != '&')
7102
0
        return;
7103
7104
    /*
7105
     * Simple case of a CharRef
7106
     */
7107
328k
    if (NXT(1) == '#') {
7108
71.5k
  int i = 0;
7109
71.5k
  xmlChar out[16];
7110
71.5k
  int value = xmlParseCharRef(ctxt);
7111
7112
71.5k
  if (value == 0)
7113
172
      return;
7114
7115
        /*
7116
         * Just encode the value in UTF-8
7117
         */
7118
71.4k
        COPY_BUF(out, i, value);
7119
71.4k
        out[i] = 0;
7120
71.4k
        if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7121
71.4k
            (!ctxt->disableSAX))
7122
71.3k
            ctxt->sax->characters(ctxt->userData, out, i);
7123
71.4k
  return;
7124
71.5k
    }
7125
7126
    /*
7127
     * We are seeing an entity reference
7128
     */
7129
256k
    name = xmlParseEntityRefInternal(ctxt);
7130
256k
    if (name == NULL)
7131
283
        return;
7132
256k
    ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0);
7133
256k
    if (ent == NULL) {
7134
        /*
7135
         * Create a reference for undeclared entities.
7136
         */
7137
943
        if ((ctxt->replaceEntities == 0) &&
7138
0
            (ctxt->sax != NULL) &&
7139
0
            (ctxt->disableSAX == 0) &&
7140
0
            (ctxt->sax->reference != NULL)) {
7141
0
            ctxt->sax->reference(ctxt->userData, name);
7142
0
        }
7143
943
        return;
7144
943
    }
7145
255k
    if (!ctxt->wellFormed)
7146
0
  return;
7147
7148
    /* special case of predefined entities */
7149
255k
    if ((ent->name == NULL) ||
7150
255k
        (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
7151
255k
  val = ent->content;
7152
255k
  if (val == NULL) return;
7153
  /*
7154
   * inline the entity.
7155
   */
7156
255k
  if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7157
255k
      (!ctxt->disableSAX))
7158
255k
      ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val));
7159
255k
  return;
7160
255k
    }
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
18.4E
    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
18.4E
    if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) ||
7211
0
        (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
7212
0
         ((ctxt->replaceEntities) ||
7213
0
          (ctxt->validate)))) {
7214
0
        if ((ent->flags & XML_ENT_PARSED) == 0) {
7215
0
            xmlCtxtParseEntity(ctxt, ent);
7216
0
        } 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
0
            xmlCtxtParseEntity(ctxt, ent);
7227
0
        }
7228
0
    }
7229
7230
    /*
7231
     * We also check for amplification if entities aren't substituted.
7232
     * They might be expanded later.
7233
     */
7234
18.4E
    if (xmlParserEntityCheck(ctxt, ent->expandedSize))
7235
0
        return;
7236
7237
18.4E
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
7238
0
        return;
7239
7240
18.4E
    if (ctxt->replaceEntities == 0) {
7241
  /*
7242
   * Create a reference
7243
   */
7244
0
        if (ctxt->sax->reference != NULL)
7245
0
      ctxt->sax->reference(ctxt->userData, ent->name);
7246
18.4E
    } else if ((ent->children != NULL) && (ctxt->node != NULL)) {
7247
0
        xmlNodePtr copy, cur;
7248
7249
        /*
7250
         * Seems we are generating the DOM content, copy the tree
7251
   */
7252
0
        cur = ent->children;
7253
7254
        /*
7255
         * Handle first text node with SAX to coalesce text efficiently
7256
         */
7257
0
        if ((cur->type == XML_TEXT_NODE) ||
7258
0
            (cur->type == XML_CDATA_SECTION_NODE)) {
7259
0
            int len = xmlStrlen(cur->content);
7260
7261
0
            if ((cur->type == XML_TEXT_NODE) ||
7262
0
                (ctxt->options & XML_PARSE_NOCDATA)) {
7263
0
                if (ctxt->sax->characters != NULL)
7264
0
                    ctxt->sax->characters(ctxt, cur->content, len);
7265
0
            } else {
7266
0
                if (ctxt->sax->cdataBlock != NULL)
7267
0
                    ctxt->sax->cdataBlock(ctxt, cur->content, len);
7268
0
            }
7269
7270
0
            cur = cur->next;
7271
0
        }
7272
7273
0
        while (cur != NULL) {
7274
0
            xmlNodePtr last;
7275
7276
            /*
7277
             * Handle last text node with SAX to coalesce text efficiently
7278
             */
7279
0
            if ((cur->next == NULL) &&
7280
0
                ((cur->type == XML_TEXT_NODE) ||
7281
0
                 (cur->type == XML_CDATA_SECTION_NODE))) {
7282
0
                int len = xmlStrlen(cur->content);
7283
7284
0
                if ((cur->type == XML_TEXT_NODE) ||
7285
0
                    (ctxt->options & XML_PARSE_NOCDATA)) {
7286
0
                    if (ctxt->sax->characters != NULL)
7287
0
                        ctxt->sax->characters(ctxt, cur->content, len);
7288
0
                } else {
7289
0
                    if (ctxt->sax->cdataBlock != NULL)
7290
0
                        ctxt->sax->cdataBlock(ctxt, cur->content, len);
7291
0
                }
7292
7293
0
                break;
7294
0
            }
7295
7296
            /*
7297
             * Reset coalesce buffer stats only for non-text nodes.
7298
             */
7299
0
            ctxt->nodemem = 0;
7300
0
            ctxt->nodelen = 0;
7301
7302
0
            copy = xmlDocCopyNode(cur, ctxt->myDoc, 1);
7303
7304
0
            if (copy == NULL) {
7305
0
                xmlErrMemory(ctxt);
7306
0
                break;
7307
0
            }
7308
7309
0
            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
0
            copy->parent = ctxt->node;
7317
0
            last = ctxt->node->last;
7318
0
            if (last == NULL) {
7319
0
                ctxt->node->children = copy;
7320
0
            } else {
7321
0
                last->next = copy;
7322
0
                copy->prev = last;
7323
0
            }
7324
0
            ctxt->node->last = copy;
7325
7326
0
            cur = cur->next;
7327
0
        }
7328
0
    }
7329
18.4E
}
7330
7331
static void
7332
31.5k
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
31.5k
    if ((ctxt->standalone == 1) ||
7355
29.4k
        ((ctxt->hasExternalSubset == 0) &&
7356
28.7k
         (ctxt->hasPErefs == 0))) {
7357
20.9k
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY,
7358
20.9k
                          "Entity '%s' not defined\n", name);
7359
20.9k
#ifdef LIBXML_VALID_ENABLED
7360
20.9k
    } 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
10.5k
    } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) ||
7372
10.5k
               ((ctxt->replaceEntities) &&
7373
10.5k
                ((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
10.5k
        xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY,
7383
10.5k
                     "Entity '%s' not defined\n", name);
7384
10.5k
    } else {
7385
0
        xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7386
0
                      "Entity '%s' not defined\n", name, NULL);
7387
0
    }
7388
7389
31.5k
    ctxt->valid = 0;
7390
31.5k
}
7391
7392
static xmlEntityPtr
7393
433k
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) {
7394
433k
    xmlEntityPtr ent = NULL;
7395
7396
    /*
7397
     * Predefined entities override any extra definition
7398
     */
7399
433k
    if ((ctxt->options & XML_PARSE_OLDSAX) == 0) {
7400
433k
        ent = xmlGetPredefinedEntity(name);
7401
433k
        if (ent != NULL)
7402
411k
            return(ent);
7403
433k
    }
7404
7405
    /*
7406
     * Ask first SAX for entity resolution, otherwise try the
7407
     * entities which may have stored in the parser context.
7408
     */
7409
22.0k
    if (ctxt->sax != NULL) {
7410
22.0k
  if (ctxt->sax->getEntity != NULL)
7411
22.0k
      ent = ctxt->sax->getEntity(ctxt->userData, name);
7412
22.0k
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7413
1.27k
      (ctxt->options & XML_PARSE_OLDSAX))
7414
0
      ent = xmlGetPredefinedEntity(name);
7415
22.0k
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7416
1.27k
      (ctxt->userData==ctxt)) {
7417
0
      ent = xmlSAX2GetEntity(ctxt, name);
7418
0
  }
7419
22.0k
    }
7420
7421
22.0k
    if (ent == NULL) {
7422
22.0k
        xmlHandleUndeclaredEntity(ctxt, name);
7423
22.0k
    }
7424
7425
    /*
7426
     * [ WFC: Parsed Entity ]
7427
     * An entity reference must not contain the name of an
7428
     * unparsed entity
7429
     */
7430
0
    else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) {
7431
0
  xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY,
7432
0
     "Entity reference to unparsed entity %s\n", name);
7433
0
        ent = NULL;
7434
0
    }
7435
7436
    /*
7437
     * [ WFC: No External Entity References ]
7438
     * Attribute values cannot contain direct or indirect
7439
     * entity references to external entities.
7440
     */
7441
0
    else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) {
7442
0
        if (inAttr) {
7443
0
            xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL,
7444
0
                 "Attribute references external entity '%s'\n", name);
7445
0
            ent = NULL;
7446
0
        }
7447
0
    }
7448
7449
22.0k
    return(ent);
7450
433k
}
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
717k
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) {
7462
717k
    const xmlChar *name;
7463
7464
717k
    GROW;
7465
7466
717k
    if (RAW != '&')
7467
0
        return(NULL);
7468
717k
    NEXT;
7469
717k
    name = xmlParseName(ctxt);
7470
717k
    if (name == NULL) {
7471
160k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7472
160k
           "xmlParseEntityRef: no name\n");
7473
160k
        return(NULL);
7474
160k
    }
7475
557k
    if (RAW != ';') {
7476
123k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7477
123k
  return(NULL);
7478
123k
    }
7479
433k
    NEXT;
7480
7481
433k
    return(name);
7482
557k
}
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
0
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) {
7535
0
    xmlChar *name;
7536
0
    const xmlChar *ptr;
7537
0
    xmlChar cur;
7538
7539
0
    if ((str == NULL) || (*str == NULL))
7540
0
        return(NULL);
7541
0
    ptr = *str;
7542
0
    cur = *ptr;
7543
0
    if (cur != '&')
7544
0
  return(NULL);
7545
7546
0
    ptr++;
7547
0
    name = xmlParseStringName(ctxt, &ptr);
7548
0
    if (name == NULL) {
7549
0
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7550
0
           "xmlParseStringEntityRef: no name\n");
7551
0
  *str = ptr;
7552
0
  return(NULL);
7553
0
    }
7554
0
    if (*ptr != ';') {
7555
0
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7556
0
        xmlFree(name);
7557
0
  *str = ptr;
7558
0
  return(NULL);
7559
0
    }
7560
0
    ptr++;
7561
7562
0
    *str = ptr;
7563
0
    return(name);
7564
0
}
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
11.7k
xmlParsePERefInternal(xmlParserCtxt *ctxt, int markupDecl) {
7598
11.7k
    const xmlChar *name;
7599
11.7k
    xmlEntityPtr entity = NULL;
7600
11.7k
    xmlParserInputPtr input;
7601
7602
11.7k
    if (RAW != '%')
7603
0
        return;
7604
11.7k
    NEXT;
7605
11.7k
    name = xmlParseName(ctxt);
7606
11.7k
    if (name == NULL) {
7607
3.64k
  xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n");
7608
3.64k
  return;
7609
3.64k
    }
7610
8.09k
    if (RAW != ';') {
7611
207
  xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL);
7612
207
        return;
7613
207
    }
7614
7615
7.88k
    NEXT;
7616
7617
    /* Must be set before xmlHandleUndeclaredEntity */
7618
7.88k
    ctxt->hasPErefs = 1;
7619
7620
    /*
7621
     * Request the entity from SAX
7622
     */
7623
7.88k
    if ((ctxt->sax != NULL) &&
7624
7.88k
  (ctxt->sax->getParameterEntity != NULL))
7625
0
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7626
7627
7.88k
    if (entity == NULL) {
7628
7.88k
        xmlHandleUndeclaredEntity(ctxt, name);
7629
7.88k
    } else {
7630
  /*
7631
   * Internal checking in case the entity quest barfed
7632
   */
7633
0
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7634
0
      (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
0
  } else {
7639
0
      if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) &&
7640
0
                ((ctxt->options & XML_PARSE_NO_XXE) ||
7641
0
     (((ctxt->loadsubset & ~XML_SKIP_IDS) == 0) &&
7642
0
      (ctxt->replaceEntities == 0) &&
7643
0
      (ctxt->validate == 0))))
7644
0
    return;
7645
7646
0
            if (entity->flags & XML_ENT_EXPANDING) {
7647
0
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
7648
0
                return;
7649
0
            }
7650
7651
0
      input = xmlNewEntityInputStream(ctxt, entity);
7652
0
      if (xmlCtxtPushInput(ctxt, input) < 0) {
7653
0
                xmlFreeInputStream(input);
7654
0
    return;
7655
0
            }
7656
7657
0
            entity->flags |= XML_ENT_EXPANDING;
7658
7659
0
            if (markupDecl)
7660
0
                input->flags |= XML_INPUT_MARKUP_DECL;
7661
7662
0
            GROW;
7663
7664
0
      if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
7665
0
                xmlDetectEncoding(ctxt);
7666
7667
0
                if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
7668
0
                    (IS_BLANK_CH(NXT(5)))) {
7669
0
                    xmlParseTextDecl(ctxt);
7670
0
                }
7671
0
            }
7672
0
  }
7673
0
    }
7674
7.88k
}
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
0
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) {
7697
0
    xmlParserInputPtr oldinput, input = NULL;
7698
0
    xmlParserInputPtr *oldinputTab;
7699
0
    xmlChar *oldencoding;
7700
0
    xmlChar *content = NULL;
7701
0
    xmlResourceType rtype;
7702
0
    size_t length, i;
7703
0
    int oldinputNr, oldinputMax;
7704
0
    int ret = -1;
7705
0
    int res;
7706
7707
0
    if ((ctxt == NULL) || (entity == NULL) ||
7708
0
        ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) &&
7709
0
   (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) ||
7710
0
  (entity->content != NULL)) {
7711
0
  xmlFatalErr(ctxt, XML_ERR_ARGUMENT,
7712
0
              "xmlLoadEntityContent parameter error");
7713
0
        return(-1);
7714
0
    }
7715
7716
0
    if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY)
7717
0
        rtype = XML_RESOURCE_PARAMETER_ENTITY;
7718
0
    else
7719
0
        rtype = XML_RESOURCE_GENERAL_ENTITY;
7720
7721
0
    input = xmlLoadResource(ctxt, (char *) entity->URI,
7722
0
                            (char *) entity->ExternalID, rtype);
7723
0
    if (input == NULL)
7724
0
        return(-1);
7725
7726
0
    oldinput = ctxt->input;
7727
0
    oldinputNr = ctxt->inputNr;
7728
0
    oldinputMax = ctxt->inputMax;
7729
0
    oldinputTab = ctxt->inputTab;
7730
0
    oldencoding = ctxt->encoding;
7731
7732
0
    ctxt->input = NULL;
7733
0
    ctxt->inputNr = 0;
7734
0
    ctxt->inputMax = 1;
7735
0
    ctxt->encoding = NULL;
7736
0
    ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr));
7737
0
    if (ctxt->inputTab == NULL) {
7738
0
        xmlErrMemory(ctxt);
7739
0
        xmlFreeInputStream(input);
7740
0
        goto error;
7741
0
    }
7742
7743
0
    xmlBufResetInput(input->buf->buffer, input);
7744
7745
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
7746
0
        xmlFreeInputStream(input);
7747
0
        goto error;
7748
0
    }
7749
7750
0
    xmlDetectEncoding(ctxt);
7751
7752
    /*
7753
     * Parse a possible text declaration first
7754
     */
7755
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
7756
0
  xmlParseTextDecl(ctxt);
7757
        /*
7758
         * An XML-1.0 document can't reference an entity not XML-1.0
7759
         */
7760
0
        if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
7761
0
            (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
7762
0
            xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
7763
0
                           "Version mismatch between document and entity\n");
7764
0
        }
7765
0
    }
7766
7767
0
    length = input->cur - input->base;
7768
0
    xmlBufShrink(input->buf->buffer, length);
7769
0
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7770
7771
0
    while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0)
7772
0
        ;
7773
7774
0
    xmlBufResetInput(input->buf->buffer, input);
7775
7776
0
    if (res < 0) {
7777
0
        xmlCtxtErrIO(ctxt, input->buf->error, NULL);
7778
0
        goto error;
7779
0
    }
7780
7781
0
    length = xmlBufUse(input->buf->buffer);
7782
0
    if (length > INT_MAX) {
7783
0
        xmlErrMemory(ctxt);
7784
0
        goto error;
7785
0
    }
7786
7787
0
    content = xmlStrndup(xmlBufContent(input->buf->buffer), length);
7788
0
    if (content == NULL) {
7789
0
        xmlErrMemory(ctxt);
7790
0
        goto error;
7791
0
    }
7792
7793
0
    for (i = 0; i < length; ) {
7794
0
        int clen = length - i;
7795
0
        int c = xmlGetUTF8Char(content + i, &clen);
7796
7797
0
        if ((c < 0) || (!IS_CHAR(c))) {
7798
0
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
7799
0
                              "xmlLoadEntityContent: invalid char value %d\n",
7800
0
                              content[i]);
7801
0
            goto error;
7802
0
        }
7803
0
        i += clen;
7804
0
    }
7805
7806
0
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7807
0
    entity->content = content;
7808
0
    entity->length = length;
7809
0
    content = NULL;
7810
0
    ret = 0;
7811
7812
0
error:
7813
0
    while (ctxt->inputNr > 0)
7814
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
7815
0
    xmlFree(ctxt->inputTab);
7816
0
    xmlFree(ctxt->encoding);
7817
7818
0
    ctxt->input = oldinput;
7819
0
    ctxt->inputNr = oldinputNr;
7820
0
    ctxt->inputMax = oldinputMax;
7821
0
    ctxt->inputTab = oldinputTab;
7822
0
    ctxt->encoding = oldencoding;
7823
7824
0
    xmlFree(content);
7825
7826
0
    return(ret);
7827
0
}
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
3.62k
xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) {
7860
3.62k
    const xmlChar *ptr;
7861
3.62k
    xmlChar cur;
7862
3.62k
    xmlChar *name;
7863
3.62k
    xmlEntityPtr entity = NULL;
7864
7865
3.62k
    if ((str == NULL) || (*str == NULL)) return(NULL);
7866
3.62k
    ptr = *str;
7867
3.62k
    cur = *ptr;
7868
3.62k
    if (cur != '%')
7869
0
        return(NULL);
7870
3.62k
    ptr++;
7871
3.62k
    name = xmlParseStringName(ctxt, &ptr);
7872
3.62k
    if (name == NULL) {
7873
1.09k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7874
1.09k
           "xmlParseStringPEReference: no name\n");
7875
1.09k
  *str = ptr;
7876
1.09k
  return(NULL);
7877
1.09k
    }
7878
2.52k
    cur = *ptr;
7879
2.52k
    if (cur != ';') {
7880
956
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7881
956
  xmlFree(name);
7882
956
  *str = ptr;
7883
956
  return(NULL);
7884
956
    }
7885
1.56k
    ptr++;
7886
7887
    /* Must be set before xmlHandleUndeclaredEntity */
7888
1.56k
    ctxt->hasPErefs = 1;
7889
7890
    /*
7891
     * Request the entity from SAX
7892
     */
7893
1.56k
    if ((ctxt->sax != NULL) &&
7894
1.56k
  (ctxt->sax->getParameterEntity != NULL))
7895
0
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7896
7897
1.56k
    if (entity == NULL) {
7898
1.56k
        xmlHandleUndeclaredEntity(ctxt, name);
7899
1.56k
    } else {
7900
  /*
7901
   * Internal checking in case the entity quest barfed
7902
   */
7903
0
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7904
0
      (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
0
    }
7910
7911
1.56k
    xmlFree(name);
7912
1.56k
    *str = ptr;
7913
1.56k
    return(entity);
7914
2.52k
}
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
7.57k
xmlParseDocTypeDecl(xmlParserCtxt *ctxt) {
7933
7.57k
    const xmlChar *name = NULL;
7934
7.57k
    xmlChar *publicId = NULL;
7935
7.57k
    xmlChar *URI = NULL;
7936
7937
    /*
7938
     * We know that '<!DOCTYPE' has been detected.
7939
     */
7940
7.57k
    SKIP(9);
7941
7942
7.57k
    if (SKIP_BLANKS == 0) {
7943
137
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7944
137
                       "Space required after 'DOCTYPE'\n");
7945
137
    }
7946
7947
    /*
7948
     * Parse the DOCTYPE name.
7949
     */
7950
7.57k
    name = xmlParseName(ctxt);
7951
7.57k
    if (name == NULL) {
7952
15
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7953
15
           "xmlParseDocTypeDecl : no DOCTYPE name !\n");
7954
15
    }
7955
7.57k
    ctxt->intSubName = name;
7956
7957
7.57k
    SKIP_BLANKS;
7958
7959
    /*
7960
     * Check for public and system identifier (URI)
7961
     */
7962
7.57k
    URI = xmlParseExternalID(ctxt, &publicId, 1);
7963
7964
7.57k
    if ((URI != NULL) || (publicId != NULL)) {
7965
330
        ctxt->hasExternalSubset = 1;
7966
330
    }
7967
7.57k
    ctxt->extSubURI = URI;
7968
7.57k
    ctxt->extSubSystem = publicId;
7969
7970
7.57k
    SKIP_BLANKS;
7971
7972
    /*
7973
     * Create and update the internal subset.
7974
     */
7975
7.57k
    if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) &&
7976
0
  (!ctxt->disableSAX))
7977
0
  ctxt->sax->internalSubset(ctxt->userData, name, publicId, URI);
7978
7979
7.57k
    if ((RAW != '[') && (RAW != '>')) {
7980
220
  xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
7981
220
    }
7982
7.57k
}
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
7.06k
xmlParseInternalSubset(xmlParserCtxtPtr ctxt) {
7993
    /*
7994
     * Is there any DTD definition ?
7995
     */
7996
7.06k
    if (RAW == '[') {
7997
7.06k
        int oldInputNr = ctxt->inputNr;
7998
7999
7.06k
        NEXT;
8000
  /*
8001
   * Parse the succession of Markup declarations and
8002
   * PEReferences.
8003
   * Subsequence (markupdecl | PEReference | S)*
8004
   */
8005
7.06k
  SKIP_BLANKS;
8006
137k
        while (1) {
8007
137k
            if (PARSER_STOPPED(ctxt)) {
8008
10
                return;
8009
137k
            } else if (ctxt->input->cur >= ctxt->input->end) {
8010
1.92k
                if (ctxt->inputNr <= oldInputNr) {
8011
1.92k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8012
1.92k
                    return;
8013
1.92k
                }
8014
0
                xmlPopPE(ctxt);
8015
135k
            } else if ((RAW == ']') && (ctxt->inputNr <= oldInputNr)) {
8016
2.44k
                NEXT;
8017
2.44k
                SKIP_BLANKS;
8018
2.44k
                break;
8019
132k
            } else if ((PARSER_EXTERNAL(ctxt)) &&
8020
0
                       (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
0
                xmlParseConditionalSections(ctxt);
8026
132k
            } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
8027
118k
                xmlParseMarkupDecl(ctxt);
8028
118k
            } else if (RAW == '%') {
8029
11.7k
                xmlParsePERefInternal(ctxt, 1);
8030
11.7k
            } else {
8031
2.68k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8032
8033
2.68k
                while (ctxt->inputNr > oldInputNr)
8034
0
                    xmlPopPE(ctxt);
8035
2.68k
                return;
8036
2.68k
            }
8037
130k
            SKIP_BLANKS;
8038
130k
            SHRINK;
8039
130k
            GROW;
8040
130k
        }
8041
7.06k
    }
8042
8043
    /*
8044
     * We should be at the end of the DOCTYPE declaration.
8045
     */
8046
2.44k
    if (RAW != '>') {
8047
42
        xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
8048
42
        return;
8049
42
    }
8050
2.40k
    NEXT;
8051
2.40k
}
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
55.5M
xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) {
8391
55.5M
    xmlHashedString l, p;
8392
55.5M
    int start, isNCName = 0;
8393
8394
55.5M
    l.name = NULL;
8395
55.5M
    p.name = NULL;
8396
8397
55.5M
    GROW;
8398
55.5M
    start = CUR_PTR - BASE_PTR;
8399
8400
55.5M
    l = xmlParseNCName(ctxt);
8401
55.5M
    if (l.name != NULL) {
8402
55.3M
        isNCName = 1;
8403
55.3M
        if (CUR == ':') {
8404
20.5M
            NEXT;
8405
20.5M
            p = l;
8406
20.5M
            l = xmlParseNCName(ctxt);
8407
20.5M
        }
8408
55.3M
    }
8409
55.5M
    if ((l.name == NULL) || (CUR == ':')) {
8410
558k
        xmlChar *tmp;
8411
8412
558k
        l.name = NULL;
8413
558k
        p.name = NULL;
8414
558k
        if ((isNCName == 0) && (CUR != ':'))
8415
58.6k
            return(l);
8416
499k
        tmp = xmlParseNmtoken(ctxt);
8417
499k
        if (tmp != NULL)
8418
419k
            xmlFree(tmp);
8419
499k
        l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start,
8420
499k
                                CUR_PTR - (BASE_PTR + start));
8421
499k
        if (l.name == NULL) {
8422
0
            xmlErrMemory(ctxt);
8423
0
            return(l);
8424
0
        }
8425
499k
        xmlNsErr(ctxt, XML_NS_ERR_QNAME,
8426
499k
                 "Failed to parse QName '%s'\n", l.name, NULL, NULL);
8427
499k
    }
8428
8429
55.5M
    *prefix = p;
8430
55.5M
    return(l);
8431
55.5M
}
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
10.9k
xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) {
8447
10.9k
    xmlHashedString n, p;
8448
8449
10.9k
    n = xmlParseQNameHashed(ctxt, &p);
8450
10.9k
    if (n.name == NULL)
8451
554
        return(NULL);
8452
10.4k
    *prefix = p.name;
8453
10.4k
    return(n.name);
8454
10.9k
}
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
5.43M
                        xmlChar const *prefix) {
8470
5.43M
    const xmlChar *cmp;
8471
5.43M
    const xmlChar *in;
8472
5.43M
    const xmlChar *ret;
8473
5.43M
    const xmlChar *prefix2;
8474
8475
5.43M
    if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name));
8476
8477
5.43M
    GROW;
8478
5.43M
    in = ctxt->input->cur;
8479
8480
5.43M
    cmp = prefix;
8481
16.6M
    while (*in != 0 && *in == *cmp) {
8482
11.2M
  ++in;
8483
11.2M
  ++cmp;
8484
11.2M
    }
8485
5.43M
    if ((*cmp == 0) && (*in == ':')) {
8486
5.42M
        in++;
8487
5.42M
  cmp = name;
8488
36.8M
  while (*in != 0 && *in == *cmp) {
8489
31.4M
      ++in;
8490
31.4M
      ++cmp;
8491
31.4M
  }
8492
5.42M
  if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
8493
      /* success */
8494
5.42M
            ctxt->input->col += in - ctxt->input->cur;
8495
5.42M
      ctxt->input->cur = in;
8496
5.42M
      return((const xmlChar*) 1);
8497
5.42M
  }
8498
5.42M
    }
8499
    /*
8500
     * all strings coms from the dictionary, equality can be done directly
8501
     */
8502
10.9k
    ret = xmlParseQName (ctxt, &prefix2);
8503
10.9k
    if (ret == NULL)
8504
554
        return(NULL);
8505
10.4k
    if ((ret == name) && (prefix == prefix2))
8506
222
  return((const xmlChar*) 1);
8507
10.1k
    return ret;
8508
10.4k
}
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
21.9M
{
8529
21.9M
    xmlHashedString hname;
8530
21.9M
    const xmlChar *prefix, *name;
8531
21.9M
    xmlChar *val = NULL, *internal_val = NULL;
8532
21.9M
    int special = 0;
8533
21.9M
    int isNamespace;
8534
21.9M
    int flags;
8535
8536
21.9M
    *value = NULL;
8537
21.9M
    GROW;
8538
21.9M
    hname = xmlParseQNameHashed(ctxt, hprefix);
8539
21.9M
    if (hname.name == NULL) {
8540
46.7k
        xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8541
46.7k
                       "error parsing attribute name\n");
8542
46.7k
        return(hname);
8543
46.7k
    }
8544
21.9M
    name = hname.name;
8545
21.9M
    prefix = hprefix->name;
8546
8547
    /*
8548
     * get the type if needed
8549
     */
8550
21.9M
    if (ctxt->attsSpecial != NULL) {
8551
27.3k
        special = XML_PTR_TO_INT(xmlHashQLookup2(ctxt->attsSpecial, pref, elem,
8552
27.3k
                                              prefix, name));
8553
27.3k
    }
8554
8555
    /*
8556
     * read the value
8557
     */
8558
21.9M
    SKIP_BLANKS;
8559
21.9M
    if (RAW != '=') {
8560
15.4k
        xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8561
15.4k
                          "Specification mandates value for attribute %s\n",
8562
15.4k
                          name);
8563
15.4k
        goto error;
8564
15.4k
    }
8565
8566
8567
21.9M
    NEXT;
8568
21.9M
    SKIP_BLANKS;
8569
21.9M
    flags = 0;
8570
21.9M
    isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) ||
8571
21.6M
                   (prefix == ctxt->str_xmlns));
8572
21.9M
    val = xmlParseAttValueInternal(ctxt, len, &flags, special,
8573
21.9M
                                   isNamespace);
8574
21.9M
    if (val == NULL)
8575
9.18k
        goto error;
8576
8577
21.9M
    *alloc = (flags & XML_ATTVAL_ALLOC) != 0;
8578
8579
21.9M
#ifdef LIBXML_VALID_ENABLED
8580
21.9M
    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
21.9M
#endif
8590
8591
21.9M
    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
91.5k
        if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) {
8598
0
            internal_val = xmlStrndup(val, *len);
8599
0
            if (internal_val == NULL)
8600
0
                goto mem_error;
8601
0
            if (!xmlCheckLanguageID(internal_val)) {
8602
0
                xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8603
0
                              "Malformed value for xml:lang : %s\n",
8604
0
                              internal_val, NULL);
8605
0
            }
8606
0
        }
8607
8608
        /*
8609
         * Check that xml:space conforms to the specification
8610
         */
8611
91.5k
        if (xmlStrEqual(name, BAD_CAST "space")) {
8612
73.8k
            internal_val = xmlStrndup(val, *len);
8613
73.8k
            if (internal_val == NULL)
8614
0
                goto mem_error;
8615
73.8k
            if (xmlStrEqual(internal_val, BAD_CAST "default"))
8616
312
                *(ctxt->space) = 0;
8617
73.5k
            else if (xmlStrEqual(internal_val, BAD_CAST "preserve"))
8618
71.4k
                *(ctxt->space) = 1;
8619
2.14k
            else {
8620
2.14k
                xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8621
2.14k
                              "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8622
2.14k
                              internal_val, NULL);
8623
2.14k
            }
8624
73.8k
        }
8625
91.5k
        if (internal_val) {
8626
73.8k
            xmlFree(internal_val);
8627
73.8k
        }
8628
91.5k
    }
8629
8630
21.9M
    *value = val;
8631
21.9M
    return (hname);
8632
8633
0
mem_error:
8634
0
    xmlErrMemory(ctxt);
8635
24.6k
error:
8636
24.6k
    if ((val != NULL) && (*alloc != 0))
8637
0
        xmlFree(val);
8638
24.6k
    return(hname);
8639
0
}
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
16.1M
                  const xmlChar *uri, unsigned hashValue, int aindex) {
8656
16.1M
    xmlAttrHashBucket *table = ctxt->attrHash;
8657
16.1M
    xmlAttrHashBucket *bucket;
8658
16.1M
    unsigned hindex;
8659
8660
16.1M
    hindex = hashValue & (size - 1);
8661
16.1M
    bucket = &table[hindex];
8662
8663
19.0M
    while (bucket->index >= 0) {
8664
2.88M
        const xmlChar **atts = &ctxt->atts[bucket->index];
8665
8666
2.88M
        if (name == atts[0]) {
8667
168k
            int nsIndex = XML_PTR_TO_INT(atts[2]);
8668
8669
168k
            if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) :
8670
168k
                (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) :
8671
3.35k
                (uri == ctxt->nsTab[nsIndex * 2 + 1]))
8672
62.6k
                return(bucket->index);
8673
168k
        }
8674
8675
2.81M
        hindex++;
8676
2.81M
        bucket++;
8677
2.81M
        if (hindex >= size) {
8678
963k
            hindex = 0;
8679
963k
            bucket = table;
8680
963k
        }
8681
2.81M
    }
8682
8683
16.1M
    bucket->index = aindex;
8684
8685
16.1M
    return(INT_MAX);
8686
16.1M
}
8687
8688
static int
8689
xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size,
8690
                       const xmlChar *name, const xmlChar *prefix,
8691
572
                       unsigned hashValue, int aindex) {
8692
572
    xmlAttrHashBucket *table = ctxt->attrHash;
8693
572
    xmlAttrHashBucket *bucket;
8694
572
    unsigned hindex;
8695
8696
572
    hindex = hashValue & (size - 1);
8697
572
    bucket = &table[hindex];
8698
8699
620
    while (bucket->index >= 0) {
8700
531
        const xmlChar **atts = &ctxt->atts[bucket->index];
8701
8702
531
        if ((name == atts[0]) && (prefix == atts[1]))
8703
483
            return(bucket->index);
8704
8705
48
        hindex++;
8706
48
        bucket++;
8707
48
        if (hindex >= size) {
8708
2
            hindex = 0;
8709
2
            bucket = table;
8710
2
        }
8711
48
    }
8712
8713
89
    bucket->index = aindex;
8714
8715
89
    return(INT_MAX);
8716
572
}
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
33.6M
                  const xmlChar **URI, int *nbNsPtr) {
8750
33.6M
    xmlHashedString hlocalname;
8751
33.6M
    xmlHashedString hprefix;
8752
33.6M
    xmlHashedString hattname;
8753
33.6M
    xmlHashedString haprefix;
8754
33.6M
    const xmlChar *localname;
8755
33.6M
    const xmlChar *prefix;
8756
33.6M
    const xmlChar *attname;
8757
33.6M
    const xmlChar *aprefix;
8758
33.6M
    const xmlChar *uri;
8759
33.6M
    xmlChar *attvalue = NULL;
8760
33.6M
    const xmlChar **atts = ctxt->atts;
8761
33.6M
    unsigned attrHashSize = 0;
8762
33.6M
    int maxatts = ctxt->maxatts;
8763
33.6M
    int nratts, nbatts, nbdef;
8764
33.6M
    int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts;
8765
33.6M
    int alloc = 0;
8766
33.6M
    int numNsErr = 0;
8767
33.6M
    int numDupErr = 0;
8768
8769
33.6M
    if (RAW != '<') return(NULL);
8770
33.6M
    NEXT1;
8771
8772
33.6M
    nbatts = 0;
8773
33.6M
    nratts = 0;
8774
33.6M
    nbdef = 0;
8775
33.6M
    nbNs = 0;
8776
33.6M
    nbTotalDef = 0;
8777
33.6M
    attval = 0;
8778
8779
33.6M
    if (xmlParserNsStartElement(ctxt->nsdb) < 0) {
8780
0
        xmlErrMemory(ctxt);
8781
0
        return(NULL);
8782
0
    }
8783
8784
33.6M
    hlocalname = xmlParseQNameHashed(ctxt, &hprefix);
8785
33.6M
    if (hlocalname.name == NULL) {
8786
11.3k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8787
11.3k
           "StartTag: invalid element name\n");
8788
11.3k
        return(NULL);
8789
11.3k
    }
8790
33.5M
    localname = hlocalname.name;
8791
33.5M
    prefix = hprefix.name;
8792
8793
    /*
8794
     * Now parse the attributes, it ends up with the ending
8795
     *
8796
     * (S Attribute)* S?
8797
     */
8798
33.5M
    SKIP_BLANKS;
8799
33.5M
    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
45.5M
    while (((RAW != '>') &&
8826
24.3M
     ((RAW != '/') || (NXT(1) != '>')) &&
8827
22.0M
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8828
21.9M
  int len = -1;
8829
8830
21.9M
  hattname = xmlParseAttribute2(ctxt, prefix, localname,
8831
21.9M
                                          &haprefix, &attvalue, &len,
8832
21.9M
                                          &alloc);
8833
21.9M
        if (hattname.name == NULL)
8834
46.7k
      break;
8835
21.9M
        if (attvalue == NULL)
8836
24.6k
            goto next_attr;
8837
21.9M
        attname = hattname.name;
8838
21.9M
        aprefix = haprefix.name;
8839
21.9M
  if (len < 0) len = xmlStrlen(attvalue);
8840
8841
21.9M
        if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
8842
330k
            xmlHashedString huri;
8843
330k
            xmlURIPtr parsedUri;
8844
8845
330k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8846
330k
            uri = huri.name;
8847
330k
            if (uri == NULL) {
8848
0
                xmlErrMemory(ctxt);
8849
0
                goto next_attr;
8850
0
            }
8851
330k
            if (*uri != 0) {
8852
326k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8853
0
                    xmlErrMemory(ctxt);
8854
0
                    goto next_attr;
8855
0
                }
8856
326k
                if (parsedUri == NULL) {
8857
123k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8858
123k
                             "xmlns: '%s' is not a valid URI\n",
8859
123k
                                       uri, NULL, NULL);
8860
202k
                } else {
8861
202k
                    if (parsedUri->scheme == NULL) {
8862
70.2k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8863
70.2k
                                  "xmlns: URI %s is not absolute\n",
8864
70.2k
                                  uri, NULL, NULL);
8865
70.2k
                    }
8866
202k
                    xmlFreeURI(parsedUri);
8867
202k
                }
8868
326k
                if (uri == ctxt->str_xml_ns) {
8869
246
                    if (attname != ctxt->str_xml) {
8870
246
                        xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8871
246
                     "xml namespace URI cannot be the default namespace\n",
8872
246
                                 NULL, NULL, NULL);
8873
246
                    }
8874
246
                    goto next_attr;
8875
246
                }
8876
326k
                if ((len == 29) &&
8877
7.25k
                    (xmlStrEqual(uri,
8878
7.25k
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8879
311
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8880
311
                         "reuse of the xmlns namespace name is forbidden\n",
8881
311
                             NULL, NULL, NULL);
8882
311
                    goto next_attr;
8883
311
                }
8884
326k
            }
8885
8886
329k
            if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0)
8887
206k
                nbNs++;
8888
21.5M
        } else if (aprefix == ctxt->str_xmlns) {
8889
852k
            xmlHashedString huri;
8890
852k
            xmlURIPtr parsedUri;
8891
8892
852k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8893
852k
            uri = huri.name;
8894
852k
            if (uri == NULL) {
8895
0
                xmlErrMemory(ctxt);
8896
0
                goto next_attr;
8897
0
            }
8898
8899
852k
            if (attname == ctxt->str_xml) {
8900
404
                if (uri != ctxt->str_xml_ns) {
8901
259
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8902
259
                             "xml namespace prefix mapped to wrong URI\n",
8903
259
                             NULL, NULL, NULL);
8904
259
                }
8905
                /*
8906
                 * Do not keep a namespace definition node
8907
                 */
8908
404
                goto next_attr;
8909
404
            }
8910
852k
            if (uri == ctxt->str_xml_ns) {
8911
17
                if (attname != ctxt->str_xml) {
8912
17
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8913
17
                             "xml namespace URI mapped to wrong prefix\n",
8914
17
                             NULL, NULL, NULL);
8915
17
                }
8916
17
                goto next_attr;
8917
17
            }
8918
852k
            if (attname == ctxt->str_xmlns) {
8919
200
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8920
200
                         "redefinition of the xmlns prefix is forbidden\n",
8921
200
                         NULL, NULL, NULL);
8922
200
                goto next_attr;
8923
200
            }
8924
851k
            if ((len == 29) &&
8925
17.6k
                (xmlStrEqual(uri,
8926
17.6k
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8927
118
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8928
118
                         "reuse of the xmlns namespace name is forbidden\n",
8929
118
                         NULL, NULL, NULL);
8930
118
                goto next_attr;
8931
118
            }
8932
851k
            if ((uri == NULL) || (uri[0] == 0)) {
8933
383
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8934
383
                         "xmlns:%s: Empty XML namespace is not allowed\n",
8935
383
                              attname, NULL, NULL);
8936
383
                goto next_attr;
8937
851k
            } else {
8938
851k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8939
0
                    xmlErrMemory(ctxt);
8940
0
                    goto next_attr;
8941
0
                }
8942
851k
                if (parsedUri == NULL) {
8943
90.7k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8944
90.7k
                         "xmlns:%s: '%s' is not a valid URI\n",
8945
90.7k
                                       attname, uri, NULL);
8946
760k
                } else {
8947
760k
                    if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) {
8948
0
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8949
0
                                  "xmlns:%s: URI %s is not absolute\n",
8950
0
                                  attname, uri, NULL);
8951
0
                    }
8952
760k
                    xmlFreeURI(parsedUri);
8953
760k
                }
8954
851k
            }
8955
8956
851k
            if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0)
8957
805k
                nbNs++;
8958
20.7M
        } else {
8959
            /*
8960
             * Populate attributes array, see above for repurposing
8961
             * of xmlChar pointers.
8962
             */
8963
20.7M
            if ((atts == NULL) || (nbatts + 5 > maxatts)) {
8964
564k
                int res = xmlCtxtGrowAttrs(ctxt);
8965
8966
564k
                maxatts = ctxt->maxatts;
8967
564k
                atts = ctxt->atts;
8968
8969
564k
                if (res < 0)
8970
0
                    goto next_attr;
8971
564k
            }
8972
20.7M
            ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) |
8973
20.7M
                                        ((unsigned) alloc << 31);
8974
20.7M
            atts[nbatts++] = attname;
8975
20.7M
            atts[nbatts++] = aprefix;
8976
20.7M
            atts[nbatts++] = XML_INT_TO_PTR(haprefix.hashValue);
8977
20.7M
            if (alloc) {
8978
90.6k
                atts[nbatts++] = attvalue;
8979
90.6k
                attvalue += len;
8980
90.6k
                atts[nbatts++] = attvalue;
8981
20.6M
            } 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
20.6M
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8988
20.6M
                attvalue += len;
8989
20.6M
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8990
20.6M
            }
8991
            /*
8992
             * tag if some deallocation is needed
8993
             */
8994
20.7M
            if (alloc != 0) attval = 1;
8995
20.7M
            attvalue = NULL; /* moved into atts */
8996
20.7M
        }
8997
8998
21.9M
next_attr:
8999
21.9M
        if ((attvalue != NULL) && (alloc != 0)) {
9000
101k
            xmlFree(attvalue);
9001
101k
            attvalue = NULL;
9002
101k
        }
9003
9004
21.9M
  GROW
9005
21.9M
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
9006
9.96M
      break;
9007
11.9M
  if (SKIP_BLANKS == 0) {
9008
32.6k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
9009
32.6k
         "attributes construct error\n");
9010
32.6k
      break;
9011
32.6k
  }
9012
11.9M
        GROW;
9013
11.9M
    }
9014
9015
    /*
9016
     * Namespaces from default attributes
9017
     */
9018
33.5M
    if (ctxt->attsDefault != NULL) {
9019
115k
        xmlDefAttrsPtr defaults;
9020
9021
115k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9022
115k
  if (defaults != NULL) {
9023
823k
      for (i = 0; i < defaults->nbAttrs; i++) {
9024
739k
                xmlDefAttr *attr = &defaults->attrs[i];
9025
9026
739k
          attname = attr->name.name;
9027
739k
    aprefix = attr->prefix.name;
9028
9029
739k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
9030
26.3k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9031
9032
26.3k
                    if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0)
9033
22.8k
                        nbNs++;
9034
712k
    } else if (aprefix == ctxt->str_xmlns) {
9035
317k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9036
9037
317k
                    if (xmlParserNsPush(ctxt, &attr->name, &attr->value,
9038
317k
                                      NULL, 1) > 0)
9039
317k
                        nbNs++;
9040
395k
    } else {
9041
395k
                    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
395k
                    nbTotalDef += 1;
9047
395k
                }
9048
739k
      }
9049
84.2k
  }
9050
115k
    }
9051
9052
    /*
9053
     * Resolve attribute namespaces
9054
     */
9055
54.3M
    for (i = 0; i < nbatts; i += 5) {
9056
20.7M
        attname = atts[i];
9057
20.7M
        aprefix = atts[i+1];
9058
9059
        /*
9060
  * The default namespace does not apply to attribute names.
9061
  */
9062
20.7M
  if (aprefix == NULL) {
9063
13.4M
            nsIndex = NS_INDEX_EMPTY;
9064
13.4M
        } else if (aprefix == ctxt->str_xml) {
9065
91.5k
            nsIndex = NS_INDEX_XML;
9066
7.16M
        } else {
9067
7.16M
            haprefix.name = aprefix;
9068
7.16M
            haprefix.hashValue = (size_t) atts[i+2];
9069
7.16M
            nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL);
9070
9071
7.16M
      if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) {
9072
371k
                xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9073
371k
        "Namespace prefix %s for %s on %s is not defined\n",
9074
371k
        aprefix, attname, localname);
9075
371k
                nsIndex = NS_INDEX_EMPTY;
9076
371k
            }
9077
7.16M
        }
9078
9079
20.7M
        atts[i+2] = XML_INT_TO_PTR(nsIndex);
9080
20.7M
    }
9081
9082
    /*
9083
     * Maximum number of attributes including default attributes.
9084
     */
9085
33.5M
    maxAtts = nratts + nbTotalDef;
9086
9087
    /*
9088
     * Verify that attribute names are unique.
9089
     */
9090
33.5M
    if (maxAtts > 1) {
9091
5.42M
        attrHashSize = 4;
9092
8.60M
        while (attrHashSize / 2 < (unsigned) maxAtts)
9093
3.17M
            attrHashSize *= 2;
9094
9095
5.42M
        if (attrHashSize > ctxt->attrHashMax) {
9096
230k
            xmlAttrHashBucket *tmp;
9097
9098
230k
            tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0]));
9099
230k
            if (tmp == NULL) {
9100
0
                xmlErrMemory(ctxt);
9101
0
                goto done;
9102
0
            }
9103
9104
230k
            ctxt->attrHash = tmp;
9105
230k
            ctxt->attrHashMax = attrHashSize;
9106
230k
        }
9107
9108
5.42M
        memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0]));
9109
9110
21.5M
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9111
16.1M
            const xmlChar *nsuri;
9112
16.1M
            unsigned hashValue, nameHashValue, uriHashValue;
9113
16.1M
            int res;
9114
9115
16.1M
            attname = atts[i];
9116
16.1M
            aprefix = atts[i+1];
9117
16.1M
            nsIndex = XML_PTR_TO_INT(atts[i+2]);
9118
            /* Hash values always have bit 31 set, see dict.c */
9119
16.1M
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9120
9121
16.1M
            if (nsIndex == NS_INDEX_EMPTY) {
9122
                /*
9123
                 * Prefix with empty namespace means an undeclared
9124
                 * prefix which was already reported above.
9125
                 */
9126
10.9M
                if (aprefix != NULL)
9127
299k
                    continue;
9128
10.6M
                nsuri = NULL;
9129
10.6M
                uriHashValue = URI_HASH_EMPTY;
9130
10.6M
            } else if (nsIndex == NS_INDEX_XML) {
9131
2.53k
                nsuri = ctxt->str_xml_ns;
9132
2.53k
                uriHashValue = URI_HASH_XML;
9133
5.14M
            } else {
9134
5.14M
                nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9135
5.14M
                uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9136
5.14M
            }
9137
9138
15.8M
            hashValue = xmlDictCombineHash(nameHashValue, uriHashValue);
9139
15.8M
            res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9140
15.8M
                                    hashValue, i);
9141
15.8M
            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
15.8M
            if (res < INT_MAX) {
9151
54.3k
                if (aprefix == atts[res+1]) {
9152
53.7k
                    xmlErrAttributeDup(ctxt, aprefix, attname);
9153
53.7k
                    numDupErr += 1;
9154
53.7k
                } else {
9155
630
                    xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9156
630
                             "Namespaced Attribute %s in '%s' redefined\n",
9157
630
                             attname, nsuri, NULL);
9158
630
                    numNsErr += 1;
9159
630
                }
9160
54.3k
            }
9161
15.8M
        }
9162
5.42M
    }
9163
9164
    /*
9165
     * Default attributes
9166
     */
9167
33.5M
    if (ctxt->attsDefault != NULL) {
9168
115k
        xmlDefAttrsPtr defaults;
9169
9170
115k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9171
115k
  if (defaults != NULL) {
9172
823k
      for (i = 0; i < defaults->nbAttrs; i++) {
9173
739k
                xmlDefAttr *attr = &defaults->attrs[i];
9174
739k
                const xmlChar *nsuri = NULL;
9175
739k
                unsigned hashValue, uriHashValue = 0;
9176
739k
                int res;
9177
9178
739k
          attname = attr->name.name;
9179
739k
    aprefix = attr->prefix.name;
9180
9181
739k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL))
9182
26.3k
                    continue;
9183
712k
    if (aprefix == ctxt->str_xmlns)
9184
317k
                    continue;
9185
9186
395k
                if (aprefix == NULL) {
9187
221k
                    nsIndex = NS_INDEX_EMPTY;
9188
221k
                    nsuri = NULL;
9189
221k
                    uriHashValue = URI_HASH_EMPTY;
9190
221k
                } else if (aprefix == ctxt->str_xml) {
9191
46.3k
                    nsIndex = NS_INDEX_XML;
9192
46.3k
                    nsuri = ctxt->str_xml_ns;
9193
46.3k
                    uriHashValue = URI_HASH_XML;
9194
127k
                } else {
9195
127k
                    nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL);
9196
127k
                    if ((nsIndex == INT_MAX) ||
9197
126k
                        (nsIndex < ctxt->nsdb->minNsIndex)) {
9198
126k
                        xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9199
126k
                                 "Namespace prefix %s for %s on %s is not "
9200
126k
                                 "defined\n",
9201
126k
                                 aprefix, attname, localname);
9202
126k
                        nsIndex = NS_INDEX_EMPTY;
9203
126k
                        nsuri = NULL;
9204
126k
                        uriHashValue = URI_HASH_EMPTY;
9205
126k
                    } else {
9206
541
                        nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9207
541
                        uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9208
541
                    }
9209
127k
                }
9210
9211
                /*
9212
                 * Check whether the attribute exists
9213
                 */
9214
395k
                if (maxAtts > 1) {
9215
380k
                    hashValue = xmlDictCombineHash(attr->name.hashValue,
9216
380k
                                                   uriHashValue);
9217
380k
                    res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9218
380k
                                            hashValue, nbatts);
9219
380k
                    if (res < 0)
9220
0
                        continue;
9221
380k
                    if (res < INT_MAX) {
9222
8.24k
                        if (aprefix == atts[res+1])
9223
954
                            continue;
9224
7.28k
                        xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9225
7.28k
                                 "Namespaced Attribute %s in '%s' redefined\n",
9226
7.28k
                                 attname, nsuri, NULL);
9227
7.28k
                    }
9228
380k
                }
9229
9230
394k
                xmlParserEntityCheck(ctxt, attr->expandedSize);
9231
9232
394k
                if ((atts == NULL) || (nbatts + 5 > maxatts)) {
9233
6.38k
                    res = xmlCtxtGrowAttrs(ctxt);
9234
9235
6.38k
                    maxatts = ctxt->maxatts;
9236
6.38k
                    atts = ctxt->atts;
9237
9238
6.38k
                    if (res < 0) {
9239
0
                        localname = NULL;
9240
0
                        goto done;
9241
0
                    }
9242
6.38k
                }
9243
9244
394k
                atts[nbatts++] = attname;
9245
394k
                atts[nbatts++] = aprefix;
9246
394k
                atts[nbatts++] = XML_INT_TO_PTR(nsIndex);
9247
394k
                atts[nbatts++] = attr->value.name;
9248
394k
                atts[nbatts++] = attr->valueEnd;
9249
9250
394k
#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
394k
                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
394k
#endif
9264
394k
                nbdef++;
9265
394k
      }
9266
84.2k
  }
9267
115k
    }
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
33.5M
    if ((numDupErr == 0) && (numNsErr > 1)) {
9282
26
        memset(ctxt->attrHash, -1,
9283
26
               attrHashSize * sizeof(ctxt->attrHash[0]));
9284
9285
630
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9286
604
            unsigned hashValue, nameHashValue, prefixHashValue;
9287
604
            int res;
9288
9289
604
            aprefix = atts[i+1];
9290
604
            if (aprefix == NULL)
9291
32
                continue;
9292
9293
572
            attname = atts[i];
9294
            /* Hash values always have bit 31 set, see dict.c */
9295
572
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9296
572
            prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix);
9297
9298
572
            hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue);
9299
572
            res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname,
9300
572
                                         aprefix, hashValue, i);
9301
572
            if (res < INT_MAX)
9302
483
                xmlErrAttributeDup(ctxt, aprefix, attname);
9303
572
        }
9304
26
    }
9305
9306
    /*
9307
     * Reconstruct attribute pointers
9308
     */
9309
54.7M
    for (i = 0, j = 0; i < nbatts; i += 5, j++) {
9310
        /* namespace URI */
9311
21.1M
        nsIndex = XML_PTR_TO_INT(atts[i+2]);
9312
21.1M
        if (nsIndex == INT_MAX)
9313
14.2M
            atts[i+2] = NULL;
9314
6.93M
        else if (nsIndex == INT_MAX - 1)
9315
137k
            atts[i+2] = ctxt->str_xml_ns;
9316
6.79M
        else
9317
6.79M
            atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1];
9318
9319
21.1M
        if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) {
9320
20.6M
            atts[i+3] = BASE_PTR + XML_PTR_TO_INT(atts[i+3]);  /* value */
9321
20.6M
            atts[i+4] = BASE_PTR + XML_PTR_TO_INT(atts[i+4]);  /* valuend */
9322
20.6M
        }
9323
21.1M
    }
9324
9325
33.5M
    uri = xmlParserNsLookupUri(ctxt, &hprefix);
9326
33.5M
    if ((prefix != NULL) && (uri == NULL)) {
9327
564k
  xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9328
564k
           "Namespace prefix %s on %s is not defined\n",
9329
564k
     prefix, localname, NULL);
9330
564k
    }
9331
33.5M
    *pref = prefix;
9332
33.5M
    *URI = uri;
9333
9334
    /*
9335
     * SAX callback
9336
     */
9337
33.5M
    if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) &&
9338
33.5M
  (!ctxt->disableSAX)) {
9339
33.4M
  if (nbNs > 0)
9340
437k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9341
437k
                          nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs),
9342
437k
        nbatts / 5, nbdef, atts);
9343
33.0M
  else
9344
33.0M
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9345
33.0M
                          0, NULL, nbatts / 5, nbdef, atts);
9346
33.4M
    }
9347
9348
33.5M
done:
9349
    /*
9350
     * Free allocated attribute values
9351
     */
9352
33.5M
    if (attval != 0) {
9353
355k
  for (i = 0, j = 0; j < nratts; i += 5, j++)
9354
274k
      if (ctxt->attallocs[j] & 0x80000000)
9355
90.6k
          xmlFree((xmlChar *) atts[i+3]);
9356
80.5k
    }
9357
9358
33.5M
    *nbNsPtr = nbNs;
9359
33.5M
    return(localname);
9360
33.5M
}
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
8.07M
xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) {
9376
8.07M
    const xmlChar *name;
9377
9378
8.07M
    GROW;
9379
8.07M
    if ((RAW != '<') || (NXT(1) != '/')) {
9380
5
  xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL);
9381
5
  return;
9382
5
    }
9383
8.07M
    SKIP(2);
9384
9385
8.07M
    if (tag->prefix == NULL)
9386
2.63M
        name = xmlParseNameAndCompare(ctxt, ctxt->name);
9387
5.43M
    else
9388
5.43M
        name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix);
9389
9390
    /*
9391
     * We should definitely be at the ending "S? '>'" part
9392
     */
9393
8.07M
    GROW;
9394
8.07M
    SKIP_BLANKS;
9395
8.07M
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
9396
7.98k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
9397
7.98k
    } else
9398
8.06M
  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
8.07M
    if (name != (xmlChar*)1) {
9407
16.0k
        if (name == NULL) name = BAD_CAST "unparsable";
9408
16.0k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
9409
16.0k
         "Opening and ending tag mismatch: %s line %d and %s\n",
9410
16.0k
                    ctxt->name, tag->line, name);
9411
16.0k
    }
9412
9413
    /*
9414
     * SAX: End of Tag
9415
     */
9416
8.07M
    if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9417
8.07M
  (!ctxt->disableSAX))
9418
8.05M
  ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix,
9419
8.05M
                                tag->URI);
9420
9421
8.07M
    spacePop(ctxt);
9422
8.07M
    if (tag->nsNr != 0)
9423
164k
  xmlParserNsPop(ctxt, tag->nsNr);
9424
8.07M
}
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
12.6k
xmlParseCDSect(xmlParserCtxt *ctxt) {
9442
12.6k
    xmlChar *buf = NULL;
9443
12.6k
    int len = 0;
9444
12.6k
    int size = XML_PARSER_BUFFER_SIZE;
9445
12.6k
    int r, rl;
9446
12.6k
    int s, sl;
9447
12.6k
    int cur, l;
9448
12.6k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9449
12.6k
                    XML_MAX_HUGE_LENGTH :
9450
12.6k
                    XML_MAX_TEXT_LENGTH;
9451
9452
12.6k
    if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '['))
9453
0
        return;
9454
12.6k
    SKIP(3);
9455
9456
12.6k
    if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '['))
9457
0
        return;
9458
12.6k
    SKIP(6);
9459
9460
12.6k
    r = xmlCurrentCharRecover(ctxt, &rl);
9461
12.6k
    if (!IS_CHAR(r)) {
9462
20
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9463
20
        goto out;
9464
20
    }
9465
12.6k
    NEXTL(rl);
9466
12.6k
    s = xmlCurrentCharRecover(ctxt, &sl);
9467
12.6k
    if (!IS_CHAR(s)) {
9468
26
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9469
26
        goto out;
9470
26
    }
9471
12.6k
    NEXTL(sl);
9472
12.6k
    cur = xmlCurrentCharRecover(ctxt, &l);
9473
12.6k
    buf = xmlMalloc(size);
9474
12.6k
    if (buf == NULL) {
9475
0
  xmlErrMemory(ctxt);
9476
0
        goto out;
9477
0
    }
9478
45.0M
    while (IS_CHAR(cur) &&
9479
45.0M
           ((r != ']') || (s != ']') || (cur != '>'))) {
9480
45.0M
  if (len + 5 >= size) {
9481
55.9k
      xmlChar *tmp;
9482
55.9k
            int newSize;
9483
9484
55.9k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9485
55.9k
            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
55.9k
      tmp = xmlRealloc(buf, newSize);
9491
55.9k
      if (tmp == NULL) {
9492
0
    xmlErrMemory(ctxt);
9493
0
                goto out;
9494
0
      }
9495
55.9k
      buf = tmp;
9496
55.9k
      size = newSize;
9497
55.9k
  }
9498
45.0M
  COPY_BUF(buf, len, r);
9499
45.0M
  r = s;
9500
45.0M
  rl = sl;
9501
45.0M
  s = cur;
9502
45.0M
  sl = l;
9503
45.0M
  NEXTL(l);
9504
45.0M
  cur = xmlCurrentCharRecover(ctxt, &l);
9505
45.0M
    }
9506
12.6k
    buf[len] = 0;
9507
12.6k
    if (cur != '>') {
9508
334
  xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9509
334
                       "CData section not finished\n%.50s\n", buf);
9510
334
        goto out;
9511
334
    }
9512
12.2k
    NEXTL(l);
9513
9514
    /*
9515
     * OK the buffer is to be consumed as cdata.
9516
     */
9517
12.2k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) {
9518
12.2k
        if ((ctxt->sax->cdataBlock != NULL) &&
9519
0
            ((ctxt->options & XML_PARSE_NOCDATA) == 0)) {
9520
0
            ctxt->sax->cdataBlock(ctxt->userData, buf, len);
9521
12.2k
        } else if (ctxt->sax->characters != NULL) {
9522
12.2k
            ctxt->sax->characters(ctxt->userData, buf, len);
9523
12.2k
        }
9524
12.2k
    }
9525
9526
12.6k
out:
9527
12.6k
    xmlFree(buf);
9528
12.6k
}
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
5.03k
xmlParseContentInternal(xmlParserCtxtPtr ctxt) {
9539
5.03k
    int oldNameNr = ctxt->nameNr;
9540
5.03k
    int oldSpaceNr = ctxt->spaceNr;
9541
5.03k
    int oldNodeNr = ctxt->nodeNr;
9542
9543
5.03k
    GROW;
9544
1.59M
    while ((ctxt->input->cur < ctxt->input->end) &&
9545
1.59M
     (PARSER_STOPPED(ctxt) == 0)) {
9546
1.59M
  const xmlChar *cur = ctxt->input->cur;
9547
9548
  /*
9549
   * First case : a Processing Instruction.
9550
   */
9551
1.59M
  if ((*cur == '<') && (cur[1] == '?')) {
9552
846
      xmlParsePI(ctxt);
9553
846
  }
9554
9555
  /*
9556
   * Second case : a CDSection
9557
   */
9558
  /* 2.6.0 test was *cur not RAW */
9559
1.59M
  else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) {
9560
0
      xmlParseCDSect(ctxt);
9561
0
  }
9562
9563
  /*
9564
   * Third case :  a comment
9565
   */
9566
1.59M
  else if ((*cur == '<') && (NXT(1) == '!') &&
9567
0
     (NXT(2) == '-') && (NXT(3) == '-')) {
9568
0
      xmlParseComment(ctxt);
9569
0
  }
9570
9571
  /*
9572
   * Fourth case :  a sub-element.
9573
   */
9574
1.59M
  else if (*cur == '<') {
9575
1.25M
            if (NXT(1) == '/') {
9576
362k
                if (ctxt->nameNr <= oldNameNr)
9577
4.81k
                    break;
9578
357k
          xmlParseElementEnd(ctxt);
9579
892k
            } else {
9580
892k
          xmlParseElementStart(ctxt);
9581
892k
            }
9582
1.25M
  }
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
335k
  else if (*cur == '&') {
9590
6
      xmlParseReference(ctxt);
9591
6
  }
9592
9593
  /*
9594
   * Last case, text. Note that References are handled directly.
9595
   */
9596
335k
  else {
9597
335k
      xmlParseCharDataInternal(ctxt, 0);
9598
335k
  }
9599
9600
1.58M
  SHRINK;
9601
1.58M
  GROW;
9602
1.58M
    }
9603
9604
5.03k
    if ((ctxt->nameNr > oldNameNr) &&
9605
207
        (ctxt->input->cur >= ctxt->input->end) &&
9606
202
        (ctxt->wellFormed)) {
9607
0
        const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9608
0
        int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9609
0
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9610
0
                "Premature end of data in tag %s line %d\n",
9611
0
                name, line, NULL);
9612
0
    }
9613
9614
    /*
9615
     * Clean up in error case
9616
     */
9617
9618
5.14k
    while (ctxt->nodeNr > oldNodeNr)
9619
110
        nodePop(ctxt);
9620
9621
16.6k
    while (ctxt->nameNr > oldNameNr) {
9622
11.5k
        xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1];
9623
9624
11.5k
        if (tag->nsNr != 0)
9625
12
            xmlParserNsPop(ctxt, tag->nsNr);
9626
9627
11.5k
        namePop(ctxt);
9628
11.5k
    }
9629
9630
16.6k
    while (ctxt->spaceNr > oldSpaceNr)
9631
11.5k
        spacePop(ctxt);
9632
5.03k
}
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
5.25k
xmlParseElement(xmlParserCtxt *ctxt) {
9669
5.25k
    if (xmlParseElementStart(ctxt) != 0)
9670
223
        return;
9671
9672
5.03k
    xmlParseContentInternal(ctxt);
9673
9674
5.03k
    if (ctxt->input->cur >= ctxt->input->end) {
9675
207
        if (ctxt->wellFormed) {
9676
0
            const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9677
0
            int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9678
0
            xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9679
0
                    "Premature end of data in tag %s line %d\n",
9680
0
                    name, line, NULL);
9681
0
        }
9682
207
        return;
9683
207
    }
9684
9685
4.82k
    xmlParseElementEnd(ctxt);
9686
4.82k
}
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
897k
xmlParseElementStart(xmlParserCtxtPtr ctxt) {
9698
897k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
9699
897k
    const xmlChar *name;
9700
897k
    const xmlChar *prefix = NULL;
9701
897k
    const xmlChar *URI = NULL;
9702
897k
    xmlParserNodeInfo node_info;
9703
897k
    int line;
9704
897k
    xmlNodePtr cur;
9705
897k
    int nbNs = 0;
9706
9707
897k
    if (ctxt->nameNr > maxDepth) {
9708
5
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
9709
5
                "Excessive depth in document: %d use XML_PARSE_HUGE option\n",
9710
5
                ctxt->nameNr);
9711
5
  return(-1);
9712
5
    }
9713
9714
    /* Capture start position */
9715
897k
    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
897k
    if (ctxt->spaceNr == 0)
9722
5.25k
  spacePush(ctxt, -1);
9723
892k
    else if (*ctxt->space == -2)
9724
0
  spacePush(ctxt, -1);
9725
892k
    else
9726
892k
  spacePush(ctxt, *ctxt->space);
9727
9728
897k
    line = ctxt->input->line;
9729
897k
#ifdef LIBXML_SAX1_ENABLED
9730
897k
    if (ctxt->sax2)
9731
897k
#endif /* LIBXML_SAX1_ENABLED */
9732
897k
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
9733
0
#ifdef LIBXML_SAX1_ENABLED
9734
0
    else
9735
0
  name = xmlParseStartTag(ctxt);
9736
897k
#endif /* LIBXML_SAX1_ENABLED */
9737
897k
    if (name == NULL) {
9738
7.05k
  spacePop(ctxt);
9739
7.05k
        return(-1);
9740
7.05k
    }
9741
890k
    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
9742
890k
    cur = ctxt->node;
9743
9744
890k
#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
890k
    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
890k
#endif /* LIBXML_VALID_ENABLED */
9754
9755
    /*
9756
     * Check for an Empty Element.
9757
     */
9758
890k
    if ((RAW == '/') && (NXT(1) == '>')) {
9759
467k
        SKIP(2);
9760
467k
  if (ctxt->sax2) {
9761
467k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9762
467k
    (!ctxt->disableSAX))
9763
452k
    ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI);
9764
467k
#ifdef LIBXML_SAX1_ENABLED
9765
467k
  } 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
467k
  namePop(ctxt);
9772
467k
  spacePop(ctxt);
9773
467k
  if (nbNs > 0)
9774
1.10k
      xmlParserNsPop(ctxt, nbNs);
9775
467k
  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
467k
  return(1);
9783
467k
    }
9784
423k
    if (RAW == '>') {
9785
373k
        NEXT1;
9786
373k
        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
373k
    } else {
9793
49.2k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED,
9794
49.2k
         "Couldn't find end of Start Tag %s line %d\n",
9795
49.2k
                    name, line, NULL);
9796
9797
  /*
9798
   * end of parsing of this node.
9799
   */
9800
49.2k
  nodePop(ctxt);
9801
49.2k
  namePop(ctxt);
9802
49.2k
  spacePop(ctxt);
9803
49.2k
  if (nbNs > 0)
9804
263
      xmlParserNsPop(ctxt, nbNs);
9805
49.2k
  return(-1);
9806
49.2k
    }
9807
9808
373k
    return(0);
9809
423k
}
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
362k
xmlParseElementEnd(xmlParserCtxtPtr ctxt) {
9818
362k
    xmlNodePtr cur = ctxt->node;
9819
9820
362k
    if (ctxt->nameNr <= 0) {
9821
0
        if ((RAW == '<') && (NXT(1) == '/'))
9822
0
            SKIP(2);
9823
0
        return;
9824
0
    }
9825
9826
    /*
9827
     * parse the end of tag: '</' should be here.
9828
     */
9829
362k
    if (ctxt->sax2) {
9830
362k
  xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
9831
362k
  namePop(ctxt);
9832
362k
    }
9833
0
#ifdef LIBXML_SAX1_ENABLED
9834
0
    else
9835
0
  xmlParseEndTag1(ctxt, 0);
9836
362k
#endif /* LIBXML_SAX1_ENABLED */
9837
9838
    /*
9839
     * Capture end position
9840
     */
9841
362k
    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
362k
}
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
162k
xmlParseVersionNum(xmlParserCtxt *ctxt) {
9867
162k
    xmlChar *buf = NULL;
9868
162k
    int len = 0;
9869
162k
    int size = 10;
9870
162k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9871
157k
                    XML_MAX_TEXT_LENGTH :
9872
162k
                    XML_MAX_NAME_LENGTH;
9873
162k
    xmlChar cur;
9874
9875
162k
    buf = xmlMalloc(size);
9876
162k
    if (buf == NULL) {
9877
0
  xmlErrMemory(ctxt);
9878
0
  return(NULL);
9879
0
    }
9880
162k
    cur = CUR;
9881
162k
    if (!((cur >= '0') && (cur <= '9'))) {
9882
102
  xmlFree(buf);
9883
102
  return(NULL);
9884
102
    }
9885
162k
    buf[len++] = cur;
9886
162k
    NEXT;
9887
162k
    cur=CUR;
9888
162k
    if (cur != '.') {
9889
46
  xmlFree(buf);
9890
46
  return(NULL);
9891
46
    }
9892
162k
    buf[len++] = cur;
9893
162k
    NEXT;
9894
162k
    cur=CUR;
9895
336k
    while ((cur >= '0') && (cur <= '9')) {
9896
174k
  if (len + 1 >= size) {
9897
364
      xmlChar *tmp;
9898
364
            int newSize;
9899
9900
364
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9901
364
            if (newSize < 0) {
9902
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "VersionNum");
9903
0
                xmlFree(buf);
9904
0
                return(NULL);
9905
0
            }
9906
364
      tmp = xmlRealloc(buf, newSize);
9907
364
      if (tmp == NULL) {
9908
0
    xmlErrMemory(ctxt);
9909
0
          xmlFree(buf);
9910
0
    return(NULL);
9911
0
      }
9912
364
      buf = tmp;
9913
364
            size = newSize;
9914
364
  }
9915
174k
  buf[len++] = cur;
9916
174k
  NEXT;
9917
174k
  cur=CUR;
9918
174k
    }
9919
162k
    buf[len] = 0;
9920
162k
    return(buf);
9921
162k
}
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
163k
xmlParseVersionInfo(xmlParserCtxt *ctxt) {
9938
163k
    xmlChar *version = NULL;
9939
9940
163k
    if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) {
9941
162k
  SKIP(7);
9942
162k
  SKIP_BLANKS;
9943
162k
  if (RAW != '=') {
9944
45
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
9945
45
      return(NULL);
9946
45
        }
9947
162k
  NEXT;
9948
162k
  SKIP_BLANKS;
9949
162k
  if (RAW == '"') {
9950
162k
      NEXT;
9951
162k
      version = xmlParseVersionNum(ctxt);
9952
162k
      if (RAW != '"') {
9953
182
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9954
182
      } else
9955
161k
          NEXT;
9956
162k
  } else if (RAW == '\''){
9957
409
      NEXT;
9958
409
      version = xmlParseVersionNum(ctxt);
9959
409
      if (RAW != '\'') {
9960
23
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9961
23
      } else
9962
386
          NEXT;
9963
409
  } else {
9964
45
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
9965
45
  }
9966
162k
    }
9967
163k
    return(version);
9968
163k
}
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
136k
xmlParseEncName(xmlParserCtxt *ctxt) {
9982
136k
    xmlChar *buf = NULL;
9983
136k
    int len = 0;
9984
136k
    int size = 10;
9985
136k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9986
131k
                    XML_MAX_TEXT_LENGTH :
9987
136k
                    XML_MAX_NAME_LENGTH;
9988
136k
    xmlChar cur;
9989
9990
136k
    cur = CUR;
9991
136k
    if (((cur >= 'a') && (cur <= 'z')) ||
9992
136k
        ((cur >= 'A') && (cur <= 'Z'))) {
9993
136k
  buf = xmlMalloc(size);
9994
136k
  if (buf == NULL) {
9995
0
      xmlErrMemory(ctxt);
9996
0
      return(NULL);
9997
0
  }
9998
9999
136k
  buf[len++] = cur;
10000
136k
  NEXT;
10001
136k
  cur = CUR;
10002
693k
  while (((cur >= 'a') && (cur <= 'z')) ||
10003
685k
         ((cur >= 'A') && (cur <= 'Z')) ||
10004
417k
         ((cur >= '0') && (cur <= '9')) ||
10005
275k
         (cur == '.') || (cur == '_') ||
10006
557k
         (cur == '-')) {
10007
557k
      if (len + 1 >= size) {
10008
1.19k
          xmlChar *tmp;
10009
1.19k
                int newSize;
10010
10011
1.19k
                newSize = xmlGrowCapacity(size, 1, 1, maxLength);
10012
1.19k
                if (newSize < 0) {
10013
0
                    xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName");
10014
0
                    xmlFree(buf);
10015
0
                    return(NULL);
10016
0
                }
10017
1.19k
    tmp = xmlRealloc(buf, newSize);
10018
1.19k
    if (tmp == NULL) {
10019
0
        xmlErrMemory(ctxt);
10020
0
        xmlFree(buf);
10021
0
        return(NULL);
10022
0
    }
10023
1.19k
    buf = tmp;
10024
1.19k
                size = newSize;
10025
1.19k
      }
10026
557k
      buf[len++] = cur;
10027
557k
      NEXT;
10028
557k
      cur = CUR;
10029
557k
        }
10030
136k
  buf[len] = 0;
10031
136k
    } else {
10032
38
  xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL);
10033
38
    }
10034
136k
    return(buf);
10035
136k
}
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
138k
xmlParseEncodingDecl(xmlParserCtxt *ctxt) {
10053
138k
    xmlChar *encoding = NULL;
10054
10055
138k
    SKIP_BLANKS;
10056
138k
    if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0)
10057
2.30k
        return(NULL);
10058
10059
136k
    SKIP(8);
10060
136k
    SKIP_BLANKS;
10061
136k
    if (RAW != '=') {
10062
37
        xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10063
37
        return(NULL);
10064
37
    }
10065
136k
    NEXT;
10066
136k
    SKIP_BLANKS;
10067
136k
    if (RAW == '"') {
10068
136k
        NEXT;
10069
136k
        encoding = xmlParseEncName(ctxt);
10070
136k
        if (RAW != '"') {
10071
178
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10072
178
            xmlFree(encoding);
10073
178
            return(NULL);
10074
178
        } else
10075
135k
            NEXT;
10076
136k
    } else if (RAW == '\''){
10077
41
        NEXT;
10078
41
        encoding = xmlParseEncName(ctxt);
10079
41
        if (RAW != '\'') {
10080
15
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10081
15
            xmlFree(encoding);
10082
15
            return(NULL);
10083
15
        } else
10084
26
            NEXT;
10085
41
    } else {
10086
14
        xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10087
14
    }
10088
10089
135k
    if (encoding == NULL)
10090
24
        return(NULL);
10091
10092
135k
    xmlSetDeclaredEncoding(ctxt, encoding);
10093
10094
135k
    return(ctxt->encoding);
10095
135k
}
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
97.4k
xmlParseSDDecl(xmlParserCtxt *ctxt) {
10130
97.4k
    int standalone = -2;
10131
10132
97.4k
    SKIP_BLANKS;
10133
97.4k
    if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) {
10134
95.4k
  SKIP(10);
10135
95.4k
        SKIP_BLANKS;
10136
95.4k
  if (RAW != '=') {
10137
11
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10138
11
      return(standalone);
10139
11
        }
10140
95.4k
  NEXT;
10141
95.4k
  SKIP_BLANKS;
10142
95.4k
        if (RAW == '\''){
10143
51
      NEXT;
10144
51
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10145
31
          standalone = 0;
10146
31
                SKIP(2);
10147
31
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10148
7
                 (NXT(2) == 's')) {
10149
4
          standalone = 1;
10150
4
    SKIP(3);
10151
16
            } else {
10152
16
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10153
16
      }
10154
51
      if (RAW != '\'') {
10155
20
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10156
20
      } else
10157
31
          NEXT;
10158
95.4k
  } else if (RAW == '"'){
10159
95.3k
      NEXT;
10160
95.3k
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10161
448
          standalone = 0;
10162
448
    SKIP(2);
10163
94.9k
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10164
94.9k
                 (NXT(2) == 's')) {
10165
94.8k
          standalone = 1;
10166
94.8k
                SKIP(3);
10167
94.8k
            } else {
10168
47
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10169
47
      }
10170
95.3k
      if (RAW != '"') {
10171
61
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10172
61
      } else
10173
95.3k
          NEXT;
10174
95.3k
  } else {
10175
13
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10176
13
        }
10177
95.4k
    }
10178
97.4k
    return(standalone);
10179
97.4k
}
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
163k
xmlParseXMLDecl(xmlParserCtxt *ctxt) {
10192
163k
    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
163k
    ctxt->standalone = -2;
10201
10202
    /*
10203
     * We know that '<?xml' is here.
10204
     */
10205
163k
    SKIP(5);
10206
10207
163k
    if (!IS_BLANK_CH(RAW)) {
10208
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
10209
0
                 "Blank needed after '<?xml'\n");
10210
0
    }
10211
163k
    SKIP_BLANKS;
10212
10213
    /*
10214
     * We must have the VersionInfo here.
10215
     */
10216
163k
    version = xmlParseVersionInfo(ctxt);
10217
163k
    if (version == NULL) {
10218
968
  xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL);
10219
162k
    } else {
10220
162k
  if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) {
10221
      /*
10222
       * Changed here for XML-1.0 5th edition
10223
       */
10224
5.56k
      if (ctxt->options & XML_PARSE_OLD10) {
10225
0
    xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10226
0
                "Unsupported version '%s'\n",
10227
0
                version);
10228
5.56k
      } else {
10229
5.56k
          if ((version[0] == '1') && ((version[1] == '.'))) {
10230
5.39k
        xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION,
10231
5.39k
                      "Unsupported version '%s'\n",
10232
5.39k
          version, NULL);
10233
5.39k
    } else {
10234
172
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10235
172
              "Unsupported version '%s'\n",
10236
172
              version);
10237
172
    }
10238
5.56k
      }
10239
5.56k
  }
10240
162k
  if (ctxt->version != NULL)
10241
0
      xmlFree(ctxt->version);
10242
162k
  ctxt->version = version;
10243
162k
    }
10244
10245
    /*
10246
     * We may have the encoding declaration
10247
     */
10248
163k
    if (!IS_BLANK_CH(RAW)) {
10249
25.9k
        if ((RAW == '?') && (NXT(1) == '>')) {
10250
24.8k
      SKIP(2);
10251
24.8k
      return;
10252
24.8k
  }
10253
1.07k
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10254
1.07k
    }
10255
138k
    xmlParseEncodingDecl(ctxt);
10256
10257
    /*
10258
     * We may have the standalone status.
10259
     */
10260
138k
    if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) {
10261
41.0k
        if ((RAW == '?') && (NXT(1) == '>')) {
10262
41.0k
      SKIP(2);
10263
41.0k
      return;
10264
41.0k
  }
10265
43
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10266
43
    }
10267
10268
    /*
10269
     * We can grow the input buffer freely at that point
10270
     */
10271
97.4k
    GROW;
10272
10273
97.4k
    SKIP_BLANKS;
10274
97.4k
    ctxt->standalone = xmlParseSDDecl(ctxt);
10275
10276
97.4k
    SKIP_BLANKS;
10277
97.4k
    if ((RAW == '?') && (NXT(1) == '>')) {
10278
95.7k
        SKIP(2);
10279
95.7k
    } else if (RAW == '>') {
10280
        /* Deprecated old WD ... */
10281
63
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10282
63
  NEXT;
10283
1.56k
    } else {
10284
1.56k
        int c;
10285
10286
1.56k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10287
660k
        while ((PARSER_STOPPED(ctxt) == 0) &&
10288
660k
               ((c = CUR) != 0)) {
10289
659k
            NEXT;
10290
659k
            if (c == '>')
10291
1.02k
                break;
10292
659k
        }
10293
1.56k
    }
10294
97.4k
}
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
10.5k
xmlParseMisc(xmlParserCtxt *ctxt) {
10335
10.5k
    while (PARSER_STOPPED(ctxt) == 0) {
10336
10.5k
        SKIP_BLANKS;
10337
10.5k
        GROW;
10338
10.5k
        if ((RAW == '<') && (NXT(1) == '?')) {
10339
74
      xmlParsePI(ctxt);
10340
10.5k
        } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) {
10341
0
      xmlParseComment(ctxt);
10342
10.5k
        } else {
10343
10.5k
            break;
10344
10.5k
        }
10345
10.5k
    }
10346
10.5k
}
10347
10348
static void
10349
144k
xmlFinishDocument(xmlParserCtxtPtr ctxt) {
10350
144k
    xmlDocPtr doc;
10351
10352
    /*
10353
     * SAX: end of the document processing.
10354
     */
10355
144k
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10356
5.25k
        ctxt->sax->endDocument(ctxt->userData);
10357
10358
    /*
10359
     * Remove locally kept entity definitions if the tree was not built
10360
     */
10361
144k
    doc = ctxt->myDoc;
10362
144k
    if ((doc != NULL) &&
10363
5.26k
        (xmlStrEqual(doc->version, SAX_COMPAT_MODE))) {
10364
13
        xmlFreeDoc(doc);
10365
13
        ctxt->myDoc = NULL;
10366
13
    }
10367
144k
}
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
5.29k
xmlParseDocument(xmlParserCtxt *ctxt) {
10380
5.29k
    if ((ctxt == NULL) || (ctxt->input == NULL))
10381
0
        return(-1);
10382
10383
5.29k
    GROW;
10384
10385
    /*
10386
     * SAX: detecting the level.
10387
     */
10388
5.29k
    xmlCtxtInitializeLate(ctxt);
10389
10390
5.29k
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10391
5.29k
        ctxt->sax->setDocumentLocator(ctxt->userData,
10392
5.29k
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10393
5.29k
    }
10394
10395
5.29k
    xmlDetectEncoding(ctxt);
10396
10397
5.29k
    if (CUR == 0) {
10398
41
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10399
41
  return(-1);
10400
41
    }
10401
10402
5.25k
    GROW;
10403
5.25k
    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
4.97k
  xmlParseXMLDecl(ctxt);
10409
4.97k
  SKIP_BLANKS;
10410
4.97k
    } else {
10411
284
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10412
284
        if (ctxt->version == NULL) {
10413
0
            xmlErrMemory(ctxt);
10414
0
            return(-1);
10415
0
        }
10416
284
    }
10417
5.25k
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10418
5.24k
        ctxt->sax->startDocument(ctxt->userData);
10419
5.25k
    if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) &&
10420
5.24k
        (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
5.25k
    xmlParseMisc(ctxt);
10428
10429
    /*
10430
     * Then possibly doc type declaration(s) and more Misc
10431
     * (doctypedecl Misc*)?
10432
     */
10433
5.25k
    GROW;
10434
5.25k
    if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) {
10435
10436
0
  ctxt->inSubset = 1;
10437
0
  xmlParseDocTypeDecl(ctxt);
10438
0
  if (RAW == '[') {
10439
0
      xmlParseInternalSubset(ctxt);
10440
0
  } else if (RAW == '>') {
10441
0
            NEXT;
10442
0
        }
10443
10444
  /*
10445
   * Create and update the external subset.
10446
   */
10447
0
  ctxt->inSubset = 2;
10448
0
  if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) &&
10449
0
      (!ctxt->disableSAX))
10450
0
      ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
10451
0
                                ctxt->extSubSystem, ctxt->extSubURI);
10452
0
  ctxt->inSubset = 0;
10453
10454
0
        xmlCleanSpecialAttr(ctxt);
10455
10456
0
  xmlParseMisc(ctxt);
10457
0
    }
10458
10459
    /*
10460
     * Time to start parsing the tree itself
10461
     */
10462
5.25k
    GROW;
10463
5.25k
    if (RAW != '<') {
10464
4
        if (ctxt->wellFormed)
10465
1
            xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10466
1
                           "Start tag expected, '<' not found\n");
10467
5.25k
    } else {
10468
5.25k
  xmlParseElement(ctxt);
10469
10470
  /*
10471
   * The Misc part at the end
10472
   */
10473
5.25k
  xmlParseMisc(ctxt);
10474
10475
5.25k
        xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
10476
5.25k
    }
10477
10478
5.25k
    ctxt->instate = XML_PARSER_EOF;
10479
5.25k
    xmlFinishDocument(ctxt);
10480
10481
5.25k
    if (! ctxt->wellFormed) {
10482
310
  ctxt->valid = 0;
10483
310
  return(-1);
10484
310
    }
10485
10486
4.94k
    return(0);
10487
5.25k
}
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
3.88M
xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) {
10575
3.88M
    const xmlChar *cur;
10576
10577
3.88M
    if (ctxt->checkIndex == 0) {
10578
3.88M
        cur = ctxt->input->cur + 1;
10579
3.88M
    } else {
10580
624
        cur = ctxt->input->cur + ctxt->checkIndex;
10581
624
    }
10582
10583
3.88M
    if (memchr(cur, c, ctxt->input->end - cur) == NULL) {
10584
822
        size_t index = ctxt->input->end - ctxt->input->cur;
10585
10586
822
        if (index > LONG_MAX) {
10587
0
            ctxt->checkIndex = 0;
10588
0
            return(1);
10589
0
        }
10590
822
        ctxt->checkIndex = index;
10591
822
        return(0);
10592
3.88M
    } else {
10593
3.88M
        ctxt->checkIndex = 0;
10594
3.88M
        return(1);
10595
3.88M
    }
10596
3.88M
}
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
84.7k
                     const char *str, size_t strLen) {
10609
84.7k
    const xmlChar *cur, *term;
10610
10611
84.7k
    if (ctxt->checkIndex == 0) {
10612
83.8k
        cur = ctxt->input->cur + startDelta;
10613
83.8k
    } else {
10614
867
        cur = ctxt->input->cur + ctxt->checkIndex;
10615
867
    }
10616
10617
84.7k
    term = BAD_CAST strstr((const char *) cur, str);
10618
84.7k
    if (term == NULL) {
10619
1.19k
        const xmlChar *end = ctxt->input->end;
10620
1.19k
        size_t index;
10621
10622
        /* Rescan (strLen - 1) characters. */
10623
1.19k
        if ((size_t) (end - cur) < strLen)
10624
27
            end = cur;
10625
1.16k
        else
10626
1.16k
            end -= strLen - 1;
10627
1.19k
        index = end - ctxt->input->cur;
10628
1.19k
        if (index > LONG_MAX) {
10629
0
            ctxt->checkIndex = 0;
10630
0
            return(ctxt->input->end - strLen);
10631
0
        }
10632
1.19k
        ctxt->checkIndex = index;
10633
83.5k
    } else {
10634
83.5k
        ctxt->checkIndex = 0;
10635
83.5k
    }
10636
10637
84.7k
    return(term);
10638
84.7k
}
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
72.2k
xmlParseLookupCharData(xmlParserCtxtPtr ctxt) {
10647
72.2k
    const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex;
10648
72.2k
    const xmlChar *end = ctxt->input->end;
10649
72.2k
    size_t index;
10650
10651
891k
    while (cur < end) {
10652
888k
        if ((*cur == '<') || (*cur == '&')) {
10653
70.0k
            ctxt->checkIndex = 0;
10654
70.0k
            return(1);
10655
70.0k
        }
10656
818k
        cur++;
10657
818k
    }
10658
10659
2.17k
    index = cur - ctxt->input->cur;
10660
2.17k
    if (index > LONG_MAX) {
10661
0
        ctxt->checkIndex = 0;
10662
0
        return(1);
10663
0
    }
10664
2.17k
    ctxt->checkIndex = index;
10665
2.17k
    return(0);
10666
2.17k
}
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
18.5M
xmlParseLookupGt(xmlParserCtxtPtr ctxt) {
10676
18.5M
    const xmlChar *cur;
10677
18.5M
    const xmlChar *end = ctxt->input->end;
10678
18.5M
    int state = ctxt->endCheckState;
10679
18.5M
    size_t index;
10680
10681
18.5M
    if (ctxt->checkIndex == 0)
10682
18.5M
        cur = ctxt->input->cur + 1;
10683
7.14k
    else
10684
7.14k
        cur = ctxt->input->cur + ctxt->checkIndex;
10685
10686
461M
    while (cur < end) {
10687
461M
        if (state) {
10688
121M
            if (*cur == state)
10689
10.8M
                state = 0;
10690
339M
        } else if (*cur == '\'' || *cur == '"') {
10691
10.8M
            state = *cur;
10692
329M
        } else if (*cur == '>') {
10693
18.5M
            ctxt->checkIndex = 0;
10694
18.5M
            ctxt->endCheckState = 0;
10695
18.5M
            return(1);
10696
18.5M
        }
10697
442M
        cur++;
10698
442M
    }
10699
10700
10.3k
    index = cur - ctxt->input->cur;
10701
10.3k
    if (index > LONG_MAX) {
10702
0
        ctxt->checkIndex = 0;
10703
0
        ctxt->endCheckState = 0;
10704
0
        return(1);
10705
0
    }
10706
10.3k
    ctxt->checkIndex = index;
10707
10.3k
    ctxt->endCheckState = state;
10708
10.3k
    return(0);
10709
10.3k
}
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
792
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
792
    const xmlChar *cur, *start;
10727
792
    const xmlChar *end = ctxt->input->end;
10728
792
    int state = ctxt->endCheckState;
10729
792
    size_t index;
10730
10731
792
    if (ctxt->checkIndex == 0) {
10732
594
        cur = ctxt->input->cur + 1;
10733
594
    } else {
10734
198
        cur = ctxt->input->cur + ctxt->checkIndex;
10735
198
    }
10736
792
    start = cur;
10737
10738
15.1M
    while (cur < end) {
10739
15.1M
        if (state == '-') {
10740
2.08M
            if ((*cur == '-') &&
10741
137k
                (cur[1] == '-') &&
10742
73.1k
                (cur[2] == '>')) {
10743
33.4k
                state = 0;
10744
33.4k
                cur += 3;
10745
33.4k
                start = cur;
10746
33.4k
                continue;
10747
33.4k
            }
10748
2.08M
        }
10749
13.0M
        else if (state == ']') {
10750
11.0k
            if (*cur == '>') {
10751
133
                ctxt->checkIndex = 0;
10752
133
                ctxt->endCheckState = 0;
10753
133
                return(1);
10754
133
            }
10755
10.8k
            if (IS_BLANK_CH(*cur)) {
10756
4.62k
                state = ' ';
10757
6.25k
            } else if (*cur != ']') {
10758
3.72k
                state = 0;
10759
3.72k
                start = cur;
10760
3.72k
                continue;
10761
3.72k
            }
10762
10.8k
        }
10763
13.0M
        else if (state == ' ') {
10764
15.9k
            if (*cur == '>') {
10765
4
                ctxt->checkIndex = 0;
10766
4
                ctxt->endCheckState = 0;
10767
4
                return(1);
10768
4
            }
10769
15.9k
            if (!IS_BLANK_CH(*cur)) {
10770
4.61k
                state = 0;
10771
4.61k
                start = cur;
10772
4.61k
                continue;
10773
4.61k
            }
10774
15.9k
        }
10775
12.9M
        else if (state != 0) {
10776
8.93M
            if (*cur == state) {
10777
76.9k
                state = 0;
10778
76.9k
                start = cur + 1;
10779
76.9k
            }
10780
8.93M
        }
10781
4.05M
        else if (*cur == '<') {
10782
162k
            if ((cur[1] == '!') &&
10783
74.7k
                (cur[2] == '-') &&
10784
33.6k
                (cur[3] == '-')) {
10785
33.5k
                state = '-';
10786
33.5k
                cur += 4;
10787
                /* Don't treat <!--> as comment */
10788
33.5k
                start = cur;
10789
33.5k
                continue;
10790
33.5k
            }
10791
162k
        }
10792
3.89M
        else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) {
10793
85.6k
            state = *cur;
10794
85.6k
        }
10795
10796
15.0M
        cur++;
10797
15.0M
    }
10798
10799
    /*
10800
     * Rescan the three last characters to detect "<!--" and "-->"
10801
     * split across chunks.
10802
     */
10803
655
    if ((state == 0) || (state == '-')) {
10804
317
        if (cur - start < 3)
10805
22
            cur = start;
10806
295
        else
10807
295
            cur -= 3;
10808
317
    }
10809
655
    index = cur - ctxt->input->cur;
10810
655
    if (index > LONG_MAX) {
10811
0
        ctxt->checkIndex = 0;
10812
0
        ctxt->endCheckState = 0;
10813
0
        return(1);
10814
0
    }
10815
655
    ctxt->checkIndex = index;
10816
655
    ctxt->endCheckState = state;
10817
655
    return(0);
10818
655
}
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
236k
xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) {
10829
236k
    int ret = 0;
10830
236k
    size_t avail;
10831
236k
    xmlChar cur, next;
10832
10833
236k
    if (ctxt->input == NULL)
10834
0
        return(0);
10835
10836
236k
    if ((ctxt->input != NULL) &&
10837
236k
        (ctxt->input->cur - ctxt->input->base > 4096)) {
10838
19.8k
        xmlParserShrink(ctxt);
10839
19.8k
    }
10840
10841
92.8M
    while (ctxt->disableSAX == 0) {
10842
92.7M
        avail = ctxt->input->end - ctxt->input->cur;
10843
92.7M
        if (avail < 1)
10844
131k
      goto done;
10845
92.6M
        switch (ctxt->instate) {
10846
1.48k
            case XML_PARSER_EOF:
10847
          /*
10848
     * Document parsing is done !
10849
     */
10850
1.48k
          goto done;
10851
213k
            case XML_PARSER_START:
10852
                /*
10853
                 * Very first chars read from the document flow.
10854
                 */
10855
213k
                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
213k
                if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) &&
10863
6
                    (!terminate) && (avail < 200))
10864
0
                    goto done;
10865
10866
213k
                xmlDetectEncoding(ctxt);
10867
213k
                ctxt->instate = XML_PARSER_XML_DECL;
10868
213k
    break;
10869
10870
213k
            case XML_PARSER_XML_DECL:
10871
213k
    if ((!terminate) && (avail < 2))
10872
0
        goto done;
10873
213k
    cur = ctxt->input->cur[0];
10874
213k
    next = ctxt->input->cur[1];
10875
213k
          if ((cur == '<') && (next == '?')) {
10876
        /* PI or XML decl */
10877
175k
        if ((!terminate) &&
10878
19.4k
                        (!xmlParseLookupString(ctxt, 2, "?>", 2)))
10879
150
      goto done;
10880
175k
        if ((ctxt->input->cur[2] == 'x') &&
10881
172k
      (ctxt->input->cur[3] == 'm') &&
10882
171k
      (ctxt->input->cur[4] == 'l') &&
10883
169k
      (IS_BLANK_CH(ctxt->input->cur[5]))) {
10884
158k
      ret += 5;
10885
158k
      xmlParseXMLDecl(ctxt);
10886
158k
        } else {
10887
17.2k
      ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10888
17.2k
                        if (ctxt->version == NULL) {
10889
0
                            xmlErrMemory(ctxt);
10890
0
                            break;
10891
0
                        }
10892
17.2k
        }
10893
175k
    } else {
10894
37.9k
        ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10895
37.9k
        if (ctxt->version == NULL) {
10896
0
            xmlErrMemory(ctxt);
10897
0
      break;
10898
0
        }
10899
37.9k
    }
10900
213k
                if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10901
0
                    ctxt->sax->setDocumentLocator(ctxt->userData,
10902
0
                            (xmlSAXLocator *) &xmlDefaultSAXLocator);
10903
0
                }
10904
213k
                if ((ctxt->sax) && (ctxt->sax->startDocument) &&
10905
0
                    (!ctxt->disableSAX))
10906
0
                    ctxt->sax->startDocument(ctxt->userData);
10907
213k
                ctxt->instate = XML_PARSER_MISC;
10908
213k
    break;
10909
32.7M
            case XML_PARSER_START_TAG: {
10910
32.7M
          const xmlChar *name;
10911
32.7M
    const xmlChar *prefix = NULL;
10912
32.7M
    const xmlChar *URI = NULL;
10913
32.7M
                int line = ctxt->input->line;
10914
32.7M
    int nbNs = 0;
10915
10916
32.7M
    if ((!terminate) && (avail < 2))
10917
2
        goto done;
10918
32.7M
    cur = ctxt->input->cur[0];
10919
32.7M
          if (cur != '<') {
10920
476
        xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10921
476
                                   "Start tag expected, '<' not found");
10922
476
                    ctxt->instate = XML_PARSER_EOF;
10923
476
                    xmlFinishDocument(ctxt);
10924
476
        goto done;
10925
476
    }
10926
32.7M
    if ((!terminate) && (!xmlParseLookupGt(ctxt)))
10927
10.5k
                    goto done;
10928
32.7M
    if (ctxt->spaceNr == 0)
10929
0
        spacePush(ctxt, -1);
10930
32.7M
    else if (*ctxt->space == -2)
10931
8.25M
        spacePush(ctxt, -1);
10932
24.4M
    else
10933
24.4M
        spacePush(ctxt, *ctxt->space);
10934
32.7M
#ifdef LIBXML_SAX1_ENABLED
10935
32.7M
    if (ctxt->sax2)
10936
32.7M
#endif /* LIBXML_SAX1_ENABLED */
10937
32.7M
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
10938
312
#ifdef LIBXML_SAX1_ENABLED
10939
312
    else
10940
312
        name = xmlParseStartTag(ctxt);
10941
32.7M
#endif /* LIBXML_SAX1_ENABLED */
10942
32.7M
    if (name == NULL) {
10943
4.32k
        spacePop(ctxt);
10944
4.32k
                    ctxt->instate = XML_PARSER_EOF;
10945
4.32k
                    xmlFinishDocument(ctxt);
10946
4.32k
        goto done;
10947
4.32k
    }
10948
32.6M
#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
32.6M
    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
32.6M
#endif /* LIBXML_VALID_ENABLED */
10958
10959
    /*
10960
     * Check for an Empty Element.
10961
     */
10962
32.6M
    if ((RAW == '/') && (NXT(1) == '>')) {
10963
8.24M
        SKIP(2);
10964
10965
8.24M
        if (ctxt->sax2) {
10966
8.24M
      if ((ctxt->sax != NULL) &&
10967
8.24M
          (ctxt->sax->endElementNs != NULL) &&
10968
8.24M
          (!ctxt->disableSAX))
10969
8.24M
          ctxt->sax->endElementNs(ctxt->userData, name,
10970
8.24M
                                  prefix, URI);
10971
8.24M
      if (nbNs > 0)
10972
140k
          xmlParserNsPop(ctxt, nbNs);
10973
8.24M
#ifdef LIBXML_SAX1_ENABLED
10974
18.4E
        } else {
10975
18.4E
      if ((ctxt->sax != NULL) &&
10976
0
          (ctxt->sax->endElement != NULL) &&
10977
0
          (!ctxt->disableSAX))
10978
0
          ctxt->sax->endElement(ctxt->userData, name);
10979
18.4E
#endif /* LIBXML_SAX1_ENABLED */
10980
18.4E
        }
10981
8.24M
        spacePop(ctxt);
10982
24.4M
    } else if (RAW == '>') {
10983
24.4M
        NEXT;
10984
24.4M
                    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
10985
24.4M
    } else {
10986
50.9k
        xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED,
10987
50.9k
           "Couldn't find end of Start Tag %s\n",
10988
50.9k
           name);
10989
50.9k
        nodePop(ctxt);
10990
50.9k
        spacePop(ctxt);
10991
50.9k
                    if (nbNs > 0)
10992
5.28k
                        xmlParserNsPop(ctxt, nbNs);
10993
50.9k
    }
10994
10995
32.6M
                if (ctxt->nameNr == 0)
10996
9.46k
                    ctxt->instate = XML_PARSER_EPILOG;
10997
32.6M
                else
10998
32.6M
                    ctxt->instate = XML_PARSER_CONTENT;
10999
32.6M
                break;
11000
32.7M
      }
11001
51.4M
            case XML_PARSER_CONTENT: {
11002
51.4M
    cur = ctxt->input->cur[0];
11003
11004
51.4M
    if (cur == '<') {
11005
40.2M
                    if ((!terminate) && (avail < 2))
11006
511
                        goto done;
11007
40.2M
        next = ctxt->input->cur[1];
11008
11009
40.2M
                    if (next == '/') {
11010
7.71M
                        ctxt->instate = XML_PARSER_END_TAG;
11011
7.71M
                        break;
11012
32.5M
                    } else if (next == '?') {
11013
12.7k
                        if ((!terminate) &&
11014
7.39k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11015
96
                            goto done;
11016
12.6k
                        xmlParsePI(ctxt);
11017
12.6k
                        ctxt->instate = XML_PARSER_CONTENT;
11018
12.6k
                        break;
11019
32.5M
                    } else if (next == '!') {
11020
63.7k
                        if ((!terminate) && (avail < 3))
11021
6
                            goto done;
11022
63.6k
                        next = ctxt->input->cur[2];
11023
11024
63.6k
                        if (next == '-') {
11025
50.0k
                            if ((!terminate) && (avail < 4))
11026
9
                                goto done;
11027
50.0k
                            if (ctxt->input->cur[3] == '-') {
11028
50.0k
                                if ((!terminate) &&
11029
33.6k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11030
103
                                    goto done;
11031
49.9k
                                xmlParseComment(ctxt);
11032
49.9k
                                ctxt->instate = XML_PARSER_CONTENT;
11033
49.9k
                                break;
11034
50.0k
                            }
11035
50.0k
                        } else if (next == '[') {
11036
13.4k
                            if ((!terminate) && (avail < 9))
11037
4
                                goto done;
11038
13.4k
                            if ((ctxt->input->cur[2] == '[') &&
11039
13.4k
                                (ctxt->input->cur[3] == 'C') &&
11040
13.4k
                                (ctxt->input->cur[4] == 'D') &&
11041
13.4k
                                (ctxt->input->cur[5] == 'A') &&
11042
13.4k
                                (ctxt->input->cur[6] == 'T') &&
11043
13.4k
                                (ctxt->input->cur[7] == 'A') &&
11044
13.4k
                                (ctxt->input->cur[8] == '[')) {
11045
13.4k
                                if ((!terminate) &&
11046
6.12k
                                    (!xmlParseLookupString(ctxt, 9, "]]>", 3)))
11047
740
                                    goto done;
11048
12.6k
                                ctxt->instate = XML_PARSER_CDATA_SECTION;
11049
12.6k
                                xmlParseCDSect(ctxt);
11050
12.6k
                                ctxt->instate = XML_PARSER_CONTENT;
11051
12.6k
                                break;
11052
13.4k
                            }
11053
13.4k
                        }
11054
63.6k
                    }
11055
40.2M
    } else if (cur == '&') {
11056
328k
        if ((!terminate) && (!xmlParseLookupChar(ctxt, ';')))
11057
107
      goto done;
11058
328k
        xmlParseReference(ctxt);
11059
328k
                    break;
11060
10.8M
    } 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
10.8M
        if (avail < XML_PARSER_BIG_BUFFER_SIZE) {
11074
422k
      if ((!terminate) && (!xmlParseLookupCharData(ctxt)))
11075
2.17k
          goto done;
11076
422k
                    }
11077
10.8M
                    ctxt->checkIndex = 0;
11078
10.8M
        xmlParseCharDataInternal(ctxt, !terminate);
11079
10.8M
                    break;
11080
10.8M
    }
11081
11082
32.4M
                ctxt->instate = XML_PARSER_START_TAG;
11083
32.4M
    break;
11084
51.4M
      }
11085
7.71M
            case XML_PARSER_END_TAG:
11086
7.71M
    if ((!terminate) && (!xmlParseLookupChar(ctxt, '>')))
11087
715
        goto done;
11088
7.71M
    if (ctxt->sax2) {
11089
7.71M
              xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
11090
7.71M
        nameNsPop(ctxt);
11091
7.71M
    }
11092
37
#ifdef LIBXML_SAX1_ENABLED
11093
37
      else
11094
37
        xmlParseEndTag1(ctxt, 0);
11095
7.71M
#endif /* LIBXML_SAX1_ENABLED */
11096
7.71M
    if (ctxt->nameNr == 0) {
11097
123k
        ctxt->instate = XML_PARSER_EPILOG;
11098
7.58M
    } else {
11099
7.58M
        ctxt->instate = XML_PARSER_CONTENT;
11100
7.58M
    }
11101
7.71M
    break;
11102
251k
            case XML_PARSER_MISC:
11103
254k
            case XML_PARSER_PROLOG:
11104
266k
            case XML_PARSER_EPILOG:
11105
266k
    SKIP_BLANKS;
11106
266k
                avail = ctxt->input->end - ctxt->input->cur;
11107
266k
    if (avail < 1)
11108
8.54k
        goto done;
11109
257k
    if (ctxt->input->cur[0] == '<') {
11110
256k
                    if ((!terminate) && (avail < 2))
11111
12
                        goto done;
11112
256k
                    next = ctxt->input->cur[1];
11113
256k
                    if (next == '?') {
11114
40.8k
                        if ((!terminate) &&
11115
16.5k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11116
58
                            goto done;
11117
40.8k
                        xmlParsePI(ctxt);
11118
40.8k
                        break;
11119
215k
                    } else if (next == '!') {
11120
11.3k
                        if ((!terminate) && (avail < 3))
11121
5
                            goto done;
11122
11123
11.3k
                        if (ctxt->input->cur[2] == '-') {
11124
3.57k
                            if ((!terminate) && (avail < 4))
11125
0
                                goto done;
11126
3.57k
                            if (ctxt->input->cur[3] == '-') {
11127
3.55k
                                if ((!terminate) &&
11128
1.52k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11129
43
                                    goto done;
11130
3.51k
                                xmlParseComment(ctxt);
11131
3.51k
                                break;
11132
3.55k
                            }
11133
7.73k
                        } else if (ctxt->instate == XML_PARSER_MISC) {
11134
7.71k
                            if ((!terminate) && (avail < 9))
11135
0
                                goto done;
11136
7.71k
                            if ((ctxt->input->cur[2] == 'D') &&
11137
7.70k
                                (ctxt->input->cur[3] == 'O') &&
11138
7.69k
                                (ctxt->input->cur[4] == 'C') &&
11139
7.68k
                                (ctxt->input->cur[5] == 'T') &&
11140
7.67k
                                (ctxt->input->cur[6] == 'Y') &&
11141
7.67k
                                (ctxt->input->cur[7] == 'P') &&
11142
7.66k
                                (ctxt->input->cur[8] == 'E')) {
11143
7.65k
                                if ((!terminate) && (!xmlParseLookupGt(ctxt)))
11144
79
                                    goto done;
11145
7.57k
                                ctxt->inSubset = 1;
11146
7.57k
                                xmlParseDocTypeDecl(ctxt);
11147
7.57k
                                if (RAW == '[') {
11148
7.14k
                                    ctxt->instate = XML_PARSER_DTD;
11149
7.14k
                                } else {
11150
439
                                    if (RAW == '>')
11151
219
                                        NEXT;
11152
                                    /*
11153
                                     * Create and update the external subset.
11154
                                     */
11155
439
                                    ctxt->inSubset = 2;
11156
439
                                    if ((ctxt->sax != NULL) &&
11157
439
                                        (!ctxt->disableSAX) &&
11158
217
                                        (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
439
                                    ctxt->inSubset = 0;
11165
439
                                    xmlCleanSpecialAttr(ctxt);
11166
439
                                    ctxt->instate = XML_PARSER_PROLOG;
11167
439
                                }
11168
7.57k
                                break;
11169
7.65k
                            }
11170
7.71k
                        }
11171
11.3k
                    }
11172
256k
                }
11173
11174
205k
                if (ctxt->instate == XML_PARSER_EPILOG) {
11175
1.32k
                    if (ctxt->errNo == XML_ERR_OK)
11176
135
                        xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL);
11177
1.32k
        ctxt->instate = XML_PARSER_EOF;
11178
1.32k
                    xmlFinishDocument(ctxt);
11179
204k
                } else {
11180
204k
        ctxt->instate = XML_PARSER_START_TAG;
11181
204k
    }
11182
205k
    break;
11183
7.72k
            case XML_PARSER_DTD: {
11184
7.72k
                if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt)))
11185
655
                    goto done;
11186
7.06k
    xmlParseInternalSubset(ctxt);
11187
7.06k
    ctxt->inSubset = 2;
11188
7.06k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
11189
2.36k
        (ctxt->sax->externalSubset != NULL))
11190
0
        ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
11191
0
          ctxt->extSubSystem, ctxt->extSubURI);
11192
7.06k
    ctxt->inSubset = 0;
11193
7.06k
    xmlCleanSpecialAttr(ctxt);
11194
7.06k
    ctxt->instate = XML_PARSER_PROLOG;
11195
7.06k
                break;
11196
7.72k
      }
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
92.6M
  }
11203
92.6M
    }
11204
236k
done:
11205
236k
    return(ret);
11206
236k
}
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
236k
              int terminate) {
11237
236k
    size_t curBase;
11238
236k
    size_t maxLength;
11239
236k
    size_t pos;
11240
236k
    int end_in_lf = 0;
11241
236k
    int res;
11242
11243
236k
    if ((ctxt == NULL) || (size < 0))
11244
0
        return(XML_ERR_ARGUMENT);
11245
236k
    if ((chunk == NULL) && (size > 0))
11246
0
        return(XML_ERR_ARGUMENT);
11247
236k
    if ((ctxt->input == NULL) || (ctxt->input->buf == NULL))
11248
0
        return(XML_ERR_ARGUMENT);
11249
236k
    if (ctxt->disableSAX != 0)
11250
0
        return(ctxt->errNo);
11251
11252
236k
    ctxt->input->flags |= XML_INPUT_PROGRESSIVE;
11253
236k
    if (ctxt->instate == XML_PARSER_START)
11254
213k
        xmlCtxtInitializeLate(ctxt);
11255
236k
    if ((size > 0) && (chunk != NULL) && (!terminate) &&
11256
33.6k
        (chunk[size - 1] == '\r')) {
11257
98
  end_in_lf = 1;
11258
98
  size--;
11259
98
    }
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
236k
    pos = ctxt->input->cur - ctxt->input->base;
11266
236k
    res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk);
11267
236k
    xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11268
236k
    if (res < 0) {
11269
4
        xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11270
4
        return(ctxt->errNo);
11271
4
    }
11272
11273
236k
    xmlParseTryOrFinish(ctxt, terminate);
11274
11275
236k
    curBase = ctxt->input->cur - ctxt->input->base;
11276
236k
    maxLength = (ctxt->options & XML_PARSE_HUGE) ?
11277
236k
                XML_MAX_HUGE_LENGTH :
11278
236k
                XML_MAX_LOOKUP_LIMIT;
11279
236k
    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
236k
    if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0))
11285
79.0k
        return(ctxt->errNo);
11286
11287
157k
    if (end_in_lf == 1) {
11288
84
  pos = ctxt->input->cur - ctxt->input->base;
11289
84
  res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r");
11290
84
  xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11291
84
        if (res < 0) {
11292
0
            xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11293
0
            return(ctxt->errNo);
11294
0
        }
11295
84
    }
11296
157k
    if (terminate) {
11297
  /*
11298
   * Check for termination
11299
   */
11300
133k
        if ((ctxt->instate != XML_PARSER_EOF) &&
11301
132k
            (ctxt->instate != XML_PARSER_EPILOG)) {
11302
9.98k
            if (ctxt->nameNr > 0) {
11303
9.82k
                const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
11304
9.82k
                int line = ctxt->pushTab[ctxt->nameNr - 1].line;
11305
9.82k
                xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
11306
9.82k
                        "Premature end of data in tag %s line %d\n",
11307
9.82k
                        name, line, NULL);
11308
9.82k
            } else if (ctxt->instate == XML_PARSER_START) {
11309
0
                xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
11310
159
            } else {
11311
159
                xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
11312
159
                               "Start tag expected, '<' not found\n");
11313
159
            }
11314
123k
        } else {
11315
123k
            xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
11316
123k
        }
11317
133k
  if (ctxt->instate != XML_PARSER_EOF) {
11318
132k
            ctxt->instate = XML_PARSER_EOF;
11319
132k
            xmlFinishDocument(ctxt);
11320
132k
  }
11321
133k
    }
11322
157k
    if (ctxt->wellFormed == 0)
11323
9.98k
  return((xmlParserErrors) ctxt->errNo);
11324
147k
    else
11325
147k
        return(0);
11326
157k
}
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
213k
                        const char *chunk, int size, const char *filename) {
11358
213k
    xmlParserCtxtPtr ctxt;
11359
213k
    xmlParserInputPtr input;
11360
11361
213k
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11362
213k
    if (ctxt == NULL)
11363
0
  return(NULL);
11364
11365
213k
    ctxt->options &= ~XML_PARSE_NODICT;
11366
213k
    ctxt->dictNames = 1;
11367
11368
213k
    input = xmlNewPushInput(filename, chunk, size);
11369
213k
    if (input == NULL) {
11370
0
  xmlFreeParserCtxt(ctxt);
11371
0
  return(NULL);
11372
0
    }
11373
213k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11374
0
        xmlFreeInputStream(input);
11375
0
        xmlFreeParserCtxt(ctxt);
11376
0
        return(NULL);
11377
0
    }
11378
11379
213k
    return(ctxt);
11380
213k
}
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
0
                            int hasTextDecl, int buildTree) {
11669
0
    xmlNodePtr root = NULL;
11670
0
    xmlNodePtr list = NULL;
11671
0
    xmlChar *rootName = BAD_CAST "#root";
11672
0
    int result;
11673
11674
0
    if (buildTree) {
11675
0
        root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL);
11676
0
        if (root == NULL) {
11677
0
            xmlErrMemory(ctxt);
11678
0
            goto error;
11679
0
        }
11680
0
    }
11681
11682
0
    if (xmlCtxtPushInput(ctxt, input) < 0)
11683
0
        goto error;
11684
11685
0
    nameNsPush(ctxt, rootName, NULL, NULL, 0, 0);
11686
0
    spacePush(ctxt, -1);
11687
11688
0
    if (buildTree)
11689
0
        nodePush(ctxt, root);
11690
11691
0
    if (hasTextDecl) {
11692
0
        xmlDetectEncoding(ctxt);
11693
11694
        /*
11695
         * Parse a possible text declaration first
11696
         */
11697
0
        if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
11698
0
            (IS_BLANK_CH(NXT(5)))) {
11699
0
            xmlParseTextDecl(ctxt);
11700
            /*
11701
             * An XML-1.0 document can't reference an entity not XML-1.0
11702
             */
11703
0
            if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
11704
0
                (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
11705
0
                xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
11706
0
                               "Version mismatch between document and "
11707
0
                               "entity\n");
11708
0
            }
11709
0
        }
11710
0
    }
11711
11712
0
    xmlParseContentInternal(ctxt);
11713
11714
0
    if (ctxt->input->cur < ctxt->input->end)
11715
0
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
11716
11717
0
    if ((ctxt->wellFormed) ||
11718
0
        ((ctxt->recovery) && (!xmlCtxtIsCatastrophicError(ctxt)))) {
11719
0
        if (root != NULL) {
11720
0
            xmlNodePtr cur;
11721
11722
            /*
11723
             * Unlink newly created node list.
11724
             */
11725
0
            list = root->children;
11726
0
            root->children = NULL;
11727
0
            root->last = NULL;
11728
0
            for (cur = list; cur != NULL; cur = cur->next)
11729
0
                cur->parent = NULL;
11730
0
        }
11731
0
    }
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
0
    do {
11738
0
        ctxt->input->cur = ctxt->input->end;
11739
0
        xmlParserShrink(ctxt);
11740
0
        result = xmlParserGrow(ctxt);
11741
0
    } while (result > 0);
11742
11743
0
    if (buildTree)
11744
0
        nodePop(ctxt);
11745
11746
0
    namePop(ctxt);
11747
0
    spacePop(ctxt);
11748
11749
0
    xmlCtxtPopInput(ctxt);
11750
11751
0
error:
11752
0
    xmlFreeNode(root);
11753
11754
0
    return(list);
11755
0
}
11756
11757
static void
11758
0
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
11759
0
    xmlParserInputPtr input;
11760
0
    xmlNodePtr list;
11761
0
    unsigned long consumed;
11762
0
    int isExternal;
11763
0
    int buildTree;
11764
0
    int oldMinNsIndex;
11765
0
    int oldNodelen, oldNodemem;
11766
11767
0
    isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY);
11768
0
    buildTree = (ctxt->node != NULL);
11769
11770
    /*
11771
     * Recursion check
11772
     */
11773
0
    if (ent->flags & XML_ENT_EXPANDING) {
11774
0
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
11775
0
        goto error;
11776
0
    }
11777
11778
    /*
11779
     * Load entity
11780
     */
11781
0
    input = xmlNewEntityInputStream(ctxt, ent);
11782
0
    if (input == NULL)
11783
0
        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
0
    oldMinNsIndex = ctxt->nsdb->minNsIndex;
11791
0
    if (buildTree)
11792
0
        ctxt->nsdb->minNsIndex = ctxt->nsNr;
11793
11794
0
    oldNodelen = ctxt->nodelen;
11795
0
    oldNodemem = ctxt->nodemem;
11796
0
    ctxt->nodelen = 0;
11797
0
    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
0
    ent->flags |= XML_ENT_EXPANDING;
11817
11818
0
    list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree);
11819
11820
0
    ent->flags &= ~XML_ENT_EXPANDING;
11821
11822
0
    ctxt->nsdb->minNsIndex = oldMinNsIndex;
11823
0
    ctxt->nodelen = oldNodelen;
11824
0
    ctxt->nodemem = oldNodemem;
11825
11826
    /*
11827
     * Entity size accounting
11828
     */
11829
0
    consumed = input->consumed;
11830
0
    xmlSaturatedAddSizeT(&consumed, input->end - input->base);
11831
11832
0
    if ((ent->flags & XML_ENT_CHECKED) == 0)
11833
0
        xmlSaturatedAdd(&ent->expandedSize, consumed);
11834
11835
0
    if ((ent->flags & XML_ENT_PARSED) == 0) {
11836
0
        if (isExternal)
11837
0
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
11838
11839
0
        ent->children = list;
11840
11841
0
        while (list != NULL) {
11842
0
            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
0
            if (list->doc != ent->doc)
11850
0
                xmlSetTreeDoc(list, ent->doc);
11851
11852
0
            if (list->next == NULL)
11853
0
                ent->last = list;
11854
0
            list = list->next;
11855
0
        }
11856
0
    } else {
11857
0
        xmlFreeNodeList(list);
11858
0
    }
11859
11860
0
    xmlFreeInputStream(input);
11861
11862
0
error:
11863
0
    ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED;
11864
0
}
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 = cur->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
5.29k
{
12999
5.29k
    xmlParserInputPtr input;
13000
13001
5.29k
    if (ctxt == NULL)
13002
0
        return;
13003
13004
5.29k
    while ((input = xmlCtxtPopInput(ctxt)) != NULL) { /* Non consuming */
13005
0
        xmlFreeInputStream(input);
13006
0
    }
13007
5.29k
    ctxt->inputNr = 0;
13008
5.29k
    ctxt->input = NULL;
13009
13010
5.29k
    ctxt->spaceNr = 0;
13011
5.29k
    if (ctxt->spaceTab != NULL) {
13012
5.29k
  ctxt->spaceTab[0] = -1;
13013
5.29k
  ctxt->space = &ctxt->spaceTab[0];
13014
5.29k
    } else {
13015
0
        ctxt->space = NULL;
13016
0
    }
13017
13018
13019
5.29k
    ctxt->nodeNr = 0;
13020
5.29k
    ctxt->node = NULL;
13021
13022
5.29k
    ctxt->nameNr = 0;
13023
5.29k
    ctxt->name = NULL;
13024
13025
5.29k
    ctxt->nsNr = 0;
13026
5.29k
    xmlParserNsReset(ctxt->nsdb);
13027
13028
5.29k
    if (ctxt->version != NULL) {
13029
0
        xmlFree(ctxt->version);
13030
0
        ctxt->version = NULL;
13031
0
    }
13032
5.29k
    if (ctxt->encoding != NULL) {
13033
0
        xmlFree(ctxt->encoding);
13034
0
        ctxt->encoding = NULL;
13035
0
    }
13036
5.29k
    if (ctxt->extSubURI != NULL) {
13037
0
        xmlFree(ctxt->extSubURI);
13038
0
        ctxt->extSubURI = NULL;
13039
0
    }
13040
5.29k
    if (ctxt->extSubSystem != NULL) {
13041
0
        xmlFree(ctxt->extSubSystem);
13042
0
        ctxt->extSubSystem = NULL;
13043
0
    }
13044
5.29k
    if (ctxt->directory != NULL) {
13045
0
        xmlFree(ctxt->directory);
13046
0
        ctxt->directory = NULL;
13047
0
    }
13048
13049
5.29k
    if (ctxt->myDoc != NULL)
13050
0
        xmlFreeDoc(ctxt->myDoc);
13051
5.29k
    ctxt->myDoc = NULL;
13052
13053
5.29k
    ctxt->standalone = -1;
13054
5.29k
    ctxt->hasExternalSubset = 0;
13055
5.29k
    ctxt->hasPErefs = 0;
13056
5.29k
    ctxt->html = ctxt->html ? 1 : 0;
13057
5.29k
    ctxt->instate = XML_PARSER_START;
13058
13059
5.29k
    ctxt->wellFormed = 1;
13060
5.29k
    ctxt->nsWellFormed = 1;
13061
5.29k
    ctxt->disableSAX = 0;
13062
5.29k
    ctxt->valid = 1;
13063
5.29k
    ctxt->record_info = 0;
13064
5.29k
    ctxt->checkIndex = 0;
13065
5.29k
    ctxt->endCheckState = 0;
13066
5.29k
    ctxt->inSubset = 0;
13067
5.29k
    ctxt->errNo = XML_ERR_OK;
13068
5.29k
    ctxt->depth = 0;
13069
5.29k
    ctxt->catalogs = NULL;
13070
5.29k
    ctxt->sizeentities = 0;
13071
5.29k
    ctxt->sizeentcopy = 0;
13072
5.29k
    xmlInitNodeInfoSeq(&ctxt->node_seq);
13073
13074
5.29k
    if (ctxt->attsDefault != NULL) {
13075
0
        xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator);
13076
0
        ctxt->attsDefault = NULL;
13077
0
    }
13078
5.29k
    if (ctxt->attsSpecial != NULL) {
13079
0
        xmlHashFree(ctxt->attsSpecial, NULL);
13080
0
        ctxt->attsSpecial = NULL;
13081
0
    }
13082
13083
5.29k
#ifdef LIBXML_CATALOG_ENABLED
13084
5.29k
    if (ctxt->catalogs != NULL)
13085
0
  xmlCatalogFreeLocal(ctxt->catalogs);
13086
5.29k
#endif
13087
5.29k
    ctxt->nbErrors = 0;
13088
5.29k
    ctxt->nbWarnings = 0;
13089
5.29k
    if (ctxt->lastError.code != XML_ERR_OK)
13090
0
        xmlResetError(&ctxt->lastError);
13091
5.29k
}
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
218k
{
13132
218k
    int allMask;
13133
13134
218k
    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
218k
    allMask = XML_PARSE_RECOVER |
13145
218k
              XML_PARSE_NOENT |
13146
218k
              XML_PARSE_DTDLOAD |
13147
218k
              XML_PARSE_DTDATTR |
13148
218k
              XML_PARSE_DTDVALID |
13149
218k
              XML_PARSE_NOERROR |
13150
218k
              XML_PARSE_NOWARNING |
13151
218k
              XML_PARSE_PEDANTIC |
13152
218k
              XML_PARSE_NOBLANKS |
13153
218k
#ifdef LIBXML_SAX1_ENABLED
13154
218k
              XML_PARSE_SAX1 |
13155
218k
#endif
13156
218k
              XML_PARSE_NONET |
13157
218k
              XML_PARSE_NODICT |
13158
218k
              XML_PARSE_NSCLEAN |
13159
218k
              XML_PARSE_NOCDATA |
13160
218k
              XML_PARSE_COMPACT |
13161
218k
              XML_PARSE_OLD10 |
13162
218k
              XML_PARSE_HUGE |
13163
218k
              XML_PARSE_OLDSAX |
13164
218k
              XML_PARSE_IGNORE_ENC |
13165
218k
              XML_PARSE_BIG_LINES |
13166
218k
              XML_PARSE_NO_XXE |
13167
218k
              XML_PARSE_UNZIP |
13168
218k
              XML_PARSE_NO_SYS_CATALOG |
13169
218k
              XML_PARSE_CATALOG_PI;
13170
13171
218k
    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
218k
    ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0;
13188
218k
    ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0;
13189
218k
    ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0;
13190
218k
    ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0;
13191
218k
    ctxt->loadsubset |= (options & XML_PARSE_SKIP_IDS) ? XML_SKIP_IDS : 0;
13192
218k
    ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0;
13193
218k
    ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0;
13194
218k
    ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1;
13195
218k
    ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1;
13196
13197
218k
    return(options & ~allMask);
13198
218k
}
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
218k
{
13267
218k
    int keepMask;
13268
13269
218k
#ifdef LIBXML_HTML_ENABLED
13270
218k
    if ((ctxt != NULL) && (ctxt->html))
13271
0
        return(htmlCtxtUseOptions(ctxt, options));
13272
218k
#endif
13273
13274
    /*
13275
     * For historic reasons, some options can only be enabled.
13276
     */
13277
218k
    keepMask = XML_PARSE_NOERROR |
13278
218k
               XML_PARSE_NOWARNING |
13279
218k
               XML_PARSE_NONET |
13280
218k
               XML_PARSE_NSCLEAN |
13281
218k
               XML_PARSE_NOCDATA |
13282
218k
               XML_PARSE_COMPACT |
13283
218k
               XML_PARSE_OLD10 |
13284
218k
               XML_PARSE_HUGE |
13285
218k
               XML_PARSE_OLDSAX |
13286
218k
               XML_PARSE_IGNORE_ENC |
13287
218k
               XML_PARSE_BIG_LINES;
13288
13289
218k
    return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask));
13290
218k
}
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
5.29k
{
13325
5.29k
    xmlDocPtr ret = NULL;
13326
13327
5.29k
    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
5.29k
    while (ctxt->inputNr > 0)
13335
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13336
13337
5.29k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13338
0
        xmlFreeInputStream(input);
13339
0
        return(NULL);
13340
0
    }
13341
13342
5.29k
    xmlParseDocument(ctxt);
13343
13344
5.29k
    ret = xmlCtxtGetDocument(ctxt);
13345
13346
    /* assert(ctxt->inputNr == 1); */
13347
10.5k
    while (ctxt->inputNr > 0)
13348
5.29k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13349
13350
5.29k
    return(ret);
13351
5.29k
}
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
0
{
13629
0
    xmlParserInputPtr input;
13630
13631
0
    if ((ctxt == NULL) || (size < 0))
13632
0
        return(NULL);
13633
13634
0
    xmlCtxtReset(ctxt);
13635
0
    xmlCtxtUseOptions(ctxt, options);
13636
13637
0
    input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding,
13638
0
                                      XML_INPUT_BUF_STATIC);
13639
0
    if (input == NULL)
13640
0
        return(NULL);
13641
13642
0
    return(xmlCtxtParseDocument(ctxt, input));
13643
0
}
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
5.29k
{
13702
5.29k
    xmlParserInputPtr input;
13703
13704
5.29k
    if (ctxt == NULL)
13705
0
        return(NULL);
13706
13707
5.29k
    xmlCtxtReset(ctxt);
13708
5.29k
    xmlCtxtUseOptions(ctxt, options);
13709
13710
5.29k
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
13711
5.29k
                                  encoding, 0);
13712
5.29k
    if (input == NULL)
13713
0
        return(NULL);
13714
13715
5.29k
    return(xmlCtxtParseDocument(ctxt, input));
13716
5.29k
}
13717