Coverage Report

Created: 2026-04-14 06:37

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/libxml2/parser.c
Line
Count
Source
1
/*
2
 * parser.c : an XML 1.0 parser, namespaces and validity support are mostly
3
 *            implemented on top of the SAX interfaces
4
 *
5
 * References:
6
 *   The XML specification:
7
 *     http://www.w3.org/TR/REC-xml
8
 *   Original 1.0 version:
9
 *     http://www.w3.org/TR/1998/REC-xml-19980210
10
 *   XML second edition working draft
11
 *     http://www.w3.org/TR/2000/WD-xml-2e-20000814
12
 *
13
 * Okay this is a big file, the parser core is around 7000 lines, then it
14
 * is followed by the progressive parser top routines, then the various
15
 * high level APIs to call the parser and a few miscellaneous functions.
16
 * A number of helper functions and deprecated ones have been moved to
17
 * parserInternals.c to reduce this file size.
18
 * As much as possible the functions are associated with their relative
19
 * production in the XML specification. A few productions defining the
20
 * different ranges of character are actually implanted either in
21
 * parserInternals.h or parserInternals.c
22
 * The DOM tree build is realized from the default SAX callbacks in
23
 * the module SAX2.c.
24
 * The routines doing the validation checks are in valid.c and called either
25
 * from the SAX callbacks or as standalone functions using a preparsed
26
 * document.
27
 *
28
 * See Copyright for the status of this software.
29
 *
30
 * Author: Daniel Veillard
31
 */
32
33
/* To avoid EBCDIC trouble when parsing on zOS */
34
#if defined(__MVS__)
35
#pragma convert("ISO8859-1")
36
#endif
37
38
#define IN_LIBXML
39
#include "libxml.h"
40
41
#if defined(_WIN32)
42
#define XML_DIR_SEP '\\'
43
#else
44
#define XML_DIR_SEP '/'
45
#endif
46
47
#include <stdlib.h>
48
#include <limits.h>
49
#include <string.h>
50
#include <stdarg.h>
51
#include <stddef.h>
52
#include <ctype.h>
53
#include <stdlib.h>
54
#include <libxml/parser.h>
55
#include <libxml/xmlmemory.h>
56
#include <libxml/tree.h>
57
#include <libxml/parserInternals.h>
58
#include <libxml/valid.h>
59
#include <libxml/entities.h>
60
#include <libxml/xmlerror.h>
61
#include <libxml/encoding.h>
62
#include <libxml/xmlIO.h>
63
#include <libxml/uri.h>
64
#include <libxml/SAX2.h>
65
#include <libxml/HTMLparser.h>
66
#ifdef LIBXML_CATALOG_ENABLED
67
#include <libxml/catalog.h>
68
#endif
69
70
#include "private/buf.h"
71
#include "private/dict.h"
72
#include "private/entities.h"
73
#include "private/error.h"
74
#include "private/html.h"
75
#include "private/io.h"
76
#include "private/memory.h"
77
#include "private/parser.h"
78
#include "private/tree.h"
79
80
3.91M
#define NS_INDEX_EMPTY  INT_MAX
81
511k
#define NS_INDEX_XML    (INT_MAX - 1)
82
1.48M
#define URI_HASH_EMPTY  0xD943A04E
83
182k
#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
1.17M
#define XML_MAX_ATTRS 100000000 /* 100 million */
94
95
3.07M
#define XML_SPECIAL_EXTERNAL    (1 << 20)
96
2.90M
#define XML_SPECIAL_TYPE_MASK   (XML_SPECIAL_EXTERNAL - 1)
97
98
3.07M
#define XML_ATTVAL_ALLOC        (1 << 0)
99
2.51M
#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
7.91M
#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
7.95M
#define XML_ENT_FIXED_COST 20
170
171
305M
#define XML_PARSER_BIG_BUFFER_SIZE 300
172
3.01M
#define XML_PARSER_BUFFER_SIZE 100
173
532k
#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
7.59k
xmlErrMemory(xmlParserCtxtPtr ctxt) {
225
7.59k
    xmlCtxtErrMemory(ctxt);
226
7.59k
}
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
243k
{
239
243k
    if (prefix == NULL)
240
152k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
241
152k
                   XML_ERR_FATAL, localname, NULL, NULL, 0,
242
152k
                   "Attribute %s redefined\n", localname);
243
90.6k
    else
244
90.6k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
245
90.6k
                   XML_ERR_FATAL, prefix, localname, NULL, 0,
246
90.6k
                   "Attribute %s:%s redefined\n", prefix, localname);
247
243k
}
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
214M
{
260
214M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
261
214M
               NULL, NULL, NULL, 0, "%s", msg);
262
214M
}
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
271k
{
277
271k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING,
278
271k
               str1, str2, NULL, 0, msg, str1, str2);
279
271k
}
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
337k
{
295
337k
    ctxt->valid = 0;
296
297
337k
    xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR,
298
337k
               str1, str2, NULL, 0, msg, str1, str2);
299
337k
}
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
77.5M
{
314
77.5M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
315
77.5M
               NULL, NULL, NULL, val, msg, val);
316
77.5M
}
317
318
/**
319
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
320
 *
321
 * @param ctxt  an XML parser context
322
 * @param error  the error number
323
 * @param msg  the error message
324
 * @param str1  an string info
325
 * @param val  an integer value
326
 * @param str2  an string info
327
 */
328
static void LIBXML_ATTR_FORMAT(3,0)
329
xmlFatalErrMsgStrIntStr(xmlParserCtxtPtr ctxt, xmlParserErrors error,
330
                  const char *msg, const xmlChar *str1, int val,
331
      const xmlChar *str2)
332
1.41M
{
333
1.41M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
334
1.41M
               str1, str2, NULL, val, msg, str1, val, str2);
335
1.41M
}
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
5.04M
{
349
5.04M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
350
5.04M
               val, NULL, NULL, 0, msg, val);
351
5.04M
}
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
196k
{
365
196k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR,
366
196k
               val, NULL, NULL, 0, msg, val);
367
196k
}
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
1.37M
{
385
1.37M
    ctxt->nsWellFormed = 0;
386
387
1.37M
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR,
388
1.37M
               info1, info2, info3, 0, msg, info1, info2, info3);
389
1.37M
}
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
85.5k
{
407
85.5k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING,
408
85.5k
               info1, info2, info3, 0, msg, info1, info2, info3);
409
85.5k
}
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
8.37M
{
436
8.37M
    unsigned long consumed;
437
8.37M
    unsigned long *expandedSize;
438
8.37M
    xmlParserInputPtr input = ctxt->input;
439
8.37M
    xmlEntityPtr entity = input->entity;
440
441
8.37M
    if ((entity) && (entity->flags & XML_ENT_CHECKED))
442
455k
        return(0);
443
444
    /*
445
     * Compute total consumed bytes so far, including input streams of
446
     * external entities.
447
     */
448
7.91M
    consumed = input->consumed;
449
7.91M
    xmlSaturatedAddSizeT(&consumed, input->cur - input->base);
450
7.91M
    xmlSaturatedAdd(&consumed, ctxt->sizeentities);
451
452
7.91M
    if (entity)
453
335k
        expandedSize = &entity->expandedSize;
454
7.58M
    else
455
7.58M
        expandedSize = &ctxt->sizeentcopy;
456
457
    /*
458
     * Add extra cost and some fixed cost.
459
     */
460
7.91M
    xmlSaturatedAdd(expandedSize, extra);
461
7.91M
    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
7.91M
    if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) &&
469
1.28M
        ((*expandedSize >= ULONG_MAX) ||
470
1.28M
         (*expandedSize / ctxt->maxAmpl > consumed))) {
471
3.23k
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
472
3.23k
                       "Maximum entity amplification factor exceeded, see "
473
3.23k
                       "xmlCtxtSetMaxAmplification.\n");
474
3.23k
        return(1);
475
3.23k
    }
476
477
7.91M
    return(0);
478
7.91M
}
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
61.4k
{
496
61.4k
    switch (feature) {
497
2.56k
  case XML_WITH_THREAD:
498
2.56k
#ifdef LIBXML_THREAD_ENABLED
499
2.56k
      return(1);
500
#else
501
      return(0);
502
#endif
503
0
        case XML_WITH_TREE:
504
0
            return(1);
505
2.56k
        case XML_WITH_OUTPUT:
506
2.56k
#ifdef LIBXML_OUTPUT_ENABLED
507
2.56k
            return(1);
508
#else
509
            return(0);
510
#endif
511
2.56k
        case XML_WITH_PUSH:
512
2.56k
#ifdef LIBXML_PUSH_ENABLED
513
2.56k
            return(1);
514
#else
515
            return(0);
516
#endif
517
2.56k
        case XML_WITH_READER:
518
2.56k
#ifdef LIBXML_READER_ENABLED
519
2.56k
            return(1);
520
#else
521
            return(0);
522
#endif
523
2.56k
        case XML_WITH_PATTERN:
524
2.56k
#ifdef LIBXML_PATTERN_ENABLED
525
2.56k
            return(1);
526
#else
527
            return(0);
528
#endif
529
2.56k
        case XML_WITH_WRITER:
530
2.56k
#ifdef LIBXML_WRITER_ENABLED
531
2.56k
            return(1);
532
#else
533
            return(0);
534
#endif
535
2.56k
        case XML_WITH_SAX1:
536
2.56k
#ifdef LIBXML_SAX1_ENABLED
537
2.56k
            return(1);
538
#else
539
            return(0);
540
#endif
541
0
        case XML_WITH_HTTP:
542
0
            return(0);
543
2.56k
        case XML_WITH_VALID:
544
2.56k
#ifdef LIBXML_VALID_ENABLED
545
2.56k
            return(1);
546
#else
547
            return(0);
548
#endif
549
2.56k
        case XML_WITH_HTML:
550
2.56k
#ifdef LIBXML_HTML_ENABLED
551
2.56k
            return(1);
552
#else
553
            return(0);
554
#endif
555
0
        case XML_WITH_LEGACY:
556
0
            return(0);
557
2.56k
        case XML_WITH_C14N:
558
2.56k
#ifdef LIBXML_C14N_ENABLED
559
2.56k
            return(1);
560
#else
561
            return(0);
562
#endif
563
2.56k
        case XML_WITH_CATALOG:
564
2.56k
#ifdef LIBXML_CATALOG_ENABLED
565
2.56k
            return(1);
566
#else
567
            return(0);
568
#endif
569
2.56k
        case XML_WITH_XPATH:
570
2.56k
#ifdef LIBXML_XPATH_ENABLED
571
2.56k
            return(1);
572
#else
573
            return(0);
574
#endif
575
2.56k
        case XML_WITH_XPTR:
576
2.56k
#ifdef LIBXML_XPTR_ENABLED
577
2.56k
            return(1);
578
#else
579
            return(0);
580
#endif
581
2.56k
        case XML_WITH_XINCLUDE:
582
2.56k
#ifdef LIBXML_XINCLUDE_ENABLED
583
2.56k
            return(1);
584
#else
585
            return(0);
586
#endif
587
2.56k
        case XML_WITH_ICONV:
588
2.56k
#ifdef LIBXML_ICONV_ENABLED
589
2.56k
            return(1);
590
#else
591
            return(0);
592
#endif
593
2.56k
        case XML_WITH_ISO8859X:
594
2.56k
#ifdef LIBXML_ISO8859X_ENABLED
595
2.56k
            return(1);
596
#else
597
            return(0);
598
#endif
599
0
        case XML_WITH_UNICODE:
600
0
            return(0);
601
2.56k
        case XML_WITH_REGEXP:
602
2.56k
#ifdef LIBXML_REGEXP_ENABLED
603
2.56k
            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
2.56k
        case XML_WITH_RELAXNG:
616
2.56k
#ifdef LIBXML_RELAXNG_ENABLED
617
2.56k
            return(1);
618
#else
619
            return(0);
620
#endif
621
2.56k
        case XML_WITH_SCHEMAS:
622
2.56k
#ifdef LIBXML_SCHEMAS_ENABLED
623
2.56k
            return(1);
624
#else
625
            return(0);
626
#endif
627
2.56k
        case XML_WITH_SCHEMATRON:
628
#ifdef LIBXML_SCHEMATRON_ENABLED
629
            return(1);
630
#else
631
2.56k
            return(0);
632
0
#endif
633
2.56k
        case XML_WITH_MODULES:
634
2.56k
#ifdef LIBXML_MODULES_ENABLED
635
2.56k
            return(1);
636
#else
637
            return(0);
638
#endif
639
2.56k
        case XML_WITH_DEBUG:
640
#ifdef LIBXML_DEBUG_ENABLED
641
            return(1);
642
#else
643
2.56k
            return(0);
644
0
#endif
645
0
        case XML_WITH_DEBUG_MEM:
646
0
            return(0);
647
2.56k
        case XML_WITH_ZLIB:
648
2.56k
#ifdef LIBXML_ZLIB_ENABLED
649
2.56k
            return(1);
650
#else
651
            return(0);
652
#endif
653
0
        case XML_WITH_LZMA:
654
0
            return(0);
655
2.56k
        case XML_WITH_ICU:
656
#ifdef LIBXML_ICU_ENABLED
657
            return(1);
658
#else
659
2.56k
            return(0);
660
0
#endif
661
0
        default:
662
0
      break;
663
61.4k
     }
664
0
     return(0);
665
61.4k
}
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
3.27M
xmlSBufInit(xmlSBuf *buf, unsigned max) {
683
3.27M
    buf->mem = NULL;
684
3.27M
    buf->size = 0;
685
3.27M
    buf->cap = 0;
686
3.27M
    buf->max = max;
687
3.27M
    buf->code = XML_ERR_OK;
688
3.27M
}
689
690
static int
691
1.33M
xmlSBufGrow(xmlSBuf *buf, unsigned len) {
692
1.33M
    xmlChar *mem;
693
1.33M
    unsigned cap;
694
695
1.33M
    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
1.33M
    cap = (buf->size + len) * 2;
702
1.33M
    if (cap < 240)
703
1.04M
        cap = 240;
704
705
1.33M
    mem = xmlRealloc(buf->mem, cap);
706
1.33M
    if (mem == NULL) {
707
1.18k
        buf->code = XML_ERR_NO_MEMORY;
708
1.18k
        return(-1);
709
1.18k
    }
710
711
1.33M
    buf->mem = mem;
712
1.33M
    buf->cap = cap;
713
714
1.33M
    return(0);
715
1.33M
}
716
717
static void
718
822M
xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) {
719
822M
    if (buf->max - buf->size < len) {
720
3.68M
        if (buf->code == XML_ERR_OK)
721
782
            buf->code = XML_ERR_RESOURCE_LIMIT;
722
3.68M
        return;
723
3.68M
    }
724
725
819M
    if (buf->cap - buf->size <= len) {
726
1.29M
        if (xmlSBufGrow(buf, len) < 0)
727
1.08k
            return;
728
1.29M
    }
729
730
819M
    if (len > 0)
731
819M
        memcpy(buf->mem + buf->size, str, len);
732
819M
    buf->size += len;
733
819M
}
734
735
static void
736
804M
xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) {
737
804M
    xmlSBufAddString(buf, (const xmlChar *) str, len);
738
804M
}
739
740
static void
741
2.32M
xmlSBufAddChar(xmlSBuf *buf, int c) {
742
2.32M
    xmlChar *end;
743
744
2.32M
    if (buf->max - buf->size < 4) {
745
2.14k
        if (buf->code == XML_ERR_OK)
746
19
            buf->code = XML_ERR_RESOURCE_LIMIT;
747
2.14k
        return;
748
2.14k
    }
749
750
2.32M
    if (buf->cap - buf->size <= 4) {
751
39.1k
        if (xmlSBufGrow(buf, 4) < 0)
752
98
            return;
753
39.1k
    }
754
755
2.32M
    end = buf->mem + buf->size;
756
757
2.32M
    if (c < 0x80) {
758
2.13M
        *end = (xmlChar) c;
759
2.13M
        buf->size += 1;
760
2.13M
    } else {
761
193k
        buf->size += xmlCopyCharMultiByte(end, c);
762
193k
    }
763
2.32M
}
764
765
static void
766
674M
xmlSBufAddReplChar(xmlSBuf *buf) {
767
674M
    xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3);
768
674M
}
769
770
static void
771
2.08k
xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
772
2.08k
    if (buf->code == XML_ERR_NO_MEMORY)
773
1.28k
        xmlCtxtErrMemory(ctxt);
774
801
    else
775
801
        xmlFatalErr(ctxt, buf->code, errMsg);
776
2.08k
}
777
778
static xmlChar *
779
xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt,
780
1.11M
              const char *errMsg) {
781
1.11M
    if (buf->mem == NULL) {
782
106k
        buf->mem = xmlMalloc(1);
783
106k
        if (buf->mem == NULL) {
784
102
            buf->code = XML_ERR_NO_MEMORY;
785
105k
        } else {
786
105k
            buf->mem[0] = 0;
787
105k
        }
788
1.00M
    } else {
789
1.00M
        buf->mem[buf->size] = 0;
790
1.00M
    }
791
792
1.11M
    if (buf->code == XML_ERR_OK) {
793
1.11M
        if (sizeOut != NULL)
794
293k
            *sizeOut = buf->size;
795
1.11M
        return(buf->mem);
796
1.11M
    }
797
798
1.13k
    xmlSBufReportError(buf, ctxt, errMsg);
799
800
1.13k
    xmlFree(buf->mem);
801
802
1.13k
    if (sizeOut != NULL)
803
305
        *sizeOut = 0;
804
1.13k
    return(NULL);
805
1.11M
}
806
807
static void
808
2.11M
xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
809
2.11M
    if (buf->code != XML_ERR_OK)
810
948
        xmlSBufReportError(buf, ctxt, errMsg);
811
812
2.11M
    xmlFree(buf->mem);
813
2.11M
}
814
815
static int
816
xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str,
817
2.00G
                    const char *errMsg) {
818
2.00G
    int c = str[0];
819
2.00G
    int c1 = str[1];
820
821
2.00G
    if ((c1 & 0xC0) != 0x80)
822
340M
        goto encoding_error;
823
824
1.66G
    if (c < 0xE0) {
825
        /* 2-byte sequence */
826
182M
        if (c < 0xC2)
827
122M
            goto encoding_error;
828
829
59.6M
        return(2);
830
1.48G
    } else {
831
1.48G
        int c2 = str[2];
832
833
1.48G
        if ((c2 & 0xC0) != 0x80)
834
68.5k
            goto encoding_error;
835
836
1.48G
        if (c < 0xF0) {
837
            /* 3-byte sequence */
838
1.48G
            if (c == 0xE0) {
839
                /* overlong */
840
7.71M
                if (c1 < 0xA0)
841
7.79k
                    goto encoding_error;
842
1.47G
            } else if (c == 0xED) {
843
                /* surrogate */
844
418k
                if (c1 >= 0xA0)
845
9.06k
                    goto encoding_error;
846
1.47G
            } else if (c == 0xEF) {
847
                /* U+FFFE and U+FFFF are invalid Chars */
848
455M
                if ((c1 == 0xBF) && (c2 >= 0xBE))
849
4.11k
                    xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg);
850
455M
            }
851
852
1.48G
            return(3);
853
1.48G
        } else {
854
            /* 4-byte sequence */
855
398k
            if ((str[3] & 0xC0) != 0x80)
856
29.6k
                goto encoding_error;
857
368k
            if (c == 0xF0) {
858
                /* overlong */
859
7.26k
                if (c1 < 0x90)
860
2.31k
                    goto encoding_error;
861
361k
            } else if (c >= 0xF4) {
862
                /* greater than 0x10FFFF */
863
72.0k
                if ((c > 0xF4) || (c1 >= 0x90))
864
70.4k
                    goto encoding_error;
865
72.0k
            }
866
867
296k
            return(4);
868
368k
        }
869
1.48G
    }
870
871
463M
encoding_error:
872
    /* Only report the first error */
873
463M
    if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) {
874
78.6k
        xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL);
875
78.6k
        ctxt->input->flags |= XML_INPUT_ENCODING_ERROR;
876
78.6k
    }
877
878
463M
    return(0);
879
1.66G
}
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
430k
xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) {
897
430k
    xmlSAXHandlerPtr sax;
898
899
    /* Avoid unused variable warning if features are disabled. */
900
430k
    (void) sax;
901
902
    /*
903
     * Changing the SAX struct directly is still widespread practice
904
     * in internal and external code.
905
     */
906
430k
    if (ctxt == NULL) return;
907
430k
    sax = ctxt->sax;
908
430k
#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
430k
    if (((ctxt->options & XML_PARSE_SAX1) == 0) &&
914
336k
        (sax) &&
915
336k
        (sax->initialized == XML_SAX2_MAGIC) &&
916
336k
        ((sax->startElementNs != NULL) ||
917
0
         (sax->endElementNs != NULL) ||
918
0
         ((sax->startElement == NULL) && (sax->endElement == NULL))))
919
336k
        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
430k
    ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3);
929
430k
    ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5);
930
430k
    ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36);
931
430k
    if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) ||
932
429k
    (ctxt->str_xml_ns == NULL)) {
933
1.38k
        xmlErrMemory(ctxt);
934
1.38k
    }
935
936
430k
    xmlDictSetLimit(ctxt->dict,
937
430k
                    (ctxt->options & XML_PARSE_HUGE) ?
938
122k
                        0 :
939
430k
                        XML_MAX_DICTIONARY_LIMIT);
940
941
430k
#ifdef LIBXML_VALID_ENABLED
942
430k
    if (ctxt->validate)
943
124k
        ctxt->vctxt.flags |= XML_VCTXT_VALIDATE;
944
306k
    else
945
306k
        ctxt->vctxt.flags &= ~XML_VCTXT_VALIDATE;
946
430k
#endif /* LIBXML_VALID_ENABLED */
947
430k
}
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
212k
{
989
212k
    if ((src == NULL) || (dst == NULL))
990
0
        return(NULL);
991
992
226k
    while (*src == 0x20) src++;
993
104M
    while (*src != 0) {
994
104M
  if (*src == 0x20) {
995
12.7M
      while (*src == 0x20) src++;
996
78.7k
      if (*src != 0)
997
74.7k
    *dst++ = 0x20;
998
104M
  } else {
999
104M
      *dst++ = *src++;
1000
104M
  }
1001
104M
    }
1002
212k
    *dst = 0;
1003
212k
    if (dst == src)
1004
202k
       return(NULL);
1005
10.5k
    return(dst);
1006
212k
}
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
204k
               const xmlChar *value) {
1021
204k
    xmlDefAttrsPtr defaults;
1022
204k
    xmlDefAttr *attr;
1023
204k
    int len, expandedSize;
1024
204k
    xmlHashedString name;
1025
204k
    xmlHashedString prefix;
1026
204k
    xmlHashedString hvalue;
1027
204k
    const xmlChar *localname;
1028
1029
    /*
1030
     * Allows to detect attribute redefinitions
1031
     */
1032
204k
    if (ctxt->attsSpecial != NULL) {
1033
177k
        if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL)
1034
116k
      return;
1035
177k
    }
1036
1037
88.2k
    if (ctxt->attsDefault == NULL) {
1038
28.0k
        ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict);
1039
28.0k
  if (ctxt->attsDefault == NULL)
1040
61
      goto mem_error;
1041
28.0k
    }
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
88.1k
    localname = xmlSplitQName3(fullname, &len);
1048
88.1k
    if (localname == NULL) {
1049
81.9k
        name = xmlDictLookupHashed(ctxt->dict, fullname, -1);
1050
81.9k
  prefix.name = NULL;
1051
81.9k
    } else {
1052
6.21k
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1053
6.21k
  prefix = xmlDictLookupHashed(ctxt->dict, fullname, len);
1054
6.21k
        if (prefix.name == NULL)
1055
9
            goto mem_error;
1056
6.21k
    }
1057
88.1k
    if (name.name == NULL)
1058
12
        goto mem_error;
1059
1060
    /*
1061
     * make sure there is some storage
1062
     */
1063
88.1k
    defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name);
1064
88.1k
    if ((defaults == NULL) ||
1065
56.1k
        (defaults->nbAttrs >= defaults->maxAttrs)) {
1066
36.4k
        xmlDefAttrsPtr temp;
1067
36.4k
        int newSize;
1068
1069
36.4k
        if (defaults == NULL) {
1070
31.9k
            newSize = 4;
1071
31.9k
        } else {
1072
4.49k
            if ((defaults->maxAttrs >= XML_MAX_ATTRS) ||
1073
4.49k
                ((size_t) defaults->maxAttrs >
1074
4.49k
                     SIZE_MAX / 2 / sizeof(temp[0]) - sizeof(*defaults)))
1075
0
                goto mem_error;
1076
1077
4.49k
            if (defaults->maxAttrs > XML_MAX_ATTRS / 2)
1078
0
                newSize = XML_MAX_ATTRS;
1079
4.49k
            else
1080
4.49k
                newSize = defaults->maxAttrs * 2;
1081
4.49k
        }
1082
36.4k
        temp = xmlRealloc(defaults,
1083
36.4k
                          sizeof(*defaults) + newSize * sizeof(xmlDefAttr));
1084
36.4k
  if (temp == NULL)
1085
44
      goto mem_error;
1086
36.4k
        if (defaults == NULL)
1087
31.9k
            temp->nbAttrs = 0;
1088
36.4k
  temp->maxAttrs = newSize;
1089
36.4k
        defaults = temp;
1090
36.4k
  if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name,
1091
36.4k
                          defaults, NULL) < 0) {
1092
2
      xmlFree(defaults);
1093
2
      goto mem_error;
1094
2
  }
1095
36.4k
    }
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
88.0k
    localname = xmlSplitQName3(fullattr, &len);
1102
88.0k
    if (localname == NULL) {
1103
59.3k
        name = xmlDictLookupHashed(ctxt->dict, fullattr, -1);
1104
59.3k
  prefix.name = NULL;
1105
59.3k
    } else {
1106
28.7k
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1107
28.7k
  prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len);
1108
28.7k
        if (prefix.name == NULL)
1109
12
            goto mem_error;
1110
28.7k
    }
1111
88.0k
    if (name.name == NULL)
1112
13
        goto mem_error;
1113
1114
    /* intern the string and precompute the end */
1115
88.0k
    len = strlen((const char *) value);
1116
88.0k
    hvalue = xmlDictLookupHashed(ctxt->dict, value, len);
1117
88.0k
    if (hvalue.name == NULL)
1118
17
        goto mem_error;
1119
1120
88.0k
    expandedSize = strlen((const char *) name.name);
1121
88.0k
    if (prefix.name != NULL)
1122
28.7k
        expandedSize += strlen((const char *) prefix.name);
1123
88.0k
    expandedSize += len;
1124
1125
88.0k
    attr = &defaults->attrs[defaults->nbAttrs++];
1126
88.0k
    attr->name = name;
1127
88.0k
    attr->prefix = prefix;
1128
88.0k
    attr->value = hvalue;
1129
88.0k
    attr->valueEnd = hvalue.name + len;
1130
88.0k
    attr->external = PARSER_EXTERNAL(ctxt);
1131
88.0k
    attr->expandedSize = expandedSize;
1132
1133
88.0k
    return;
1134
1135
170
mem_error:
1136
170
    xmlErrMemory(ctxt);
1137
170
}
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
301k
{
1153
301k
    if (ctxt->attsSpecial == NULL) {
1154
36.0k
        ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict);
1155
36.0k
  if (ctxt->attsSpecial == NULL)
1156
73
      goto mem_error;
1157
36.0k
    }
1158
1159
301k
    if (PARSER_EXTERNAL(ctxt))
1160
160k
        type |= XML_SPECIAL_EXTERNAL;
1161
1162
301k
    if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr,
1163
301k
                    XML_INT_TO_PTR(type)) < 0)
1164
11
        goto mem_error;
1165
301k
    return;
1166
1167
301k
mem_error:
1168
84
    xmlErrMemory(ctxt);
1169
84
}
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
162k
                            const xmlChar *unused ATTRIBUTE_UNUSED) {
1178
162k
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data;
1179
1180
162k
    if (XML_PTR_TO_INT(payload) == XML_ATTRIBUTE_CDATA) {
1181
20.2k
        xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL);
1182
20.2k
    }
1183
162k
}
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
212k
{
1195
212k
    if (ctxt->attsSpecial == NULL)
1196
176k
        return;
1197
1198
35.9k
    xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt);
1199
1200
35.9k
    if (xmlHashSize(ctxt->attsSpecial) == 0) {
1201
6.42k
        xmlHashFree(ctxt->attsSpecial, NULL);
1202
6.42k
        ctxt->attsSpecial = NULL;
1203
6.42k
    }
1204
35.9k
}
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
106k
{
1268
106k
    const xmlChar *cur = lang, *nxt;
1269
1270
106k
    if (cur == NULL)
1271
1.01k
        return (0);
1272
105k
    if (((cur[0] == 'i') && (cur[1] == '-')) ||
1273
104k
        ((cur[0] == 'I') && (cur[1] == '-')) ||
1274
100k
        ((cur[0] == 'x') && (cur[1] == '-')) ||
1275
100k
        ((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
5.78k
        cur += 2;
1282
14.5k
        while (((cur[0] >= 'A') && (cur[0] <= 'Z')) ||
1283
9.51k
               ((cur[0] >= 'a') && (cur[0] <= 'z')))
1284
8.72k
            cur++;
1285
5.78k
        return(cur[0] == 0);
1286
5.78k
    }
1287
99.3k
    nxt = cur;
1288
567k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1289
270k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1290
467k
           nxt++;
1291
99.3k
    if (nxt - cur >= 4) {
1292
        /*
1293
         * Reserved
1294
         */
1295
6.35k
        if ((nxt - cur > 8) || (nxt[0] != 0))
1296
4.40k
            return(0);
1297
1.94k
        return(1);
1298
6.35k
    }
1299
93.0k
    if (nxt - cur < 2)
1300
7.04k
        return(0);
1301
    /* we got an ISO 639 code */
1302
85.9k
    if (nxt[0] == 0)
1303
3.16k
        return(1);
1304
82.8k
    if (nxt[0] != '-')
1305
5.65k
        return(0);
1306
1307
77.1k
    nxt++;
1308
77.1k
    cur = nxt;
1309
    /* now we can have extlang or script or region or variant */
1310
77.1k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1311
18.2k
        goto region_m49;
1312
1313
1.32M
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1314
1.19M
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1315
1.26M
           nxt++;
1316
58.9k
    if (nxt - cur == 4)
1317
7.20k
        goto script;
1318
51.7k
    if (nxt - cur == 2)
1319
4.33k
        goto region;
1320
47.4k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1321
1.96k
        goto variant;
1322
45.4k
    if (nxt - cur != 3)
1323
7.42k
        return(0);
1324
    /* we parsed an extlang */
1325
38.0k
    if (nxt[0] == 0)
1326
1.22k
        return(1);
1327
36.7k
    if (nxt[0] != '-')
1328
12.6k
        return(0);
1329
1330
24.1k
    nxt++;
1331
24.1k
    cur = nxt;
1332
    /* now we can have script or region or variant */
1333
24.1k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1334
1.76k
        goto region_m49;
1335
1336
283k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1337
46.7k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1338
260k
           nxt++;
1339
22.3k
    if (nxt - cur == 2)
1340
15.1k
        goto region;
1341
7.16k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1342
1.62k
        goto variant;
1343
5.54k
    if (nxt - cur != 4)
1344
3.10k
        return(0);
1345
    /* we parsed a script */
1346
9.64k
script:
1347
9.64k
    if (nxt[0] == 0)
1348
1.89k
        return(1);
1349
7.75k
    if (nxt[0] != '-')
1350
1.05k
        return(0);
1351
1352
6.69k
    nxt++;
1353
6.69k
    cur = nxt;
1354
    /* now we can have region or variant */
1355
6.69k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1356
1.14k
        goto region_m49;
1357
1358
29.2k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1359
21.1k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1360
23.7k
           nxt++;
1361
1362
5.55k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1363
1.08k
        goto variant;
1364
4.46k
    if (nxt - cur != 2)
1365
2.98k
        return(0);
1366
    /* we parsed a region */
1367
22.1k
region:
1368
22.1k
    if (nxt[0] == 0)
1369
1.62k
        return(1);
1370
20.5k
    if (nxt[0] != '-')
1371
16.7k
        return(0);
1372
1373
3.79k
    nxt++;
1374
3.79k
    cur = nxt;
1375
    /* now we can just have a variant */
1376
44.6k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1377
37.7k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1378
40.8k
           nxt++;
1379
1380
3.79k
    if ((nxt - cur < 5) || (nxt - cur > 8))
1381
2.66k
        return(0);
1382
1383
    /* we parsed a variant */
1384
5.80k
variant:
1385
5.80k
    if (nxt[0] == 0)
1386
1.73k
        return(1);
1387
4.07k
    if (nxt[0] != '-')
1388
2.91k
        return(0);
1389
    /* extensions and private use subtags not checked */
1390
1.15k
    return (1);
1391
1392
21.1k
region_m49:
1393
21.1k
    if (((nxt[1] >= '0') && (nxt[1] <= '9')) &&
1394
14.5k
        ((nxt[2] >= '0') && (nxt[2] <= '9'))) {
1395
1.14k
        nxt += 3;
1396
1.14k
        goto region;
1397
1.14k
    }
1398
19.9k
    return(0);
1399
21.1k
}
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
438k
xmlParserNsCreate(void) {
1417
438k
    xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb));
1418
1419
438k
    if (nsdb == NULL)
1420
52
        return(NULL);
1421
438k
    memset(nsdb, 0, sizeof(*nsdb));
1422
438k
    nsdb->defaultNsIndex = INT_MAX;
1423
1424
438k
    return(nsdb);
1425
438k
}
1426
1427
/**
1428
 * Free a namespace database.
1429
 *
1430
 * @param nsdb  namespace database
1431
 */
1432
void
1433
438k
xmlParserNsFree(xmlParserNsData *nsdb) {
1434
438k
    if (nsdb == NULL)
1435
0
        return;
1436
1437
438k
    xmlFree(nsdb->extra);
1438
438k
    xmlFree(nsdb->hash);
1439
438k
    xmlFree(nsdb);
1440
438k
}
1441
1442
/**
1443
 * Reset a namespace database.
1444
 *
1445
 * @param nsdb  namespace database
1446
 */
1447
static void
1448
161k
xmlParserNsReset(xmlParserNsData *nsdb) {
1449
161k
    if (nsdb == NULL)
1450
12.0k
        return;
1451
1452
149k
    nsdb->hashElems = 0;
1453
149k
    nsdb->elementId = 0;
1454
149k
    nsdb->defaultNsIndex = INT_MAX;
1455
1456
149k
    if (nsdb->hash)
1457
895
        memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0]));
1458
149k
}
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
4.83M
xmlParserNsStartElement(xmlParserNsData *nsdb) {
1468
4.83M
    if (nsdb->elementId == UINT_MAX)
1469
0
        return(-1);
1470
4.83M
    nsdb->elementId++;
1471
1472
4.83M
    return(0);
1473
4.83M
}
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
10.5M
                  xmlParserNsBucket **bucketPtr) {
1489
10.5M
    xmlParserNsBucket *bucket, *tombstone;
1490
10.5M
    unsigned index, hashValue;
1491
1492
10.5M
    if (prefix->name == NULL)
1493
4.22M
        return(ctxt->nsdb->defaultNsIndex);
1494
1495
6.35M
    if (ctxt->nsdb->hashSize == 0)
1496
480k
        return(INT_MAX);
1497
1498
5.87M
    hashValue = prefix->hashValue;
1499
5.87M
    index = hashValue & (ctxt->nsdb->hashSize - 1);
1500
5.87M
    bucket = &ctxt->nsdb->hash[index];
1501
5.87M
    tombstone = NULL;
1502
1503
7.49M
    while (bucket->hashValue) {
1504
6.61M
        if (bucket->index == INT_MAX) {
1505
558k
            if (tombstone == NULL)
1506
489k
                tombstone = bucket;
1507
6.05M
        } else if (bucket->hashValue == hashValue) {
1508
4.99M
            if (ctxt->nsTab[bucket->index * 2] == prefix->name) {
1509
4.99M
                if (bucketPtr != NULL)
1510
2.64M
                    *bucketPtr = bucket;
1511
4.99M
                return(bucket->index);
1512
4.99M
            }
1513
4.99M
        }
1514
1515
1.61M
        index++;
1516
1.61M
        bucket++;
1517
1.61M
        if (index == ctxt->nsdb->hashSize) {
1518
236k
            index = 0;
1519
236k
            bucket = ctxt->nsdb->hash;
1520
236k
        }
1521
1.61M
    }
1522
1523
881k
    if (bucketPtr != NULL)
1524
449k
        *bucketPtr = tombstone ? tombstone : bucket;
1525
881k
    return(INT_MAX);
1526
5.87M
}
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
4.30M
xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) {
1537
4.30M
    const xmlChar *ret;
1538
4.30M
    int nsIndex;
1539
1540
4.30M
    if (prefix->name == ctxt->str_xml)
1541
7.33k
        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
4.29M
    nsIndex = xmlParserNsLookup(ctxt, prefix, NULL);
1548
4.29M
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1549
2.62M
        return(NULL);
1550
1551
1.67M
    ret = ctxt->nsTab[nsIndex * 2 + 1];
1552
1.67M
    if (ret[0] == 0)
1553
36.2k
        ret = NULL;
1554
1.67M
    return(ret);
1555
4.29M
}
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
1.36M
xmlParserNsLookupSax(xmlParserCtxt *ctxt, const xmlChar *prefix) {
1567
1.36M
    xmlHashedString hprefix;
1568
1.36M
    int nsIndex;
1569
1570
1.36M
    if (prefix == ctxt->str_xml)
1571
197k
        return(NULL);
1572
1573
1.17M
    hprefix.name = prefix;
1574
1.17M
    if (prefix != NULL)
1575
595k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1576
575k
    else
1577
575k
        hprefix.hashValue = 0;
1578
1.17M
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1579
1.17M
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1580
0
        return(NULL);
1581
1582
1.17M
    return(ctxt->nsdb->extra[nsIndex].saxData);
1583
1.17M
}
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
1.13M
                     void *saxData) {
1598
1.13M
    xmlHashedString hprefix;
1599
1.13M
    int nsIndex;
1600
1601
1.13M
    if (prefix == ctxt->str_xml)
1602
0
        return(-1);
1603
1604
1.13M
    hprefix.name = prefix;
1605
1.13M
    if (prefix != NULL)
1606
981k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1607
151k
    else
1608
151k
        hprefix.hashValue = 0;
1609
1.13M
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1610
1.13M
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1611
0
        return(-1);
1612
1613
1.13M
    ctxt->nsdb->extra[nsIndex].saxData = saxData;
1614
1.13M
    return(0);
1615
1.13M
}
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
146k
xmlParserNsGrow(xmlParserCtxtPtr ctxt) {
1625
146k
    const xmlChar **table;
1626
146k
    xmlParserNsExtra *extra;
1627
146k
    int newSize;
1628
1629
146k
    newSize = xmlGrowCapacity(ctxt->nsMax,
1630
146k
                              sizeof(table[0]) + sizeof(extra[0]),
1631
146k
                              16, XML_MAX_ITEMS);
1632
146k
    if (newSize < 0)
1633
0
        goto error;
1634
1635
146k
    table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0]));
1636
146k
    if (table == NULL)
1637
116
        goto error;
1638
145k
    ctxt->nsTab = table;
1639
1640
145k
    extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0]));
1641
145k
    if (extra == NULL)
1642
117
        goto error;
1643
145k
    ctxt->nsdb->extra = extra;
1644
1645
145k
    ctxt->nsMax = newSize;
1646
145k
    return(0);
1647
1648
233
error:
1649
233
    xmlErrMemory(ctxt);
1650
233
    return(-1);
1651
145k
}
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
2.02M
                const xmlHashedString *uri, void *saxData, int defAttr) {
1667
2.02M
    xmlParserNsBucket *bucket = NULL;
1668
2.02M
    xmlParserNsExtra *extra;
1669
2.02M
    const xmlChar **ns;
1670
2.02M
    unsigned hashValue, nsIndex, oldIndex;
1671
1672
2.02M
    if ((prefix != NULL) && (prefix->name == ctxt->str_xml))
1673
2.48k
        return(0);
1674
1675
2.02M
    if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) {
1676
233
        xmlErrMemory(ctxt);
1677
233
        return(-1);
1678
233
    }
1679
1680
    /*
1681
     * Default namespace and 'xml' namespace
1682
     */
1683
2.02M
    if ((prefix == NULL) || (prefix->name == NULL)) {
1684
356k
        oldIndex = ctxt->nsdb->defaultNsIndex;
1685
1686
356k
        if (oldIndex != INT_MAX) {
1687
316k
            extra = &ctxt->nsdb->extra[oldIndex];
1688
1689
316k
            if (extra->elementId == ctxt->nsdb->elementId) {
1690
16.5k
                if (defAttr == 0)
1691
15.0k
                    xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns");
1692
16.5k
                return(0);
1693
16.5k
            }
1694
1695
299k
            if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1696
64.5k
                (uri->name == ctxt->nsTab[oldIndex * 2 + 1]))
1697
35.3k
                return(0);
1698
299k
        }
1699
1700
304k
        ctxt->nsdb->defaultNsIndex = ctxt->nsNr;
1701
304k
        goto populate_entry;
1702
356k
    }
1703
1704
    /*
1705
     * Hash table lookup
1706
     */
1707
1.66M
    oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket);
1708
1.66M
    if (oldIndex != INT_MAX) {
1709
1.16M
        extra = &ctxt->nsdb->extra[oldIndex];
1710
1711
        /*
1712
         * Check for duplicate definitions on the same element.
1713
         */
1714
1.16M
        if (extra->elementId == ctxt->nsdb->elementId) {
1715
6.38k
            if (defAttr == 0)
1716
4.81k
                xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name);
1717
6.38k
            return(0);
1718
6.38k
        }
1719
1720
1.15M
        if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1721
53.5k
            (uri->name == ctxt->nsTab[bucket->index * 2 + 1]))
1722
46.0k
            return(0);
1723
1724
1.11M
        bucket->index = ctxt->nsNr;
1725
1.11M
        goto populate_entry;
1726
1.15M
    }
1727
1728
    /*
1729
     * Insert new bucket
1730
     */
1731
1732
499k
    hashValue = prefix->hashValue;
1733
1734
    /*
1735
     * Grow hash table, 50% fill factor
1736
     */
1737
499k
    if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) {
1738
58.2k
        xmlParserNsBucket *newHash;
1739
58.2k
        unsigned newSize, i, index;
1740
1741
58.2k
        if (ctxt->nsdb->hashSize > UINT_MAX / 2) {
1742
0
            xmlErrMemory(ctxt);
1743
0
            return(-1);
1744
0
        }
1745
58.2k
        newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16;
1746
58.2k
        newHash = xmlMalloc(newSize * sizeof(newHash[0]));
1747
58.2k
        if (newHash == NULL) {
1748
29
            xmlErrMemory(ctxt);
1749
29
            return(-1);
1750
29
        }
1751
58.1k
        memset(newHash, 0, newSize * sizeof(newHash[0]));
1752
1753
1.45M
        for (i = 0; i < ctxt->nsdb->hashSize; i++) {
1754
1.40M
            unsigned hv = ctxt->nsdb->hash[i].hashValue;
1755
1.40M
            unsigned newIndex;
1756
1757
1.40M
            if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX))
1758
1.37M
                continue;
1759
23.7k
            newIndex = hv & (newSize - 1);
1760
1761
34.3k
            while (newHash[newIndex].hashValue != 0) {
1762
10.5k
                newIndex++;
1763
10.5k
                if (newIndex == newSize)
1764
3.32k
                    newIndex = 0;
1765
10.5k
            }
1766
1767
23.7k
            newHash[newIndex] = ctxt->nsdb->hash[i];
1768
23.7k
        }
1769
1770
58.1k
        xmlFree(ctxt->nsdb->hash);
1771
58.1k
        ctxt->nsdb->hash = newHash;
1772
58.1k
        ctxt->nsdb->hashSize = newSize;
1773
1774
        /*
1775
         * Relookup
1776
         */
1777
58.1k
        index = hashValue & (newSize - 1);
1778
1779
64.3k
        while (newHash[index].hashValue != 0) {
1780
6.12k
            index++;
1781
6.12k
            if (index == newSize)
1782
1.36k
                index = 0;
1783
6.12k
        }
1784
1785
58.1k
        bucket = &newHash[index];
1786
58.1k
    }
1787
1788
499k
    bucket->hashValue = hashValue;
1789
499k
    bucket->index = ctxt->nsNr;
1790
499k
    ctxt->nsdb->hashElems++;
1791
499k
    oldIndex = INT_MAX;
1792
1793
1.91M
populate_entry:
1794
1.91M
    nsIndex = ctxt->nsNr;
1795
1796
1.91M
    ns = &ctxt->nsTab[nsIndex * 2];
1797
1.91M
    ns[0] = prefix ? prefix->name : NULL;
1798
1.91M
    ns[1] = uri->name;
1799
1800
1.91M
    extra = &ctxt->nsdb->extra[nsIndex];
1801
1.91M
    extra->saxData = saxData;
1802
1.91M
    extra->prefixHashValue = prefix ? prefix->hashValue : 0;
1803
1.91M
    extra->uriHashValue = uri->hashValue;
1804
1.91M
    extra->elementId = ctxt->nsdb->elementId;
1805
1.91M
    extra->oldIndex = oldIndex;
1806
1807
1.91M
    ctxt->nsNr++;
1808
1809
1.91M
    return(1);
1810
499k
}
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
837k
{
1822
837k
    int i;
1823
1824
    /* assert(nr <= ctxt->nsNr); */
1825
1826
2.58M
    for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) {
1827
1.74M
        const xmlChar *prefix = ctxt->nsTab[i * 2];
1828
1.74M
        xmlParserNsExtra *extra = &ctxt->nsdb->extra[i];
1829
1830
1.74M
        if (prefix == NULL) {
1831
268k
            ctxt->nsdb->defaultNsIndex = extra->oldIndex;
1832
1.47M
        } else {
1833
1.47M
            xmlHashedString hprefix;
1834
1.47M
            xmlParserNsBucket *bucket = NULL;
1835
1836
1.47M
            hprefix.name = prefix;
1837
1.47M
            hprefix.hashValue = extra->prefixHashValue;
1838
1.47M
            xmlParserNsLookup(ctxt, &hprefix, &bucket);
1839
            /* assert(bucket && bucket->hashValue); */
1840
1.47M
            bucket->index = extra->oldIndex;
1841
1.47M
        }
1842
1.74M
    }
1843
1844
837k
    ctxt->nsNr -= nr;
1845
837k
    return(nr);
1846
837k
}
1847
1848
static int
1849
136k
xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt) {
1850
136k
    const xmlChar **atts;
1851
136k
    unsigned *attallocs;
1852
136k
    int newSize;
1853
1854
136k
    newSize = xmlGrowCapacity(ctxt->maxatts / 5,
1855
136k
                              sizeof(atts[0]) * 5 + sizeof(attallocs[0]),
1856
136k
                              10, XML_MAX_ATTRS);
1857
136k
    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
136k
    atts = xmlRealloc(ctxt->atts, newSize * sizeof(atts[0]) * 5);
1864
136k
    if (atts == NULL)
1865
90
        goto mem_error;
1866
136k
    ctxt->atts = atts;
1867
1868
136k
    attallocs = xmlRealloc(ctxt->attallocs,
1869
136k
                           newSize * sizeof(attallocs[0]));
1870
136k
    if (attallocs == NULL)
1871
82
        goto mem_error;
1872
136k
    ctxt->attallocs = attallocs;
1873
1874
136k
    ctxt->maxatts = newSize * 5;
1875
1876
136k
    return(0);
1877
1878
172
mem_error:
1879
172
    xmlErrMemory(ctxt);
1880
172
    return(-1);
1881
136k
}
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
1.09M
{
1893
1.09M
    char *directory = NULL;
1894
1.09M
    int maxDepth;
1895
1896
1.09M
    if ((ctxt == NULL) || (value == NULL))
1897
13.8k
        return(-1);
1898
1899
1.08M
    maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
1900
1901
1.08M
    if (ctxt->inputNr >= ctxt->inputMax) {
1902
48.9k
        xmlParserInputPtr *tmp;
1903
48.9k
        int newSize;
1904
1905
48.9k
        newSize = xmlGrowCapacity(ctxt->inputMax, sizeof(tmp[0]),
1906
48.9k
                                  5, maxDepth);
1907
48.9k
        if (newSize < 0) {
1908
11
            xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
1909
11
                           "Maximum entity nesting depth exceeded");
1910
11
            return(-1);
1911
11
        }
1912
48.9k
        tmp = xmlRealloc(ctxt->inputTab, newSize * sizeof(tmp[0]));
1913
48.9k
        if (tmp == NULL) {
1914
138
            xmlErrMemory(ctxt);
1915
138
            return(-1);
1916
138
        }
1917
48.8k
        ctxt->inputTab = tmp;
1918
48.8k
        ctxt->inputMax = newSize;
1919
48.8k
    }
1920
1921
1.08M
    if ((ctxt->inputNr == 0) && (value->filename != NULL)) {
1922
517k
        directory = xmlParserGetDirectory(value->filename);
1923
517k
        if (directory == NULL) {
1924
189
            xmlErrMemory(ctxt);
1925
189
            return(-1);
1926
189
        }
1927
517k
    }
1928
1929
1.08M
    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
1.08M
    ctxt->inputTab[ctxt->inputNr] = value;
1935
1.08M
    ctxt->input = value;
1936
1937
1.08M
    if (ctxt->inputNr == 0) {
1938
630k
        xmlFree(ctxt->directory);
1939
630k
        ctxt->directory = directory;
1940
630k
    }
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
1.08M
    value->id = ctxt->input_id++;
1948
1949
1.08M
    return(ctxt->inputNr++);
1950
1.08M
}
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
2.13M
{
1961
2.13M
    xmlParserInputPtr ret;
1962
1963
2.13M
    if (ctxt == NULL)
1964
0
        return(NULL);
1965
2.13M
    if (ctxt->inputNr <= 0)
1966
1.06M
        return (NULL);
1967
1.07M
    ctxt->inputNr--;
1968
1.07M
    if (ctxt->inputNr > 0)
1969
450k
        ctxt->input = ctxt->inputTab[ctxt->inputNr - 1];
1970
622k
    else
1971
622k
        ctxt->input = NULL;
1972
1.07M
    ret = ctxt->inputTab[ctxt->inputNr];
1973
1.07M
    ctxt->inputTab[ctxt->inputNr] = NULL;
1974
1.07M
    return (ret);
1975
2.13M
}
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
4.37M
{
1989
4.37M
    if (ctxt == NULL)
1990
0
        return(0);
1991
1992
4.37M
    if (ctxt->nodeNr >= ctxt->nodeMax) {
1993
349k
        int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
1994
349k
        xmlNodePtr *tmp;
1995
349k
        int newSize;
1996
1997
349k
        newSize = xmlGrowCapacity(ctxt->nodeMax, sizeof(tmp[0]),
1998
349k
                                  10, maxDepth);
1999
349k
        if (newSize < 0) {
2000
762
            xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
2001
762
                    "Excessive depth in document: %d,"
2002
762
                    " use XML_PARSE_HUGE option\n",
2003
762
                    ctxt->nodeNr);
2004
762
            return(-1);
2005
762
        }
2006
2007
348k
  tmp = xmlRealloc(ctxt->nodeTab, newSize * sizeof(tmp[0]));
2008
348k
        if (tmp == NULL) {
2009
284
            xmlErrMemory(ctxt);
2010
284
            return (-1);
2011
284
        }
2012
348k
        ctxt->nodeTab = tmp;
2013
348k
  ctxt->nodeMax = newSize;
2014
348k
    }
2015
2016
4.36M
    ctxt->nodeTab[ctxt->nodeNr] = value;
2017
4.36M
    ctxt->node = value;
2018
4.36M
    return (ctxt->nodeNr++);
2019
4.37M
}
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
4.39M
{
2032
4.39M
    xmlNodePtr ret;
2033
2034
4.39M
    if (ctxt == NULL) return(NULL);
2035
4.39M
    if (ctxt->nodeNr <= 0)
2036
440k
        return (NULL);
2037
3.95M
    ctxt->nodeNr--;
2038
3.95M
    if (ctxt->nodeNr > 0)
2039
3.82M
        ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1];
2040
128k
    else
2041
128k
        ctxt->node = NULL;
2042
3.95M
    ret = ctxt->nodeTab[ctxt->nodeNr];
2043
3.95M
    ctxt->nodeTab[ctxt->nodeNr] = NULL;
2044
3.95M
    return (ret);
2045
4.39M
}
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
4.86M
{
2062
4.86M
    xmlStartTag *tag;
2063
2064
4.86M
    if (ctxt->nameNr >= ctxt->nameMax) {
2065
351k
        const xmlChar **tmp;
2066
351k
        xmlStartTag *tmp2;
2067
351k
        int newSize;
2068
2069
351k
        newSize = xmlGrowCapacity(ctxt->nameMax,
2070
351k
                                  sizeof(tmp[0]) + sizeof(tmp2[0]),
2071
351k
                                  10, XML_MAX_ITEMS);
2072
351k
        if (newSize < 0)
2073
0
            goto mem_error;
2074
2075
351k
        tmp = xmlRealloc(ctxt->nameTab, newSize * sizeof(tmp[0]));
2076
351k
        if (tmp == NULL)
2077
151
      goto mem_error;
2078
351k
  ctxt->nameTab = tmp;
2079
2080
351k
        tmp2 = xmlRealloc(ctxt->pushTab, newSize * sizeof(tmp2[0]));
2081
351k
        if (tmp2 == NULL)
2082
155
      goto mem_error;
2083
351k
  ctxt->pushTab = tmp2;
2084
2085
351k
        ctxt->nameMax = newSize;
2086
4.51M
    } else if (ctxt->pushTab == NULL) {
2087
208k
        ctxt->pushTab = xmlMalloc(ctxt->nameMax * sizeof(ctxt->pushTab[0]));
2088
208k
        if (ctxt->pushTab == NULL)
2089
419
            goto mem_error;
2090
208k
    }
2091
4.86M
    ctxt->nameTab[ctxt->nameNr] = value;
2092
4.86M
    ctxt->name = value;
2093
4.86M
    tag = &ctxt->pushTab[ctxt->nameNr];
2094
4.86M
    tag->prefix = prefix;
2095
4.86M
    tag->URI = URI;
2096
4.86M
    tag->line = line;
2097
4.86M
    tag->nsNr = nsNr;
2098
4.86M
    return (ctxt->nameNr++);
2099
725
mem_error:
2100
725
    xmlErrMemory(ctxt);
2101
725
    return (-1);
2102
4.86M
}
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
51.6k
{
2113
51.6k
    const xmlChar *ret;
2114
2115
51.6k
    if (ctxt->nameNr <= 0)
2116
0
        return (NULL);
2117
51.6k
    ctxt->nameNr--;
2118
51.6k
    if (ctxt->nameNr > 0)
2119
49.7k
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2120
1.94k
    else
2121
1.94k
        ctxt->name = NULL;
2122
51.6k
    ret = ctxt->nameTab[ctxt->nameNr];
2123
51.6k
    ctxt->nameTab[ctxt->nameNr] = NULL;
2124
51.6k
    return (ret);
2125
51.6k
}
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
4.46M
{
2139
4.46M
    const xmlChar *ret;
2140
2141
4.46M
    if ((ctxt == NULL) || (ctxt->nameNr <= 0))
2142
183
        return (NULL);
2143
4.46M
    ctxt->nameNr--;
2144
4.46M
    if (ctxt->nameNr > 0)
2145
4.35M
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2146
115k
    else
2147
115k
        ctxt->name = NULL;
2148
4.46M
    ret = ctxt->nameTab[ctxt->nameNr];
2149
4.46M
    ctxt->nameTab[ctxt->nameNr] = NULL;
2150
4.46M
    return (ret);
2151
4.46M
}
2152
2153
5.70M
static int spacePush(xmlParserCtxtPtr ctxt, int val) {
2154
5.70M
    if (ctxt->spaceNr >= ctxt->spaceMax) {
2155
496k
        int *tmp;
2156
496k
        int newSize;
2157
2158
496k
        newSize = xmlGrowCapacity(ctxt->spaceMax, sizeof(tmp[0]),
2159
496k
                                  10, XML_MAX_ITEMS);
2160
496k
        if (newSize < 0) {
2161
0
      xmlErrMemory(ctxt);
2162
0
      return(-1);
2163
0
        }
2164
2165
496k
        tmp = xmlRealloc(ctxt->spaceTab, newSize * sizeof(tmp[0]));
2166
496k
        if (tmp == NULL) {
2167
312
      xmlErrMemory(ctxt);
2168
312
      return(-1);
2169
312
  }
2170
495k
  ctxt->spaceTab = tmp;
2171
2172
495k
        ctxt->spaceMax = newSize;
2173
495k
    }
2174
5.70M
    ctxt->spaceTab[ctxt->spaceNr] = val;
2175
5.70M
    ctxt->space = &ctxt->spaceTab[ctxt->spaceNr];
2176
5.70M
    return(ctxt->spaceNr++);
2177
5.70M
}
2178
2179
5.34M
static int spacePop(xmlParserCtxtPtr ctxt) {
2180
5.34M
    int ret;
2181
5.34M
    if (ctxt->spaceNr <= 0) return(0);
2182
5.34M
    ctxt->spaceNr--;
2183
5.34M
    if (ctxt->spaceNr > 0)
2184
5.30M
  ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1];
2185
47.5k
    else
2186
47.5k
        ctxt->space = &ctxt->spaceTab[0];
2187
5.34M
    ret = ctxt->spaceTab[ctxt->spaceNr];
2188
5.34M
    ctxt->spaceTab[ctxt->spaceNr] = -1;
2189
5.34M
    return(ret);
2190
5.34M
}
2191
2192
/*
2193
 * Macros for accessing the content. Those should be used only by the parser,
2194
 * and not exported.
2195
 *
2196
 * Dirty macros, i.e. one often need to make assumption on the context to
2197
 * use them
2198
 *
2199
 *   CUR_PTR return the current pointer to the xmlChar to be parsed.
2200
 *           To be used with extreme caution since operations consuming
2201
 *           characters may move the input buffer to a different location !
2202
 *   CUR     returns the current xmlChar value, i.e. a 8 bit value if compiled
2203
 *           This should be used internally by the parser
2204
 *           only to compare to ASCII values otherwise it would break when
2205
 *           running with UTF-8 encoding.
2206
 *   RAW     same as CUR but in the input buffer, bypass any token
2207
 *           extraction that may have been done
2208
 *   NXT(n)  returns the n'th next xmlChar. Same as CUR is should be used only
2209
 *           to compare on ASCII based substring.
2210
 *   SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined
2211
 *           strings without newlines within the parser.
2212
 *   NEXT1(l) Skip 1 xmlChar, and must also be used only to skip 1 non-newline ASCII
2213
 *           defined char within the parser.
2214
 * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding
2215
 *
2216
 *   NEXT    Skip to the next character, this does the proper decoding
2217
 *           in UTF-8 mode. It also pop-up unfinished entities on the fly.
2218
 *   NEXTL(l) Skip the current unicode character of l xmlChars long.
2219
 *   COPY_BUF  copy the current unicode char to the target buffer, increment
2220
 *            the index
2221
 *   GROW, SHRINK  handling of input buffers
2222
 */
2223
2224
86.5M
#define RAW (*ctxt->input->cur)
2225
5.35G
#define CUR (*ctxt->input->cur)
2226
47.0M
#define NXT(val) ctxt->input->cur[(val)]
2227
7.48G
#define CUR_PTR ctxt->input->cur
2228
14.0M
#define BASE_PTR ctxt->input->base
2229
2230
#define CMP4( s, c1, c2, c3, c4 ) \
2231
179M
  ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \
2232
90.1M
    ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 )
2233
#define CMP5( s, c1, c2, c3, c4, c5 ) \
2234
176M
  ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 )
2235
#define CMP6( s, c1, c2, c3, c4, c5, c6 ) \
2236
172M
  ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 )
2237
#define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \
2238
168M
  ( 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
166M
  ( 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
82.7M
  ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \
2243
82.7M
    ((unsigned char *) s)[ 8 ] == c9 )
2244
#define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \
2245
112k
  ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \
2246
112k
    ((unsigned char *) s)[ 9 ] == c10 )
2247
2248
12.9M
#define SKIP(val) do {             \
2249
12.9M
    ctxt->input->cur += (val),ctxt->input->col+=(val);      \
2250
12.9M
    if (*ctxt->input->cur == 0)           \
2251
12.9M
        xmlParserGrow(ctxt);           \
2252
12.9M
  } 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
173M
    if (!PARSER_PROGRESSIVE(ctxt)) \
2268
173M
  xmlParserShrink(ctxt);
2269
2270
#define GROW \
2271
308M
    if ((!PARSER_PROGRESSIVE(ctxt)) && \
2272
308M
        (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \
2273
18.7M
  xmlParserGrow(ctxt);
2274
2275
24.9M
#define SKIP_BLANKS xmlSkipBlankChars(ctxt)
2276
2277
4.21M
#define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt)
2278
2279
1.86G
#define NEXT xmlNextChar(ctxt)
2280
2281
9.31M
#define NEXT1 {               \
2282
9.31M
  ctxt->input->col++;           \
2283
9.31M
  ctxt->input->cur++;           \
2284
9.31M
  if (*ctxt->input->cur == 0)         \
2285
9.31M
      xmlParserGrow(ctxt);           \
2286
9.31M
    }
2287
2288
4.11G
#define NEXTL(l) do {             \
2289
4.11G
    if (*(ctxt->input->cur) == '\n') {         \
2290
72.9M
  ctxt->input->line++; ctxt->input->col = 1;      \
2291
4.03G
    } else ctxt->input->col++;           \
2292
4.11G
    ctxt->input->cur += l;        \
2293
4.11G
  } while (0)
2294
2295
#define COPY_BUF(b, i, v)           \
2296
480M
    if (v < 0x80) b[i++] = v;           \
2297
480M
    else i += xmlCopyCharMultiByte(&b[i],v)
2298
2299
static int
2300
524M
xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) {
2301
524M
    int c = xmlCurrentChar(ctxt, len);
2302
2303
524M
    if (c == XML_INVALID_CHAR)
2304
89.1M
        c = 0xFFFD; /* replacement character */
2305
2306
524M
    return(c);
2307
524M
}
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
26.3M
xmlSkipBlankChars(xmlParserCtxt *ctxt) {
2319
26.3M
    const xmlChar *cur;
2320
26.3M
    int res = 0;
2321
2322
26.3M
    cur = ctxt->input->cur;
2323
26.3M
    while (IS_BLANK_CH(*cur)) {
2324
12.1M
        if (*cur == '\n') {
2325
4.80M
            ctxt->input->line++; ctxt->input->col = 1;
2326
7.37M
        } else {
2327
7.37M
            ctxt->input->col++;
2328
7.37M
        }
2329
12.1M
        cur++;
2330
12.1M
        if (res < INT_MAX)
2331
12.1M
            res++;
2332
12.1M
        if (*cur == 0) {
2333
110k
            ctxt->input->cur = cur;
2334
110k
            xmlParserGrow(ctxt);
2335
110k
            cur = ctxt->input->cur;
2336
110k
        }
2337
12.1M
    }
2338
26.3M
    ctxt->input->cur = cur;
2339
2340
26.3M
    if (res > 4)
2341
134k
        GROW;
2342
2343
26.3M
    return(res);
2344
26.3M
}
2345
2346
static void
2347
412k
xmlPopPE(xmlParserCtxtPtr ctxt) {
2348
412k
    unsigned long consumed;
2349
412k
    xmlEntityPtr ent;
2350
2351
412k
    ent = ctxt->input->entity;
2352
2353
412k
    ent->flags &= ~XML_ENT_EXPANDING;
2354
2355
412k
    if ((ent->flags & XML_ENT_CHECKED) == 0) {
2356
31.4k
        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
164k
        do {
2363
164k
            ctxt->input->cur = ctxt->input->end;
2364
164k
            xmlParserShrink(ctxt);
2365
164k
            result = xmlParserGrow(ctxt);
2366
164k
        } while (result > 0);
2367
2368
31.4k
        consumed = ctxt->input->consumed;
2369
31.4k
        xmlSaturatedAddSizeT(&consumed,
2370
31.4k
                             ctxt->input->end - ctxt->input->base);
2371
2372
31.4k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
2373
2374
        /*
2375
         * Add to sizeentities when parsing an external entity
2376
         * for the first time.
2377
         */
2378
31.4k
        if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
2379
17.0k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
2380
17.0k
        }
2381
2382
31.4k
        ent->flags |= XML_ENT_CHECKED;
2383
31.4k
    }
2384
2385
412k
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
2386
2387
412k
    xmlParserEntityCheck(ctxt, ent->expandedSize);
2388
2389
412k
    GROW;
2390
412k
}
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
4.21M
xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) {
2401
4.21M
    int res = 0;
2402
4.21M
    int inParam;
2403
4.21M
    int expandParam;
2404
2405
4.21M
    inParam = PARSER_IN_PE(ctxt);
2406
4.21M
    expandParam = PARSER_EXTERNAL(ctxt);
2407
2408
4.21M
    if (!inParam && !expandParam)
2409
1.32M
        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
7.43M
    while (PARSER_STOPPED(ctxt) == 0) {
2416
7.43M
        if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */
2417
4.44M
            NEXT;
2418
4.44M
        } else if (CUR == '%') {
2419
178k
            if ((expandParam == 0) ||
2420
176k
                (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0))
2421
118k
                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
59.8k
            xmlParsePERefInternal(ctxt, 0);
2430
2431
59.8k
            inParam = PARSER_IN_PE(ctxt);
2432
59.8k
            expandParam = PARSER_EXTERNAL(ctxt);
2433
2.80M
        } else if (CUR == 0) {
2434
203k
            if (inParam == 0)
2435
622
                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
202k
            if (ctxt->input->flags & XML_INPUT_MARKUP_DECL)
2443
170k
                break;
2444
2445
31.3k
            xmlPopPE(ctxt);
2446
2447
31.3k
            inParam = PARSER_IN_PE(ctxt);
2448
31.3k
            expandParam = PARSER_EXTERNAL(ctxt);
2449
2.60M
        } else {
2450
2.60M
            break;
2451
2.60M
        }
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
4.53M
        if (res < INT_MAX)
2461
4.53M
            res++;
2462
4.53M
    }
2463
2464
2.89M
    return(res);
2465
4.21M
}
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
600k
xmlParseCharRef(xmlParserCtxt *ctxt) {
2530
600k
    int val = 0;
2531
600k
    int count = 0;
2532
2533
    /*
2534
     * Using RAW/CUR/NEXT is okay since we are working on ASCII range here
2535
     */
2536
600k
    if ((RAW == '&') && (NXT(1) == '#') &&
2537
600k
        (NXT(2) == 'x')) {
2538
304k
  SKIP(3);
2539
304k
  GROW;
2540
1.16M
  while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) {
2541
922k
      if (count++ > 20) {
2542
30.1k
    count = 0;
2543
30.1k
    GROW;
2544
30.1k
      }
2545
922k
      if ((RAW >= '0') && (RAW <= '9'))
2546
507k
          val = val * 16 + (CUR - '0');
2547
415k
      else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20))
2548
202k
          val = val * 16 + (CUR - 'a') + 10;
2549
212k
      else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20))
2550
153k
          val = val * 16 + (CUR - 'A') + 10;
2551
59.6k
      else {
2552
59.6k
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2553
59.6k
    val = 0;
2554
59.6k
    break;
2555
59.6k
      }
2556
862k
      if (val > 0x110000)
2557
357k
          val = 0x110000;
2558
2559
862k
      NEXT;
2560
862k
      count++;
2561
862k
  }
2562
304k
  if (RAW == ';') {
2563
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2564
244k
      ctxt->input->col++;
2565
244k
      ctxt->input->cur++;
2566
244k
  }
2567
304k
    } else if  ((RAW == '&') && (NXT(1) == '#')) {
2568
295k
  SKIP(2);
2569
295k
  GROW;
2570
941k
  while (RAW != ';') { /* loop blocked by count */
2571
680k
      if (count++ > 20) {
2572
10.7k
    count = 0;
2573
10.7k
    GROW;
2574
10.7k
      }
2575
680k
      if ((RAW >= '0') && (RAW <= '9'))
2576
645k
          val = val * 10 + (CUR - '0');
2577
34.7k
      else {
2578
34.7k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2579
34.7k
    val = 0;
2580
34.7k
    break;
2581
34.7k
      }
2582
645k
      if (val > 0x110000)
2583
107k
          val = 0x110000;
2584
2585
645k
      NEXT;
2586
645k
      count++;
2587
645k
  }
2588
295k
  if (RAW == ';') {
2589
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2590
260k
      ctxt->input->col++;
2591
260k
      ctxt->input->cur++;
2592
260k
  }
2593
295k
    } 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
600k
    if (val >= 0x110000) {
2605
3.86k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2606
3.86k
                "xmlParseCharRef: character reference out of bounds\n",
2607
3.86k
          val);
2608
3.86k
        val = 0xFFFD;
2609
596k
    } else if (!IS_CHAR(val)) {
2610
127k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2611
127k
                          "xmlParseCharRef: invalid xmlChar value %d\n",
2612
127k
                    val);
2613
127k
    }
2614
600k
    return(val);
2615
600k
}
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
2.19M
xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) {
2635
2.19M
    const xmlChar *ptr;
2636
2.19M
    xmlChar cur;
2637
2.19M
    int val = 0;
2638
2639
2.19M
    if ((str == NULL) || (*str == NULL)) return(0);
2640
2.19M
    ptr = *str;
2641
2.19M
    cur = *ptr;
2642
2.19M
    if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) {
2643
93.3k
  ptr += 3;
2644
93.3k
  cur = *ptr;
2645
442k
  while (cur != ';') { /* Non input consuming loop */
2646
354k
      if ((cur >= '0') && (cur <= '9'))
2647
111k
          val = val * 16 + (cur - '0');
2648
242k
      else if ((cur >= 'a') && (cur <= 'f'))
2649
106k
          val = val * 16 + (cur - 'a') + 10;
2650
136k
      else if ((cur >= 'A') && (cur <= 'F'))
2651
131k
          val = val * 16 + (cur - 'A') + 10;
2652
4.91k
      else {
2653
4.91k
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2654
4.91k
    val = 0;
2655
4.91k
    break;
2656
4.91k
      }
2657
349k
      if (val > 0x110000)
2658
8.94k
          val = 0x110000;
2659
2660
349k
      ptr++;
2661
349k
      cur = *ptr;
2662
349k
  }
2663
93.3k
  if (cur == ';')
2664
88.4k
      ptr++;
2665
2.10M
    } else if  ((cur == '&') && (ptr[1] == '#')){
2666
2.10M
  ptr += 2;
2667
2.10M
  cur = *ptr;
2668
6.58M
  while (cur != ';') { /* Non input consuming loops */
2669
4.48M
      if ((cur >= '0') && (cur <= '9'))
2670
4.48M
          val = val * 10 + (cur - '0');
2671
5.74k
      else {
2672
5.74k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2673
5.74k
    val = 0;
2674
5.74k
    break;
2675
5.74k
      }
2676
4.48M
      if (val > 0x110000)
2677
6.79k
          val = 0x110000;
2678
2679
4.48M
      ptr++;
2680
4.48M
      cur = *ptr;
2681
4.48M
  }
2682
2.10M
  if (cur == ';')
2683
2.09M
      ptr++;
2684
2.10M
    } else {
2685
0
  xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2686
0
  return(0);
2687
0
    }
2688
2.19M
    *str = ptr;
2689
2690
    /*
2691
     * [ WFC: Legal Character ]
2692
     * Characters referred to using character references must match the
2693
     * production for Char.
2694
     */
2695
2.19M
    if (val >= 0x110000) {
2696
1.86k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2697
1.86k
                "xmlParseStringCharRef: character reference out of bounds\n",
2698
1.86k
                val);
2699
2.19M
    } else if (IS_CHAR(val)) {
2700
2.17M
        return(val);
2701
2.17M
    } else {
2702
16.0k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2703
16.0k
        "xmlParseStringCharRef: invalid xmlChar value %d\n",
2704
16.0k
        val);
2705
16.0k
    }
2706
17.8k
    return(0);
2707
2.19M
}
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
2.06M
                     int blank_chars) {
2816
2.06M
    int i;
2817
2.06M
    xmlNodePtr lastChild;
2818
2819
    /*
2820
     * Check for xml:space value.
2821
     */
2822
2.06M
    if ((ctxt->space == NULL) || (*(ctxt->space) == 1) ||
2823
2.06M
        (*(ctxt->space) == -2))
2824
1.63M
  return(0);
2825
2826
    /*
2827
     * Check that the string is made of blanks
2828
     */
2829
430k
    if (blank_chars == 0) {
2830
793k
  for (i = 0;i < len;i++)
2831
777k
      if (!(IS_BLANK_CH(str[i]))) return(0);
2832
385k
    }
2833
2834
    /*
2835
     * Look if the element is mixed content in the DTD if available
2836
     */
2837
60.8k
    if (ctxt->node == NULL) return(0);
2838
60.8k
    if (ctxt->myDoc != NULL) {
2839
60.8k
        xmlElementPtr elemDecl = NULL;
2840
60.8k
        xmlDocPtr doc = ctxt->myDoc;
2841
60.8k
        const xmlChar *prefix = NULL;
2842
2843
60.8k
        if (ctxt->node->ns)
2844
10.5k
            prefix = ctxt->node->ns->prefix;
2845
60.8k
        if (doc->intSubset != NULL)
2846
36.9k
            elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name,
2847
36.9k
                                      prefix);
2848
60.8k
        if ((elemDecl == NULL) && (doc->extSubset != NULL))
2849
2.82k
            elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name,
2850
2.82k
                                      prefix);
2851
60.8k
        if (elemDecl != NULL) {
2852
11.9k
            if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)
2853
4.61k
                return(1);
2854
7.29k
            if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) ||
2855
6.45k
                (elemDecl->etype == XML_ELEMENT_TYPE_MIXED))
2856
1.74k
                return(0);
2857
7.29k
        }
2858
60.8k
    }
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
54.4k
    if ((RAW != '<') && (RAW != 0xD)) return(0);
2868
47.0k
    if ((ctxt->node->children == NULL) &&
2869
28.2k
  (RAW == '<') && (NXT(1) == '/')) return(0);
2870
2871
45.1k
    lastChild = xmlGetLastChild(ctxt->node);
2872
45.1k
    if (lastChild == NULL) {
2873
26.4k
        if ((ctxt->node->type != XML_ELEMENT_NODE) &&
2874
0
            (ctxt->node->content != NULL)) return(0);
2875
26.4k
    } else if (xmlNodeIsText(lastChild))
2876
1.20k
        return(0);
2877
17.5k
    else if ((ctxt->node->children != NULL) &&
2878
17.5k
             (xmlNodeIsText(ctxt->node->children)))
2879
979
        return(0);
2880
42.9k
    return(1);
2881
45.1k
}
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
39.6M
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
39.6M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2953
39.5M
        (((c >= 'a') && (c <= 'z')) ||
2954
23.6M
         ((c >= 'A') && (c <= 'Z')) ||
2955
21.9M
         (c == '_') || (c == ':') ||
2956
21.7M
         ((c >= 0xC0) && (c <= 0xD6)) ||
2957
21.7M
         ((c >= 0xD8) && (c <= 0xF6)) ||
2958
21.7M
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2959
21.6M
         ((c >= 0x370) && (c <= 0x37D)) ||
2960
21.6M
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2961
21.6M
         ((c >= 0x200C) && (c <= 0x200D)) ||
2962
21.6M
         ((c >= 0x2070) && (c <= 0x218F)) ||
2963
21.6M
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2964
21.6M
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2965
21.6M
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2966
21.6M
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2967
21.5M
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2968
18.0M
        return(1);
2969
21.6M
    return(0);
2970
39.6M
}
2971
2972
static int
2973
215M
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
215M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2979
206M
        (((c >= 'a') && (c <= 'z')) ||
2980
173M
         ((c >= 'A') && (c <= 'Z')) ||
2981
159M
         ((c >= '0') && (c <= '9')) || /* !start */
2982
157M
         (c == '_') || (c == ':') ||
2983
156M
         (c == '-') || (c == '.') || (c == 0xB7) || /* !start */
2984
154M
         ((c >= 0xC0) && (c <= 0xD6)) ||
2985
151M
         ((c >= 0xD8) && (c <= 0xF6)) ||
2986
150M
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2987
147M
         ((c >= 0x300) && (c <= 0x36F)) || /* !start */
2988
147M
         ((c >= 0x370) && (c <= 0x37D)) ||
2989
147M
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2990
142M
         ((c >= 0x200C) && (c <= 0x200D)) ||
2991
142M
         ((c >= 0x203F) && (c <= 0x2040)) || /* !start */
2992
142M
         ((c >= 0x2070) && (c <= 0x218F)) ||
2993
34.2M
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2994
34.2M
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2995
31.0M
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2996
31.0M
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2997
9.38M
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2998
197M
         return(1);
2999
17.8M
    return(0);
3000
215M
}
3001
3002
static int
3003
3.11M
xmlIsNameStartCharOld(int c) {
3004
3.11M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3005
3.09M
        ((IS_LETTER(c) || (c == '_') || (c == ':'))))
3006
2.39M
        return(1);
3007
725k
    return(0);
3008
3.11M
}
3009
3010
static int
3011
25.9M
xmlIsNameCharOld(int c) {
3012
25.9M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3013
25.9M
        ((IS_LETTER(c)) || (IS_DIGIT(c)) ||
3014
3.31M
         (c == '.') || (c == '-') ||
3015
3.07M
         (c == '_') || (c == ':') ||
3016
3.03M
         (IS_COMBINING(c)) ||
3017
2.38M
         (IS_EXTENDER(c))))
3018
23.5M
        return(1);
3019
2.40M
    return(0);
3020
25.9M
}
3021
3022
static int
3023
42.7M
xmlIsNameStartChar(int c, int old10) {
3024
42.7M
    if (!old10)
3025
39.6M
        return(xmlIsNameStartCharNew(c));
3026
3.11M
    else
3027
3.11M
        return(xmlIsNameStartCharOld(c));
3028
42.7M
}
3029
3030
static int
3031
241M
xmlIsNameChar(int c, int old10) {
3032
241M
    if (!old10)
3033
215M
        return(xmlIsNameCharNew(c));
3034
25.9M
    else
3035
25.9M
        return(xmlIsNameCharOld(c));
3036
241M
}
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
40.0M
xmlScanName(const xmlChar *ptr, size_t maxSize, int flags) {
3052
40.0M
    int stop = flags & XML_SCAN_NC ? ':' : 0;
3053
40.0M
    int old10 = flags & XML_SCAN_OLD10 ? 1 : 0;
3054
3055
101M
    while (1) {
3056
101M
        int c, len;
3057
3058
101M
        c = *ptr;
3059
101M
        if (c < 0x80) {
3060
80.1M
            if (c == stop)
3061
418k
                break;
3062
79.6M
            len = 1;
3063
79.6M
        } else {
3064
21.3M
            len = 4;
3065
21.3M
            c = xmlGetUTF8Char(ptr, &len);
3066
21.3M
            if (c < 0)
3067
21.8k
                break;
3068
21.3M
        }
3069
3070
100M
        if (flags & XML_SCAN_NMTOKEN ?
3071
61.0M
                !xmlIsNameChar(c, old10) :
3072
100M
                !xmlIsNameStartChar(c, old10))
3073
39.6M
            break;
3074
3075
61.3M
        if ((size_t) len > maxSize)
3076
479
            return(NULL);
3077
61.3M
        ptr += len;
3078
61.3M
        maxSize -= len;
3079
61.3M
        flags |= XML_SCAN_NMTOKEN;
3080
61.3M
    }
3081
3082
40.0M
    return(ptr);
3083
40.0M
}
3084
3085
static const xmlChar *
3086
1.07M
xmlParseNameComplex(xmlParserCtxtPtr ctxt) {
3087
1.07M
    const xmlChar *ret;
3088
1.07M
    int len = 0, l;
3089
1.07M
    int c;
3090
1.07M
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3091
474k
                    XML_MAX_TEXT_LENGTH :
3092
1.07M
                    XML_MAX_NAME_LENGTH;
3093
1.07M
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3094
3095
    /*
3096
     * Handler for more complex cases
3097
     */
3098
1.07M
    c = xmlCurrentChar(ctxt, &l);
3099
1.07M
    if (!xmlIsNameStartChar(c, old10))
3100
760k
        return(NULL);
3101
312k
    len += l;
3102
312k
    NEXTL(l);
3103
312k
    c = xmlCurrentChar(ctxt, &l);
3104
51.9M
    while (xmlIsNameChar(c, old10)) {
3105
51.6M
        if (len <= INT_MAX - l)
3106
51.6M
            len += l;
3107
51.6M
        NEXTL(l);
3108
51.6M
        c = xmlCurrentChar(ctxt, &l);
3109
51.6M
    }
3110
312k
    if (len > maxLength) {
3111
731
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3112
731
        return(NULL);
3113
731
    }
3114
311k
    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
311k
    if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r'))
3125
1.95k
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len);
3126
309k
    else
3127
309k
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len);
3128
311k
    if (ret == NULL)
3129
24
        xmlErrMemory(ctxt);
3130
311k
    return(ret);
3131
311k
}
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
8.90M
xmlParseName(xmlParserCtxt *ctxt) {
3151
8.90M
    const xmlChar *in;
3152
8.90M
    const xmlChar *ret;
3153
8.90M
    size_t count = 0;
3154
8.90M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3155
3.79M
                       XML_MAX_TEXT_LENGTH :
3156
8.90M
                       XML_MAX_NAME_LENGTH;
3157
3158
8.90M
    GROW;
3159
3160
    /*
3161
     * Accelerator for simple ASCII names
3162
     */
3163
8.90M
    in = ctxt->input->cur;
3164
8.90M
    if (((*in >= 0x61) && (*in <= 0x7A)) ||
3165
1.48M
  ((*in >= 0x41) && (*in <= 0x5A)) ||
3166
8.04M
  (*in == '_') || (*in == ':')) {
3167
8.04M
  in++;
3168
72.0M
  while (((*in >= 0x61) && (*in <= 0x7A)) ||
3169
27.7M
         ((*in >= 0x41) && (*in <= 0x5A)) ||
3170
12.4M
         ((*in >= 0x30) && (*in <= 0x39)) ||
3171
9.72M
         (*in == '_') || (*in == '-') ||
3172
9.33M
         (*in == ':') || (*in == '.'))
3173
64.0M
      in++;
3174
8.04M
  if ((*in > 0) && (*in < 0x80)) {
3175
7.82M
      count = in - ctxt->input->cur;
3176
7.82M
            if (count > maxLength) {
3177
449
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3178
449
                return(NULL);
3179
449
            }
3180
7.82M
      ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count);
3181
7.82M
      ctxt->input->cur = in;
3182
7.82M
      ctxt->input->col += count;
3183
7.82M
      if (ret == NULL)
3184
39
          xmlErrMemory(ctxt);
3185
7.82M
      return(ret);
3186
7.82M
  }
3187
8.04M
    }
3188
    /* accelerator for special cases */
3189
1.07M
    return(xmlParseNameComplex(ctxt));
3190
8.90M
}
3191
3192
static xmlHashedString
3193
1.86M
xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) {
3194
1.86M
    xmlHashedString ret;
3195
1.86M
    int len = 0, l;
3196
1.86M
    int c;
3197
1.86M
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3198
820k
                    XML_MAX_TEXT_LENGTH :
3199
1.86M
                    XML_MAX_NAME_LENGTH;
3200
1.86M
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3201
1.86M
    size_t startPosition = 0;
3202
3203
1.86M
    ret.name = NULL;
3204
1.86M
    ret.hashValue = 0;
3205
3206
    /*
3207
     * Handler for more complex cases
3208
     */
3209
1.86M
    startPosition = CUR_PTR - BASE_PTR;
3210
1.86M
    c = xmlCurrentChar(ctxt, &l);
3211
1.86M
    if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */
3212
1.79M
  (!xmlIsNameStartChar(c, old10) || (c == ':'))) {
3213
1.67M
  return(ret);
3214
1.67M
    }
3215
3216
92.7M
    while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */
3217
92.6M
     (xmlIsNameChar(c, old10) && (c != ':'))) {
3218
92.5M
        if (len <= INT_MAX - l)
3219
92.5M
      len += l;
3220
92.5M
  NEXTL(l);
3221
92.5M
  c = xmlCurrentChar(ctxt, &l);
3222
92.5M
    }
3223
188k
    if (len > maxLength) {
3224
679
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3225
679
        return(ret);
3226
679
    }
3227
187k
    ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len);
3228
187k
    if (ret.name == NULL)
3229
21
        xmlErrMemory(ctxt);
3230
187k
    return(ret);
3231
188k
}
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
10.0M
xmlParseNCName(xmlParserCtxtPtr ctxt) {
3247
10.0M
    const xmlChar *in, *e;
3248
10.0M
    xmlHashedString ret;
3249
10.0M
    size_t count = 0;
3250
10.0M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3251
3.42M
                       XML_MAX_TEXT_LENGTH :
3252
10.0M
                       XML_MAX_NAME_LENGTH;
3253
3254
10.0M
    ret.name = NULL;
3255
3256
    /*
3257
     * Accelerator for simple ASCII names
3258
     */
3259
10.0M
    in = ctxt->input->cur;
3260
10.0M
    e = ctxt->input->end;
3261
10.0M
    if ((((*in >= 0x61) && (*in <= 0x7A)) ||
3262
2.10M
   ((*in >= 0x41) && (*in <= 0x5A)) ||
3263
8.35M
   (*in == '_')) && (in < e)) {
3264
8.35M
  in++;
3265
58.8M
  while ((((*in >= 0x61) && (*in <= 0x7A)) ||
3266
23.3M
          ((*in >= 0x41) && (*in <= 0x5A)) ||
3267
10.1M
          ((*in >= 0x30) && (*in <= 0x39)) ||
3268
9.65M
          (*in == '_') || (*in == '-') ||
3269
50.5M
          (*in == '.')) && (in < e))
3270
50.5M
      in++;
3271
8.35M
  if (in >= e)
3272
12.6k
      goto complex;
3273
8.33M
  if ((*in > 0) && (*in < 0x80)) {
3274
8.22M
      count = in - ctxt->input->cur;
3275
8.22M
            if (count > maxLength) {
3276
299
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3277
299
                return(ret);
3278
299
            }
3279
8.22M
      ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count);
3280
8.22M
      ctxt->input->cur = in;
3281
8.22M
      ctxt->input->col += count;
3282
8.22M
      if (ret.name == NULL) {
3283
23
          xmlErrMemory(ctxt);
3284
23
      }
3285
8.22M
      return(ret);
3286
8.22M
  }
3287
8.33M
    }
3288
1.86M
complex:
3289
1.86M
    return(xmlParseNCNameComplex(ctxt));
3290
10.0M
}
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
1.08M
xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) {
3304
1.08M
    register const xmlChar *cmp = other;
3305
1.08M
    register const xmlChar *in;
3306
1.08M
    const xmlChar *ret;
3307
3308
1.08M
    GROW;
3309
3310
1.08M
    in = ctxt->input->cur;
3311
3.68M
    while (*in != 0 && *in == *cmp) {
3312
2.59M
  ++in;
3313
2.59M
  ++cmp;
3314
2.59M
    }
3315
1.08M
    if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
3316
  /* success */
3317
992k
  ctxt->input->col += in - ctxt->input->cur;
3318
992k
  ctxt->input->cur = in;
3319
992k
  return (const xmlChar*) 1;
3320
992k
    }
3321
    /* failure (or end of input buffer), check with full function */
3322
89.7k
    ret = xmlParseName (ctxt);
3323
    /* strings coming from the dictionary direct compare possible */
3324
89.7k
    if (ret == other) {
3325
16.0k
  return (const xmlChar*) 1;
3326
16.0k
    }
3327
73.6k
    return ret;
3328
89.7k
}
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
6.92M
xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) {
3341
6.92M
    xmlChar *ret;
3342
6.92M
    const xmlChar *cur = *str;
3343
6.92M
    int flags = 0;
3344
6.92M
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3345
2.85M
                    XML_MAX_TEXT_LENGTH :
3346
6.92M
                    XML_MAX_NAME_LENGTH;
3347
3348
6.92M
    if (ctxt->options & XML_PARSE_OLD10)
3349
2.24M
        flags |= XML_SCAN_OLD10;
3350
3351
6.92M
    cur = xmlScanName(*str, maxLength, flags);
3352
6.92M
    if (cur == NULL) {
3353
419
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3354
419
        return(NULL);
3355
419
    }
3356
6.92M
    if (cur == *str)
3357
25.0k
        return(NULL);
3358
3359
6.90M
    ret = xmlStrndup(*str, cur - *str);
3360
6.90M
    if (ret == NULL)
3361
252
        xmlErrMemory(ctxt);
3362
6.90M
    *str = cur;
3363
6.90M
    return(ret);
3364
6.92M
}
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
350k
xmlParseNmtoken(xmlParserCtxt *ctxt) {
3381
350k
    xmlChar buf[XML_MAX_NAMELEN + 5];
3382
350k
    xmlChar *ret;
3383
350k
    int len = 0, l;
3384
350k
    int c;
3385
350k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3386
67.8k
                    XML_MAX_TEXT_LENGTH :
3387
350k
                    XML_MAX_NAME_LENGTH;
3388
350k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3389
3390
350k
    c = xmlCurrentChar(ctxt, &l);
3391
3392
1.26M
    while (xmlIsNameChar(c, old10)) {
3393
923k
  COPY_BUF(buf, len, c);
3394
923k
  NEXTL(l);
3395
923k
  c = xmlCurrentChar(ctxt, &l);
3396
923k
  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
5.51k
      xmlChar *buffer;
3402
5.51k
      int max = len * 2;
3403
3404
5.51k
      buffer = xmlMalloc(max);
3405
5.51k
      if (buffer == NULL) {
3406
38
          xmlErrMemory(ctxt);
3407
38
    return(NULL);
3408
38
      }
3409
5.47k
      memcpy(buffer, buf, len);
3410
34.3M
      while (xmlIsNameChar(c, old10)) {
3411
34.3M
    if (len + 10 > max) {
3412
20.7k
        xmlChar *tmp;
3413
20.7k
                    int newSize;
3414
3415
20.7k
                    newSize = xmlGrowCapacity(max, 1, 1, maxLength);
3416
20.7k
                    if (newSize < 0) {
3417
722
                        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3418
722
                        xmlFree(buffer);
3419
722
                        return(NULL);
3420
722
                    }
3421
20.0k
        tmp = xmlRealloc(buffer, newSize);
3422
20.0k
        if (tmp == NULL) {
3423
17
      xmlErrMemory(ctxt);
3424
17
      xmlFree(buffer);
3425
17
      return(NULL);
3426
17
        }
3427
20.0k
        buffer = tmp;
3428
20.0k
                    max = newSize;
3429
20.0k
    }
3430
34.3M
    COPY_BUF(buffer, len, c);
3431
34.3M
    NEXTL(l);
3432
34.3M
    c = xmlCurrentChar(ctxt, &l);
3433
34.3M
      }
3434
4.73k
      buffer[len] = 0;
3435
4.73k
      return(buffer);
3436
5.47k
  }
3437
923k
    }
3438
345k
    if (len == 0)
3439
50.5k
        return(NULL);
3440
294k
    if (len > maxLength) {
3441
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3442
0
        return(NULL);
3443
0
    }
3444
294k
    ret = xmlStrndup(buf, len);
3445
294k
    if (ret == NULL)
3446
50
        xmlErrMemory(ctxt);
3447
294k
    return(ret);
3448
294k
}
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
519k
                          const xmlChar *str, int length, int depth) {
3462
519k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3463
519k
    const xmlChar *end, *chunk;
3464
519k
    int c, l;
3465
3466
519k
    if (str == NULL)
3467
196k
        return;
3468
3469
322k
    depth += 1;
3470
322k
    if (depth > maxDepth) {
3471
7
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3472
7
                       "Maximum entity nesting depth exceeded");
3473
7
  return;
3474
7
    }
3475
3476
322k
    end = str + length;
3477
322k
    chunk = str;
3478
3479
1.54G
    while ((str < end) && (!PARSER_STOPPED(ctxt))) {
3480
1.54G
        c = *str;
3481
3482
1.54G
        if (c >= 0x80) {
3483
1.35G
            l = xmlUTF8MultibyteLen(ctxt, str,
3484
1.35G
                    "invalid character in entity value\n");
3485
1.35G
            if (l == 0) {
3486
208M
                if (chunk < str)
3487
421k
                    xmlSBufAddString(buf, chunk, str - chunk);
3488
208M
                xmlSBufAddReplChar(buf);
3489
208M
                str += 1;
3490
208M
                chunk = str;
3491
1.14G
            } else {
3492
1.14G
                str += l;
3493
1.14G
            }
3494
1.35G
        } else if (c == '&') {
3495
384k
            if (str[1] == '#') {
3496
105k
                if (chunk < str)
3497
68.3k
                    xmlSBufAddString(buf, chunk, str - chunk);
3498
3499
105k
                c = xmlParseStringCharRef(ctxt, &str);
3500
105k
                if (c == 0)
3501
17.7k
                    return;
3502
3503
87.5k
                xmlSBufAddChar(buf, c);
3504
3505
87.5k
                chunk = str;
3506
279k
            } else {
3507
279k
                xmlChar *name;
3508
3509
                /*
3510
                 * General entity references are checked for
3511
                 * syntactic validity.
3512
                 */
3513
279k
                str++;
3514
279k
                name = xmlParseStringName(ctxt, &str);
3515
3516
279k
                if ((name == NULL) || (*str++ != ';')) {
3517
26.5k
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR,
3518
26.5k
                            "EntityValue: '&' forbidden except for entities "
3519
26.5k
                            "references\n");
3520
26.5k
                    xmlFree(name);
3521
26.5k
                    return;
3522
26.5k
                }
3523
3524
253k
                xmlFree(name);
3525
253k
            }
3526
193M
        } else if (c == '%') {
3527
314k
            xmlEntityPtr ent;
3528
3529
314k
            if (chunk < str)
3530
125k
                xmlSBufAddString(buf, chunk, str - chunk);
3531
3532
314k
            ent = xmlParseStringPEReference(ctxt, &str);
3533
314k
            if (ent == NULL)
3534
68.6k
                return;
3535
3536
245k
            if (!PARSER_EXTERNAL(ctxt)) {
3537
1.91k
                xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL);
3538
1.91k
                return;
3539
1.91k
            }
3540
3541
243k
            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
197k
                if (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
3549
197k
                    ((ctxt->replaceEntities) ||
3550
194k
                     (ctxt->validate))) {
3551
194k
                    xmlLoadEntityContent(ctxt, ent);
3552
194k
                } else {
3553
3.70k
                    xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING,
3554
3.70k
                                  "not validating will not read content for "
3555
3.70k
                                  "PE entity %s\n", ent->name, NULL);
3556
3.70k
                }
3557
197k
            }
3558
3559
            /*
3560
             * TODO: Skip if ent->content is still NULL.
3561
             */
3562
3563
243k
            if (xmlParserEntityCheck(ctxt, ent->length))
3564
48
                return;
3565
3566
243k
            if (ent->flags & XML_ENT_EXPANDING) {
3567
492
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3568
492
                return;
3569
492
            }
3570
3571
243k
            ent->flags |= XML_ENT_EXPANDING;
3572
243k
            xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length,
3573
243k
                                      depth);
3574
243k
            ent->flags &= ~XML_ENT_EXPANDING;
3575
3576
243k
            chunk = str;
3577
192M
        } else {
3578
            /* Normal ASCII char */
3579
192M
            if (!IS_BYTE_CHAR(c)) {
3580
17.9M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3581
17.9M
                        "invalid character in entity value\n");
3582
17.9M
                if (chunk < str)
3583
64.6k
                    xmlSBufAddString(buf, chunk, str - chunk);
3584
17.9M
                xmlSBufAddReplChar(buf);
3585
17.9M
                str += 1;
3586
17.9M
                chunk = str;
3587
175M
            } else {
3588
175M
                str += 1;
3589
175M
            }
3590
192M
        }
3591
1.54G
    }
3592
3593
207k
    if (chunk < str)
3594
174k
        xmlSBufAddString(buf, chunk, str - chunk);
3595
207k
}
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
286k
xmlParseEntityValue(xmlParserCtxt *ctxt, xmlChar **orig) {
3611
286k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3612
90.6k
                         XML_MAX_HUGE_LENGTH :
3613
286k
                         XML_MAX_TEXT_LENGTH;
3614
286k
    xmlSBuf buf;
3615
286k
    const xmlChar *start;
3616
286k
    int quote, length;
3617
3618
286k
    xmlSBufInit(&buf, maxLength);
3619
3620
286k
    GROW;
3621
3622
286k
    quote = CUR;
3623
286k
    if ((quote != '"') && (quote != '\'')) {
3624
0
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
3625
0
  return(NULL);
3626
0
    }
3627
286k
    CUR_PTR++;
3628
3629
286k
    length = 0;
3630
3631
    /*
3632
     * Copy raw content of the entity into a buffer
3633
     */
3634
2.39G
    while (1) {
3635
2.39G
        int c;
3636
3637
2.39G
        if (PARSER_STOPPED(ctxt))
3638
40
            goto error;
3639
3640
2.39G
        if (CUR_PTR >= ctxt->input->end) {
3641
9.70k
            xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL);
3642
9.70k
            goto error;
3643
9.70k
        }
3644
3645
2.39G
        c = CUR;
3646
3647
2.39G
        if (c == 0) {
3648
657
            xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3649
657
                    "invalid character in entity value\n");
3650
657
            goto error;
3651
657
        }
3652
2.39G
        if (c == quote)
3653
275k
            break;
3654
2.39G
        NEXTL(1);
3655
2.39G
        length += 1;
3656
3657
        /*
3658
         * TODO: Check growth threshold
3659
         */
3660
2.39G
        if (ctxt->input->end - CUR_PTR < 10)
3661
408k
            GROW;
3662
2.39G
    }
3663
3664
275k
    start = CUR_PTR - length;
3665
3666
275k
    if (orig != NULL) {
3667
275k
        *orig = xmlStrndup(start, length);
3668
275k
        if (*orig == NULL)
3669
151
            xmlErrMemory(ctxt);
3670
275k
    }
3671
3672
275k
    xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr);
3673
3674
275k
    NEXTL(1);
3675
3676
275k
    return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long"));
3677
3678
10.4k
error:
3679
10.4k
    xmlSBufCleanup(&buf, ctxt, "entity length too long");
3680
10.4k
    return(NULL);
3681
286k
}
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
26.3k
xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) {
3693
26.3k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3694
26.3k
    const xmlChar *str;
3695
26.3k
    unsigned long expandedSize = pent->length;
3696
26.3k
    int c, flags;
3697
3698
26.3k
    depth += 1;
3699
26.3k
    if (depth > maxDepth) {
3700
11
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3701
11
                       "Maximum entity nesting depth exceeded");
3702
11
  return;
3703
11
    }
3704
3705
26.3k
    if (pent->flags & XML_ENT_EXPANDING) {
3706
599
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3707
599
        return;
3708
599
    }
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
25.7k
    if (ctxt->inSubset == 0)
3716
25.0k
        flags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
3717
779
    else
3718
779
        flags = XML_ENT_VALIDATED;
3719
3720
25.7k
    str = pent->content;
3721
25.7k
    if (str == NULL)
3722
132
        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
276M
    while (!PARSER_STOPPED(ctxt)) {
3731
276M
        c = *str;
3732
3733
276M
  if (c != '&') {
3734
276M
            if (c == 0)
3735
24.3k
                break;
3736
3737
276M
            if (c == '<')
3738
11.5k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3739
11.5k
                        "'<' in entity '%s' is not allowed in attributes "
3740
11.5k
                        "values\n", pent->name);
3741
3742
276M
            str += 1;
3743
276M
        } else if (str[1] == '#') {
3744
7.33k
            int val;
3745
3746
7.33k
      val = xmlParseStringCharRef(ctxt, &str);
3747
7.33k
      if (val == 0) {
3748
57
                pent->content[0] = 0;
3749
57
                break;
3750
57
            }
3751
70.6k
  } else {
3752
70.6k
            xmlChar *name;
3753
70.6k
            xmlEntityPtr ent;
3754
3755
70.6k
      name = xmlParseStringEntityRef(ctxt, &str);
3756
70.6k
      if (name == NULL) {
3757
96
                pent->content[0] = 0;
3758
96
                break;
3759
96
            }
3760
3761
70.5k
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3762
70.5k
            xmlFree(name);
3763
3764
70.5k
            if ((ent != NULL) &&
3765
63.5k
                (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) {
3766
41.0k
                if ((ent->flags & flags) != flags) {
3767
19.2k
                    pent->flags |= XML_ENT_EXPANDING;
3768
19.2k
                    xmlCheckEntityInAttValue(ctxt, ent, depth);
3769
19.2k
                    pent->flags &= ~XML_ENT_EXPANDING;
3770
19.2k
                }
3771
3772
41.0k
                xmlSaturatedAdd(&expandedSize, ent->expandedSize);
3773
41.0k
                xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST);
3774
41.0k
            }
3775
70.5k
        }
3776
276M
    }
3777
3778
25.7k
done:
3779
25.7k
    if (ctxt->inSubset == 0)
3780
25.0k
        pent->expandedSize = expandedSize;
3781
3782
25.7k
    pent->flags |= flags;
3783
25.7k
}
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
4.47M
                          int *inSpace, int depth, int check) {
3803
4.47M
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3804
4.47M
    int c, chunkSize;
3805
4.47M
    int normChange = 0;
3806
3807
4.47M
    if (str == NULL)
3808
1.08k
        return(0);
3809
3810
4.47M
    depth += 1;
3811
4.47M
    if (depth > maxDepth) {
3812
12
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3813
12
                       "Maximum entity nesting depth exceeded");
3814
12
  return(0);
3815
12
    }
3816
3817
4.47M
    if (pent != NULL) {
3818
4.39M
        if (pent->flags & XML_ENT_EXPANDING) {
3819
282
            xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3820
282
            return(0);
3821
282
        }
3822
3823
4.39M
        if (check) {
3824
4.37M
            if (xmlParserEntityCheck(ctxt, pent->length))
3825
1.27k
                return(0);
3826
4.37M
        }
3827
4.39M
    }
3828
3829
4.47M
    chunkSize = 0;
3830
3831
    /*
3832
     * Note that entity values are already validated. No special
3833
     * handling for multi-byte characters is needed.
3834
     */
3835
13.5G
    while (!PARSER_STOPPED(ctxt)) {
3836
13.5G
        c = *str;
3837
3838
13.5G
  if (c != '&') {
3839
13.5G
            if (c == 0)
3840
4.17M
                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
13.5G
            if ((pent != NULL) && (c == '<')) {
3848
292k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3849
292k
                        "'<' in entity '%s' is not allowed in attributes "
3850
292k
                        "values\n", pent->name);
3851
292k
                break;
3852
292k
            }
3853
3854
13.5G
            if (c <= 0x20) {
3855
113M
                if ((normalize) && (*inSpace)) {
3856
                    /* Skip char */
3857
1.85M
                    if (chunkSize > 0) {
3858
439k
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3859
439k
                        chunkSize = 0;
3860
439k
                    }
3861
1.85M
                    normChange = 1;
3862
112M
                } else if (c < 0x20) {
3863
95.0M
                    if (chunkSize > 0) {
3864
1.84M
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3865
1.84M
                        chunkSize = 0;
3866
1.84M
                    }
3867
3868
95.0M
                    xmlSBufAddCString(buf, " ", 1);
3869
95.0M
                } else {
3870
17.0M
                    chunkSize += 1;
3871
17.0M
                }
3872
3873
113M
                *inSpace = 1;
3874
13.4G
            } else {
3875
13.4G
                chunkSize += 1;
3876
13.4G
                *inSpace = 0;
3877
13.4G
            }
3878
3879
13.5G
            str += 1;
3880
13.5G
        } else if (str[1] == '#') {
3881
2.08M
            int val;
3882
3883
2.08M
            if (chunkSize > 0) {
3884
2.02M
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3885
2.02M
                chunkSize = 0;
3886
2.02M
            }
3887
3888
2.08M
      val = xmlParseStringCharRef(ctxt, &str);
3889
2.08M
      if (val == 0) {
3890
72
                if (pent != NULL)
3891
72
                    pent->content[0] = 0;
3892
72
                break;
3893
72
            }
3894
3895
2.08M
            if (val == ' ') {
3896
46.7k
                if ((normalize) && (*inSpace))
3897
10.8k
                    normChange = 1;
3898
35.8k
                else
3899
35.8k
                    xmlSBufAddCString(buf, " ", 1);
3900
46.7k
                *inSpace = 1;
3901
2.03M
            } else {
3902
2.03M
                xmlSBufAddChar(buf, val);
3903
2.03M
                *inSpace = 0;
3904
2.03M
            }
3905
6.26M
  } else {
3906
6.26M
            xmlChar *name;
3907
6.26M
            xmlEntityPtr ent;
3908
3909
6.26M
            if (chunkSize > 0) {
3910
3.07M
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3911
3.07M
                chunkSize = 0;
3912
3.07M
            }
3913
3914
6.26M
      name = xmlParseStringEntityRef(ctxt, &str);
3915
6.26M
            if (name == NULL) {
3916
207
                if (pent != NULL)
3917
196
                    pent->content[0] = 0;
3918
207
                break;
3919
207
            }
3920
3921
6.26M
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3922
6.26M
            xmlFree(name);
3923
3924
6.26M
      if ((ent != NULL) &&
3925
5.90M
    (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
3926
2.60M
    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
2.60M
                xmlSBufAddString(buf, ent->content, ent->length);
3933
3934
2.60M
                *inSpace = 0;
3935
3.65M
      } else if ((ent != NULL) && (ent->content != NULL)) {
3936
3.29M
                if (pent != NULL)
3937
3.28M
                    pent->flags |= XML_ENT_EXPANDING;
3938
3.29M
    normChange |= xmlExpandEntityInAttValue(ctxt, buf,
3939
3.29M
                        ent->content, ent, normalize, inSpace, depth, check);
3940
3.29M
                if (pent != NULL)
3941
3.28M
                    pent->flags &= ~XML_ENT_EXPANDING;
3942
3.29M
      }
3943
6.26M
        }
3944
13.5G
    }
3945
3946
4.47M
    if (chunkSize > 0)
3947
2.67M
        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3948
3949
4.47M
    return(normChange);
3950
4.47M
}
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
78.6k
                            int normalize) {
3964
78.6k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3965
52.8k
                         XML_MAX_HUGE_LENGTH :
3966
78.6k
                         XML_MAX_TEXT_LENGTH;
3967
78.6k
    xmlSBuf buf;
3968
78.6k
    int inSpace = 1;
3969
3970
78.6k
    xmlSBufInit(&buf, maxLength);
3971
3972
78.6k
    xmlExpandEntityInAttValue(ctxt, &buf, str, NULL, normalize, &inSpace,
3973
78.6k
                              ctxt->inputNr, /* check */ 0);
3974
3975
78.6k
    if ((normalize) && (inSpace) && (buf.size > 0))
3976
0
        buf.size--;
3977
3978
78.6k
    return(xmlSBufFinish(&buf, NULL, ctxt, "AttValue length too long"));
3979
78.6k
}
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
2.90M
                         int special, int isNamespace) {
4020
2.90M
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4021
698k
                         XML_MAX_HUGE_LENGTH :
4022
2.90M
                         XML_MAX_TEXT_LENGTH;
4023
2.90M
    xmlSBuf buf;
4024
2.90M
    xmlChar *ret;
4025
2.90M
    int c, l, quote, entFlags, chunkSize;
4026
2.90M
    int inSpace = 1;
4027
2.90M
    int replaceEntities;
4028
2.90M
    int normalize = (special & XML_SPECIAL_TYPE_MASK) > XML_ATTRIBUTE_CDATA;
4029
2.90M
    int attvalFlags = 0;
4030
4031
    /* Always expand namespace URIs */
4032
2.90M
    replaceEntities = (ctxt->replaceEntities) || (isNamespace);
4033
4034
2.90M
    xmlSBufInit(&buf, maxLength);
4035
4036
2.90M
    GROW;
4037
4038
2.90M
    quote = CUR;
4039
2.90M
    if ((quote != '"') && (quote != '\'')) {
4040
44.1k
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
4041
44.1k
  return(NULL);
4042
44.1k
    }
4043
2.86M
    NEXTL(1);
4044
4045
2.86M
    if (ctxt->inSubset == 0)
4046
2.56M
        entFlags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
4047
294k
    else
4048
294k
        entFlags = XML_ENT_VALIDATED;
4049
4050
2.86M
    inSpace = 1;
4051
2.86M
    chunkSize = 0;
4052
4053
1.01G
    while (1) {
4054
1.01G
        if (PARSER_STOPPED(ctxt))
4055
2.84k
            goto error;
4056
4057
1.01G
        if (CUR_PTR >= ctxt->input->end) {
4058
69.0k
            xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED,
4059
69.0k
                           "AttValue: ' expected\n");
4060
69.0k
            goto error;
4061
69.0k
        }
4062
4063
        /*
4064
         * TODO: Check growth threshold
4065
         */
4066
1.01G
        if (ctxt->input->end - CUR_PTR < 10)
4067
935k
            GROW;
4068
4069
1.01G
        c = CUR;
4070
4071
1.01G
        if (c >= 0x80) {
4072
653M
            l = xmlUTF8MultibyteLen(ctxt, CUR_PTR,
4073
653M
                    "invalid character in attribute value\n");
4074
653M
            if (l == 0) {
4075
254M
                if (chunkSize > 0) {
4076
661k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4077
661k
                    chunkSize = 0;
4078
661k
                }
4079
254M
                xmlSBufAddReplChar(&buf);
4080
254M
                NEXTL(1);
4081
398M
            } else {
4082
398M
                chunkSize += l;
4083
398M
                NEXTL(l);
4084
398M
            }
4085
4086
653M
            inSpace = 0;
4087
653M
        } else if (c != '&') {
4088
359M
            if (c > 0x20) {
4089
126M
                if (c == quote)
4090
2.78M
                    break;
4091
4092
123M
                if (c == '<')
4093
1.40M
                    xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL);
4094
4095
123M
                chunkSize += 1;
4096
123M
                inSpace = 0;
4097
232M
            } else if (!IS_BYTE_CHAR(c)) {
4098
193M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
4099
193M
                        "invalid character in attribute value\n");
4100
193M
                if (chunkSize > 0) {
4101
319k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4102
319k
                    chunkSize = 0;
4103
319k
                }
4104
193M
                xmlSBufAddReplChar(&buf);
4105
193M
                inSpace = 0;
4106
193M
            } else {
4107
                /* Whitespace */
4108
39.3M
                if ((normalize) && (inSpace)) {
4109
                    /* Skip char */
4110
2.34M
                    if (chunkSize > 0) {
4111
63.8k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4112
63.8k
                        chunkSize = 0;
4113
63.8k
                    }
4114
2.34M
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4115
36.9M
                } else if (c < 0x20) {
4116
                    /* Convert to space */
4117
32.7M
                    if (chunkSize > 0) {
4118
1.04M
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4119
1.04M
                        chunkSize = 0;
4120
1.04M
                    }
4121
4122
32.7M
                    xmlSBufAddCString(&buf, " ", 1);
4123
32.7M
                } else {
4124
4.19M
                    chunkSize += 1;
4125
4.19M
                }
4126
4127
39.3M
                inSpace = 1;
4128
4129
39.3M
                if ((c == 0xD) && (NXT(1) == 0xA))
4130
85.7k
                    CUR_PTR++;
4131
39.3M
            }
4132
4133
356M
            NEXTL(1);
4134
356M
        } else if (NXT(1) == '#') {
4135
264k
            int val;
4136
4137
264k
            if (chunkSize > 0) {
4138
169k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4139
169k
                chunkSize = 0;
4140
169k
            }
4141
4142
264k
            val = xmlParseCharRef(ctxt);
4143
264k
            if (val == 0)
4144
7.41k
                goto error;
4145
4146
257k
            if ((val == '&') && (!replaceEntities)) {
4147
                /*
4148
                 * The reparsing will be done in xmlNodeParseContent()
4149
                 * called from SAX2.c
4150
                 */
4151
18.3k
                xmlSBufAddCString(&buf, "&#38;", 5);
4152
18.3k
                inSpace = 0;
4153
239k
            } else if (val == ' ') {
4154
32.4k
                if ((normalize) && (inSpace))
4155
3.65k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4156
28.7k
                else
4157
28.7k
                    xmlSBufAddCString(&buf, " ", 1);
4158
32.4k
                inSpace = 1;
4159
206k
            } else {
4160
206k
                xmlSBufAddChar(&buf, val);
4161
206k
                inSpace = 0;
4162
206k
            }
4163
2.83M
        } else {
4164
2.83M
            const xmlChar *name;
4165
2.83M
            xmlEntityPtr ent;
4166
4167
2.83M
            if (chunkSize > 0) {
4168
1.07M
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4169
1.07M
                chunkSize = 0;
4170
1.07M
            }
4171
4172
2.83M
            name = xmlParseEntityRefInternal(ctxt);
4173
2.83M
            if (name == NULL) {
4174
                /*
4175
                 * Probably a literal '&' which wasn't escaped.
4176
                 * TODO: Handle gracefully in recovery mode.
4177
                 */
4178
314k
                continue;
4179
314k
            }
4180
4181
2.51M
            ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1);
4182
2.51M
            if (ent == NULL)
4183
349k
                continue;
4184
4185
2.16M
            if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) {
4186
333k
                if ((ent->content[0] == '&') && (!replaceEntities))
4187
51.7k
                    xmlSBufAddCString(&buf, "&#38;", 5);
4188
281k
                else
4189
281k
                    xmlSBufAddString(&buf, ent->content, ent->length);
4190
333k
                inSpace = 0;
4191
1.83M
            } else if (replaceEntities) {
4192
1.09M
                if (xmlExpandEntityInAttValue(ctxt, &buf,
4193
1.09M
                        ent->content, ent, normalize, &inSpace, ctxt->inputNr,
4194
1.09M
                        /* check */ 1) > 0)
4195
160k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4196
1.09M
            } else {
4197
738k
                if ((ent->flags & entFlags) != entFlags)
4198
7.16k
                    xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr);
4199
4200
738k
                if (xmlParserEntityCheck(ctxt, ent->expandedSize)) {
4201
676
                    ent->content[0] = 0;
4202
676
                    goto error;
4203
676
                }
4204
4205
                /*
4206
                 * Just output the reference
4207
                 */
4208
737k
                xmlSBufAddCString(&buf, "&", 1);
4209
737k
                xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name));
4210
737k
                xmlSBufAddCString(&buf, ";", 1);
4211
4212
737k
                inSpace = 0;
4213
737k
            }
4214
2.16M
  }
4215
1.01G
    }
4216
4217
2.78M
    if ((buf.mem == NULL) && (outFlags != NULL)) {
4218
2.02M
        ret = (xmlChar *) CUR_PTR - chunkSize;
4219
4220
2.02M
        if (attlen != NULL)
4221
2.02M
            *attlen = chunkSize;
4222
2.02M
        if ((normalize) && (inSpace) && (chunkSize > 0)) {
4223
1.90k
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4224
1.90k
            *attlen -= 1;
4225
1.90k
        }
4226
4227
        /* Report potential error */
4228
2.02M
        xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4229
2.02M
    } else {
4230
757k
        if (chunkSize > 0)
4231
569k
            xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4232
4233
757k
        if ((normalize) && (inSpace) && (buf.size > 0)) {
4234
6.96k
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4235
6.96k
            buf.size--;
4236
6.96k
        }
4237
4238
757k
        ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long");
4239
757k
        attvalFlags |= XML_ATTVAL_ALLOC;
4240
4241
757k
        if (ret != NULL) {
4242
757k
            if (attlen != NULL)
4243
293k
                *attlen = buf.size;
4244
757k
        }
4245
757k
    }
4246
4247
2.78M
    if (outFlags != NULL)
4248
2.31M
        *outFlags = attvalFlags;
4249
4250
2.78M
    NEXTL(1);
4251
4252
2.78M
    return(ret);
4253
4254
79.9k
error:
4255
79.9k
    xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4256
79.9k
    return(NULL);
4257
2.86M
}
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
548k
xmlParseAttValue(xmlParserCtxt *ctxt) {
4297
548k
    if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL);
4298
548k
    return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0));
4299
548k
}
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
214k
xmlParseSystemLiteral(xmlParserCtxt *ctxt) {
4314
214k
    xmlChar *buf = NULL;
4315
214k
    int len = 0;
4316
214k
    int size = XML_PARSER_BUFFER_SIZE;
4317
214k
    int cur, l;
4318
214k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4319
54.4k
                    XML_MAX_TEXT_LENGTH :
4320
214k
                    XML_MAX_NAME_LENGTH;
4321
214k
    xmlChar stop;
4322
4323
214k
    if (RAW == '"') {
4324
160k
        NEXT;
4325
160k
  stop = '"';
4326
160k
    } else if (RAW == '\'') {
4327
41.5k
        NEXT;
4328
41.5k
  stop = '\'';
4329
41.5k
    } else {
4330
12.6k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4331
12.6k
  return(NULL);
4332
12.6k
    }
4333
4334
202k
    buf = xmlMalloc(size);
4335
202k
    if (buf == NULL) {
4336
84
        xmlErrMemory(ctxt);
4337
84
  return(NULL);
4338
84
    }
4339
201k
    cur = xmlCurrentCharRecover(ctxt, &l);
4340
47.9M
    while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */
4341
47.7M
  if (len + 5 >= size) {
4342
30.9k
      xmlChar *tmp;
4343
30.9k
            int newSize;
4344
4345
30.9k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4346
30.9k
            if (newSize < 0) {
4347
13
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral");
4348
13
                xmlFree(buf);
4349
13
                return(NULL);
4350
13
            }
4351
30.9k
      tmp = xmlRealloc(buf, newSize);
4352
30.9k
      if (tmp == NULL) {
4353
22
          xmlFree(buf);
4354
22
    xmlErrMemory(ctxt);
4355
22
    return(NULL);
4356
22
      }
4357
30.9k
      buf = tmp;
4358
30.9k
            size = newSize;
4359
30.9k
  }
4360
47.7M
  COPY_BUF(buf, len, cur);
4361
47.7M
  NEXTL(l);
4362
47.7M
  cur = xmlCurrentCharRecover(ctxt, &l);
4363
47.7M
    }
4364
201k
    buf[len] = 0;
4365
201k
    if (!IS_CHAR(cur)) {
4366
13.7k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4367
188k
    } else {
4368
188k
  NEXT;
4369
188k
    }
4370
201k
    return(buf);
4371
201k
}
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
94.9k
xmlParsePubidLiteral(xmlParserCtxt *ctxt) {
4386
94.9k
    xmlChar *buf = NULL;
4387
94.9k
    int len = 0;
4388
94.9k
    int size = XML_PARSER_BUFFER_SIZE;
4389
94.9k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4390
23.7k
                    XML_MAX_TEXT_LENGTH :
4391
94.9k
                    XML_MAX_NAME_LENGTH;
4392
94.9k
    xmlChar cur;
4393
94.9k
    xmlChar stop;
4394
4395
94.9k
    if (RAW == '"') {
4396
62.7k
        NEXT;
4397
62.7k
  stop = '"';
4398
62.7k
    } else if (RAW == '\'') {
4399
25.9k
        NEXT;
4400
25.9k
  stop = '\'';
4401
25.9k
    } else {
4402
6.18k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4403
6.18k
  return(NULL);
4404
6.18k
    }
4405
88.7k
    buf = xmlMalloc(size);
4406
88.7k
    if (buf == NULL) {
4407
48
  xmlErrMemory(ctxt);
4408
48
  return(NULL);
4409
48
    }
4410
88.7k
    cur = CUR;
4411
2.99M
    while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) &&
4412
2.90M
           (PARSER_STOPPED(ctxt) == 0)) { /* checked */
4413
2.90M
  if (len + 1 >= size) {
4414
3.47k
      xmlChar *tmp;
4415
3.47k
            int newSize;
4416
4417
3.47k
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4418
3.47k
            if (newSize < 0) {
4419
12
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID");
4420
12
                xmlFree(buf);
4421
12
                return(NULL);
4422
12
            }
4423
3.45k
      tmp = xmlRealloc(buf, newSize);
4424
3.45k
      if (tmp == NULL) {
4425
13
    xmlErrMemory(ctxt);
4426
13
    xmlFree(buf);
4427
13
    return(NULL);
4428
13
      }
4429
3.44k
      buf = tmp;
4430
3.44k
            size = newSize;
4431
3.44k
  }
4432
2.90M
  buf[len++] = cur;
4433
2.90M
  NEXT;
4434
2.90M
  cur = CUR;
4435
2.90M
    }
4436
88.6k
    buf[len] = 0;
4437
88.6k
    if (cur != stop) {
4438
13.1k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4439
75.5k
    } else {
4440
75.5k
  NEXTL(1);
4441
75.5k
    }
4442
88.6k
    return(buf);
4443
88.7k
}
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
11.5M
              int isBlank) {
4488
11.5M
    int checkBlanks;
4489
4490
11.5M
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
4491
2.09M
        return;
4492
4493
9.42M
    checkBlanks = (!ctxt->keepBlanks) ||
4494
7.35M
                  (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
9.42M
    if ((checkBlanks) &&
4502
2.06M
        (areBlanks(ctxt, buf, size, isBlank))) {
4503
47.5k
        if ((ctxt->sax->ignorableWhitespace != NULL) &&
4504
47.5k
            (ctxt->keepBlanks))
4505
0
            ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size);
4506
9.38M
    } else {
4507
9.38M
        if (ctxt->sax->characters != NULL)
4508
9.38M
            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
9.38M
        if ((checkBlanks) && (*ctxt->space == -1))
4515
378k
            *ctxt->space = -2;
4516
9.38M
    }
4517
9.42M
}
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
81.0M
xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) {
4534
81.0M
    const xmlChar *in;
4535
81.0M
    int line = ctxt->input->line;
4536
81.0M
    int col = ctxt->input->col;
4537
81.0M
    int ccol;
4538
81.0M
    int terminate = 0;
4539
4540
81.0M
    GROW;
4541
    /*
4542
     * Accelerated common case where input don't need to be
4543
     * modified before passing it to the handler.
4544
     */
4545
81.0M
    in = ctxt->input->cur;
4546
81.1M
    do {
4547
82.1M
get_more_space:
4548
85.5M
        while (*in == 0x20) { in++; ctxt->input->col++; }
4549
82.1M
        if (*in == 0xA) {
4550
28.7M
            do {
4551
28.7M
                ctxt->input->line++; ctxt->input->col = 1;
4552
28.7M
                in++;
4553
28.7M
            } while (*in == 0xA);
4554
996k
            goto get_more_space;
4555
996k
        }
4556
81.1M
        if (*in == '<') {
4557
1.50M
            while (in > ctxt->input->cur) {
4558
752k
                const xmlChar *tmp = ctxt->input->cur;
4559
752k
                size_t nbchar = in - tmp;
4560
4561
752k
                if (nbchar > XML_MAX_ITEMS)
4562
0
                    nbchar = XML_MAX_ITEMS;
4563
752k
                ctxt->input->cur += nbchar;
4564
4565
752k
                xmlCharacters(ctxt, tmp, nbchar, 1);
4566
752k
            }
4567
752k
            return;
4568
752k
        }
4569
4570
82.5M
get_more:
4571
82.5M
        ccol = ctxt->input->col;
4572
146M
        while (test_char_data[*in]) {
4573
63.8M
            in++;
4574
63.8M
            ccol++;
4575
63.8M
        }
4576
82.5M
        ctxt->input->col = ccol;
4577
82.5M
        if (*in == 0xA) {
4578
16.4M
            do {
4579
16.4M
                ctxt->input->line++; ctxt->input->col = 1;
4580
16.4M
                in++;
4581
16.4M
            } while (*in == 0xA);
4582
379k
            goto get_more;
4583
379k
        }
4584
82.2M
        if (*in == ']') {
4585
1.78M
            size_t avail = ctxt->input->end - in;
4586
4587
1.78M
            if (partial && avail < 2) {
4588
459
                terminate = 1;
4589
459
                goto invoke_callback;
4590
459
            }
4591
1.78M
            if (in[1] == ']') {
4592
1.36M
                if (partial && avail < 3) {
4593
1.86k
                    terminate = 1;
4594
1.86k
                    goto invoke_callback;
4595
1.86k
                }
4596
1.35M
                if (in[2] == '>')
4597
78.4k
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4598
1.35M
            }
4599
4600
1.78M
            in++;
4601
1.78M
            ctxt->input->col++;
4602
1.78M
            goto get_more;
4603
1.78M
        }
4604
4605
80.4M
invoke_callback:
4606
85.3M
        while (in > ctxt->input->cur) {
4607
4.96M
            const xmlChar *tmp = ctxt->input->cur;
4608
4.96M
            size_t nbchar = in - tmp;
4609
4610
4.96M
            if (nbchar > XML_MAX_ITEMS)
4611
0
                nbchar = XML_MAX_ITEMS;
4612
4.96M
            ctxt->input->cur += nbchar;
4613
4614
4.96M
            xmlCharacters(ctxt, tmp, nbchar, 0);
4615
4616
4.96M
            line = ctxt->input->line;
4617
4.96M
            col = ctxt->input->col;
4618
4.96M
        }
4619
80.4M
        ctxt->input->cur = in;
4620
80.4M
        if (*in == 0xD) {
4621
258k
            in++;
4622
258k
            if (*in == 0xA) {
4623
146k
                ctxt->input->cur = in;
4624
146k
                in++;
4625
146k
                ctxt->input->line++; ctxt->input->col = 1;
4626
146k
                continue; /* while */
4627
146k
            }
4628
112k
            in--;
4629
112k
        }
4630
80.2M
        if (*in == '<') {
4631
1.91M
            return;
4632
1.91M
        }
4633
78.3M
        if (*in == '&') {
4634
457k
            return;
4635
457k
        }
4636
77.8M
        if (terminate) {
4637
2.32k
            return;
4638
2.32k
        }
4639
77.8M
        SHRINK;
4640
77.8M
        GROW;
4641
77.8M
        in = ctxt->input->cur;
4642
78.0M
    } while (((*in >= 0x20) && (*in <= 0x7F)) ||
4643
77.9M
             (*in == 0x09) || (*in == 0x0a));
4644
77.8M
    ctxt->input->line = line;
4645
77.8M
    ctxt->input->col = col;
4646
77.8M
    xmlParseCharDataComplex(ctxt, partial);
4647
77.8M
}
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
77.8M
xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) {
4661
77.8M
    xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5];
4662
77.8M
    int nbchar = 0;
4663
77.8M
    int cur, l;
4664
4665
77.8M
    cur = xmlCurrentCharRecover(ctxt, &l);
4666
376M
    while ((cur != '<') && /* checked */
4667
376M
           (cur != '&') &&
4668
376M
     (IS_CHAR(cur))) {
4669
298M
        if (cur == ']') {
4670
2.91M
            size_t avail = ctxt->input->end - ctxt->input->cur;
4671
4672
2.91M
            if (partial && avail < 2)
4673
572
                break;
4674
2.90M
            if (NXT(1) == ']') {
4675
2.35M
                if (partial && avail < 3)
4676
2.85k
                    break;
4677
2.35M
                if (NXT(2) == '>')
4678
246k
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4679
2.35M
            }
4680
2.90M
        }
4681
4682
298M
  COPY_BUF(buf, nbchar, cur);
4683
  /* move current position before possible calling of ctxt->sax->characters */
4684
298M
  NEXTL(l);
4685
298M
  if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) {
4686
2.50M
      buf[nbchar] = 0;
4687
4688
2.50M
            xmlCharacters(ctxt, buf, nbchar, 0);
4689
2.50M
      nbchar = 0;
4690
2.50M
            SHRINK;
4691
2.50M
  }
4692
298M
  cur = xmlCurrentCharRecover(ctxt, &l);
4693
298M
    }
4694
77.8M
    if (nbchar != 0) {
4695
3.30M
        buf[nbchar] = 0;
4696
4697
3.30M
        xmlCharacters(ctxt, buf, nbchar, 0);
4698
3.30M
    }
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
77.8M
    if (ctxt->input->cur < ctxt->input->end) {
4707
77.8M
        if ((cur == 0) && (CUR != 0)) {
4708
5.53k
            if (partial == 0) {
4709
4.92k
                xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4710
4.92k
                        "Incomplete UTF-8 sequence starting with %02X\n", CUR);
4711
4.92k
                NEXTL(1);
4712
4.92k
            }
4713
77.8M
        } else if ((cur != '<') && (cur != '&') && (cur != ']')) {
4714
            /* Generate the error and skip the offending character */
4715
77.2M
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4716
77.2M
                              "PCDATA invalid Char value %d\n", cur);
4717
77.2M
            NEXTL(l);
4718
77.2M
        }
4719
77.8M
    }
4720
77.8M
}
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
347k
xmlParseExternalID(xmlParserCtxt *ctxt, xmlChar **publicId, int strict) {
4756
347k
    xmlChar *URI = NULL;
4757
4758
347k
    *publicId = NULL;
4759
347k
    if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) {
4760
134k
        SKIP(6);
4761
134k
  if (SKIP_BLANKS == 0) {
4762
5.58k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4763
5.58k
                     "Space required after 'SYSTEM'\n");
4764
5.58k
  }
4765
134k
  URI = xmlParseSystemLiteral(ctxt);
4766
134k
  if (URI == NULL) {
4767
2.61k
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4768
2.61k
        }
4769
213k
    } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) {
4770
94.9k
        SKIP(6);
4771
94.9k
  if (SKIP_BLANKS == 0) {
4772
9.72k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4773
9.72k
        "Space required after 'PUBLIC'\n");
4774
9.72k
  }
4775
94.9k
  *publicId = xmlParsePubidLiteral(ctxt);
4776
94.9k
  if (*publicId == NULL) {
4777
6.25k
      xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL);
4778
6.25k
  }
4779
94.9k
  if (strict) {
4780
      /*
4781
       * We don't handle [83] so "S SystemLiteral" is required.
4782
       */
4783
77.0k
      if (SKIP_BLANKS == 0) {
4784
14.4k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4785
14.4k
      "Space required after the Public Identifier\n");
4786
14.4k
      }
4787
77.0k
  } 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
17.8k
      if (SKIP_BLANKS == 0) return(NULL);
4795
8.26k
      if ((CUR != '\'') && (CUR != '"')) return(NULL);
4796
8.26k
  }
4797
80.6k
  URI = xmlParseSystemLiteral(ctxt);
4798
80.6k
  if (URI == NULL) {
4799
10.1k
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4800
10.1k
        }
4801
80.6k
    }
4802
333k
    return(URI);
4803
347k
}
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
263k
                       size_t len, size_t size) {
4820
263k
    int q, ql;
4821
263k
    int r, rl;
4822
263k
    int cur, l;
4823
263k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4824
79.1k
                    XML_MAX_HUGE_LENGTH :
4825
263k
                    XML_MAX_TEXT_LENGTH;
4826
4827
263k
    if (buf == NULL) {
4828
56.4k
        len = 0;
4829
56.4k
  size = XML_PARSER_BUFFER_SIZE;
4830
56.4k
  buf = xmlMalloc(size);
4831
56.4k
  if (buf == NULL) {
4832
146
      xmlErrMemory(ctxt);
4833
146
      return;
4834
146
  }
4835
56.4k
    }
4836
263k
    q = xmlCurrentCharRecover(ctxt, &ql);
4837
263k
    if (q == 0)
4838
17.0k
        goto not_terminated;
4839
246k
    if (!IS_CHAR(q)) {
4840
7.00k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4841
7.00k
                          "xmlParseComment: invalid xmlChar value %d\n",
4842
7.00k
                    q);
4843
7.00k
  xmlFree (buf);
4844
7.00k
  return;
4845
7.00k
    }
4846
239k
    NEXTL(ql);
4847
239k
    r = xmlCurrentCharRecover(ctxt, &rl);
4848
239k
    if (r == 0)
4849
3.43k
        goto not_terminated;
4850
235k
    if (!IS_CHAR(r)) {
4851
4.43k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4852
4.43k
                          "xmlParseComment: invalid xmlChar value %d\n",
4853
4.43k
                    r);
4854
4.43k
  xmlFree (buf);
4855
4.43k
  return;
4856
4.43k
    }
4857
231k
    NEXTL(rl);
4858
231k
    cur = xmlCurrentCharRecover(ctxt, &l);
4859
231k
    if (cur == 0)
4860
3.69k
        goto not_terminated;
4861
59.4M
    while (IS_CHAR(cur) && /* checked */
4862
59.4M
           ((cur != '>') ||
4863
59.2M
      (r != '-') || (q != '-'))) {
4864
59.2M
  if ((r == '-') && (q == '-')) {
4865
3.47M
      xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL);
4866
3.47M
  }
4867
59.2M
  if (len + 5 >= size) {
4868
101k
      xmlChar *tmp;
4869
101k
            int newSize;
4870
4871
101k
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4872
101k
            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
101k
      tmp = xmlRealloc(buf, newSize);
4879
101k
      if (tmp == NULL) {
4880
32
    xmlErrMemory(ctxt);
4881
32
    xmlFree(buf);
4882
32
    return;
4883
32
      }
4884
101k
      buf = tmp;
4885
101k
            size = newSize;
4886
101k
  }
4887
59.2M
  COPY_BUF(buf, len, q);
4888
4889
59.2M
  q = r;
4890
59.2M
  ql = rl;
4891
59.2M
  r = cur;
4892
59.2M
  rl = l;
4893
4894
59.2M
  NEXTL(l);
4895
59.2M
  cur = xmlCurrentCharRecover(ctxt, &l);
4896
4897
59.2M
    }
4898
227k
    buf[len] = 0;
4899
227k
    if (cur == 0) {
4900
33.2k
  xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4901
33.2k
                       "Comment not terminated \n<!--%.50s\n", buf);
4902
194k
    } else if (!IS_CHAR(cur)) {
4903
5.85k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4904
5.85k
                          "xmlParseComment: invalid xmlChar value %d\n",
4905
5.85k
                    cur);
4906
188k
    } else {
4907
188k
        NEXT;
4908
188k
  if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
4909
188k
      (!ctxt->disableSAX))
4910
172k
      ctxt->sax->comment(ctxt->userData, buf);
4911
188k
    }
4912
227k
    xmlFree(buf);
4913
227k
    return;
4914
24.1k
not_terminated:
4915
24.1k
    xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4916
24.1k
       "Comment not terminated\n", NULL);
4917
24.1k
    xmlFree(buf);
4918
24.1k
}
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
1.81M
xmlParseComment(xmlParserCtxt *ctxt) {
4933
1.81M
    xmlChar *buf = NULL;
4934
1.81M
    size_t size = XML_PARSER_BUFFER_SIZE;
4935
1.81M
    size_t len = 0;
4936
1.81M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4937
327k
                       XML_MAX_HUGE_LENGTH :
4938
1.81M
                       XML_MAX_TEXT_LENGTH;
4939
1.81M
    const xmlChar *in;
4940
1.81M
    size_t nbchar = 0;
4941
1.81M
    int ccol;
4942
4943
    /*
4944
     * Check that there is a comment right here.
4945
     */
4946
1.81M
    if ((RAW != '<') || (NXT(1) != '!'))
4947
0
        return;
4948
1.81M
    SKIP(2);
4949
1.81M
    if ((RAW != '-') || (NXT(1) != '-'))
4950
293
        return;
4951
1.81M
    SKIP(2);
4952
1.81M
    GROW;
4953
4954
    /*
4955
     * Accelerated common case where input don't need to be
4956
     * modified before passing it to the handler.
4957
     */
4958
1.81M
    in = ctxt->input->cur;
4959
1.81M
    do {
4960
1.81M
  if (*in == 0xA) {
4961
1.02M
      do {
4962
1.02M
    ctxt->input->line++; ctxt->input->col = 1;
4963
1.02M
    in++;
4964
1.02M
      } while (*in == 0xA);
4965
11.1k
  }
4966
7.77M
get_more:
4967
7.77M
        ccol = ctxt->input->col;
4968
40.7M
  while (((*in > '-') && (*in <= 0x7F)) ||
4969
15.6M
         ((*in >= 0x20) && (*in < '-')) ||
4970
33.0M
         (*in == 0x09)) {
4971
33.0M
        in++;
4972
33.0M
        ccol++;
4973
33.0M
  }
4974
7.77M
  ctxt->input->col = ccol;
4975
7.77M
  if (*in == 0xA) {
4976
1.52M
      do {
4977
1.52M
    ctxt->input->line++; ctxt->input->col = 1;
4978
1.52M
    in++;
4979
1.52M
      } while (*in == 0xA);
4980
87.6k
      goto get_more;
4981
87.6k
  }
4982
7.68M
  nbchar = in - ctxt->input->cur;
4983
  /*
4984
   * save current set of data
4985
   */
4986
7.68M
  if (nbchar > 0) {
4987
6.17M
            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
6.17M
            if (buf == NULL) {
4994
404k
                if ((*in == '-') && (in[1] == '-'))
4995
188k
                    size = nbchar + 1;
4996
215k
                else
4997
215k
                    size = XML_PARSER_BUFFER_SIZE + nbchar;
4998
404k
                buf = xmlMalloc(size);
4999
404k
                if (buf == NULL) {
5000
90
                    xmlErrMemory(ctxt);
5001
90
                    return;
5002
90
                }
5003
404k
                len = 0;
5004
5.77M
            } else if (len + nbchar + 1 >= size) {
5005
103k
                xmlChar *new_buf;
5006
103k
                size += len + nbchar + XML_PARSER_BUFFER_SIZE;
5007
103k
                new_buf = xmlRealloc(buf, size);
5008
103k
                if (new_buf == NULL) {
5009
17
                    xmlErrMemory(ctxt);
5010
17
                    xmlFree(buf);
5011
17
                    return;
5012
17
                }
5013
103k
                buf = new_buf;
5014
103k
            }
5015
6.17M
            memcpy(&buf[len], ctxt->input->cur, nbchar);
5016
6.17M
            len += nbchar;
5017
6.17M
            buf[len] = 0;
5018
6.17M
  }
5019
7.68M
  ctxt->input->cur = in;
5020
7.68M
  if (*in == 0xA) {
5021
0
      in++;
5022
0
      ctxt->input->line++; ctxt->input->col = 1;
5023
0
  }
5024
7.68M
  if (*in == 0xD) {
5025
123k
      in++;
5026
123k
      if (*in == 0xA) {
5027
41.1k
    ctxt->input->cur = in;
5028
41.1k
    in++;
5029
41.1k
    ctxt->input->line++; ctxt->input->col = 1;
5030
41.1k
    goto get_more;
5031
41.1k
      }
5032
81.9k
      in--;
5033
81.9k
  }
5034
7.64M
  SHRINK;
5035
7.64M
  GROW;
5036
7.64M
  in = ctxt->input->cur;
5037
7.64M
  if (*in == '-') {
5038
7.37M
      if (in[1] == '-') {
5039
5.12M
          if (in[2] == '>') {
5040
1.55M
        SKIP(3);
5041
1.55M
        if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
5042
1.55M
            (!ctxt->disableSAX)) {
5043
1.50M
      if (buf != NULL)
5044
171k
          ctxt->sax->comment(ctxt->userData, buf);
5045
1.32M
      else
5046
1.32M
          ctxt->sax->comment(ctxt->userData, BAD_CAST "");
5047
1.50M
        }
5048
1.55M
        if (buf != NULL)
5049
197k
            xmlFree(buf);
5050
1.55M
        return;
5051
1.55M
    }
5052
3.56M
    if (buf != NULL) {
5053
3.52M
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5054
3.52M
                          "Double hyphen within comment: "
5055
3.52M
                                      "<!--%.50s\n",
5056
3.52M
              buf);
5057
3.52M
    } else
5058
43.9k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5059
43.9k
                          "Double hyphen within comment\n", NULL);
5060
3.56M
    in++;
5061
3.56M
    ctxt->input->col++;
5062
3.56M
      }
5063
5.82M
      in++;
5064
5.82M
      ctxt->input->col++;
5065
5.82M
      goto get_more;
5066
7.37M
  }
5067
7.64M
    } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a));
5068
263k
    xmlParseCommentComplex(ctxt, buf, len, size);
5069
263k
}
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
344k
xmlParsePITarget(xmlParserCtxt *ctxt) {
5085
344k
    const xmlChar *name;
5086
5087
344k
    name = xmlParseName(ctxt);
5088
344k
    if ((name != NULL) &&
5089
302k
        ((name[0] == 'x') || (name[0] == 'X')) &&
5090
148k
        ((name[1] == 'm') || (name[1] == 'M')) &&
5091
126k
        ((name[2] == 'l') || (name[2] == 'L'))) {
5092
58.5k
  int i;
5093
58.5k
  if ((name[0] == 'x') && (name[1] == 'm') &&
5094
55.0k
      (name[2] == 'l') && (name[3] == 0)) {
5095
48.7k
      xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5096
48.7k
     "XML declaration allowed only at the start of the document\n");
5097
48.7k
      return(name);
5098
48.7k
  } else if (name[3] == 0) {
5099
3.95k
      xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL);
5100
3.95k
      return(name);
5101
3.95k
  }
5102
15.5k
  for (i = 0;;i++) {
5103
15.5k
      if (xmlW3CPIs[i] == NULL) break;
5104
11.3k
      if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i]))
5105
1.65k
          return(name);
5106
11.3k
  }
5107
4.19k
  xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5108
4.19k
          "xmlParsePITarget: invalid name prefix 'xml'\n",
5109
4.19k
          NULL, NULL);
5110
4.19k
    }
5111
290k
    if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) {
5112
5.21k
  xmlNsErr(ctxt, XML_NS_ERR_COLON,
5113
5.21k
     "colons are forbidden from PI names '%s'\n", name, NULL, NULL);
5114
5.21k
    }
5115
290k
    return(name);
5116
344k
}
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
344k
xmlParsePI(xmlParserCtxt *ctxt) {
5199
344k
    xmlChar *buf = NULL;
5200
344k
    size_t len = 0;
5201
344k
    size_t size = XML_PARSER_BUFFER_SIZE;
5202
344k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
5203
115k
                       XML_MAX_HUGE_LENGTH :
5204
344k
                       XML_MAX_TEXT_LENGTH;
5205
344k
    int cur, l;
5206
344k
    const xmlChar *target;
5207
5208
344k
    if ((RAW == '<') && (NXT(1) == '?')) {
5209
  /*
5210
   * this is a Processing Instruction.
5211
   */
5212
344k
  SKIP(2);
5213
5214
  /*
5215
   * Parse the target name and check for special support like
5216
   * namespace.
5217
   */
5218
344k
        target = xmlParsePITarget(ctxt);
5219
344k
  if (target != NULL) {
5220
302k
      if ((RAW == '?') && (NXT(1) == '>')) {
5221
152k
    SKIP(2);
5222
5223
    /*
5224
     * SAX: PI detected.
5225
     */
5226
152k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5227
147k
        (ctxt->sax->processingInstruction != NULL))
5228
147k
        ctxt->sax->processingInstruction(ctxt->userData,
5229
147k
                                         target, NULL);
5230
152k
    return;
5231
152k
      }
5232
149k
      buf = xmlMalloc(size);
5233
149k
      if (buf == NULL) {
5234
186
    xmlErrMemory(ctxt);
5235
186
    return;
5236
186
      }
5237
149k
      if (SKIP_BLANKS == 0) {
5238
76.4k
    xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED,
5239
76.4k
        "ParsePI: PI %s space expected\n", target);
5240
76.4k
      }
5241
149k
      cur = xmlCurrentCharRecover(ctxt, &l);
5242
29.1M
      while (IS_CHAR(cur) && /* checked */
5243
29.1M
       ((cur != '?') || (NXT(1) != '>'))) {
5244
29.0M
    if (len + 5 >= size) {
5245
57.9k
        xmlChar *tmp;
5246
57.9k
                    int newSize;
5247
5248
57.9k
                    newSize = xmlGrowCapacity(size, 1, 1, maxLength);
5249
57.9k
                    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
57.9k
        tmp = xmlRealloc(buf, newSize);
5256
57.9k
        if (tmp == NULL) {
5257
36
      xmlErrMemory(ctxt);
5258
36
      xmlFree(buf);
5259
36
      return;
5260
36
        }
5261
57.8k
        buf = tmp;
5262
57.8k
                    size = newSize;
5263
57.8k
    }
5264
29.0M
    COPY_BUF(buf, len, cur);
5265
29.0M
    NEXTL(l);
5266
29.0M
    cur = xmlCurrentCharRecover(ctxt, &l);
5267
29.0M
      }
5268
149k
      buf[len] = 0;
5269
149k
      if (cur != '?') {
5270
50.6k
    xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5271
50.6k
          "ParsePI: PI %s never end ...\n", target);
5272
99.1k
      } else {
5273
99.1k
    SKIP(2);
5274
5275
99.1k
#ifdef LIBXML_CATALOG_ENABLED
5276
99.1k
    if ((ctxt->inSubset == 0) &&
5277
82.7k
        (xmlStrEqual(target, XML_CATALOG_PI))) {
5278
2.33k
        xmlCatalogAllow allow = xmlCatalogGetDefaults();
5279
5280
2.33k
        if ((ctxt->options & XML_PARSE_CATALOG_PI) &&
5281
900
                        ((allow == XML_CATA_ALLOW_DOCUMENT) ||
5282
900
       (allow == XML_CATA_ALLOW_ALL)))
5283
0
      xmlParseCatalogPI(ctxt, buf);
5284
2.33k
    }
5285
99.1k
#endif
5286
5287
    /*
5288
     * SAX: PI detected.
5289
     */
5290
99.1k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5291
75.9k
        (ctxt->sax->processingInstruction != NULL))
5292
75.9k
        ctxt->sax->processingInstruction(ctxt->userData,
5293
75.9k
                                         target, buf);
5294
99.1k
      }
5295
149k
      xmlFree(buf);
5296
149k
  } else {
5297
42.5k
      xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL);
5298
42.5k
  }
5299
344k
    }
5300
344k
}
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
43.7k
xmlParseNotationDecl(xmlParserCtxt *ctxt) {
5323
43.7k
    const xmlChar *name;
5324
43.7k
    xmlChar *Pubid;
5325
43.7k
    xmlChar *Systemid;
5326
5327
43.7k
    if ((CUR != '<') || (NXT(1) != '!'))
5328
0
        return;
5329
43.7k
    SKIP(2);
5330
5331
43.7k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5332
42.1k
#ifdef LIBXML_VALID_ENABLED
5333
42.1k
  int oldInputNr = ctxt->inputNr;
5334
42.1k
#endif
5335
5336
42.1k
  SKIP(8);
5337
42.1k
  if (SKIP_BLANKS_PE == 0) {
5338
1.31k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5339
1.31k
         "Space required after '<!NOTATION'\n");
5340
1.31k
      return;
5341
1.31k
  }
5342
5343
40.7k
        name = xmlParseName(ctxt);
5344
40.7k
  if (name == NULL) {
5345
3.81k
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5346
3.81k
      return;
5347
3.81k
  }
5348
36.9k
  if (xmlStrchr(name, ':') != NULL) {
5349
3.35k
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5350
3.35k
         "colons are forbidden from notation names '%s'\n",
5351
3.35k
         name, NULL, NULL);
5352
3.35k
  }
5353
36.9k
  if (SKIP_BLANKS_PE == 0) {
5354
2.13k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5355
2.13k
         "Space required after the NOTATION name'\n");
5356
2.13k
      return;
5357
2.13k
  }
5358
5359
  /*
5360
   * Parse the IDs.
5361
   */
5362
34.8k
  Systemid = xmlParseExternalID(ctxt, &Pubid, 0);
5363
34.8k
  SKIP_BLANKS_PE;
5364
5365
34.8k
  if (RAW == '>') {
5366
24.3k
#ifdef LIBXML_VALID_ENABLED
5367
24.3k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
5368
10
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
5369
10
                           "Notation declaration doesn't start and stop"
5370
10
                                 " in the same entity\n",
5371
10
                                 NULL, NULL);
5372
10
      }
5373
24.3k
#endif
5374
24.3k
      NEXT;
5375
24.3k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5376
21.9k
    (ctxt->sax->notationDecl != NULL))
5377
21.9k
    ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid);
5378
24.3k
  } else {
5379
10.5k
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5380
10.5k
  }
5381
34.8k
  if (Systemid != NULL) xmlFree(Systemid);
5382
34.8k
  if (Pubid != NULL) xmlFree(Pubid);
5383
34.8k
    }
5384
43.7k
}
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
425k
xmlParseEntityDecl(xmlParserCtxt *ctxt) {
5411
425k
    const xmlChar *name = NULL;
5412
425k
    xmlChar *value = NULL;
5413
425k
    xmlChar *URI = NULL, *literal = NULL;
5414
425k
    const xmlChar *ndata = NULL;
5415
425k
    int isParameter = 0;
5416
425k
    xmlChar *orig = NULL;
5417
5418
425k
    if ((CUR != '<') || (NXT(1) != '!'))
5419
0
        return;
5420
425k
    SKIP(2);
5421
5422
    /* GROW; done in the caller */
5423
425k
    if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5424
423k
#ifdef LIBXML_VALID_ENABLED
5425
423k
  int oldInputNr = ctxt->inputNr;
5426
423k
#endif
5427
5428
423k
  SKIP(6);
5429
423k
  if (SKIP_BLANKS_PE == 0) {
5430
31.6k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5431
31.6k
         "Space required after '<!ENTITY'\n");
5432
31.6k
  }
5433
5434
423k
  if (RAW == '%') {
5435
166k
      NEXT;
5436
166k
      if (SKIP_BLANKS_PE == 0) {
5437
7.06k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5438
7.06k
             "Space required after '%%'\n");
5439
7.06k
      }
5440
166k
      isParameter = 1;
5441
166k
  }
5442
5443
423k
        name = xmlParseName(ctxt);
5444
423k
  if (name == NULL) {
5445
39.0k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5446
39.0k
                     "xmlParseEntityDecl: no name\n");
5447
39.0k
            return;
5448
39.0k
  }
5449
384k
  if (xmlStrchr(name, ':') != NULL) {
5450
7.89k
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5451
7.89k
         "colons are forbidden from entities names '%s'\n",
5452
7.89k
         name, NULL, NULL);
5453
7.89k
  }
5454
384k
  if (SKIP_BLANKS_PE == 0) {
5455
33.4k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5456
33.4k
         "Space required after the entity name\n");
5457
33.4k
  }
5458
5459
  /*
5460
   * handle the various case of definitions...
5461
   */
5462
384k
  if (isParameter) {
5463
161k
      if ((RAW == '"') || (RAW == '\'')) {
5464
120k
          value = xmlParseEntityValue(ctxt, &orig);
5465
120k
    if (value) {
5466
116k
        if ((ctxt->sax != NULL) &&
5467
116k
      (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5468
93.0k
      ctxt->sax->entityDecl(ctxt->userData, name,
5469
93.0k
                        XML_INTERNAL_PARAMETER_ENTITY,
5470
93.0k
            NULL, NULL, value);
5471
116k
    }
5472
120k
      } else {
5473
40.8k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5474
40.8k
    if ((URI == NULL) && (literal == NULL)) {
5475
3.98k
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5476
3.98k
    }
5477
40.8k
    if (URI) {
5478
35.7k
                    if (xmlStrchr(URI, '#')) {
5479
2.17k
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5480
33.5k
                    } else {
5481
33.5k
                        if ((ctxt->sax != NULL) &&
5482
33.5k
                            (!ctxt->disableSAX) &&
5483
30.3k
                            (ctxt->sax->entityDecl != NULL))
5484
30.3k
                            ctxt->sax->entityDecl(ctxt->userData, name,
5485
30.3k
                                        XML_EXTERNAL_PARAMETER_ENTITY,
5486
30.3k
                                        literal, URI, NULL);
5487
33.5k
                    }
5488
35.7k
    }
5489
40.8k
      }
5490
223k
  } else {
5491
223k
      if ((RAW == '"') || (RAW == '\'')) {
5492
165k
          value = xmlParseEntityValue(ctxt, &orig);
5493
165k
    if ((ctxt->sax != NULL) &&
5494
165k
        (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5495
138k
        ctxt->sax->entityDecl(ctxt->userData, name,
5496
138k
        XML_INTERNAL_GENERAL_ENTITY,
5497
138k
        NULL, NULL, value);
5498
    /*
5499
     * For expat compatibility in SAX mode.
5500
     */
5501
165k
    if ((ctxt->myDoc == NULL) ||
5502
163k
        (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) {
5503
15.5k
        if (ctxt->myDoc == NULL) {
5504
2.54k
      ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5505
2.54k
      if (ctxt->myDoc == NULL) {
5506
7
          xmlErrMemory(ctxt);
5507
7
          goto done;
5508
7
      }
5509
2.53k
      ctxt->myDoc->properties = XML_DOC_INTERNAL;
5510
2.53k
        }
5511
15.5k
        if (ctxt->myDoc->intSubset == NULL) {
5512
2.53k
      ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5513
2.53k
              BAD_CAST "fake", NULL, NULL);
5514
2.53k
                        if (ctxt->myDoc->intSubset == NULL) {
5515
8
                            xmlErrMemory(ctxt);
5516
8
                            goto done;
5517
8
                        }
5518
2.53k
                    }
5519
5520
15.5k
        xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY,
5521
15.5k
                    NULL, NULL, value);
5522
15.5k
    }
5523
165k
      } else {
5524
57.5k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5525
57.5k
    if ((URI == NULL) && (literal == NULL)) {
5526
21.5k
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5527
21.5k
    }
5528
57.5k
    if (URI) {
5529
33.6k
                    if (xmlStrchr(URI, '#')) {
5530
3.17k
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5531
3.17k
                    }
5532
33.6k
    }
5533
57.5k
    if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) {
5534
20.0k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5535
20.0k
           "Space required before 'NDATA'\n");
5536
20.0k
    }
5537
57.5k
    if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) {
5538
5.74k
        SKIP(5);
5539
5.74k
        if (SKIP_BLANKS_PE == 0) {
5540
1.89k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5541
1.89k
               "Space required after 'NDATA'\n");
5542
1.89k
        }
5543
5.74k
        ndata = xmlParseName(ctxt);
5544
5.74k
        if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5545
3.79k
            (ctxt->sax->unparsedEntityDecl != NULL))
5546
3.79k
      ctxt->sax->unparsedEntityDecl(ctxt->userData, name,
5547
3.79k
            literal, URI, ndata);
5548
51.8k
    } else {
5549
51.8k
        if ((ctxt->sax != NULL) &&
5550
51.8k
            (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5551
41.6k
      ctxt->sax->entityDecl(ctxt->userData, name,
5552
41.6k
            XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5553
41.6k
            literal, URI, NULL);
5554
        /*
5555
         * For expat compatibility in SAX mode.
5556
         * assuming the entity replacement was asked for
5557
         */
5558
51.8k
        if ((ctxt->replaceEntities != 0) &&
5559
27.1k
      ((ctxt->myDoc == NULL) ||
5560
25.6k
      (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) {
5561
4.08k
      if (ctxt->myDoc == NULL) {
5562
1.53k
          ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5563
1.53k
          if (ctxt->myDoc == NULL) {
5564
6
              xmlErrMemory(ctxt);
5565
6
        goto done;
5566
6
          }
5567
1.52k
          ctxt->myDoc->properties = XML_DOC_INTERNAL;
5568
1.52k
      }
5569
5570
4.07k
      if (ctxt->myDoc->intSubset == NULL) {
5571
1.52k
          ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5572
1.52k
            BAD_CAST "fake", NULL, NULL);
5573
1.52k
                            if (ctxt->myDoc->intSubset == NULL) {
5574
7
                                xmlErrMemory(ctxt);
5575
7
                                goto done;
5576
7
                            }
5577
1.52k
                        }
5578
4.07k
      xmlSAX2EntityDecl(ctxt, name,
5579
4.07k
                  XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5580
4.07k
                  literal, URI, NULL);
5581
4.07k
        }
5582
51.8k
    }
5583
57.5k
      }
5584
223k
  }
5585
384k
  SKIP_BLANKS_PE;
5586
384k
  if (RAW != '>') {
5587
62.2k
      xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED,
5588
62.2k
              "xmlParseEntityDecl: entity %s not terminated\n", name);
5589
322k
  } else {
5590
322k
#ifdef LIBXML_VALID_ENABLED
5591
322k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
5592
221
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
5593
221
                           "Entity declaration doesn't start and stop in"
5594
221
                                 " the same entity\n",
5595
221
                                 NULL, NULL);
5596
221
      }
5597
322k
#endif
5598
322k
      NEXT;
5599
322k
  }
5600
384k
  if (orig != NULL) {
5601
      /*
5602
       * Ugly mechanism to save the raw entity value.
5603
       */
5604
275k
      xmlEntityPtr cur = NULL;
5605
5606
275k
      if (isParameter) {
5607
116k
          if ((ctxt->sax != NULL) &&
5608
116k
        (ctxt->sax->getParameterEntity != NULL))
5609
116k
        cur = ctxt->sax->getParameterEntity(ctxt->userData, name);
5610
159k
      } else {
5611
159k
          if ((ctxt->sax != NULL) &&
5612
159k
        (ctxt->sax->getEntity != NULL))
5613
159k
        cur = ctxt->sax->getEntity(ctxt->userData, name);
5614
159k
    if ((cur == NULL) && (ctxt->userData==ctxt)) {
5615
7.69k
        cur = xmlSAX2GetEntity(ctxt, name);
5616
7.69k
    }
5617
159k
      }
5618
275k
            if ((cur != NULL) && (cur->orig == NULL)) {
5619
102k
    cur->orig = orig;
5620
102k
                orig = NULL;
5621
102k
      }
5622
275k
  }
5623
5624
384k
done:
5625
384k
  if (value != NULL) xmlFree(value);
5626
384k
  if (URI != NULL) xmlFree(URI);
5627
384k
  if (literal != NULL) xmlFree(literal);
5628
384k
        if (orig != NULL) xmlFree(orig);
5629
384k
    }
5630
425k
}
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
419k
xmlParseDefaultDecl(xmlParserCtxt *ctxt, xmlChar **value) {
5663
419k
    int val;
5664
419k
    xmlChar *ret;
5665
5666
419k
    *value = NULL;
5667
419k
    if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) {
5668
15.0k
  SKIP(9);
5669
15.0k
  return(XML_ATTRIBUTE_REQUIRED);
5670
15.0k
    }
5671
404k
    if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) {
5672
95.0k
  SKIP(8);
5673
95.0k
  return(XML_ATTRIBUTE_IMPLIED);
5674
95.0k
    }
5675
309k
    val = XML_ATTRIBUTE_NONE;
5676
309k
    if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) {
5677
28.9k
  SKIP(6);
5678
28.9k
  val = XML_ATTRIBUTE_FIXED;
5679
28.9k
  if (SKIP_BLANKS_PE == 0) {
5680
3.18k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5681
3.18k
         "Space required after '#FIXED'\n");
5682
3.18k
  }
5683
28.9k
    }
5684
309k
    ret = xmlParseAttValue(ctxt);
5685
309k
    if (ret == NULL) {
5686
75.1k
  xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo,
5687
75.1k
           "Attribute default value declaration error\n");
5688
75.1k
    } else
5689
234k
        *value = ret;
5690
309k
    return(val);
5691
404k
}
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
8.90k
xmlParseNotationType(xmlParserCtxt *ctxt) {
5712
8.90k
    const xmlChar *name;
5713
8.90k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5714
5715
8.90k
    if (RAW != '(') {
5716
1.45k
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5717
1.45k
  return(NULL);
5718
1.45k
    }
5719
12.3k
    do {
5720
12.3k
        NEXT;
5721
12.3k
  SKIP_BLANKS_PE;
5722
12.3k
        name = xmlParseName(ctxt);
5723
12.3k
  if (name == NULL) {
5724
1.45k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5725
1.45k
         "Name expected in NOTATION declaration\n");
5726
1.45k
            xmlFreeEnumeration(ret);
5727
1.45k
      return(NULL);
5728
1.45k
  }
5729
10.9k
        tmp = NULL;
5730
10.9k
#ifdef LIBXML_VALID_ENABLED
5731
10.9k
        if (ctxt->validate) {
5732
4.45k
            tmp = ret;
5733
12.1k
            while (tmp != NULL) {
5734
8.80k
                if (xmlStrEqual(name, tmp->name)) {
5735
1.10k
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5736
1.10k
              "standalone: attribute notation value token %s duplicated\n",
5737
1.10k
                                     name, NULL);
5738
1.10k
                    if (!xmlDictOwns(ctxt->dict, name))
5739
0
                        xmlFree((xmlChar *) name);
5740
1.10k
                    break;
5741
1.10k
                }
5742
7.69k
                tmp = tmp->next;
5743
7.69k
            }
5744
4.45k
        }
5745
10.9k
#endif /* LIBXML_VALID_ENABLED */
5746
10.9k
  if (tmp == NULL) {
5747
9.79k
      cur = xmlCreateEnumeration(name);
5748
9.79k
      if (cur == NULL) {
5749
33
                xmlErrMemory(ctxt);
5750
33
                xmlFreeEnumeration(ret);
5751
33
                return(NULL);
5752
33
            }
5753
9.76k
      if (last == NULL) ret = last = cur;
5754
3.77k
      else {
5755
3.77k
    last->next = cur;
5756
3.77k
    last = cur;
5757
3.77k
      }
5758
9.76k
  }
5759
10.8k
  SKIP_BLANKS_PE;
5760
10.8k
    } while (RAW == '|');
5761
5.96k
    if (RAW != ')') {
5762
2.27k
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5763
2.27k
        xmlFreeEnumeration(ret);
5764
2.27k
  return(NULL);
5765
2.27k
    }
5766
3.68k
    NEXT;
5767
3.68k
    return(ret);
5768
5.96k
}
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
92.2k
xmlParseEnumerationType(xmlParserCtxt *ctxt) {
5787
92.2k
    xmlChar *name;
5788
92.2k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5789
5790
92.2k
    if (RAW != '(') {
5791
12.2k
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL);
5792
12.2k
  return(NULL);
5793
12.2k
    }
5794
115k
    do {
5795
115k
        NEXT;
5796
115k
  SKIP_BLANKS_PE;
5797
115k
        name = xmlParseNmtoken(ctxt);
5798
115k
  if (name == NULL) {
5799
1.50k
      xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL);
5800
1.50k
      return(ret);
5801
1.50k
  }
5802
114k
        tmp = NULL;
5803
114k
#ifdef LIBXML_VALID_ENABLED
5804
114k
        if (ctxt->validate) {
5805
46.5k
            tmp = ret;
5806
95.5k
            while (tmp != NULL) {
5807
50.2k
                if (xmlStrEqual(name, tmp->name)) {
5808
1.20k
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5809
1.20k
              "standalone: attribute enumeration value token %s duplicated\n",
5810
1.20k
                                     name, NULL);
5811
1.20k
                    if (!xmlDictOwns(ctxt->dict, name))
5812
1.20k
                        xmlFree(name);
5813
1.20k
                    break;
5814
1.20k
                }
5815
49.0k
                tmp = tmp->next;
5816
49.0k
            }
5817
46.5k
        }
5818
114k
#endif /* LIBXML_VALID_ENABLED */
5819
114k
  if (tmp == NULL) {
5820
113k
      cur = xmlCreateEnumeration(name);
5821
113k
      if (!xmlDictOwns(ctxt->dict, name))
5822
113k
    xmlFree(name);
5823
113k
      if (cur == NULL) {
5824
67
                xmlErrMemory(ctxt);
5825
67
                xmlFreeEnumeration(ret);
5826
67
                return(NULL);
5827
67
            }
5828
113k
      if (last == NULL) ret = last = cur;
5829
34.5k
      else {
5830
34.5k
    last->next = cur;
5831
34.5k
    last = cur;
5832
34.5k
      }
5833
113k
  }
5834
114k
  SKIP_BLANKS_PE;
5835
114k
    } while (RAW == '|');
5836
78.4k
    if (RAW != ')') {
5837
4.44k
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL);
5838
4.44k
  return(ret);
5839
4.44k
    }
5840
74.0k
    NEXT;
5841
74.0k
    return(ret);
5842
78.4k
}
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
102k
xmlParseEnumeratedType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5860
102k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5861
10.4k
  SKIP(8);
5862
10.4k
  if (SKIP_BLANKS_PE == 0) {
5863
1.50k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5864
1.50k
         "Space required after 'NOTATION'\n");
5865
1.50k
      return(0);
5866
1.50k
  }
5867
8.90k
  *tree = xmlParseNotationType(ctxt);
5868
8.90k
  if (*tree == NULL) return(0);
5869
3.68k
  return(XML_ATTRIBUTE_NOTATION);
5870
8.90k
    }
5871
92.2k
    *tree = xmlParseEnumerationType(ctxt);
5872
92.2k
    if (*tree == NULL) return(0);
5873
78.5k
    return(XML_ATTRIBUTE_ENUMERATION);
5874
92.2k
}
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
465k
xmlParseAttributeType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5923
465k
    if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) {
5924
73.4k
  SKIP(5);
5925
73.4k
  return(XML_ATTRIBUTE_CDATA);
5926
392k
     } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) {
5927
40.9k
  SKIP(6);
5928
40.9k
  return(XML_ATTRIBUTE_IDREFS);
5929
351k
     } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) {
5930
11.8k
  SKIP(5);
5931
11.8k
  return(XML_ATTRIBUTE_IDREF);
5932
339k
     } else if ((RAW == 'I') && (NXT(1) == 'D')) {
5933
79.9k
        SKIP(2);
5934
79.9k
  return(XML_ATTRIBUTE_ID);
5935
259k
     } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5936
66.5k
  SKIP(6);
5937
66.5k
  return(XML_ATTRIBUTE_ENTITY);
5938
193k
     } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) {
5939
18.8k
  SKIP(8);
5940
18.8k
  return(XML_ATTRIBUTE_ENTITIES);
5941
174k
     } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) {
5942
44.4k
  SKIP(8);
5943
44.4k
  return(XML_ATTRIBUTE_NMTOKENS);
5944
129k
     } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) {
5945
27.2k
  SKIP(7);
5946
27.2k
  return(XML_ATTRIBUTE_NMTOKEN);
5947
27.2k
     }
5948
102k
     return(xmlParseEnumeratedType(ctxt, tree));
5949
465k
}
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
385k
xmlParseAttributeListDecl(xmlParserCtxt *ctxt) {
5963
385k
    const xmlChar *elemName;
5964
385k
    const xmlChar *attrName;
5965
385k
    xmlEnumerationPtr tree;
5966
5967
385k
    if ((CUR != '<') || (NXT(1) != '!'))
5968
0
        return;
5969
385k
    SKIP(2);
5970
5971
385k
    if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) {
5972
383k
#ifdef LIBXML_VALID_ENABLED
5973
383k
  int oldInputNr = ctxt->inputNr;
5974
383k
#endif
5975
5976
383k
  SKIP(7);
5977
383k
  if (SKIP_BLANKS_PE == 0) {
5978
46.7k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5979
46.7k
                     "Space required after '<!ATTLIST'\n");
5980
46.7k
  }
5981
383k
        elemName = xmlParseName(ctxt);
5982
383k
  if (elemName == NULL) {
5983
13.1k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5984
13.1k
         "ATTLIST: no name for Element\n");
5985
13.1k
      return;
5986
13.1k
  }
5987
370k
  SKIP_BLANKS_PE;
5988
370k
  GROW;
5989
709k
  while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) {
5990
518k
      int type;
5991
518k
      int def;
5992
518k
      xmlChar *defaultValue = NULL;
5993
5994
518k
      GROW;
5995
518k
            tree = NULL;
5996
518k
      attrName = xmlParseName(ctxt);
5997
518k
      if (attrName == NULL) {
5998
32.3k
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5999
32.3k
             "ATTLIST: no name for Attribute\n");
6000
32.3k
    break;
6001
32.3k
      }
6002
486k
      GROW;
6003
486k
      if (SKIP_BLANKS_PE == 0) {
6004
20.2k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6005
20.2k
            "Space required after the attribute name\n");
6006
20.2k
    break;
6007
20.2k
      }
6008
6009
465k
      type = xmlParseAttributeType(ctxt, &tree);
6010
465k
      if (type <= 0) {
6011
20.4k
          break;
6012
20.4k
      }
6013
6014
445k
      GROW;
6015
445k
      if (SKIP_BLANKS_PE == 0) {
6016
25.5k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6017
25.5k
             "Space required after the attribute type\n");
6018
25.5k
          if (tree != NULL)
6019
5.03k
        xmlFreeEnumeration(tree);
6020
25.5k
    break;
6021
25.5k
      }
6022
6023
419k
      def = xmlParseDefaultDecl(ctxt, &defaultValue);
6024
419k
      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
419k
      if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL))
6032
212k
          xmlAttrNormalizeSpace(defaultValue, defaultValue);
6033
6034
419k
      GROW;
6035
419k
            if (RAW != '>') {
6036
261k
    if (SKIP_BLANKS_PE == 0) {
6037
81.1k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6038
81.1k
      "Space required after the attribute default value\n");
6039
81.1k
        if (defaultValue != NULL)
6040
7.73k
      xmlFree(defaultValue);
6041
81.1k
        if (tree != NULL)
6042
9.81k
      xmlFreeEnumeration(tree);
6043
81.1k
        break;
6044
81.1k
    }
6045
261k
      }
6046
338k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6047
313k
    (ctxt->sax->attributeDecl != NULL))
6048
313k
    ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName,
6049
313k
                          type, def, defaultValue, tree);
6050
24.9k
      else if (tree != NULL)
6051
8.69k
    xmlFreeEnumeration(tree);
6052
6053
338k
      if ((ctxt->sax2) && (defaultValue != NULL) &&
6054
204k
          (def != XML_ATTRIBUTE_IMPLIED) &&
6055
204k
    (def != XML_ATTRIBUTE_REQUIRED)) {
6056
204k
    xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue);
6057
204k
      }
6058
338k
      if (ctxt->sax2) {
6059
301k
    xmlAddSpecialAttr(ctxt, elemName, attrName, type);
6060
301k
      }
6061
338k
      if (defaultValue != NULL)
6062
226k
          xmlFree(defaultValue);
6063
338k
      GROW;
6064
338k
  }
6065
370k
  if (RAW == '>') {
6066
198k
#ifdef LIBXML_VALID_ENABLED
6067
198k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6068
366
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6069
366
                                 "Attribute list declaration doesn't start and"
6070
366
                                 " stop in the same entity\n",
6071
366
                                 NULL, NULL);
6072
366
      }
6073
198k
#endif
6074
198k
      NEXT;
6075
198k
  }
6076
370k
    }
6077
385k
}
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
2.84M
xmlSkipBlankCharsPEBalanced(xmlParserCtxt *ctxt, int openInputNr) {
6088
2.84M
    SKIP_BLANKS;
6089
2.84M
    GROW;
6090
6091
2.84M
    (void) openInputNr;
6092
6093
2.84M
    if (!PARSER_EXTERNAL(ctxt) && !PARSER_IN_PE(ctxt))
6094
2.35M
        return;
6095
6096
543k
    while (!PARSER_STOPPED(ctxt)) {
6097
542k
        if (ctxt->input->cur >= ctxt->input->end) {
6098
22.2k
#ifdef LIBXML_VALID_ENABLED
6099
22.2k
            if ((ctxt->validate) && (ctxt->inputNr <= openInputNr)) {
6100
3.50k
                xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6101
3.50k
                                 "Element content declaration doesn't start "
6102
3.50k
                                 "and stop in the same entity\n",
6103
3.50k
                                 NULL, NULL);
6104
3.50k
            }
6105
22.2k
#endif
6106
22.2k
            if (PARSER_IN_PE(ctxt))
6107
21.6k
                xmlPopPE(ctxt);
6108
673
            else
6109
673
                break;
6110
519k
        } else if (RAW == '%') {
6111
31.3k
            xmlParsePERefInternal(ctxt, 0);
6112
488k
        } else {
6113
488k
            break;
6114
488k
        }
6115
6116
52.9k
        SKIP_BLANKS;
6117
52.9k
        GROW;
6118
52.9k
    }
6119
490k
}
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
35.5k
xmlParseElementMixedContentDecl(xmlParserCtxt *ctxt, int openInputNr) {
6143
35.5k
    xmlElementContentPtr ret = NULL, cur = NULL, n;
6144
35.5k
    const xmlChar *elem = NULL;
6145
6146
35.5k
    GROW;
6147
35.5k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6148
35.5k
  SKIP(7);
6149
35.5k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6150
35.5k
  if (RAW == ')') {
6151
19.2k
#ifdef LIBXML_VALID_ENABLED
6152
19.2k
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6153
13
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6154
13
                                 "Element content declaration doesn't start "
6155
13
                                 "and stop in the same entity\n",
6156
13
                                 NULL, NULL);
6157
13
      }
6158
19.2k
#endif
6159
19.2k
      NEXT;
6160
19.2k
      ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6161
19.2k
      if (ret == NULL)
6162
22
                goto mem_error;
6163
19.2k
      if (RAW == '*') {
6164
1.71k
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6165
1.71k
    NEXT;
6166
1.71k
      }
6167
19.2k
      return(ret);
6168
19.2k
  }
6169
16.3k
  if ((RAW == '(') || (RAW == '|')) {
6170
13.9k
      ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6171
13.9k
      if (ret == NULL)
6172
19
                goto mem_error;
6173
13.9k
  }
6174
98.5k
  while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) {
6175
84.0k
      NEXT;
6176
84.0k
            n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6177
84.0k
            if (n == NULL)
6178
25
                goto mem_error;
6179
84.0k
      if (elem == NULL) {
6180
13.6k
    n->c1 = cur;
6181
13.6k
    if (cur != NULL)
6182
13.6k
        cur->parent = n;
6183
13.6k
    ret = cur = n;
6184
70.3k
      } else {
6185
70.3k
          cur->c2 = n;
6186
70.3k
    n->parent = cur;
6187
70.3k
    n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6188
70.3k
                if (n->c1 == NULL)
6189
15
                    goto mem_error;
6190
70.3k
    n->c1->parent = n;
6191
70.3k
    cur = n;
6192
70.3k
      }
6193
83.9k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6194
83.9k
      elem = xmlParseName(ctxt);
6195
83.9k
      if (elem == NULL) {
6196
1.74k
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6197
1.74k
      "xmlParseElementMixedContentDecl : Name expected\n");
6198
1.74k
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6199
1.74k
    return(NULL);
6200
1.74k
      }
6201
82.2k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6202
82.2k
  }
6203
14.5k
  if ((RAW == ')') && (NXT(1) == '*')) {
6204
10.3k
      if (elem != NULL) {
6205
10.3k
    cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem,
6206
10.3k
                                   XML_ELEMENT_CONTENT_ELEMENT);
6207
10.3k
    if (cur->c2 == NULL)
6208
23
                    goto mem_error;
6209
10.3k
    cur->c2->parent = cur;
6210
10.3k
            }
6211
10.3k
            if (ret != NULL)
6212
10.3k
                ret->ocur = XML_ELEMENT_CONTENT_MULT;
6213
10.3k
#ifdef LIBXML_VALID_ENABLED
6214
10.3k
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6215
6
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6216
6
                                 "Element content declaration doesn't start "
6217
6
                                 "and stop in the same entity\n",
6218
6
                                 NULL, NULL);
6219
6
      }
6220
10.3k
#endif
6221
10.3k
      SKIP(2);
6222
10.3k
  } else {
6223
4.13k
      xmlFreeDocElementContent(ctxt->myDoc, ret);
6224
4.13k
      xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL);
6225
4.13k
      return(NULL);
6226
4.13k
  }
6227
6228
14.5k
    } else {
6229
0
  xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL);
6230
0
    }
6231
10.3k
    return(ret);
6232
6233
104
mem_error:
6234
104
    xmlErrMemory(ctxt);
6235
104
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6236
104
    return(NULL);
6237
35.5k
}
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
336k
                                       int depth) {
6272
336k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
6273
336k
    xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL;
6274
336k
    const xmlChar *elem;
6275
336k
    xmlChar type = 0;
6276
6277
336k
    if (depth > maxDepth) {
6278
17
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
6279
17
                "xmlParseElementChildrenContentDecl : depth %d too deep, "
6280
17
                "use XML_PARSE_HUGE\n", depth);
6281
17
  return(NULL);
6282
17
    }
6283
336k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6284
336k
    if (RAW == '(') {
6285
205k
        int newInputNr = ctxt->inputNr;
6286
6287
        /* Recurse on first child */
6288
205k
  NEXT;
6289
205k
        cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6290
205k
                                                           depth + 1);
6291
205k
        if (cur == NULL)
6292
173k
            return(NULL);
6293
205k
    } else {
6294
131k
  elem = xmlParseName(ctxt);
6295
131k
  if (elem == NULL) {
6296
4.89k
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6297
4.89k
      return(NULL);
6298
4.89k
  }
6299
126k
        cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6300
126k
  if (cur == NULL) {
6301
118
      xmlErrMemory(ctxt);
6302
118
      return(NULL);
6303
118
  }
6304
126k
  GROW;
6305
126k
  if (RAW == '?') {
6306
10.8k
      cur->ocur = XML_ELEMENT_CONTENT_OPT;
6307
10.8k
      NEXT;
6308
115k
  } else if (RAW == '*') {
6309
11.8k
      cur->ocur = XML_ELEMENT_CONTENT_MULT;
6310
11.8k
      NEXT;
6311
103k
  } else if (RAW == '+') {
6312
6.35k
      cur->ocur = XML_ELEMENT_CONTENT_PLUS;
6313
6.35k
      NEXT;
6314
96.9k
  } else {
6315
96.9k
      cur->ocur = XML_ELEMENT_CONTENT_ONCE;
6316
96.9k
  }
6317
126k
  GROW;
6318
126k
    }
6319
1.18M
    while (!PARSER_STOPPED(ctxt)) {
6320
1.17M
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6321
1.17M
        if (RAW == ')')
6322
123k
            break;
6323
        /*
6324
   * Each loop we parse one separator and one element.
6325
   */
6326
1.05M
        if (RAW == ',') {
6327
736k
      if (type == 0) type = CUR;
6328
6329
      /*
6330
       * Detect "Name | Name , Name" error
6331
       */
6332
705k
      else if (type != CUR) {
6333
208
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6334
208
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6335
208
                      type);
6336
208
    if ((last != NULL) && (last != ret))
6337
208
        xmlFreeDocElementContent(ctxt->myDoc, last);
6338
208
    if (ret != NULL)
6339
208
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6340
208
    return(NULL);
6341
208
      }
6342
736k
      NEXT;
6343
6344
736k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ);
6345
736k
      if (op == NULL) {
6346
44
                xmlErrMemory(ctxt);
6347
44
    if ((last != NULL) && (last != ret))
6348
20
        xmlFreeDocElementContent(ctxt->myDoc, last);
6349
44
          xmlFreeDocElementContent(ctxt->myDoc, ret);
6350
44
    return(NULL);
6351
44
      }
6352
736k
      if (last == NULL) {
6353
30.6k
    op->c1 = ret;
6354
30.6k
    if (ret != NULL)
6355
30.6k
        ret->parent = op;
6356
30.6k
    ret = cur = op;
6357
705k
      } else {
6358
705k
          cur->c2 = op;
6359
705k
    if (op != NULL)
6360
705k
        op->parent = cur;
6361
705k
    op->c1 = last;
6362
705k
    if (last != NULL)
6363
705k
        last->parent = op;
6364
705k
    cur =op;
6365
705k
    last = NULL;
6366
705k
      }
6367
736k
  } else if (RAW == '|') {
6368
301k
      if (type == 0) type = CUR;
6369
6370
      /*
6371
       * Detect "Name , Name | Name" error
6372
       */
6373
234k
      else if (type != CUR) {
6374
223
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6375
223
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6376
223
          type);
6377
223
    if ((last != NULL) && (last != ret))
6378
223
        xmlFreeDocElementContent(ctxt->myDoc, last);
6379
223
    if (ret != NULL)
6380
223
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6381
223
    return(NULL);
6382
223
      }
6383
300k
      NEXT;
6384
6385
300k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6386
300k
      if (op == NULL) {
6387
42
                xmlErrMemory(ctxt);
6388
42
    if ((last != NULL) && (last != ret))
6389
17
        xmlFreeDocElementContent(ctxt->myDoc, last);
6390
42
    if (ret != NULL)
6391
42
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6392
42
    return(NULL);
6393
42
      }
6394
300k
      if (last == NULL) {
6395
66.0k
    op->c1 = ret;
6396
66.0k
    if (ret != NULL)
6397
66.0k
        ret->parent = op;
6398
66.0k
    ret = cur = op;
6399
234k
      } else {
6400
234k
          cur->c2 = op;
6401
234k
    if (op != NULL)
6402
234k
        op->parent = cur;
6403
234k
    op->c1 = last;
6404
234k
    if (last != NULL)
6405
234k
        last->parent = op;
6406
234k
    cur =op;
6407
234k
    last = NULL;
6408
234k
      }
6409
300k
  } else {
6410
14.3k
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL);
6411
14.3k
      if ((last != NULL) && (last != ret))
6412
8.41k
          xmlFreeDocElementContent(ctxt->myDoc, last);
6413
14.3k
      if (ret != NULL)
6414
14.3k
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6415
14.3k
      return(NULL);
6416
14.3k
  }
6417
1.03M
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6418
1.03M
        if (RAW == '(') {
6419
68.3k
            int newInputNr = ctxt->inputNr;
6420
6421
      /* Recurse on second child */
6422
68.3k
      NEXT;
6423
68.3k
      last = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6424
68.3k
                                                          depth + 1);
6425
68.3k
            if (last == NULL) {
6426
10.2k
    if (ret != NULL)
6427
10.2k
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6428
10.2k
    return(NULL);
6429
10.2k
            }
6430
968k
  } else {
6431
968k
      elem = xmlParseName(ctxt);
6432
968k
      if (elem == NULL) {
6433
2.77k
    xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6434
2.77k
    if (ret != NULL)
6435
2.77k
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6436
2.77k
    return(NULL);
6437
2.77k
      }
6438
965k
      last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6439
965k
      if (last == NULL) {
6440
66
                xmlErrMemory(ctxt);
6441
66
    if (ret != NULL)
6442
66
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6443
66
    return(NULL);
6444
66
      }
6445
965k
      if (RAW == '?') {
6446
94.5k
    last->ocur = XML_ELEMENT_CONTENT_OPT;
6447
94.5k
    NEXT;
6448
871k
      } else if (RAW == '*') {
6449
21.6k
    last->ocur = XML_ELEMENT_CONTENT_MULT;
6450
21.6k
    NEXT;
6451
849k
      } else if (RAW == '+') {
6452
16.4k
    last->ocur = XML_ELEMENT_CONTENT_PLUS;
6453
16.4k
    NEXT;
6454
833k
      } else {
6455
833k
    last->ocur = XML_ELEMENT_CONTENT_ONCE;
6456
833k
      }
6457
965k
  }
6458
1.03M
    }
6459
130k
    if ((cur != NULL) && (last != NULL)) {
6460
74.7k
        cur->c2 = last;
6461
74.7k
  if (last != NULL)
6462
74.7k
      last->parent = cur;
6463
74.7k
    }
6464
130k
#ifdef LIBXML_VALID_ENABLED
6465
130k
    if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6466
131
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6467
131
                         "Element content declaration doesn't start "
6468
131
                         "and stop in the same entity\n",
6469
131
                         NULL, NULL);
6470
131
    }
6471
130k
#endif
6472
130k
    NEXT;
6473
130k
    if (RAW == '?') {
6474
20.6k
  if (ret != NULL) {
6475
20.6k
      if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) ||
6476
19.6k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6477
2.21k
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6478
18.4k
      else
6479
18.4k
          ret->ocur = XML_ELEMENT_CONTENT_OPT;
6480
20.6k
  }
6481
20.6k
  NEXT;
6482
109k
    } else if (RAW == '*') {
6483
31.3k
  if (ret != NULL) {
6484
31.3k
      ret->ocur = XML_ELEMENT_CONTENT_MULT;
6485
31.3k
      cur = ret;
6486
      /*
6487
       * Some normalization:
6488
       * (a | b* | c?)* == (a | b | c)*
6489
       */
6490
104k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6491
73.2k
    if ((cur->c1 != NULL) &&
6492
73.2k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6493
69.2k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT)))
6494
8.69k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6495
73.2k
    if ((cur->c2 != NULL) &&
6496
73.2k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6497
71.7k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT)))
6498
4.14k
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6499
73.2k
    cur = cur->c2;
6500
73.2k
      }
6501
31.3k
  }
6502
31.3k
  NEXT;
6503
78.1k
    } else if (RAW == '+') {
6504
23.4k
  if (ret != NULL) {
6505
23.4k
      int found = 0;
6506
6507
23.4k
      if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) ||
6508
21.3k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6509
4.15k
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6510
19.3k
      else
6511
19.3k
          ret->ocur = XML_ELEMENT_CONTENT_PLUS;
6512
      /*
6513
       * Some normalization:
6514
       * (a | b*)+ == (a | b)*
6515
       * (a | b?)+ == (a | b)*
6516
       */
6517
47.4k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6518
23.9k
    if ((cur->c1 != NULL) &&
6519
23.9k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6520
19.8k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) {
6521
7.13k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6522
7.13k
        found = 1;
6523
7.13k
    }
6524
23.9k
    if ((cur->c2 != NULL) &&
6525
23.9k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6526
22.6k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) {
6527
5.01k
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6528
5.01k
        found = 1;
6529
5.01k
    }
6530
23.9k
    cur = cur->c2;
6531
23.9k
      }
6532
23.4k
      if (found)
6533
6.89k
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6534
23.4k
  }
6535
23.4k
  NEXT;
6536
23.4k
    }
6537
130k
    return(ret);
6538
158k
}
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
98.2k
                           xmlElementContent **result) {
6593
6594
98.2k
    xmlElementContentPtr tree = NULL;
6595
98.2k
    int openInputNr = ctxt->inputNr;
6596
98.2k
    int res;
6597
6598
98.2k
    *result = NULL;
6599
6600
98.2k
    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
98.2k
    NEXT;
6606
98.2k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6607
98.2k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6608
35.5k
        tree = xmlParseElementMixedContentDecl(ctxt, openInputNr);
6609
35.5k
  res = XML_ELEMENT_TYPE_MIXED;
6610
62.6k
    } else {
6611
62.6k
        tree = xmlParseElementChildrenContentDeclPriv(ctxt, openInputNr, 1);
6612
62.6k
  res = XML_ELEMENT_TYPE_ELEMENT;
6613
62.6k
    }
6614
98.2k
    if (tree == NULL)
6615
28.7k
        return(-1);
6616
69.5k
    SKIP_BLANKS_PE;
6617
69.5k
    *result = tree;
6618
69.5k
    return(res);
6619
98.2k
}
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
134k
xmlParseElementDecl(xmlParserCtxt *ctxt) {
6636
134k
    const xmlChar *name;
6637
134k
    int ret = -1;
6638
134k
    xmlElementContentPtr content  = NULL;
6639
6640
134k
    if ((CUR != '<') || (NXT(1) != '!'))
6641
0
        return(ret);
6642
134k
    SKIP(2);
6643
6644
    /* GROW; done in the caller */
6645
134k
    if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) {
6646
132k
#ifdef LIBXML_VALID_ENABLED
6647
132k
  int oldInputNr = ctxt->inputNr;
6648
132k
#endif
6649
6650
132k
  SKIP(7);
6651
132k
  if (SKIP_BLANKS_PE == 0) {
6652
1.71k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6653
1.71k
               "Space required after 'ELEMENT'\n");
6654
1.71k
      return(-1);
6655
1.71k
  }
6656
131k
        name = xmlParseName(ctxt);
6657
131k
  if (name == NULL) {
6658
3.22k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6659
3.22k
         "xmlParseElementDecl: no name for Element\n");
6660
3.22k
      return(-1);
6661
3.22k
  }
6662
128k
  if (SKIP_BLANKS_PE == 0) {
6663
31.0k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6664
31.0k
         "Space required after the element name\n");
6665
31.0k
  }
6666
128k
  if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) {
6667
17.0k
      SKIP(5);
6668
      /*
6669
       * Element must always be empty.
6670
       */
6671
17.0k
      ret = XML_ELEMENT_TYPE_EMPTY;
6672
111k
  } else if ((RAW == 'A') && (NXT(1) == 'N') &&
6673
7.03k
             (NXT(2) == 'Y')) {
6674
6.82k
      SKIP(3);
6675
      /*
6676
       * Element is a generic container.
6677
       */
6678
6.82k
      ret = XML_ELEMENT_TYPE_ANY;
6679
104k
  } else if (RAW == '(') {
6680
98.2k
      ret = xmlParseElementContentDecl(ctxt, name, &content);
6681
98.2k
            if (ret <= 0)
6682
28.7k
                return(-1);
6683
98.2k
  } else {
6684
      /*
6685
       * [ WFC: PEs in Internal Subset ] error handling.
6686
       */
6687
5.97k
            xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6688
5.97k
                  "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n");
6689
5.97k
      return(-1);
6690
5.97k
  }
6691
6692
93.3k
  SKIP_BLANKS_PE;
6693
6694
93.3k
  if (RAW != '>') {
6695
10.0k
      xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
6696
10.0k
      if (content != NULL) {
6697
7.97k
    xmlFreeDocElementContent(ctxt->myDoc, content);
6698
7.97k
      }
6699
83.2k
  } else {
6700
83.2k
#ifdef LIBXML_VALID_ENABLED
6701
83.2k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6702
83
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6703
83
                                 "Element declaration doesn't start and stop in"
6704
83
                                 " the same entity\n",
6705
83
                                 NULL, NULL);
6706
83
      }
6707
83.2k
#endif
6708
6709
83.2k
      NEXT;
6710
83.2k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6711
74.1k
    (ctxt->sax->elementDecl != NULL)) {
6712
74.1k
    if (content != NULL)
6713
54.2k
        content->parent = NULL;
6714
74.1k
          ctxt->sax->elementDecl(ctxt->userData, name, ret,
6715
74.1k
                           content);
6716
74.1k
    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
12.0k
        xmlFreeDocElementContent(ctxt->myDoc, content);
6724
12.0k
    }
6725
74.1k
      } else if (content != NULL) {
6726
7.27k
    xmlFreeDocElementContent(ctxt->myDoc, content);
6727
7.27k
      }
6728
83.2k
  }
6729
93.3k
    }
6730
94.5k
    return(ret);
6731
134k
}
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
28.3k
xmlParseConditionalSections(xmlParserCtxtPtr ctxt) {
6747
28.3k
    size_t depth = 0;
6748
28.3k
    int isFreshPE = 0;
6749
28.3k
    int oldInputNr = ctxt->inputNr;
6750
28.3k
    int declInputNr = ctxt->inputNr;
6751
6752
58.1k
    while (!PARSER_STOPPED(ctxt)) {
6753
58.0k
        if (ctxt->input->cur >= ctxt->input->end) {
6754
3.70k
            if (ctxt->inputNr <= oldInputNr) {
6755
1.76k
                xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6756
1.76k
                return;
6757
1.76k
            }
6758
6759
1.94k
            xmlPopPE(ctxt);
6760
1.94k
            declInputNr = ctxt->inputNr;
6761
54.3k
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6762
30.8k
            SKIP(3);
6763
30.8k
            SKIP_BLANKS_PE;
6764
6765
30.8k
            isFreshPE = 0;
6766
6767
30.8k
            if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) {
6768
9.41k
                SKIP(7);
6769
9.41k
                SKIP_BLANKS_PE;
6770
9.41k
                if (RAW != '[') {
6771
1.08k
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6772
1.08k
                    return;
6773
1.08k
                }
6774
8.33k
#ifdef LIBXML_VALID_ENABLED
6775
8.33k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6776
81
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6777
81
                                     "All markup of the conditional section is"
6778
81
                                     " not in the same entity\n",
6779
81
                                     NULL, NULL);
6780
81
                }
6781
8.33k
#endif
6782
8.33k
                NEXT;
6783
6784
8.33k
                depth++;
6785
21.4k
            } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) {
6786
9.91k
                size_t ignoreDepth = 0;
6787
6788
9.91k
                SKIP(6);
6789
9.91k
                SKIP_BLANKS_PE;
6790
9.91k
                if (RAW != '[') {
6791
1.42k
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6792
1.42k
                    return;
6793
1.42k
                }
6794
8.49k
#ifdef LIBXML_VALID_ENABLED
6795
8.49k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6796
52
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6797
52
                                     "All markup of the conditional section is"
6798
52
                                     " not in the same entity\n",
6799
52
                                     NULL, NULL);
6800
52
                }
6801
8.49k
#endif
6802
8.49k
                NEXT;
6803
6804
1.43M
                while (PARSER_STOPPED(ctxt) == 0) {
6805
1.43M
                    if (RAW == 0) {
6806
3.34k
                        xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL);
6807
3.34k
                        return;
6808
3.34k
                    }
6809
1.43M
                    if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6810
4.29k
                        SKIP(3);
6811
4.29k
                        ignoreDepth++;
6812
                        /* Check for integer overflow */
6813
4.29k
                        if (ignoreDepth == 0) {
6814
0
                            xmlErrMemory(ctxt);
6815
0
                            return;
6816
0
                        }
6817
1.42M
                    } else if ((RAW == ']') && (NXT(1) == ']') &&
6818
9.15k
                               (NXT(2) == '>')) {
6819
6.48k
                        SKIP(3);
6820
6.48k
                        if (ignoreDepth == 0)
6821
5.11k
                            break;
6822
1.37k
                        ignoreDepth--;
6823
1.41M
                    } else {
6824
1.41M
                        NEXT;
6825
1.41M
                    }
6826
1.43M
                }
6827
6828
5.14k
#ifdef LIBXML_VALID_ENABLED
6829
5.14k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6830
44
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6831
44
                                     "All markup of the conditional section is"
6832
44
                                     " not in the same entity\n",
6833
44
                                     NULL, NULL);
6834
44
                }
6835
5.14k
#endif
6836
11.4k
            } else {
6837
11.4k
                xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL);
6838
11.4k
                return;
6839
11.4k
            }
6840
30.8k
        } else if ((depth > 0) &&
6841
23.5k
                   (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) {
6842
3.89k
            if (isFreshPE) {
6843
10
                xmlFatalErrMsg(ctxt, XML_ERR_CONDSEC_INVALID,
6844
10
                               "Parameter entity must match "
6845
10
                               "extSubsetDecl\n");
6846
10
                return;
6847
10
            }
6848
6849
3.88k
            depth--;
6850
3.88k
#ifdef LIBXML_VALID_ENABLED
6851
3.88k
            if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6852
30
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6853
30
                                 "All markup of the conditional section is not"
6854
30
                                 " in the same entity\n",
6855
30
                                 NULL, NULL);
6856
30
            }
6857
3.88k
#endif
6858
3.88k
            SKIP(3);
6859
19.6k
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
6860
12.9k
            isFreshPE = 0;
6861
12.9k
            xmlParseMarkupDecl(ctxt);
6862
12.9k
        } else if (RAW == '%') {
6863
6.54k
            xmlParsePERefInternal(ctxt, 1);
6864
6.54k
            if (ctxt->inputNr > declInputNr) {
6865
1.99k
                isFreshPE = 1;
6866
1.99k
                declInputNr = ctxt->inputNr;
6867
1.99k
            }
6868
6.54k
        } else {
6869
197
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6870
197
            return;
6871
197
        }
6872
6873
38.7k
        if (depth == 0)
6874
8.95k
            break;
6875
6876
29.8k
        SKIP_BLANKS;
6877
29.8k
        SHRINK;
6878
29.8k
        GROW;
6879
29.8k
    }
6880
28.3k
}
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
2.39M
xmlParseMarkupDecl(xmlParserCtxt *ctxt) {
6907
2.39M
    GROW;
6908
2.39M
    if (CUR == '<') {
6909
2.39M
        if (NXT(1) == '!') {
6910
2.31M
      switch (NXT(2)) {
6911
559k
          case 'E':
6912
559k
        if (NXT(3) == 'L')
6913
134k
      xmlParseElementDecl(ctxt);
6914
425k
        else if (NXT(3) == 'N')
6915
425k
      xmlParseEntityDecl(ctxt);
6916
370
                    else
6917
370
                        SKIP(2);
6918
559k
        break;
6919
385k
          case 'A':
6920
385k
        xmlParseAttributeListDecl(ctxt);
6921
385k
        break;
6922
43.7k
          case 'N':
6923
43.7k
        xmlParseNotationDecl(ctxt);
6924
43.7k
        break;
6925
1.28M
          case '-':
6926
1.28M
        xmlParseComment(ctxt);
6927
1.28M
        break;
6928
42.3k
    default:
6929
42.3k
                    xmlFatalErr(ctxt,
6930
42.3k
                                ctxt->inSubset == 2 ?
6931
4.14k
                                    XML_ERR_EXT_SUBSET_NOT_FINISHED :
6932
42.3k
                                    XML_ERR_INT_SUBSET_NOT_FINISHED,
6933
42.3k
                                NULL);
6934
42.3k
                    SKIP(2);
6935
42.3k
        break;
6936
2.31M
      }
6937
2.31M
  } else if (NXT(1) == '?') {
6938
72.7k
      xmlParsePI(ctxt);
6939
72.7k
  }
6940
2.39M
    }
6941
2.39M
}
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
251k
xmlParseTextDecl(xmlParserCtxt *ctxt) {
6954
251k
    xmlChar *version;
6955
6956
    /*
6957
     * We know that '<?xml' is here.
6958
     */
6959
251k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
6960
251k
  SKIP(5);
6961
251k
    } else {
6962
40
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL);
6963
40
  return;
6964
40
    }
6965
6966
251k
    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
251k
    version = xmlParseVersionInfo(ctxt);
6975
251k
    if (version == NULL) {
6976
193k
  version = xmlCharStrdup(XML_DEFAULT_VERSION);
6977
193k
        if (version == NULL) {
6978
71
            xmlErrMemory(ctxt);
6979
71
            return;
6980
71
        }
6981
193k
    } else {
6982
58.0k
  if (SKIP_BLANKS == 0) {
6983
2.62k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6984
2.62k
               "Space needed here\n");
6985
2.62k
  }
6986
58.0k
    }
6987
251k
    ctxt->input->version = version;
6988
6989
    /*
6990
     * We must have the encoding declaration
6991
     */
6992
251k
    xmlParseEncodingDecl(ctxt);
6993
6994
251k
    SKIP_BLANKS;
6995
251k
    if ((RAW == '?') && (NXT(1) == '>')) {
6996
8.93k
        SKIP(2);
6997
242k
    } else if (RAW == '>') {
6998
        /* Deprecated old WD ... */
6999
1.95k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7000
1.95k
  NEXT;
7001
240k
    } else {
7002
240k
        int c;
7003
7004
240k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7005
1.75G
        while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) {
7006
1.75G
            NEXT;
7007
1.75G
            if (c == '>')
7008
175k
                break;
7009
1.75G
        }
7010
240k
    }
7011
251k
}
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
10.4k
                       const xmlChar *systemId) {
7029
10.4k
    int oldInputNr;
7030
7031
10.4k
    xmlCtxtInitializeLate(ctxt);
7032
7033
10.4k
    xmlDetectEncoding(ctxt);
7034
7035
10.4k
    if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) {
7036
589
  xmlParseTextDecl(ctxt);
7037
589
    }
7038
10.4k
    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
10.4k
    if ((ctxt->myDoc->intSubset == NULL) &&
7047
2.11k
        (xmlCreateIntSubset(ctxt->myDoc, NULL, publicId, systemId) == NULL)) {
7048
13
        xmlErrMemory(ctxt);
7049
13
    }
7050
7051
10.4k
    ctxt->inSubset = 2;
7052
10.4k
    oldInputNr = ctxt->inputNr;
7053
7054
10.4k
    SKIP_BLANKS;
7055
1.28M
    while (!PARSER_STOPPED(ctxt)) {
7056
1.28M
        if (ctxt->input->cur >= ctxt->input->end) {
7057
4.73k
            if (ctxt->inputNr <= oldInputNr) {
7058
2.62k
                xmlParserCheckEOF(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED);
7059
2.62k
                break;
7060
2.62k
            }
7061
7062
2.11k
            xmlPopPE(ctxt);
7063
1.27M
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
7064
15.8k
            xmlParseConditionalSections(ctxt);
7065
1.26M
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
7066
1.25M
            xmlParseMarkupDecl(ctxt);
7067
1.25M
        } else if (RAW == '%') {
7068
5.05k
            xmlParsePERefInternal(ctxt, 1);
7069
5.05k
        } else {
7070
4.74k
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
7071
7072
5.45k
            while (ctxt->inputNr > oldInputNr)
7073
712
                xmlPopPE(ctxt);
7074
4.74k
            break;
7075
4.74k
        }
7076
1.27M
        SKIP_BLANKS;
7077
1.27M
        SHRINK;
7078
1.27M
        GROW;
7079
1.27M
    }
7080
10.4k
}
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
1.10M
xmlParseReference(xmlParserCtxt *ctxt) {
7097
1.10M
    xmlEntityPtr ent = NULL;
7098
1.10M
    const xmlChar *name;
7099
1.10M
    xmlChar *val;
7100
7101
1.10M
    if (RAW != '&')
7102
0
        return;
7103
7104
    /*
7105
     * Simple case of a CharRef
7106
     */
7107
1.10M
    if (NXT(1) == '#') {
7108
335k
  int i = 0;
7109
335k
  xmlChar out[16];
7110
335k
  int value = xmlParseCharRef(ctxt);
7111
7112
335k
  if (value == 0)
7113
108k
      return;
7114
7115
        /*
7116
         * Just encode the value in UTF-8
7117
         */
7118
226k
        COPY_BUF(out, i, value);
7119
226k
        out[i] = 0;
7120
226k
        if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7121
226k
            (!ctxt->disableSAX))
7122
178k
            ctxt->sax->characters(ctxt->userData, out, i);
7123
226k
  return;
7124
335k
    }
7125
7126
    /*
7127
     * We are seeing an entity reference
7128
     */
7129
772k
    name = xmlParseEntityRefInternal(ctxt);
7130
772k
    if (name == NULL)
7131
262k
        return;
7132
509k
    ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0);
7133
509k
    if (ent == NULL) {
7134
        /*
7135
         * Create a reference for undeclared entities.
7136
         */
7137
266k
        if ((ctxt->replaceEntities == 0) &&
7138
211k
            (ctxt->sax != NULL) &&
7139
211k
            (ctxt->disableSAX == 0) &&
7140
199k
            (ctxt->sax->reference != NULL)) {
7141
199k
            ctxt->sax->reference(ctxt->userData, name);
7142
199k
        }
7143
266k
        return;
7144
266k
    }
7145
242k
    if (!ctxt->wellFormed)
7146
124k
  return;
7147
7148
    /* special case of predefined entities */
7149
117k
    if ((ent->name == NULL) ||
7150
117k
        (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
7151
31.3k
  val = ent->content;
7152
31.3k
  if (val == NULL) return;
7153
  /*
7154
   * inline the entity.
7155
   */
7156
31.3k
  if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7157
31.3k
      (!ctxt->disableSAX))
7158
31.3k
      ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val));
7159
31.3k
  return;
7160
31.3k
    }
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
86.1k
    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
86.1k
    if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) ||
7211
30.8k
        (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
7212
29.8k
         ((ctxt->replaceEntities) ||
7213
83.6k
          (ctxt->validate)))) {
7214
83.6k
        if ((ent->flags & XML_ENT_PARSED) == 0) {
7215
23.2k
            xmlCtxtParseEntity(ctxt, ent);
7216
60.3k
        } 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
16.2k
            xmlCtxtParseEntity(ctxt, ent);
7227
16.2k
        }
7228
83.6k
    }
7229
7230
    /*
7231
     * We also check for amplification if entities aren't substituted.
7232
     * They might be expanded later.
7233
     */
7234
86.1k
    if (xmlParserEntityCheck(ctxt, ent->expandedSize))
7235
476
        return;
7236
7237
85.6k
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
7238
8.85k
        return;
7239
7240
76.8k
    if (ctxt->replaceEntities == 0) {
7241
  /*
7242
   * Create a reference
7243
   */
7244
21.4k
        if (ctxt->sax->reference != NULL)
7245
21.4k
      ctxt->sax->reference(ctxt->userData, ent->name);
7246
55.3k
    } else if ((ent->children != NULL) && (ctxt->node != NULL)) {
7247
40.4k
        xmlNodePtr copy, cur;
7248
7249
        /*
7250
         * Seems we are generating the DOM content, copy the tree
7251
   */
7252
40.4k
        cur = ent->children;
7253
7254
        /*
7255
         * Handle first text node with SAX to coalesce text efficiently
7256
         */
7257
40.4k
        if ((cur->type == XML_TEXT_NODE) ||
7258
28.8k
            (cur->type == XML_CDATA_SECTION_NODE)) {
7259
28.8k
            int len = xmlStrlen(cur->content);
7260
7261
28.8k
            if ((cur->type == XML_TEXT_NODE) ||
7262
27.7k
                (ctxt->options & XML_PARSE_NOCDATA)) {
7263
27.7k
                if (ctxt->sax->characters != NULL)
7264
27.7k
                    ctxt->sax->characters(ctxt, cur->content, len);
7265
27.7k
            } else {
7266
1.07k
                if (ctxt->sax->cdataBlock != NULL)
7267
1.07k
                    ctxt->sax->cdataBlock(ctxt, cur->content, len);
7268
1.07k
            }
7269
7270
28.8k
            cur = cur->next;
7271
28.8k
        }
7272
7273
430k
        while (cur != NULL) {
7274
400k
            xmlNodePtr last;
7275
7276
            /*
7277
             * Handle last text node with SAX to coalesce text efficiently
7278
             */
7279
400k
            if ((cur->next == NULL) &&
7280
25.1k
                ((cur->type == XML_TEXT_NODE) ||
7281
16.7k
                 (cur->type == XML_CDATA_SECTION_NODE))) {
7282
9.98k
                int len = xmlStrlen(cur->content);
7283
7284
9.98k
                if ((cur->type == XML_TEXT_NODE) ||
7285
8.40k
                    (ctxt->options & XML_PARSE_NOCDATA)) {
7286
8.40k
                    if (ctxt->sax->characters != NULL)
7287
8.40k
                        ctxt->sax->characters(ctxt, cur->content, len);
7288
8.40k
                } else {
7289
1.58k
                    if (ctxt->sax->cdataBlock != NULL)
7290
1.58k
                        ctxt->sax->cdataBlock(ctxt, cur->content, len);
7291
1.58k
                }
7292
7293
9.98k
                break;
7294
9.98k
            }
7295
7296
            /*
7297
             * Reset coalesce buffer stats only for non-text nodes.
7298
             */
7299
390k
            ctxt->nodemem = 0;
7300
390k
            ctxt->nodelen = 0;
7301
7302
390k
            copy = xmlDocCopyNode(cur, ctxt->myDoc, 1);
7303
7304
390k
            if (copy == NULL) {
7305
640
                xmlErrMemory(ctxt);
7306
640
                break;
7307
640
            }
7308
7309
390k
            if (ctxt->parseMode == XML_PARSE_READER) {
7310
                /* Needed for reader */
7311
31.3k
                copy->extra = cur->extra;
7312
                /* Maybe needed for reader */
7313
31.3k
                copy->_private = cur->_private;
7314
31.3k
            }
7315
7316
390k
            copy->parent = ctxt->node;
7317
390k
            last = ctxt->node->last;
7318
390k
            if (last == NULL) {
7319
1.20k
                ctxt->node->children = copy;
7320
389k
            } else {
7321
389k
                last->next = copy;
7322
389k
                copy->prev = last;
7323
389k
            }
7324
390k
            ctxt->node->last = copy;
7325
7326
390k
            cur = cur->next;
7327
390k
        }
7328
40.4k
    }
7329
76.8k
}
7330
7331
static void
7332
1.18M
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
1.18M
    if ((ctxt->standalone == 1) ||
7355
1.18M
        ((ctxt->hasExternalSubset == 0) &&
7356
1.15M
         (ctxt->hasPErefs == 0))) {
7357
585k
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY,
7358
585k
                          "Entity '%s' not defined\n", name);
7359
585k
#ifdef LIBXML_VALID_ENABLED
7360
599k
    } 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
329k
        xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY,
7369
329k
                         "Entity '%s' not defined\n", name, NULL);
7370
329k
#endif
7371
329k
    } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) ||
7372
173k
               ((ctxt->replaceEntities) &&
7373
196k
                ((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
196k
        xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY,
7383
196k
                     "Entity '%s' not defined\n", name);
7384
196k
    } else {
7385
72.4k
        xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7386
72.4k
                      "Entity '%s' not defined\n", name, NULL);
7387
72.4k
    }
7388
7389
1.18M
    ctxt->valid = 0;
7390
1.18M
}
7391
7392
static xmlEntityPtr
7393
9.35M
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) {
7394
9.35M
    xmlEntityPtr ent = NULL;
7395
7396
    /*
7397
     * Predefined entities override any extra definition
7398
     */
7399
9.35M
    if ((ctxt->options & XML_PARSE_OLDSAX) == 0) {
7400
5.68M
        ent = xmlGetPredefinedEntity(name);
7401
5.68M
        if (ent != NULL)
7402
1.81M
            return(ent);
7403
5.68M
    }
7404
7405
    /*
7406
     * Ask first SAX for entity resolution, otherwise try the
7407
     * entities which may have stored in the parser context.
7408
     */
7409
7.54M
    if (ctxt->sax != NULL) {
7410
7.54M
  if (ctxt->sax->getEntity != NULL)
7411
7.54M
      ent = ctxt->sax->getEntity(ctxt->userData, name);
7412
7.54M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7413
88.3k
      (ctxt->options & XML_PARSE_OLDSAX))
7414
4.14k
      ent = xmlGetPredefinedEntity(name);
7415
7.54M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7416
88.3k
      (ctxt->userData==ctxt)) {
7417
88.3k
      ent = xmlSAX2GetEntity(ctxt, name);
7418
88.3k
  }
7419
7.54M
    }
7420
7421
7.54M
    if (ent == NULL) {
7422
967k
        xmlHandleUndeclaredEntity(ctxt, name);
7423
967k
    }
7424
7425
    /*
7426
     * [ WFC: Parsed Entity ]
7427
     * An entity reference must not contain the name of an
7428
     * unparsed entity
7429
     */
7430
6.57M
    else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) {
7431
1.42k
  xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY,
7432
1.42k
     "Entity reference to unparsed entity %s\n", name);
7433
1.42k
        ent = NULL;
7434
1.42k
    }
7435
7436
    /*
7437
     * [ WFC: No External Entity References ]
7438
     * Attribute values cannot contain direct or indirect
7439
     * entity references to external entities.
7440
     */
7441
6.57M
    else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) {
7442
76.9k
        if (inAttr) {
7443
9.16k
            xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL,
7444
9.16k
                 "Attribute references external entity '%s'\n", name);
7445
9.16k
            ent = NULL;
7446
9.16k
        }
7447
76.9k
    }
7448
7449
7.54M
    return(ent);
7450
9.35M
}
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
3.60M
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) {
7462
3.60M
    const xmlChar *name;
7463
7464
3.60M
    GROW;
7465
7466
3.60M
    if (RAW != '&')
7467
0
        return(NULL);
7468
3.60M
    NEXT;
7469
3.60M
    name = xmlParseName(ctxt);
7470
3.60M
    if (name == NULL) {
7471
336k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7472
336k
           "xmlParseEntityRef: no name\n");
7473
336k
        return(NULL);
7474
336k
    }
7475
3.26M
    if (RAW != ';') {
7476
240k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7477
240k
  return(NULL);
7478
240k
    }
7479
3.02M
    NEXT;
7480
7481
3.02M
    return(name);
7482
3.26M
}
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
6.33M
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) {
7535
6.33M
    xmlChar *name;
7536
6.33M
    const xmlChar *ptr;
7537
6.33M
    xmlChar cur;
7538
7539
6.33M
    if ((str == NULL) || (*str == NULL))
7540
0
        return(NULL);
7541
6.33M
    ptr = *str;
7542
6.33M
    cur = *ptr;
7543
6.33M
    if (cur != '&')
7544
0
  return(NULL);
7545
7546
6.33M
    ptr++;
7547
6.33M
    name = xmlParseStringName(ctxt, &ptr);
7548
6.33M
    if (name == NULL) {
7549
223
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7550
223
           "xmlParseStringEntityRef: no name\n");
7551
223
  *str = ptr;
7552
223
  return(NULL);
7553
223
    }
7554
6.33M
    if (*ptr != ';') {
7555
80
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7556
80
        xmlFree(name);
7557
80
  *str = ptr;
7558
80
  return(NULL);
7559
80
    }
7560
6.33M
    ptr++;
7561
7562
6.33M
    *str = ptr;
7563
6.33M
    return(name);
7564
6.33M
}
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
736k
xmlParsePERefInternal(xmlParserCtxt *ctxt, int markupDecl) {
7598
736k
    const xmlChar *name;
7599
736k
    xmlEntityPtr entity = NULL;
7600
736k
    xmlParserInputPtr input;
7601
7602
736k
    if (RAW != '%')
7603
0
        return;
7604
736k
    NEXT;
7605
736k
    name = xmlParseName(ctxt);
7606
736k
    if (name == NULL) {
7607
64.2k
  xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n");
7608
64.2k
  return;
7609
64.2k
    }
7610
672k
    if (RAW != ';') {
7611
58.6k
  xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL);
7612
58.6k
        return;
7613
58.6k
    }
7614
7615
613k
    NEXT;
7616
7617
    /* Must be set before xmlHandleUndeclaredEntity */
7618
613k
    ctxt->hasPErefs = 1;
7619
7620
    /*
7621
     * Request the entity from SAX
7622
     */
7623
613k
    if ((ctxt->sax != NULL) &&
7624
613k
  (ctxt->sax->getParameterEntity != NULL))
7625
613k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7626
7627
613k
    if (entity == NULL) {
7628
179k
        xmlHandleUndeclaredEntity(ctxt, name);
7629
434k
    } else {
7630
  /*
7631
   * Internal checking in case the entity quest barfed
7632
   */
7633
434k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7634
286k
      (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
434k
  } else {
7639
434k
      if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) &&
7640
286k
                ((ctxt->options & XML_PARSE_NO_XXE) ||
7641
285k
     (((ctxt->loadsubset & ~XML_SKIP_IDS) == 0) &&
7642
114k
      (ctxt->replaceEntities == 0) &&
7643
14.5k
      (ctxt->validate == 0))))
7644
2.59k
    return;
7645
7646
431k
            if (entity->flags & XML_ENT_EXPANDING) {
7647
654
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
7648
654
                return;
7649
654
            }
7650
7651
431k
      input = xmlNewEntityInputStream(ctxt, entity);
7652
431k
      if (xmlCtxtPushInput(ctxt, input) < 0) {
7653
13.9k
                xmlFreeInputStream(input);
7654
13.9k
    return;
7655
13.9k
            }
7656
7657
417k
            entity->flags |= XML_ENT_EXPANDING;
7658
7659
417k
            if (markupDecl)
7660
366k
                input->flags |= XML_INPUT_MARKUP_DECL;
7661
7662
417k
            GROW;
7663
7664
417k
      if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
7665
269k
                xmlDetectEncoding(ctxt);
7666
7667
269k
                if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
7668
95.8k
                    (IS_BLANK_CH(NXT(5)))) {
7669
92.8k
                    xmlParseTextDecl(ctxt);
7670
92.8k
                }
7671
269k
            }
7672
417k
  }
7673
434k
    }
7674
613k
}
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
194k
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) {
7697
194k
    xmlParserInputPtr oldinput, input = NULL;
7698
194k
    xmlParserInputPtr *oldinputTab;
7699
194k
    xmlChar *oldencoding;
7700
194k
    xmlChar *content = NULL;
7701
194k
    xmlResourceType rtype;
7702
194k
    size_t length, i;
7703
194k
    int oldinputNr, oldinputMax;
7704
194k
    int ret = -1;
7705
194k
    int res;
7706
7707
194k
    if ((ctxt == NULL) || (entity == NULL) ||
7708
194k
        ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) &&
7709
0
   (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) ||
7710
194k
  (entity->content != NULL)) {
7711
0
  xmlFatalErr(ctxt, XML_ERR_ARGUMENT,
7712
0
              "xmlLoadEntityContent parameter error");
7713
0
        return(-1);
7714
0
    }
7715
7716
194k
    if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY)
7717
194k
        rtype = XML_RESOURCE_PARAMETER_ENTITY;
7718
0
    else
7719
0
        rtype = XML_RESOURCE_GENERAL_ENTITY;
7720
7721
194k
    input = xmlLoadResource(ctxt, (char *) entity->URI,
7722
194k
                            (char *) entity->ExternalID, rtype);
7723
194k
    if (input == NULL)
7724
3.10k
        return(-1);
7725
7726
191k
    oldinput = ctxt->input;
7727
191k
    oldinputNr = ctxt->inputNr;
7728
191k
    oldinputMax = ctxt->inputMax;
7729
191k
    oldinputTab = ctxt->inputTab;
7730
191k
    oldencoding = ctxt->encoding;
7731
7732
191k
    ctxt->input = NULL;
7733
191k
    ctxt->inputNr = 0;
7734
191k
    ctxt->inputMax = 1;
7735
191k
    ctxt->encoding = NULL;
7736
191k
    ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr));
7737
191k
    if (ctxt->inputTab == NULL) {
7738
22
        xmlErrMemory(ctxt);
7739
22
        xmlFreeInputStream(input);
7740
22
        goto error;
7741
22
    }
7742
7743
190k
    xmlBufResetInput(input->buf->buffer, input);
7744
7745
190k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
7746
57
        xmlFreeInputStream(input);
7747
57
        goto error;
7748
57
    }
7749
7750
190k
    xmlDetectEncoding(ctxt);
7751
7752
    /*
7753
     * Parse a possible text declaration first
7754
     */
7755
190k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
7756
157k
  xmlParseTextDecl(ctxt);
7757
        /*
7758
         * An XML-1.0 document can't reference an entity not XML-1.0
7759
         */
7760
157k
        if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
7761
51.8k
            (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
7762
16.8k
            xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
7763
16.8k
                           "Version mismatch between document and entity\n");
7764
16.8k
        }
7765
157k
    }
7766
7767
190k
    length = input->cur - input->base;
7768
190k
    xmlBufShrink(input->buf->buffer, length);
7769
190k
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7770
7771
306k
    while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0)
7772
115k
        ;
7773
7774
190k
    xmlBufResetInput(input->buf->buffer, input);
7775
7776
190k
    if (res < 0) {
7777
11.6k
        xmlCtxtErrIO(ctxt, input->buf->error, NULL);
7778
11.6k
        goto error;
7779
11.6k
    }
7780
7781
179k
    length = xmlBufUse(input->buf->buffer);
7782
179k
    if (length > INT_MAX) {
7783
0
        xmlErrMemory(ctxt);
7784
0
        goto error;
7785
0
    }
7786
7787
179k
    content = xmlStrndup(xmlBufContent(input->buf->buffer), length);
7788
179k
    if (content == NULL) {
7789
62
        xmlErrMemory(ctxt);
7790
62
        goto error;
7791
62
    }
7792
7793
80.1M
    for (i = 0; i < length; ) {
7794
80.1M
        int clen = length - i;
7795
80.1M
        int c = xmlGetUTF8Char(content + i, &clen);
7796
7797
80.1M
        if ((c < 0) || (!IS_CHAR(c))) {
7798
178k
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
7799
178k
                              "xmlLoadEntityContent: invalid char value %d\n",
7800
178k
                              content[i]);
7801
178k
            goto error;
7802
178k
        }
7803
79.9M
        i += clen;
7804
79.9M
    }
7805
7806
690
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7807
690
    entity->content = content;
7808
690
    entity->length = length;
7809
690
    content = NULL;
7810
690
    ret = 0;
7811
7812
191k
error:
7813
381k
    while (ctxt->inputNr > 0)
7814
190k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
7815
191k
    xmlFree(ctxt->inputTab);
7816
191k
    xmlFree(ctxt->encoding);
7817
7818
191k
    ctxt->input = oldinput;
7819
191k
    ctxt->inputNr = oldinputNr;
7820
191k
    ctxt->inputMax = oldinputMax;
7821
191k
    ctxt->inputTab = oldinputTab;
7822
191k
    ctxt->encoding = oldencoding;
7823
7824
191k
    xmlFree(content);
7825
7826
191k
    return(ret);
7827
690
}
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
314k
xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) {
7860
314k
    const xmlChar *ptr;
7861
314k
    xmlChar cur;
7862
314k
    xmlChar *name;
7863
314k
    xmlEntityPtr entity = NULL;
7864
7865
314k
    if ((str == NULL) || (*str == NULL)) return(NULL);
7866
314k
    ptr = *str;
7867
314k
    cur = *ptr;
7868
314k
    if (cur != '%')
7869
0
        return(NULL);
7870
314k
    ptr++;
7871
314k
    name = xmlParseStringName(ctxt, &ptr);
7872
314k
    if (name == NULL) {
7873
12.5k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7874
12.5k
           "xmlParseStringPEReference: no name\n");
7875
12.5k
  *str = ptr;
7876
12.5k
  return(NULL);
7877
12.5k
    }
7878
301k
    cur = *ptr;
7879
301k
    if (cur != ';') {
7880
17.6k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7881
17.6k
  xmlFree(name);
7882
17.6k
  *str = ptr;
7883
17.6k
  return(NULL);
7884
17.6k
    }
7885
284k
    ptr++;
7886
7887
    /* Must be set before xmlHandleUndeclaredEntity */
7888
284k
    ctxt->hasPErefs = 1;
7889
7890
    /*
7891
     * Request the entity from SAX
7892
     */
7893
284k
    if ((ctxt->sax != NULL) &&
7894
284k
  (ctxt->sax->getParameterEntity != NULL))
7895
284k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7896
7897
284k
    if (entity == NULL) {
7898
38.3k
        xmlHandleUndeclaredEntity(ctxt, name);
7899
245k
    } else {
7900
  /*
7901
   * Internal checking in case the entity quest barfed
7902
   */
7903
245k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7904
202k
      (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
245k
    }
7910
7911
284k
    xmlFree(name);
7912
284k
    *str = ptr;
7913
284k
    return(entity);
7914
301k
}
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
214k
xmlParseDocTypeDecl(xmlParserCtxt *ctxt) {
7933
214k
    const xmlChar *name = NULL;
7934
214k
    xmlChar *publicId = NULL;
7935
214k
    xmlChar *URI = NULL;
7936
7937
    /*
7938
     * We know that '<!DOCTYPE' has been detected.
7939
     */
7940
214k
    SKIP(9);
7941
7942
214k
    if (SKIP_BLANKS == 0) {
7943
36.3k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7944
36.3k
                       "Space required after 'DOCTYPE'\n");
7945
36.3k
    }
7946
7947
    /*
7948
     * Parse the DOCTYPE name.
7949
     */
7950
214k
    name = xmlParseName(ctxt);
7951
214k
    if (name == NULL) {
7952
31.9k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7953
31.9k
           "xmlParseDocTypeDecl : no DOCTYPE name !\n");
7954
31.9k
    }
7955
214k
    ctxt->intSubName = name;
7956
7957
214k
    SKIP_BLANKS;
7958
7959
    /*
7960
     * Check for public and system identifier (URI)
7961
     */
7962
214k
    URI = xmlParseExternalID(ctxt, &publicId, 1);
7963
7964
214k
    if ((URI != NULL) || (publicId != NULL)) {
7965
116k
        ctxt->hasExternalSubset = 1;
7966
116k
    }
7967
214k
    ctxt->extSubURI = URI;
7968
214k
    ctxt->extSubSystem = publicId;
7969
7970
214k
    SKIP_BLANKS;
7971
7972
    /*
7973
     * Create and update the internal subset.
7974
     */
7975
214k
    if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) &&
7976
214k
  (!ctxt->disableSAX))
7977
188k
  ctxt->sax->internalSubset(ctxt->userData, name, publicId, URI);
7978
7979
214k
    if ((RAW != '[') && (RAW != '>')) {
7980
10.3k
  xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
7981
10.3k
    }
7982
214k
}
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
171k
xmlParseInternalSubset(xmlParserCtxtPtr ctxt) {
7993
    /*
7994
     * Is there any DTD definition ?
7995
     */
7996
171k
    if (RAW == '[') {
7997
171k
        int oldInputNr = ctxt->inputNr;
7998
7999
171k
        NEXT;
8000
  /*
8001
   * Parse the succession of Markup declarations and
8002
   * PEReferences.
8003
   * Subsequence (markupdecl | PEReference | S)*
8004
   */
8005
171k
  SKIP_BLANKS;
8006
2.29M
        while (1) {
8007
2.29M
            if (PARSER_STOPPED(ctxt)) {
8008
9.64k
                return;
8009
2.28M
            } else if (ctxt->input->cur >= ctxt->input->end) {
8010
370k
                if (ctxt->inputNr <= oldInputNr) {
8011
20.1k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8012
20.1k
                    return;
8013
20.1k
                }
8014
349k
                xmlPopPE(ctxt);
8015
1.91M
            } else if ((RAW == ']') && (ctxt->inputNr <= oldInputNr)) {
8016
72.8k
                NEXT;
8017
72.8k
                SKIP_BLANKS;
8018
72.8k
                break;
8019
1.84M
            } else if ((PARSER_EXTERNAL(ctxt)) &&
8020
1.06M
                       (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
12.5k
                xmlParseConditionalSections(ctxt);
8026
1.82M
            } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
8027
1.12M
                xmlParseMarkupDecl(ctxt);
8028
1.12M
            } else if (RAW == '%') {
8029
633k
                xmlParsePERefInternal(ctxt, 1);
8030
633k
            } else {
8031
68.4k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8032
8033
72.8k
                while (ctxt->inputNr > oldInputNr)
8034
4.42k
                    xmlPopPE(ctxt);
8035
68.4k
                return;
8036
68.4k
            }
8037
2.12M
            SKIP_BLANKS;
8038
2.12M
            SHRINK;
8039
2.12M
            GROW;
8040
2.12M
        }
8041
171k
    }
8042
8043
    /*
8044
     * We should be at the end of the DOCTYPE declaration.
8045
     */
8046
72.8k
    if (RAW != '>') {
8047
1.93k
        xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
8048
1.93k
        return;
8049
1.93k
    }
8050
70.8k
    NEXT;
8051
70.8k
}
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
372k
xmlParseAttribute(xmlParserCtxt *ctxt, xmlChar **value) {
8089
372k
    const xmlChar *name;
8090
372k
    xmlChar *val;
8091
8092
372k
    *value = NULL;
8093
372k
    GROW;
8094
372k
    name = xmlParseName(ctxt);
8095
372k
    if (name == NULL) {
8096
92.3k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8097
92.3k
                 "error parsing attribute name\n");
8098
92.3k
        return(NULL);
8099
92.3k
    }
8100
8101
    /*
8102
     * read the value
8103
     */
8104
280k
    SKIP_BLANKS;
8105
280k
    if (RAW == '=') {
8106
238k
        NEXT;
8107
238k
  SKIP_BLANKS;
8108
238k
  val = xmlParseAttValue(ctxt);
8109
238k
    } else {
8110
41.3k
  xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8111
41.3k
         "Specification mandates value for attribute %s\n", name);
8112
41.3k
  return(name);
8113
41.3k
    }
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
238k
    if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) {
8121
26.5k
  if (!xmlCheckLanguageID(val)) {
8122
19.5k
      xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8123
19.5k
              "Malformed value for xml:lang : %s\n",
8124
19.5k
        val, NULL);
8125
19.5k
  }
8126
26.5k
    }
8127
8128
    /*
8129
     * Check that xml:space conforms to the specification
8130
     */
8131
238k
    if (xmlStrEqual(name, BAD_CAST "xml:space")) {
8132
4.13k
  if (xmlStrEqual(val, BAD_CAST "default"))
8133
1.51k
      *(ctxt->space) = 0;
8134
2.62k
  else if (xmlStrEqual(val, BAD_CAST "preserve"))
8135
1.54k
      *(ctxt->space) = 1;
8136
1.08k
  else {
8137
1.08k
    xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8138
1.08k
"Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8139
1.08k
                                 val, NULL);
8140
1.08k
  }
8141
4.13k
    }
8142
8143
238k
    *value = val;
8144
238k
    return(name);
8145
280k
}
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
840k
xmlParseStartTag(xmlParserCtxt *ctxt) {
8176
840k
    const xmlChar *name;
8177
840k
    const xmlChar *attname;
8178
840k
    xmlChar *attvalue;
8179
840k
    const xmlChar **atts = ctxt->atts;
8180
840k
    int nbatts = 0;
8181
840k
    int maxatts = ctxt->maxatts;
8182
840k
    int i;
8183
8184
840k
    if (RAW != '<') return(NULL);
8185
840k
    NEXT1;
8186
8187
840k
    name = xmlParseName(ctxt);
8188
840k
    if (name == NULL) {
8189
62.5k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8190
62.5k
       "xmlParseStartTag: invalid element name\n");
8191
62.5k
        return(NULL);
8192
62.5k
    }
8193
8194
    /*
8195
     * Now parse the attributes, it ends up with the ending
8196
     *
8197
     * (S Attribute)* S?
8198
     */
8199
778k
    SKIP_BLANKS;
8200
778k
    GROW;
8201
8202
928k
    while (((RAW != '>') &&
8203
462k
     ((RAW != '/') || (NXT(1) != '>')) &&
8204
399k
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8205
372k
  attname = xmlParseAttribute(ctxt, &attvalue);
8206
372k
        if (attname == NULL)
8207
92.3k
      break;
8208
280k
        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
657k
      for (i = 0; i < nbatts;i += 2) {
8215
446k
          if (xmlStrEqual(atts[i], attname)) {
8216
18.6k
        xmlErrAttributeDup(ctxt, NULL, attname);
8217
18.6k
        goto failed;
8218
18.6k
    }
8219
446k
      }
8220
      /*
8221
       * Add the pair to atts
8222
       */
8223
210k
      if (nbatts + 4 > maxatts) {
8224
41.5k
          const xmlChar **n;
8225
41.5k
                int newSize;
8226
8227
41.5k
                newSize = xmlGrowCapacity(maxatts, sizeof(n[0]) * 2,
8228
41.5k
                                          11, XML_MAX_ATTRS);
8229
41.5k
                if (newSize < 0) {
8230
0
        xmlErrMemory(ctxt);
8231
0
        goto failed;
8232
0
    }
8233
41.5k
#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
8234
41.5k
                if (newSize < 2)
8235
30.7k
                    newSize = 2;
8236
41.5k
#endif
8237
41.5k
          n = xmlRealloc(atts, newSize * sizeof(n[0]) * 2);
8238
41.5k
    if (n == NULL) {
8239
22
        xmlErrMemory(ctxt);
8240
22
        goto failed;
8241
22
    }
8242
41.4k
    atts = n;
8243
41.4k
                maxatts = newSize * 2;
8244
41.4k
    ctxt->atts = atts;
8245
41.4k
    ctxt->maxatts = maxatts;
8246
41.4k
      }
8247
8248
210k
      atts[nbatts++] = attname;
8249
210k
      atts[nbatts++] = attvalue;
8250
210k
      atts[nbatts] = NULL;
8251
210k
      atts[nbatts + 1] = NULL;
8252
8253
210k
            attvalue = NULL;
8254
210k
  }
8255
8256
280k
failed:
8257
8258
280k
        if (attvalue != NULL)
8259
18.6k
            xmlFree(attvalue);
8260
8261
280k
  GROW
8262
280k
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
8263
130k
      break;
8264
149k
  if (SKIP_BLANKS == 0) {
8265
71.0k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
8266
71.0k
         "attributes construct error\n");
8267
71.0k
  }
8268
149k
  SHRINK;
8269
149k
        GROW;
8270
149k
    }
8271
8272
    /*
8273
     * SAX: Start of Element !
8274
     */
8275
778k
    if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) &&
8276
778k
  (!ctxt->disableSAX)) {
8277
734k
  if (nbatts > 0)
8278
163k
      ctxt->sax->startElement(ctxt->userData, name, atts);
8279
570k
  else
8280
570k
      ctxt->sax->startElement(ctxt->userData, name, NULL);
8281
734k
    }
8282
8283
778k
    if (atts != NULL) {
8284
        /* Free only the content strings */
8285
577k
        for (i = 1;i < nbatts;i+=2)
8286
210k
      if (atts[i] != NULL)
8287
210k
         xmlFree((xmlChar *) atts[i]);
8288
366k
    }
8289
778k
    return(name);
8290
778k
}
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
163k
xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) {
8306
163k
    const xmlChar *name;
8307
8308
163k
    GROW;
8309
163k
    if ((RAW != '<') || (NXT(1) != '/')) {
8310
2.41k
  xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED,
8311
2.41k
           "xmlParseEndTag: '</' not found\n");
8312
2.41k
  return;
8313
2.41k
    }
8314
161k
    SKIP(2);
8315
8316
161k
    name = xmlParseNameAndCompare(ctxt,ctxt->name);
8317
8318
    /*
8319
     * We should definitely be at the ending "S? '>'" part
8320
     */
8321
161k
    GROW;
8322
161k
    SKIP_BLANKS;
8323
161k
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
8324
21.4k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
8325
21.4k
    } else
8326
140k
  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
161k
    if (name != (xmlChar*)1) {
8335
17.1k
        if (name == NULL) name = BAD_CAST "unparsable";
8336
17.1k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
8337
17.1k
         "Opening and ending tag mismatch: %s line %d and %s\n",
8338
17.1k
                    ctxt->name, line, name);
8339
17.1k
    }
8340
8341
    /*
8342
     * SAX: End of Tag
8343
     */
8344
161k
    if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
8345
161k
  (!ctxt->disableSAX))
8346
148k
        ctxt->sax->endElement(ctxt->userData, ctxt->name);
8347
8348
161k
    namePop(ctxt);
8349
161k
    spacePop(ctxt);
8350
161k
}
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
8.27M
xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) {
8391
8.27M
    xmlHashedString l, p;
8392
8.27M
    int start, isNCName = 0;
8393
8394
8.27M
    l.name = NULL;
8395
8.27M
    p.name = NULL;
8396
8397
8.27M
    GROW;
8398
8.27M
    start = CUR_PTR - BASE_PTR;
8399
8400
8.27M
    l = xmlParseNCName(ctxt);
8401
8.27M
    if (l.name != NULL) {
8402
6.64M
        isNCName = 1;
8403
6.64M
        if (CUR == ':') {
8404
1.82M
            NEXT;
8405
1.82M
            p = l;
8406
1.82M
            l = xmlParseNCName(ctxt);
8407
1.82M
        }
8408
6.64M
    }
8409
8.27M
    if ((l.name == NULL) || (CUR == ':')) {
8410
1.69M
        xmlChar *tmp;
8411
8412
1.69M
        l.name = NULL;
8413
1.69M
        p.name = NULL;
8414
1.69M
        if ((isNCName == 0) && (CUR != ':'))
8415
1.46M
            return(l);
8416
234k
        tmp = xmlParseNmtoken(ctxt);
8417
234k
        if (tmp != NULL)
8418
184k
            xmlFree(tmp);
8419
234k
        l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start,
8420
234k
                                CUR_PTR - (BASE_PTR + start));
8421
234k
        if (l.name == NULL) {
8422
13
            xmlErrMemory(ctxt);
8423
13
            return(l);
8424
13
        }
8425
234k
        xmlNsErr(ctxt, XML_NS_ERR_QNAME,
8426
234k
                 "Failed to parse QName '%s'\n", l.name, NULL, NULL);
8427
234k
    }
8428
8429
6.80M
    *prefix = p;
8430
6.80M
    return(l);
8431
8.27M
}
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
33.6k
xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) {
8447
33.6k
    xmlHashedString n, p;
8448
8449
33.6k
    n = xmlParseQNameHashed(ctxt, &p);
8450
33.6k
    if (n.name == NULL)
8451
7.45k
        return(NULL);
8452
26.1k
    *prefix = p.name;
8453
26.1k
    return(n.name);
8454
33.6k
}
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
239k
                        xmlChar const *prefix) {
8470
239k
    const xmlChar *cmp;
8471
239k
    const xmlChar *in;
8472
239k
    const xmlChar *ret;
8473
239k
    const xmlChar *prefix2;
8474
8475
239k
    if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name));
8476
8477
239k
    GROW;
8478
239k
    in = ctxt->input->cur;
8479
8480
239k
    cmp = prefix;
8481
780k
    while (*in != 0 && *in == *cmp) {
8482
541k
  ++in;
8483
541k
  ++cmp;
8484
541k
    }
8485
239k
    if ((*cmp == 0) && (*in == ':')) {
8486
220k
        in++;
8487
220k
  cmp = name;
8488
1.81M
  while (*in != 0 && *in == *cmp) {
8489
1.58M
      ++in;
8490
1.58M
      ++cmp;
8491
1.58M
  }
8492
220k
  if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
8493
      /* success */
8494
205k
            ctxt->input->col += in - ctxt->input->cur;
8495
205k
      ctxt->input->cur = in;
8496
205k
      return((const xmlChar*) 1);
8497
205k
  }
8498
220k
    }
8499
    /*
8500
     * all strings coms from the dictionary, equality can be done directly
8501
     */
8502
33.6k
    ret = xmlParseQName (ctxt, &prefix2);
8503
33.6k
    if (ret == NULL)
8504
7.45k
        return(NULL);
8505
26.1k
    if ((ret == name) && (prefix == prefix2))
8506
6.31k
  return((const xmlChar*) 1);
8507
19.8k
    return ret;
8508
26.1k
}
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
3.40M
{
8529
3.40M
    xmlHashedString hname;
8530
3.40M
    const xmlChar *prefix, *name;
8531
3.40M
    xmlChar *val = NULL, *internal_val = NULL;
8532
3.40M
    int special = 0;
8533
3.40M
    int isNamespace;
8534
3.40M
    int flags;
8535
8536
3.40M
    *value = NULL;
8537
3.40M
    GROW;
8538
3.40M
    hname = xmlParseQNameHashed(ctxt, hprefix);
8539
3.40M
    if (hname.name == NULL) {
8540
929k
        xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8541
929k
                       "error parsing attribute name\n");
8542
929k
        return(hname);
8543
929k
    }
8544
2.47M
    name = hname.name;
8545
2.47M
    prefix = hprefix->name;
8546
8547
    /*
8548
     * get the type if needed
8549
     */
8550
2.47M
    if (ctxt->attsSpecial != NULL) {
8551
344k
        special = XML_PTR_TO_INT(xmlHashQLookup2(ctxt->attsSpecial, pref, elem,
8552
344k
                                              prefix, name));
8553
344k
    }
8554
8555
    /*
8556
     * read the value
8557
     */
8558
2.47M
    SKIP_BLANKS;
8559
2.47M
    if (RAW != '=') {
8560
120k
        xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8561
120k
                          "Specification mandates value for attribute %s\n",
8562
120k
                          name);
8563
120k
        goto error;
8564
120k
    }
8565
8566
8567
2.35M
    NEXT;
8568
2.35M
    SKIP_BLANKS;
8569
2.35M
    flags = 0;
8570
2.35M
    isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) ||
8571
2.11M
                   (prefix == ctxt->str_xmlns));
8572
2.35M
    val = xmlParseAttValueInternal(ctxt, len, &flags, special,
8573
2.35M
                                   isNamespace);
8574
2.35M
    if (val == NULL)
8575
39.6k
        goto error;
8576
8577
2.31M
    *alloc = (flags & XML_ATTVAL_ALLOC) != 0;
8578
8579
2.31M
#ifdef LIBXML_VALID_ENABLED
8580
2.31M
    if ((ctxt->validate) &&
8581
175k
        (ctxt->standalone == 1) &&
8582
7.83k
        (special & XML_SPECIAL_EXTERNAL) &&
8583
518
        (flags & XML_ATTVAL_NORM_CHANGE)) {
8584
233
        xmlValidityError(ctxt, XML_DTD_NOT_STANDALONE,
8585
233
                         "standalone: normalization of attribute %s on %s "
8586
233
                         "by external subset declaration\n",
8587
233
                         name, elem);
8588
233
    }
8589
2.31M
#endif
8590
8591
2.31M
    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
168k
        if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) {
8598
79.6k
            internal_val = xmlStrndup(val, *len);
8599
79.6k
            if (internal_val == NULL)
8600
27
                goto mem_error;
8601
79.6k
            if (!xmlCheckLanguageID(internal_val)) {
8602
70.6k
                xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8603
70.6k
                              "Malformed value for xml:lang : %s\n",
8604
70.6k
                              internal_val, NULL);
8605
70.6k
            }
8606
79.6k
        }
8607
8608
        /*
8609
         * Check that xml:space conforms to the specification
8610
         */
8611
168k
        if (xmlStrEqual(name, BAD_CAST "space")) {
8612
7.90k
            internal_val = xmlStrndup(val, *len);
8613
7.90k
            if (internal_val == NULL)
8614
16
                goto mem_error;
8615
7.88k
            if (xmlStrEqual(internal_val, BAD_CAST "default"))
8616
4.08k
                *(ctxt->space) = 0;
8617
3.80k
            else if (xmlStrEqual(internal_val, BAD_CAST "preserve"))
8618
2.17k
                *(ctxt->space) = 1;
8619
1.63k
            else {
8620
1.63k
                xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8621
1.63k
                              "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8622
1.63k
                              internal_val, NULL);
8623
1.63k
            }
8624
7.88k
        }
8625
168k
        if (internal_val) {
8626
87.5k
            xmlFree(internal_val);
8627
87.5k
        }
8628
168k
    }
8629
8630
2.31M
    *value = val;
8631
2.31M
    return (hname);
8632
8633
43
mem_error:
8634
43
    xmlErrMemory(ctxt);
8635
160k
error:
8636
160k
    if ((val != NULL) && (*alloc != 0))
8637
14
        xmlFree(val);
8638
160k
    return(hname);
8639
43
}
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
1.52M
                  const xmlChar *uri, unsigned hashValue, int aindex) {
8656
1.52M
    xmlAttrHashBucket *table = ctxt->attrHash;
8657
1.52M
    xmlAttrHashBucket *bucket;
8658
1.52M
    unsigned hindex;
8659
8660
1.52M
    hindex = hashValue & (size - 1);
8661
1.52M
    bucket = &table[hindex];
8662
8663
1.79M
    while (bucket->index >= 0) {
8664
526k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8665
8666
526k
        if (name == atts[0]) {
8667
265k
            int nsIndex = XML_PTR_TO_INT(atts[2]);
8668
8669
265k
            if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) :
8670
265k
                (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) :
8671
103k
                (uri == ctxt->nsTab[nsIndex * 2 + 1]))
8672
253k
                return(bucket->index);
8673
265k
        }
8674
8675
273k
        hindex++;
8676
273k
        bucket++;
8677
273k
        if (hindex >= size) {
8678
35.0k
            hindex = 0;
8679
35.0k
            bucket = table;
8680
35.0k
        }
8681
273k
    }
8682
8683
1.26M
    bucket->index = aindex;
8684
8685
1.26M
    return(INT_MAX);
8686
1.52M
}
8687
8688
static int
8689
xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size,
8690
                       const xmlChar *name, const xmlChar *prefix,
8691
23.9k
                       unsigned hashValue, int aindex) {
8692
23.9k
    xmlAttrHashBucket *table = ctxt->attrHash;
8693
23.9k
    xmlAttrHashBucket *bucket;
8694
23.9k
    unsigned hindex;
8695
8696
23.9k
    hindex = hashValue & (size - 1);
8697
23.9k
    bucket = &table[hindex];
8698
8699
35.8k
    while (bucket->index >= 0) {
8700
21.9k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8701
8702
21.9k
        if ((name == atts[0]) && (prefix == atts[1]))
8703
10.0k
            return(bucket->index);
8704
8705
11.8k
        hindex++;
8706
11.8k
        bucket++;
8707
11.8k
        if (hindex >= size) {
8708
1.65k
            hindex = 0;
8709
1.65k
            bucket = table;
8710
1.65k
        }
8711
11.8k
    }
8712
8713
13.8k
    bucket->index = aindex;
8714
8715
13.8k
    return(INT_MAX);
8716
23.9k
}
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
4.83M
                  const xmlChar **URI, int *nbNsPtr) {
8750
4.83M
    xmlHashedString hlocalname;
8751
4.83M
    xmlHashedString hprefix;
8752
4.83M
    xmlHashedString hattname;
8753
4.83M
    xmlHashedString haprefix;
8754
4.83M
    const xmlChar *localname;
8755
4.83M
    const xmlChar *prefix;
8756
4.83M
    const xmlChar *attname;
8757
4.83M
    const xmlChar *aprefix;
8758
4.83M
    const xmlChar *uri;
8759
4.83M
    xmlChar *attvalue = NULL;
8760
4.83M
    const xmlChar **atts = ctxt->atts;
8761
4.83M
    unsigned attrHashSize = 0;
8762
4.83M
    int maxatts = ctxt->maxatts;
8763
4.83M
    int nratts, nbatts, nbdef;
8764
4.83M
    int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts;
8765
4.83M
    int alloc = 0;
8766
4.83M
    int numNsErr = 0;
8767
4.83M
    int numDupErr = 0;
8768
8769
4.83M
    if (RAW != '<') return(NULL);
8770
4.83M
    NEXT1;
8771
8772
4.83M
    nbatts = 0;
8773
4.83M
    nratts = 0;
8774
4.83M
    nbdef = 0;
8775
4.83M
    nbNs = 0;
8776
4.83M
    nbTotalDef = 0;
8777
4.83M
    attval = 0;
8778
8779
4.83M
    if (xmlParserNsStartElement(ctxt->nsdb) < 0) {
8780
0
        xmlErrMemory(ctxt);
8781
0
        return(NULL);
8782
0
    }
8783
8784
4.83M
    hlocalname = xmlParseQNameHashed(ctxt, &hprefix);
8785
4.83M
    if (hlocalname.name == NULL) {
8786
527k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8787
527k
           "StartTag: invalid element name\n");
8788
527k
        return(NULL);
8789
527k
    }
8790
4.30M
    localname = hlocalname.name;
8791
4.30M
    prefix = hprefix.name;
8792
8793
    /*
8794
     * Now parse the attributes, it ends up with the ending
8795
     *
8796
     * (S Attribute)* S?
8797
     */
8798
4.30M
    SKIP_BLANKS;
8799
4.30M
    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
5.60M
    while (((RAW != '>') &&
8826
3.88M
     ((RAW != '/') || (NXT(1) != '>')) &&
8827
3.48M
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8828
3.40M
  int len = -1;
8829
8830
3.40M
  hattname = xmlParseAttribute2(ctxt, prefix, localname,
8831
3.40M
                                          &haprefix, &attvalue, &len,
8832
3.40M
                                          &alloc);
8833
3.40M
        if (hattname.name == NULL)
8834
929k
      break;
8835
2.47M
        if (attvalue == NULL)
8836
160k
            goto next_attr;
8837
2.31M
        attname = hattname.name;
8838
2.31M
        aprefix = haprefix.name;
8839
2.31M
  if (len < 0) len = xmlStrlen(attvalue);
8840
8841
2.31M
        if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
8842
232k
            xmlHashedString huri;
8843
232k
            xmlURIPtr parsedUri;
8844
8845
232k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8846
232k
            uri = huri.name;
8847
232k
            if (uri == NULL) {
8848
17
                xmlErrMemory(ctxt);
8849
17
                goto next_attr;
8850
17
            }
8851
232k
            if (*uri != 0) {
8852
220k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8853
214
                    xmlErrMemory(ctxt);
8854
214
                    goto next_attr;
8855
214
                }
8856
220k
                if (parsedUri == NULL) {
8857
139k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8858
139k
                             "xmlns: '%s' is not a valid URI\n",
8859
139k
                                       uri, NULL, NULL);
8860
139k
                } else {
8861
80.9k
                    if (parsedUri->scheme == NULL) {
8862
49.9k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8863
49.9k
                                  "xmlns: URI %s is not absolute\n",
8864
49.9k
                                  uri, NULL, NULL);
8865
49.9k
                    }
8866
80.9k
                    xmlFreeURI(parsedUri);
8867
80.9k
                }
8868
220k
                if (uri == ctxt->str_xml_ns) {
8869
1.40k
                    if (attname != ctxt->str_xml) {
8870
1.40k
                        xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8871
1.40k
                     "xml namespace URI cannot be the default namespace\n",
8872
1.40k
                                 NULL, NULL, NULL);
8873
1.40k
                    }
8874
1.40k
                    goto next_attr;
8875
1.40k
                }
8876
219k
                if ((len == 29) &&
8877
5.90k
                    (xmlStrEqual(uri,
8878
5.90k
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8879
1.49k
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8880
1.49k
                         "reuse of the xmlns namespace name is forbidden\n",
8881
1.49k
                             NULL, NULL, NULL);
8882
1.49k
                    goto next_attr;
8883
1.49k
                }
8884
219k
            }
8885
8886
229k
            if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0)
8887
185k
                nbNs++;
8888
2.08M
        } else if (aprefix == ctxt->str_xmlns) {
8889
259k
            xmlHashedString huri;
8890
259k
            xmlURIPtr parsedUri;
8891
8892
259k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8893
259k
            uri = huri.name;
8894
259k
            if (uri == NULL) {
8895
13
                xmlErrMemory(ctxt);
8896
13
                goto next_attr;
8897
13
            }
8898
8899
259k
            if (attname == ctxt->str_xml) {
8900
2.76k
                if (uri != ctxt->str_xml_ns) {
8901
1.93k
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8902
1.93k
                             "xml namespace prefix mapped to wrong URI\n",
8903
1.93k
                             NULL, NULL, NULL);
8904
1.93k
                }
8905
                /*
8906
                 * Do not keep a namespace definition node
8907
                 */
8908
2.76k
                goto next_attr;
8909
2.76k
            }
8910
256k
            if (uri == ctxt->str_xml_ns) {
8911
1.27k
                if (attname != ctxt->str_xml) {
8912
1.27k
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8913
1.27k
                             "xml namespace URI mapped to wrong prefix\n",
8914
1.27k
                             NULL, NULL, NULL);
8915
1.27k
                }
8916
1.27k
                goto next_attr;
8917
1.27k
            }
8918
255k
            if (attname == ctxt->str_xmlns) {
8919
1.86k
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8920
1.86k
                         "redefinition of the xmlns prefix is forbidden\n",
8921
1.86k
                         NULL, NULL, NULL);
8922
1.86k
                goto next_attr;
8923
1.86k
            }
8924
253k
            if ((len == 29) &&
8925
3.34k
                (xmlStrEqual(uri,
8926
3.34k
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8927
1.23k
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8928
1.23k
                         "reuse of the xmlns namespace name is forbidden\n",
8929
1.23k
                         NULL, NULL, NULL);
8930
1.23k
                goto next_attr;
8931
1.23k
            }
8932
252k
            if ((uri == NULL) || (uri[0] == 0)) {
8933
4.69k
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8934
4.69k
                         "xmlns:%s: Empty XML namespace is not allowed\n",
8935
4.69k
                              attname, NULL, NULL);
8936
4.69k
                goto next_attr;
8937
247k
            } else {
8938
247k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8939
89
                    xmlErrMemory(ctxt);
8940
89
                    goto next_attr;
8941
89
                }
8942
247k
                if (parsedUri == NULL) {
8943
68.8k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8944
68.8k
                         "xmlns:%s: '%s' is not a valid URI\n",
8945
68.8k
                                       attname, uri, NULL);
8946
178k
                } else {
8947
178k
                    if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) {
8948
35.5k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8949
35.5k
                                  "xmlns:%s: URI %s is not absolute\n",
8950
35.5k
                                  attname, uri, NULL);
8951
35.5k
                    }
8952
178k
                    xmlFreeURI(parsedUri);
8953
178k
                }
8954
247k
            }
8955
8956
247k
            if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0)
8957
231k
                nbNs++;
8958
1.82M
        } else {
8959
            /*
8960
             * Populate attributes array, see above for repurposing
8961
             * of xmlChar pointers.
8962
             */
8963
1.82M
            if ((atts == NULL) || (nbatts + 5 > maxatts)) {
8964
115k
                int res = xmlCtxtGrowAttrs(ctxt);
8965
8966
115k
                maxatts = ctxt->maxatts;
8967
115k
                atts = ctxt->atts;
8968
8969
115k
                if (res < 0)
8970
120
                    goto next_attr;
8971
115k
            }
8972
1.82M
            ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) |
8973
1.82M
                                        ((unsigned) alloc << 31);
8974
1.82M
            atts[nbatts++] = attname;
8975
1.82M
            atts[nbatts++] = aprefix;
8976
1.82M
            atts[nbatts++] = XML_INT_TO_PTR(haprefix.hashValue);
8977
1.82M
            if (alloc) {
8978
193k
                atts[nbatts++] = attvalue;
8979
193k
                attvalue += len;
8980
193k
                atts[nbatts++] = attvalue;
8981
1.63M
            } 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
1.63M
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8988
1.63M
                attvalue += len;
8989
1.63M
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8990
1.63M
            }
8991
            /*
8992
             * tag if some deallocation is needed
8993
             */
8994
1.82M
            if (alloc != 0) attval = 1;
8995
1.82M
            attvalue = NULL; /* moved into atts */
8996
1.82M
        }
8997
8998
2.47M
next_attr:
8999
2.47M
        if ((attvalue != NULL) && (alloc != 0)) {
9000
100k
            xmlFree(attvalue);
9001
100k
            attvalue = NULL;
9002
100k
        }
9003
9004
2.47M
  GROW
9005
2.47M
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
9006
895k
      break;
9007
1.58M
  if (SKIP_BLANKS == 0) {
9008
282k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
9009
282k
         "attributes construct error\n");
9010
282k
      break;
9011
282k
  }
9012
1.29M
        GROW;
9013
1.29M
    }
9014
9015
    /*
9016
     * Namespaces from default attributes
9017
     */
9018
4.30M
    if (ctxt->attsDefault != NULL) {
9019
1.03M
        xmlDefAttrsPtr defaults;
9020
9021
1.03M
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9022
1.03M
  if (defaults != NULL) {
9023
3.33M
      for (i = 0; i < defaults->nbAttrs; i++) {
9024
2.53M
                xmlDefAttr *attr = &defaults->attrs[i];
9025
9026
2.53M
          attname = attr->name.name;
9027
2.53M
    aprefix = attr->prefix.name;
9028
9029
2.53M
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
9030
126k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9031
9032
126k
                    if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0)
9033
119k
                        nbNs++;
9034
2.40M
    } else if (aprefix == ctxt->str_xmlns) {
9035
1.42M
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9036
9037
1.42M
                    if (xmlParserNsPush(ctxt, &attr->name, &attr->value,
9038
1.42M
                                      NULL, 1) > 0)
9039
1.38M
                        nbNs++;
9040
1.42M
    } else {
9041
988k
                    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
988k
                    nbTotalDef += 1;
9047
988k
                }
9048
2.53M
      }
9049
795k
  }
9050
1.03M
    }
9051
9052
    /*
9053
     * Resolve attribute namespaces
9054
     */
9055
6.12M
    for (i = 0; i < nbatts; i += 5) {
9056
1.82M
        attname = atts[i];
9057
1.82M
        aprefix = atts[i+1];
9058
9059
        /*
9060
  * The default namespace does not apply to attribute names.
9061
  */
9062
1.82M
  if (aprefix == NULL) {
9063
1.17M
            nsIndex = NS_INDEX_EMPTY;
9064
1.17M
        } else if (aprefix == ctxt->str_xml) {
9065
168k
            nsIndex = NS_INDEX_XML;
9066
484k
        } else {
9067
484k
            haprefix.name = aprefix;
9068
484k
            haprefix.hashValue = (size_t) atts[i+2];
9069
484k
            nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL);
9070
9071
484k
      if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) {
9072
417k
                xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9073
417k
        "Namespace prefix %s for %s on %s is not defined\n",
9074
417k
        aprefix, attname, localname);
9075
417k
                nsIndex = NS_INDEX_EMPTY;
9076
417k
            }
9077
484k
        }
9078
9079
1.82M
        atts[i+2] = XML_INT_TO_PTR(nsIndex);
9080
1.82M
    }
9081
9082
    /*
9083
     * Maximum number of attributes including default attributes.
9084
     */
9085
4.30M
    maxAtts = nratts + nbTotalDef;
9086
9087
    /*
9088
     * Verify that attribute names are unique.
9089
     */
9090
4.30M
    if (maxAtts > 1) {
9091
375k
        attrHashSize = 4;
9092
582k
        while (attrHashSize / 2 < (unsigned) maxAtts)
9093
206k
            attrHashSize *= 2;
9094
9095
375k
        if (attrHashSize > ctxt->attrHashMax) {
9096
39.0k
            xmlAttrHashBucket *tmp;
9097
9098
39.0k
            tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0]));
9099
39.0k
            if (tmp == NULL) {
9100
28
                xmlErrMemory(ctxt);
9101
28
                goto done;
9102
28
            }
9103
9104
39.0k
            ctxt->attrHash = tmp;
9105
39.0k
            ctxt->attrHashMax = attrHashSize;
9106
39.0k
        }
9107
9108
375k
        memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0]));
9109
9110
1.55M
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9111
1.18M
            const xmlChar *nsuri;
9112
1.18M
            unsigned hashValue, nameHashValue, uriHashValue;
9113
1.18M
            int res;
9114
9115
1.18M
            attname = atts[i];
9116
1.18M
            aprefix = atts[i+1];
9117
1.18M
            nsIndex = XML_PTR_TO_INT(atts[i+2]);
9118
            /* Hash values always have bit 31 set, see dict.c */
9119
1.18M
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9120
9121
1.18M
            if (nsIndex == NS_INDEX_EMPTY) {
9122
                /*
9123
                 * Prefix with empty namespace means an undeclared
9124
                 * prefix which was already reported above.
9125
                 */
9126
1.01M
                if (aprefix != NULL)
9127
405k
                    continue;
9128
605k
                nsuri = NULL;
9129
605k
                uriHashValue = URI_HASH_EMPTY;
9130
605k
            } else if (nsIndex == NS_INDEX_XML) {
9131
113k
                nsuri = ctxt->str_xml_ns;
9132
113k
                uriHashValue = URI_HASH_XML;
9133
113k
            } else {
9134
57.9k
                nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9135
57.9k
                uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9136
57.9k
            }
9137
9138
777k
            hashValue = xmlDictCombineHash(nameHashValue, uriHashValue);
9139
777k
            res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9140
777k
                                    hashValue, i);
9141
777k
            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
777k
            if (res < INT_MAX) {
9151
208k
                if (aprefix == atts[res+1]) {
9152
194k
                    xmlErrAttributeDup(ctxt, aprefix, attname);
9153
194k
                    numDupErr += 1;
9154
194k
                } else {
9155
14.2k
                    xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9156
14.2k
                             "Namespaced Attribute %s in '%s' redefined\n",
9157
14.2k
                             attname, nsuri, NULL);
9158
14.2k
                    numNsErr += 1;
9159
14.2k
                }
9160
208k
            }
9161
777k
        }
9162
375k
    }
9163
9164
    /*
9165
     * Default attributes
9166
     */
9167
4.30M
    if (ctxt->attsDefault != NULL) {
9168
1.03M
        xmlDefAttrsPtr defaults;
9169
9170
1.03M
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9171
1.03M
  if (defaults != NULL) {
9172
3.33M
      for (i = 0; i < defaults->nbAttrs; i++) {
9173
2.53M
                xmlDefAttr *attr = &defaults->attrs[i];
9174
2.53M
                const xmlChar *nsuri = NULL;
9175
2.53M
                unsigned hashValue, uriHashValue = 0;
9176
2.53M
                int res;
9177
9178
2.53M
          attname = attr->name.name;
9179
2.53M
    aprefix = attr->prefix.name;
9180
9181
2.53M
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL))
9182
126k
                    continue;
9183
2.40M
    if (aprefix == ctxt->str_xmlns)
9184
1.42M
                    continue;
9185
9186
988k
                if (aprefix == NULL) {
9187
565k
                    nsIndex = NS_INDEX_EMPTY;
9188
565k
                    nsuri = NULL;
9189
565k
                    uriHashValue = URI_HASH_EMPTY;
9190
565k
                } else if (aprefix == ctxt->str_xml) {
9191
68.3k
                    nsIndex = NS_INDEX_XML;
9192
68.3k
                    nsuri = ctxt->str_xml_ns;
9193
68.3k
                    uriHashValue = URI_HASH_XML;
9194
354k
                } else {
9195
354k
                    nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL);
9196
354k
                    if ((nsIndex == INT_MAX) ||
9197
313k
                        (nsIndex < ctxt->nsdb->minNsIndex)) {
9198
313k
                        xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9199
313k
                                 "Namespace prefix %s for %s on %s is not "
9200
313k
                                 "defined\n",
9201
313k
                                 aprefix, attname, localname);
9202
313k
                        nsIndex = NS_INDEX_EMPTY;
9203
313k
                        nsuri = NULL;
9204
313k
                        uriHashValue = URI_HASH_EMPTY;
9205
313k
                    } else {
9206
41.5k
                        nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9207
41.5k
                        uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9208
41.5k
                    }
9209
354k
                }
9210
9211
                /*
9212
                 * Check whether the attribute exists
9213
                 */
9214
988k
                if (maxAtts > 1) {
9215
743k
                    hashValue = xmlDictCombineHash(attr->name.hashValue,
9216
743k
                                                   uriHashValue);
9217
743k
                    res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9218
743k
                                            hashValue, nbatts);
9219
743k
                    if (res < 0)
9220
0
                        continue;
9221
743k
                    if (res < INT_MAX) {
9222
44.2k
                        if (aprefix == atts[res+1])
9223
18.2k
                            continue;
9224
25.9k
                        xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9225
25.9k
                                 "Namespaced Attribute %s in '%s' redefined\n",
9226
25.9k
                                 attname, nsuri, NULL);
9227
25.9k
                    }
9228
743k
                }
9229
9230
970k
                xmlParserEntityCheck(ctxt, attr->expandedSize);
9231
9232
970k
                if ((atts == NULL) || (nbatts + 5 > maxatts)) {
9233
20.4k
                    res = xmlCtxtGrowAttrs(ctxt);
9234
9235
20.4k
                    maxatts = ctxt->maxatts;
9236
20.4k
                    atts = ctxt->atts;
9237
9238
20.4k
                    if (res < 0) {
9239
52
                        localname = NULL;
9240
52
                        goto done;
9241
52
                    }
9242
20.4k
                }
9243
9244
970k
                atts[nbatts++] = attname;
9245
970k
                atts[nbatts++] = aprefix;
9246
970k
                atts[nbatts++] = XML_INT_TO_PTR(nsIndex);
9247
970k
                atts[nbatts++] = attr->value.name;
9248
970k
                atts[nbatts++] = attr->valueEnd;
9249
9250
970k
#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
970k
                if ((ctxt->validate) &&
9256
311k
                    (ctxt->standalone == 1) &&
9257
18.1k
                    (attr->external != 0)) {
9258
763
                    xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED,
9259
763
                            "standalone: attribute %s on %s defaulted "
9260
763
                            "from external subset\n",
9261
763
                            attname, localname);
9262
763
                }
9263
970k
#endif
9264
970k
                nbdef++;
9265
970k
      }
9266
795k
  }
9267
1.03M
    }
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
4.30M
    if ((numDupErr == 0) && (numNsErr > 1)) {
9282
4.14k
        memset(ctxt->attrHash, -1,
9283
4.14k
               attrHashSize * sizeof(ctxt->attrHash[0]));
9284
9285
30.1k
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9286
25.9k
            unsigned hashValue, nameHashValue, prefixHashValue;
9287
25.9k
            int res;
9288
9289
25.9k
            aprefix = atts[i+1];
9290
25.9k
            if (aprefix == NULL)
9291
2.05k
                continue;
9292
9293
23.9k
            attname = atts[i];
9294
            /* Hash values always have bit 31 set, see dict.c */
9295
23.9k
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9296
23.9k
            prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix);
9297
9298
23.9k
            hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue);
9299
23.9k
            res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname,
9300
23.9k
                                         aprefix, hashValue, i);
9301
23.9k
            if (res < INT_MAX)
9302
10.0k
                xmlErrAttributeDup(ctxt, aprefix, attname);
9303
23.9k
        }
9304
4.14k
    }
9305
9306
    /*
9307
     * Reconstruct attribute pointers
9308
     */
9309
7.09M
    for (i = 0, j = 0; i < nbatts; i += 5, j++) {
9310
        /* namespace URI */
9311
2.79M
        nsIndex = XML_PTR_TO_INT(atts[i+2]);
9312
2.79M
        if (nsIndex == INT_MAX)
9313
2.45M
            atts[i+2] = NULL;
9314
339k
        else if (nsIndex == INT_MAX - 1)
9315
235k
            atts[i+2] = ctxt->str_xml_ns;
9316
104k
        else
9317
104k
            atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1];
9318
9319
2.79M
        if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) {
9320
1.63M
            atts[i+3] = BASE_PTR + XML_PTR_TO_INT(atts[i+3]);  /* value */
9321
1.63M
            atts[i+4] = BASE_PTR + XML_PTR_TO_INT(atts[i+4]);  /* valuend */
9322
1.63M
        }
9323
2.79M
    }
9324
9325
4.30M
    uri = xmlParserNsLookupUri(ctxt, &hprefix);
9326
4.30M
    if ((prefix != NULL) && (uri == NULL)) {
9327
132k
  xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9328
132k
           "Namespace prefix %s on %s is not defined\n",
9329
132k
     prefix, localname, NULL);
9330
132k
    }
9331
4.30M
    *pref = prefix;
9332
4.30M
    *URI = uri;
9333
9334
    /*
9335
     * SAX callback
9336
     */
9337
4.30M
    if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) &&
9338
4.30M
  (!ctxt->disableSAX)) {
9339
3.31M
  if (nbNs > 0)
9340
469k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9341
469k
                          nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs),
9342
469k
        nbatts / 5, nbdef, atts);
9343
2.84M
  else
9344
2.84M
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9345
2.84M
                          0, NULL, nbatts / 5, nbdef, atts);
9346
3.31M
    }
9347
9348
4.30M
done:
9349
    /*
9350
     * Free allocated attribute values
9351
     */
9352
4.30M
    if (attval != 0) {
9353
598k
  for (i = 0, j = 0; j < nratts; i += 5, j++)
9354
430k
      if (ctxt->attallocs[j] & 0x80000000)
9355
193k
          xmlFree((xmlChar *) atts[i+3]);
9356
167k
    }
9357
9358
4.30M
    *nbNsPtr = nbNs;
9359
4.30M
    return(localname);
9360
4.30M
}
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
1.16M
xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) {
9376
1.16M
    const xmlChar *name;
9377
9378
1.16M
    GROW;
9379
1.16M
    if ((RAW != '<') || (NXT(1) != '/')) {
9380
5.27k
  xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL);
9381
5.27k
  return;
9382
5.27k
    }
9383
1.15M
    SKIP(2);
9384
9385
1.15M
    if (tag->prefix == NULL)
9386
920k
        name = xmlParseNameAndCompare(ctxt, ctxt->name);
9387
239k
    else
9388
239k
        name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix);
9389
9390
    /*
9391
     * We should definitely be at the ending "S? '>'" part
9392
     */
9393
1.15M
    GROW;
9394
1.15M
    SKIP_BLANKS;
9395
1.15M
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
9396
79.2k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
9397
79.2k
    } else
9398
1.08M
  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
1.15M
    if (name != (xmlChar*)1) {
9407
83.8k
        if (name == NULL) name = BAD_CAST "unparsable";
9408
83.8k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
9409
83.8k
         "Opening and ending tag mismatch: %s line %d and %s\n",
9410
83.8k
                    ctxt->name, tag->line, name);
9411
83.8k
    }
9412
9413
    /*
9414
     * SAX: End of Tag
9415
     */
9416
1.15M
    if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9417
1.15M
  (!ctxt->disableSAX))
9418
924k
  ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix,
9419
924k
                                tag->URI);
9420
9421
1.15M
    spacePop(ctxt);
9422
1.15M
    if (tag->nsNr != 0)
9423
89.4k
  xmlParserNsPop(ctxt, tag->nsNr);
9424
1.15M
}
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
167k
xmlParseCDSect(xmlParserCtxt *ctxt) {
9442
167k
    xmlChar *buf = NULL;
9443
167k
    int len = 0;
9444
167k
    int size = XML_PARSER_BUFFER_SIZE;
9445
167k
    int r, rl;
9446
167k
    int s, sl;
9447
167k
    int cur, l;
9448
167k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9449
87.4k
                    XML_MAX_HUGE_LENGTH :
9450
167k
                    XML_MAX_TEXT_LENGTH;
9451
9452
167k
    if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '['))
9453
0
        return;
9454
167k
    SKIP(3);
9455
9456
167k
    if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '['))
9457
0
        return;
9458
167k
    SKIP(6);
9459
9460
167k
    r = xmlCurrentCharRecover(ctxt, &rl);
9461
167k
    if (!IS_CHAR(r)) {
9462
7.84k
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9463
7.84k
        goto out;
9464
7.84k
    }
9465
159k
    NEXTL(rl);
9466
159k
    s = xmlCurrentCharRecover(ctxt, &sl);
9467
159k
    if (!IS_CHAR(s)) {
9468
5.48k
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9469
5.48k
        goto out;
9470
5.48k
    }
9471
154k
    NEXTL(sl);
9472
154k
    cur = xmlCurrentCharRecover(ctxt, &l);
9473
154k
    buf = xmlMalloc(size);
9474
154k
    if (buf == NULL) {
9475
32
  xmlErrMemory(ctxt);
9476
32
        goto out;
9477
32
    }
9478
10.1M
    while (IS_CHAR(cur) &&
9479
10.0M
           ((r != ']') || (s != ']') || (cur != '>'))) {
9480
9.96M
  if (len + 5 >= size) {
9481
18.5k
      xmlChar *tmp;
9482
18.5k
            int newSize;
9483
9484
18.5k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9485
18.5k
            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
18.5k
      tmp = xmlRealloc(buf, newSize);
9491
18.5k
      if (tmp == NULL) {
9492
15
    xmlErrMemory(ctxt);
9493
15
                goto out;
9494
15
      }
9495
18.5k
      buf = tmp;
9496
18.5k
      size = newSize;
9497
18.5k
  }
9498
9.96M
  COPY_BUF(buf, len, r);
9499
9.96M
  r = s;
9500
9.96M
  rl = sl;
9501
9.96M
  s = cur;
9502
9.96M
  sl = l;
9503
9.96M
  NEXTL(l);
9504
9.96M
  cur = xmlCurrentCharRecover(ctxt, &l);
9505
9.96M
    }
9506
154k
    buf[len] = 0;
9507
154k
    if (cur != '>') {
9508
48.4k
  xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9509
48.4k
                       "CData section not finished\n%.50s\n", buf);
9510
48.4k
        goto out;
9511
48.4k
    }
9512
105k
    NEXTL(l);
9513
9514
    /*
9515
     * OK the buffer is to be consumed as cdata.
9516
     */
9517
105k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) {
9518
102k
        if ((ctxt->sax->cdataBlock != NULL) &&
9519
102k
            ((ctxt->options & XML_PARSE_NOCDATA) == 0)) {
9520
86.4k
            ctxt->sax->cdataBlock(ctxt->userData, buf, len);
9521
86.4k
        } else if (ctxt->sax->characters != NULL) {
9522
15.8k
            ctxt->sax->characters(ctxt->userData, buf, len);
9523
15.8k
        }
9524
102k
    }
9525
9526
167k
out:
9527
167k
    xmlFree(buf);
9528
167k
}
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
156k
xmlParseContentInternal(xmlParserCtxtPtr ctxt) {
9539
156k
    int oldNameNr = ctxt->nameNr;
9540
156k
    int oldSpaceNr = ctxt->spaceNr;
9541
156k
    int oldNodeNr = ctxt->nodeNr;
9542
9543
156k
    GROW;
9544
81.9M
    while ((ctxt->input->cur < ctxt->input->end) &&
9545
81.8M
     (PARSER_STOPPED(ctxt) == 0)) {
9546
81.8M
  const xmlChar *cur = ctxt->input->cur;
9547
9548
  /*
9549
   * First case : a Processing Instruction.
9550
   */
9551
81.8M
  if ((*cur == '<') && (cur[1] == '?')) {
9552
96.4k
      xmlParsePI(ctxt);
9553
96.4k
  }
9554
9555
  /*
9556
   * Second case : a CDSection
9557
   */
9558
  /* 2.6.0 test was *cur not RAW */
9559
81.7M
  else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) {
9560
149k
      xmlParseCDSect(ctxt);
9561
149k
  }
9562
9563
  /*
9564
   * Third case :  a comment
9565
   */
9566
81.5M
  else if ((*cur == '<') && (NXT(1) == '!') &&
9567
354k
     (NXT(2) == '-') && (NXT(3) == '-')) {
9568
194k
      xmlParseComment(ctxt);
9569
194k
  }
9570
9571
  /*
9572
   * Fourth case :  a sub-element.
9573
   */
9574
81.3M
  else if (*cur == '<') {
9575
6.13M
            if (NXT(1) == '/') {
9576
1.25M
                if (ctxt->nameNr <= oldNameNr)
9577
62.5k
                    break;
9578
1.19M
          xmlParseElementEnd(ctxt);
9579
4.87M
            } else {
9580
4.87M
          xmlParseElementStart(ctxt);
9581
4.87M
            }
9582
6.13M
  }
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
75.2M
  else if (*cur == '&') {
9590
1.04M
      xmlParseReference(ctxt);
9591
1.04M
  }
9592
9593
  /*
9594
   * Last case, text. Note that References are handled directly.
9595
   */
9596
74.2M
  else {
9597
74.2M
      xmlParseCharDataInternal(ctxt, 0);
9598
74.2M
  }
9599
9600
81.7M
  SHRINK;
9601
81.7M
  GROW;
9602
81.7M
    }
9603
9604
156k
    if ((ctxt->nameNr > oldNameNr) &&
9605
36.6k
        (ctxt->input->cur >= ctxt->input->end) &&
9606
31.9k
        (ctxt->wellFormed)) {
9607
1.67k
        const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9608
1.67k
        int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9609
1.67k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9610
1.67k
                "Premature end of data in tag %s line %d\n",
9611
1.67k
                name, line, NULL);
9612
1.67k
    }
9613
9614
    /*
9615
     * Clean up in error case
9616
     */
9617
9618
1.00M
    while (ctxt->nodeNr > oldNodeNr)
9619
851k
        nodePop(ctxt);
9620
9621
1.26M
    while (ctxt->nameNr > oldNameNr) {
9622
1.10M
        xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1];
9623
9624
1.10M
        if (tag->nsNr != 0)
9625
239k
            xmlParserNsPop(ctxt, tag->nsNr);
9626
9627
1.10M
        namePop(ctxt);
9628
1.10M
    }
9629
9630
1.26M
    while (ctxt->spaceNr > oldSpaceNr)
9631
1.10M
        spacePop(ctxt);
9632
156k
}
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
178k
xmlParseElement(xmlParserCtxt *ctxt) {
9669
178k
    if (xmlParseElementStart(ctxt) != 0)
9670
54.7k
        return;
9671
9672
123k
    xmlParseContentInternal(ctxt);
9673
9674
123k
    if (ctxt->input->cur >= ctxt->input->end) {
9675
52.7k
        if (ctxt->wellFormed) {
9676
2.76k
            const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9677
2.76k
            int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9678
2.76k
            xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9679
2.76k
                    "Premature end of data in tag %s line %d\n",
9680
2.76k
                    name, line, NULL);
9681
2.76k
        }
9682
52.7k
        return;
9683
52.7k
    }
9684
9685
71.0k
    xmlParseElementEnd(ctxt);
9686
71.0k
}
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
5.05M
xmlParseElementStart(xmlParserCtxtPtr ctxt) {
9698
5.05M
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
9699
5.05M
    const xmlChar *name;
9700
5.05M
    const xmlChar *prefix = NULL;
9701
5.05M
    const xmlChar *URI = NULL;
9702
5.05M
    xmlParserNodeInfo node_info;
9703
5.05M
    int line;
9704
5.05M
    xmlNodePtr cur;
9705
5.05M
    int nbNs = 0;
9706
9707
5.05M
    if (ctxt->nameNr > maxDepth) {
9708
263
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
9709
263
                "Excessive depth in document: %d use XML_PARSE_HUGE option\n",
9710
263
                ctxt->nameNr);
9711
263
  return(-1);
9712
263
    }
9713
9714
    /* Capture start position */
9715
5.05M
    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
5.05M
    if (ctxt->spaceNr == 0)
9722
86.3k
  spacePush(ctxt, -1);
9723
4.97M
    else if (*ctxt->space == -2)
9724
775k
  spacePush(ctxt, -1);
9725
4.19M
    else
9726
4.19M
  spacePush(ctxt, *ctxt->space);
9727
9728
5.05M
    line = ctxt->input->line;
9729
5.05M
#ifdef LIBXML_SAX1_ENABLED
9730
5.05M
    if (ctxt->sax2)
9731
4.37M
#endif /* LIBXML_SAX1_ENABLED */
9732
4.37M
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
9733
679k
#ifdef LIBXML_SAX1_ENABLED
9734
679k
    else
9735
679k
  name = xmlParseStartTag(ctxt);
9736
5.05M
#endif /* LIBXML_SAX1_ENABLED */
9737
5.05M
    if (name == NULL) {
9738
581k
  spacePop(ctxt);
9739
581k
        return(-1);
9740
581k
    }
9741
4.47M
    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
9742
4.47M
    cur = ctxt->node;
9743
9744
4.47M
#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
4.47M
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
9751
69.4k
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
9752
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
9753
4.47M
#endif /* LIBXML_VALID_ENABLED */
9754
9755
    /*
9756
     * Check for an Empty Element.
9757
     */
9758
4.47M
    if ((RAW == '/') && (NXT(1) == '>')) {
9759
756k
        SKIP(2);
9760
756k
  if (ctxt->sax2) {
9761
676k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9762
676k
    (!ctxt->disableSAX))
9763
598k
    ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI);
9764
676k
#ifdef LIBXML_SAX1_ENABLED
9765
676k
  } else {
9766
79.5k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
9767
79.5k
    (!ctxt->disableSAX))
9768
77.5k
    ctxt->sax->endElement(ctxt->userData, name);
9769
79.5k
#endif /* LIBXML_SAX1_ENABLED */
9770
79.5k
  }
9771
756k
  namePop(ctxt);
9772
756k
  spacePop(ctxt);
9773
756k
  if (nbNs > 0)
9774
77.1k
      xmlParserNsPop(ctxt, nbNs);
9775
756k
  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
756k
  return(1);
9783
756k
    }
9784
3.71M
    if (RAW == '>') {
9785
2.42M
        NEXT1;
9786
2.42M
        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
2.42M
    } else {
9793
1.29M
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED,
9794
1.29M
         "Couldn't find end of Start Tag %s line %d\n",
9795
1.29M
                    name, line, NULL);
9796
9797
  /*
9798
   * end of parsing of this node.
9799
   */
9800
1.29M
  nodePop(ctxt);
9801
1.29M
  namePop(ctxt);
9802
1.29M
  spacePop(ctxt);
9803
1.29M
  if (nbNs > 0)
9804
382k
      xmlParserNsPop(ctxt, nbNs);
9805
1.29M
  return(-1);
9806
1.29M
    }
9807
9808
2.42M
    return(0);
9809
3.71M
}
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
1.26M
xmlParseElementEnd(xmlParserCtxtPtr ctxt) {
9818
1.26M
    xmlNodePtr cur = ctxt->node;
9819
9820
1.26M
    if (ctxt->nameNr <= 0) {
9821
176
        if ((RAW == '<') && (NXT(1) == '/'))
9822
57
            SKIP(2);
9823
176
        return;
9824
176
    }
9825
9826
    /*
9827
     * parse the end of tag: '</' should be here.
9828
     */
9829
1.26M
    if (ctxt->sax2) {
9830
1.11M
  xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
9831
1.11M
  namePop(ctxt);
9832
1.11M
    }
9833
149k
#ifdef LIBXML_SAX1_ENABLED
9834
149k
    else
9835
149k
  xmlParseEndTag1(ctxt, 0);
9836
1.26M
#endif /* LIBXML_SAX1_ENABLED */
9837
9838
    /*
9839
     * Capture end position
9840
     */
9841
1.26M
    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
1.26M
}
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
185k
xmlParseVersionNum(xmlParserCtxt *ctxt) {
9867
185k
    xmlChar *buf = NULL;
9868
185k
    int len = 0;
9869
185k
    int size = 10;
9870
185k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9871
37.2k
                    XML_MAX_TEXT_LENGTH :
9872
185k
                    XML_MAX_NAME_LENGTH;
9873
185k
    xmlChar cur;
9874
9875
185k
    buf = xmlMalloc(size);
9876
185k
    if (buf == NULL) {
9877
144
  xmlErrMemory(ctxt);
9878
144
  return(NULL);
9879
144
    }
9880
185k
    cur = CUR;
9881
185k
    if (!((cur >= '0') && (cur <= '9'))) {
9882
7.96k
  xmlFree(buf);
9883
7.96k
  return(NULL);
9884
7.96k
    }
9885
177k
    buf[len++] = cur;
9886
177k
    NEXT;
9887
177k
    cur=CUR;
9888
177k
    if (cur != '.') {
9889
5.02k
  xmlFree(buf);
9890
5.02k
  return(NULL);
9891
5.02k
    }
9892
172k
    buf[len++] = cur;
9893
172k
    NEXT;
9894
172k
    cur=CUR;
9895
5.01M
    while ((cur >= '0') && (cur <= '9')) {
9896
4.83M
  if (len + 1 >= size) {
9897
6.11k
      xmlChar *tmp;
9898
6.11k
            int newSize;
9899
9900
6.11k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9901
6.11k
            if (newSize < 0) {
9902
25
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "VersionNum");
9903
25
                xmlFree(buf);
9904
25
                return(NULL);
9905
25
            }
9906
6.08k
      tmp = xmlRealloc(buf, newSize);
9907
6.08k
      if (tmp == NULL) {
9908
13
    xmlErrMemory(ctxt);
9909
13
          xmlFree(buf);
9910
13
    return(NULL);
9911
13
      }
9912
6.07k
      buf = tmp;
9913
6.07k
            size = newSize;
9914
6.07k
  }
9915
4.83M
  buf[len++] = cur;
9916
4.83M
  NEXT;
9917
4.83M
  cur=CUR;
9918
4.83M
    }
9919
172k
    buf[len] = 0;
9920
172k
    return(buf);
9921
172k
}
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
393k
xmlParseVersionInfo(xmlParserCtxt *ctxt) {
9938
393k
    xmlChar *version = NULL;
9939
9940
393k
    if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) {
9941
192k
  SKIP(7);
9942
192k
  SKIP_BLANKS;
9943
192k
  if (RAW != '=') {
9944
4.41k
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
9945
4.41k
      return(NULL);
9946
4.41k
        }
9947
187k
  NEXT;
9948
187k
  SKIP_BLANKS;
9949
187k
  if (RAW == '"') {
9950
164k
      NEXT;
9951
164k
      version = xmlParseVersionNum(ctxt);
9952
164k
      if (RAW != '"') {
9953
9.25k
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9954
9.25k
      } else
9955
154k
          NEXT;
9956
164k
  } else if (RAW == '\''){
9957
21.4k
      NEXT;
9958
21.4k
      version = xmlParseVersionNum(ctxt);
9959
21.4k
      if (RAW != '\'') {
9960
5.74k
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9961
5.74k
      } else
9962
15.6k
          NEXT;
9963
21.4k
  } else {
9964
2.33k
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
9965
2.33k
  }
9966
187k
    }
9967
388k
    return(version);
9968
393k
}
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
91.8k
xmlParseEncName(xmlParserCtxt *ctxt) {
9982
91.8k
    xmlChar *buf = NULL;
9983
91.8k
    int len = 0;
9984
91.8k
    int size = 10;
9985
91.8k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9986
15.9k
                    XML_MAX_TEXT_LENGTH :
9987
91.8k
                    XML_MAX_NAME_LENGTH;
9988
91.8k
    xmlChar cur;
9989
9990
91.8k
    cur = CUR;
9991
91.8k
    if (((cur >= 'a') && (cur <= 'z')) ||
9992
82.3k
        ((cur >= 'A') && (cur <= 'Z'))) {
9993
81.6k
  buf = xmlMalloc(size);
9994
81.6k
  if (buf == NULL) {
9995
104
      xmlErrMemory(ctxt);
9996
104
      return(NULL);
9997
104
  }
9998
9999
81.5k
  buf[len++] = cur;
10000
81.5k
  NEXT;
10001
81.5k
  cur = CUR;
10002
79.6M
  while (((cur >= 'a') && (cur <= 'z')) ||
10003
27.0M
         ((cur >= 'A') && (cur <= 'Z')) ||
10004
445k
         ((cur >= '0') && (cur <= '9')) ||
10005
189k
         (cur == '.') || (cur == '_') ||
10006
79.6M
         (cur == '-')) {
10007
79.6M
      if (len + 1 >= size) {
10008
91.0k
          xmlChar *tmp;
10009
91.0k
                int newSize;
10010
10011
91.0k
                newSize = xmlGrowCapacity(size, 1, 1, maxLength);
10012
91.0k
                if (newSize < 0) {
10013
764
                    xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName");
10014
764
                    xmlFree(buf);
10015
764
                    return(NULL);
10016
764
                }
10017
90.2k
    tmp = xmlRealloc(buf, newSize);
10018
90.2k
    if (tmp == NULL) {
10019
45
        xmlErrMemory(ctxt);
10020
45
        xmlFree(buf);
10021
45
        return(NULL);
10022
45
    }
10023
90.2k
    buf = tmp;
10024
90.2k
                size = newSize;
10025
90.2k
      }
10026
79.6M
      buf[len++] = cur;
10027
79.6M
      NEXT;
10028
79.6M
      cur = CUR;
10029
79.6M
        }
10030
80.7k
  buf[len] = 0;
10031
80.7k
    } else {
10032
10.2k
  xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL);
10033
10.2k
    }
10034
90.9k
    return(buf);
10035
91.8k
}
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
378k
xmlParseEncodingDecl(xmlParserCtxt *ctxt) {
10053
378k
    xmlChar *encoding = NULL;
10054
10055
378k
    SKIP_BLANKS;
10056
378k
    if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0)
10057
282k
        return(NULL);
10058
10059
95.9k
    SKIP(8);
10060
95.9k
    SKIP_BLANKS;
10061
95.9k
    if (RAW != '=') {
10062
2.30k
        xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10063
2.30k
        return(NULL);
10064
2.30k
    }
10065
93.5k
    NEXT;
10066
93.5k
    SKIP_BLANKS;
10067
93.5k
    if (RAW == '"') {
10068
74.4k
        NEXT;
10069
74.4k
        encoding = xmlParseEncName(ctxt);
10070
74.4k
        if (RAW != '"') {
10071
9.28k
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10072
9.28k
            xmlFree(encoding);
10073
9.28k
            return(NULL);
10074
9.28k
        } else
10075
65.2k
            NEXT;
10076
74.4k
    } else if (RAW == '\''){
10077
17.3k
        NEXT;
10078
17.3k
        encoding = xmlParseEncName(ctxt);
10079
17.3k
        if (RAW != '\'') {
10080
2.61k
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10081
2.61k
            xmlFree(encoding);
10082
2.61k
            return(NULL);
10083
2.61k
        } else
10084
14.7k
            NEXT;
10085
17.3k
    } else {
10086
1.71k
        xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10087
1.71k
    }
10088
10089
81.6k
    if (encoding == NULL)
10090
5.34k
        return(NULL);
10091
10092
76.3k
    xmlSetDeclaredEncoding(ctxt, encoding);
10093
10094
76.3k
    return(ctxt->encoding);
10095
81.6k
}
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
112k
xmlParseSDDecl(xmlParserCtxt *ctxt) {
10130
112k
    int standalone = -2;
10131
10132
112k
    SKIP_BLANKS;
10133
112k
    if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) {
10134
81.8k
  SKIP(10);
10135
81.8k
        SKIP_BLANKS;
10136
81.8k
  if (RAW != '=') {
10137
375
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10138
375
      return(standalone);
10139
375
        }
10140
81.4k
  NEXT;
10141
81.4k
  SKIP_BLANKS;
10142
81.4k
        if (RAW == '\''){
10143
2.21k
      NEXT;
10144
2.21k
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10145
559
          standalone = 0;
10146
559
                SKIP(2);
10147
1.65k
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10148
897
                 (NXT(2) == 's')) {
10149
690
          standalone = 1;
10150
690
    SKIP(3);
10151
967
            } else {
10152
967
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10153
967
      }
10154
2.21k
      if (RAW != '\'') {
10155
1.59k
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10156
1.59k
      } else
10157
620
          NEXT;
10158
79.2k
  } else if (RAW == '"'){
10159
79.0k
      NEXT;
10160
79.0k
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10161
727
          standalone = 0;
10162
727
    SKIP(2);
10163
78.3k
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10164
77.8k
                 (NXT(2) == 's')) {
10165
77.6k
          standalone = 1;
10166
77.6k
                SKIP(3);
10167
77.6k
            } else {
10168
696
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10169
696
      }
10170
79.0k
      if (RAW != '"') {
10171
1.19k
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10172
1.19k
      } else
10173
77.8k
          NEXT;
10174
79.0k
  } else {
10175
222
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10176
222
        }
10177
81.4k
    }
10178
112k
    return(standalone);
10179
112k
}
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
141k
xmlParseXMLDecl(xmlParserCtxt *ctxt) {
10192
141k
    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
141k
    ctxt->standalone = -2;
10201
10202
    /*
10203
     * We know that '<?xml' is here.
10204
     */
10205
141k
    SKIP(5);
10206
10207
141k
    if (!IS_BLANK_CH(RAW)) {
10208
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
10209
0
                 "Blank needed after '<?xml'\n");
10210
0
    }
10211
141k
    SKIP_BLANKS;
10212
10213
    /*
10214
     * We must have the VersionInfo here.
10215
     */
10216
141k
    version = xmlParseVersionInfo(ctxt);
10217
141k
    if (version == NULL) {
10218
26.8k
  xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL);
10219
114k
    } else {
10220
114k
  if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) {
10221
      /*
10222
       * Changed here for XML-1.0 5th edition
10223
       */
10224
8.15k
      if (ctxt->options & XML_PARSE_OLD10) {
10225
844
    xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10226
844
                "Unsupported version '%s'\n",
10227
844
                version);
10228
7.30k
      } else {
10229
7.30k
          if ((version[0] == '1') && ((version[1] == '.'))) {
10230
5.16k
        xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION,
10231
5.16k
                      "Unsupported version '%s'\n",
10232
5.16k
          version, NULL);
10233
5.16k
    } else {
10234
2.14k
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10235
2.14k
              "Unsupported version '%s'\n",
10236
2.14k
              version);
10237
2.14k
    }
10238
7.30k
      }
10239
8.15k
  }
10240
114k
  if (ctxt->version != NULL)
10241
0
      xmlFree(ctxt->version);
10242
114k
  ctxt->version = version;
10243
114k
    }
10244
10245
    /*
10246
     * We may have the encoding declaration
10247
     */
10248
141k
    if (!IS_BLANK_CH(RAW)) {
10249
42.1k
        if ((RAW == '?') && (NXT(1) == '>')) {
10250
14.7k
      SKIP(2);
10251
14.7k
      return;
10252
14.7k
  }
10253
27.3k
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10254
27.3k
    }
10255
126k
    xmlParseEncodingDecl(ctxt);
10256
10257
    /*
10258
     * We may have the standalone status.
10259
     */
10260
126k
    if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) {
10261
16.6k
        if ((RAW == '?') && (NXT(1) == '>')) {
10262
13.5k
      SKIP(2);
10263
13.5k
      return;
10264
13.5k
  }
10265
3.06k
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10266
3.06k
    }
10267
10268
    /*
10269
     * We can grow the input buffer freely at that point
10270
     */
10271
112k
    GROW;
10272
10273
112k
    SKIP_BLANKS;
10274
112k
    ctxt->standalone = xmlParseSDDecl(ctxt);
10275
10276
112k
    SKIP_BLANKS;
10277
112k
    if ((RAW == '?') && (NXT(1) == '>')) {
10278
80.3k
        SKIP(2);
10279
80.3k
    } else if (RAW == '>') {
10280
        /* Deprecated old WD ... */
10281
2.61k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10282
2.61k
  NEXT;
10283
29.9k
    } else {
10284
29.9k
        int c;
10285
10286
29.9k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10287
4.86M
        while ((PARSER_STOPPED(ctxt) == 0) &&
10288
4.86M
               ((c = CUR) != 0)) {
10289
4.84M
            NEXT;
10290
4.84M
            if (c == '>')
10291
15.1k
                break;
10292
4.84M
        }
10293
29.9k
    }
10294
112k
}
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
663k
xmlParseMisc(xmlParserCtxt *ctxt) {
10335
867k
    while (PARSER_STOPPED(ctxt) == 0) {
10336
835k
        SKIP_BLANKS;
10337
835k
        GROW;
10338
835k
        if ((RAW == '<') && (NXT(1) == '?')) {
10339
89.8k
      xmlParsePI(ctxt);
10340
745k
        } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) {
10341
114k
      xmlParseComment(ctxt);
10342
631k
        } else {
10343
631k
            break;
10344
631k
        }
10345
835k
    }
10346
663k
}
10347
10348
static void
10349
362k
xmlFinishDocument(xmlParserCtxtPtr ctxt) {
10350
362k
    xmlDocPtr doc;
10351
10352
    /*
10353
     * SAX: end of the document processing.
10354
     */
10355
362k
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10356
362k
        ctxt->sax->endDocument(ctxt->userData);
10357
10358
    /*
10359
     * Remove locally kept entity definitions if the tree was not built
10360
     */
10361
362k
    doc = ctxt->myDoc;
10362
362k
    if ((doc != NULL) &&
10363
339k
        (xmlStrEqual(doc->version, SAX_COMPAT_MODE))) {
10364
4.06k
        xmlFreeDoc(doc);
10365
4.06k
        ctxt->myDoc = NULL;
10366
4.06k
    }
10367
362k
}
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
319k
xmlParseDocument(xmlParserCtxt *ctxt) {
10380
319k
    if ((ctxt == NULL) || (ctxt->input == NULL))
10381
0
        return(-1);
10382
10383
319k
    GROW;
10384
10385
    /*
10386
     * SAX: detecting the level.
10387
     */
10388
319k
    xmlCtxtInitializeLate(ctxt);
10389
10390
319k
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10391
319k
        ctxt->sax->setDocumentLocator(ctxt->userData,
10392
319k
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10393
319k
    }
10394
10395
319k
    xmlDetectEncoding(ctxt);
10396
10397
319k
    if (CUR == 0) {
10398
3.17k
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10399
3.17k
  return(-1);
10400
3.17k
    }
10401
10402
316k
    GROW;
10403
316k
    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
131k
  xmlParseXMLDecl(ctxt);
10409
131k
  SKIP_BLANKS;
10410
184k
    } else {
10411
184k
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10412
184k
        if (ctxt->version == NULL) {
10413
66
            xmlErrMemory(ctxt);
10414
66
            return(-1);
10415
66
        }
10416
184k
    }
10417
315k
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10418
289k
        ctxt->sax->startDocument(ctxt->userData);
10419
315k
    if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) &&
10420
288k
        (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
315k
    xmlParseMisc(ctxt);
10428
10429
    /*
10430
     * Then possibly doc type declaration(s) and more Misc
10431
     * (doctypedecl Misc*)?
10432
     */
10433
315k
    GROW;
10434
315k
    if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) {
10435
10436
168k
  ctxt->inSubset = 1;
10437
168k
  xmlParseDocTypeDecl(ctxt);
10438
168k
  if (RAW == '[') {
10439
136k
      xmlParseInternalSubset(ctxt);
10440
136k
  } else if (RAW == '>') {
10441
23.7k
            NEXT;
10442
23.7k
        }
10443
10444
  /*
10445
   * Create and update the external subset.
10446
   */
10447
168k
  ctxt->inSubset = 2;
10448
168k
  if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) &&
10449
168k
      (!ctxt->disableSAX))
10450
123k
      ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
10451
123k
                                ctxt->extSubSystem, ctxt->extSubURI);
10452
168k
  ctxt->inSubset = 0;
10453
10454
168k
        xmlCleanSpecialAttr(ctxt);
10455
10456
168k
  xmlParseMisc(ctxt);
10457
168k
    }
10458
10459
    /*
10460
     * Time to start parsing the tree itself
10461
     */
10462
315k
    GROW;
10463
315k
    if (RAW != '<') {
10464
137k
        if (ctxt->wellFormed)
10465
22.3k
            xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10466
22.3k
                           "Start tag expected, '<' not found\n");
10467
178k
    } else {
10468
178k
  xmlParseElement(ctxt);
10469
10470
  /*
10471
   * The Misc part at the end
10472
   */
10473
178k
  xmlParseMisc(ctxt);
10474
10475
178k
        xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
10476
178k
    }
10477
10478
315k
    ctxt->instate = XML_PARSER_EOF;
10479
315k
    xmlFinishDocument(ctxt);
10480
10481
315k
    if (! ctxt->wellFormed) {
10482
258k
  ctxt->valid = 0;
10483
258k
  return(-1);
10484
258k
    }
10485
10486
57.1k
    return(0);
10487
315k
}
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
196k
xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) {
10575
196k
    const xmlChar *cur;
10576
10577
196k
    if (ctxt->checkIndex == 0) {
10578
102k
        cur = ctxt->input->cur + 1;
10579
102k
    } else {
10580
93.9k
        cur = ctxt->input->cur + ctxt->checkIndex;
10581
93.9k
    }
10582
10583
196k
    if (memchr(cur, c, ctxt->input->end - cur) == NULL) {
10584
98.7k
        size_t index = ctxt->input->end - ctxt->input->cur;
10585
10586
98.7k
        if (index > LONG_MAX) {
10587
0
            ctxt->checkIndex = 0;
10588
0
            return(1);
10589
0
        }
10590
98.7k
        ctxt->checkIndex = index;
10591
98.7k
        return(0);
10592
98.7k
    } else {
10593
97.2k
        ctxt->checkIndex = 0;
10594
97.2k
        return(1);
10595
97.2k
    }
10596
196k
}
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
669k
                     const char *str, size_t strLen) {
10609
669k
    const xmlChar *cur, *term;
10610
10611
669k
    if (ctxt->checkIndex == 0) {
10612
313k
        cur = ctxt->input->cur + startDelta;
10613
355k
    } else {
10614
355k
        cur = ctxt->input->cur + ctxt->checkIndex;
10615
355k
    }
10616
10617
669k
    term = BAD_CAST strstr((const char *) cur, str);
10618
669k
    if (term == NULL) {
10619
361k
        const xmlChar *end = ctxt->input->end;
10620
361k
        size_t index;
10621
10622
        /* Rescan (strLen - 1) characters. */
10623
361k
        if ((size_t) (end - cur) < strLen)
10624
7.61k
            end = cur;
10625
354k
        else
10626
354k
            end -= strLen - 1;
10627
361k
        index = end - ctxt->input->cur;
10628
361k
        if (index > LONG_MAX) {
10629
0
            ctxt->checkIndex = 0;
10630
0
            return(ctxt->input->end - strLen);
10631
0
        }
10632
361k
        ctxt->checkIndex = index;
10633
361k
    } else {
10634
307k
        ctxt->checkIndex = 0;
10635
307k
    }
10636
10637
669k
    return(term);
10638
669k
}
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
225k
xmlParseLookupCharData(xmlParserCtxtPtr ctxt) {
10647
225k
    const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex;
10648
225k
    const xmlChar *end = ctxt->input->end;
10649
225k
    size_t index;
10650
10651
7.18M
    while (cur < end) {
10652
7.13M
        if ((*cur == '<') || (*cur == '&')) {
10653
175k
            ctxt->checkIndex = 0;
10654
175k
            return(1);
10655
175k
        }
10656
6.96M
        cur++;
10657
6.96M
    }
10658
10659
50.0k
    index = cur - ctxt->input->cur;
10660
50.0k
    if (index > LONG_MAX) {
10661
0
        ctxt->checkIndex = 0;
10662
0
        return(1);
10663
0
    }
10664
50.0k
    ctxt->checkIndex = index;
10665
50.0k
    return(0);
10666
50.0k
}
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
1.97M
xmlParseLookupGt(xmlParserCtxtPtr ctxt) {
10676
1.97M
    const xmlChar *cur;
10677
1.97M
    const xmlChar *end = ctxt->input->end;
10678
1.97M
    int state = ctxt->endCheckState;
10679
1.97M
    size_t index;
10680
10681
1.97M
    if (ctxt->checkIndex == 0)
10682
500k
        cur = ctxt->input->cur + 1;
10683
1.47M
    else
10684
1.47M
        cur = ctxt->input->cur + ctxt->checkIndex;
10685
10686
583M
    while (cur < end) {
10687
581M
        if (state) {
10688
503M
            if (*cur == state)
10689
519k
                state = 0;
10690
503M
        } else if (*cur == '\'' || *cur == '"') {
10691
528k
            state = *cur;
10692
77.6M
        } else if (*cur == '>') {
10693
472k
            ctxt->checkIndex = 0;
10694
472k
            ctxt->endCheckState = 0;
10695
472k
            return(1);
10696
472k
        }
10697
581M
        cur++;
10698
581M
    }
10699
10700
1.49M
    index = cur - ctxt->input->cur;
10701
1.49M
    if (index > LONG_MAX) {
10702
0
        ctxt->checkIndex = 0;
10703
0
        ctxt->endCheckState = 0;
10704
0
        return(1);
10705
0
    }
10706
1.49M
    ctxt->checkIndex = index;
10707
1.49M
    ctxt->endCheckState = state;
10708
1.49M
    return(0);
10709
1.49M
}
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
1.03M
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
1.03M
    const xmlChar *cur, *start;
10727
1.03M
    const xmlChar *end = ctxt->input->end;
10728
1.03M
    int state = ctxt->endCheckState;
10729
1.03M
    size_t index;
10730
10731
1.03M
    if (ctxt->checkIndex == 0) {
10732
25.7k
        cur = ctxt->input->cur + 1;
10733
1.00M
    } else {
10734
1.00M
        cur = ctxt->input->cur + ctxt->checkIndex;
10735
1.00M
    }
10736
1.03M
    start = cur;
10737
10738
605M
    while (cur < end) {
10739
604M
        if (state == '-') {
10740
2.52M
            if ((*cur == '-') &&
10741
39.3k
                (cur[1] == '-') &&
10742
28.9k
                (cur[2] == '>')) {
10743
15.5k
                state = 0;
10744
15.5k
                cur += 3;
10745
15.5k
                start = cur;
10746
15.5k
                continue;
10747
15.5k
            }
10748
2.52M
        }
10749
602M
        else if (state == ']') {
10750
23.0k
            if (*cur == '>') {
10751
15.3k
                ctxt->checkIndex = 0;
10752
15.3k
                ctxt->endCheckState = 0;
10753
15.3k
                return(1);
10754
15.3k
            }
10755
7.64k
            if (IS_BLANK_CH(*cur)) {
10756
3.95k
                state = ' ';
10757
3.95k
            } else if (*cur != ']') {
10758
2.30k
                state = 0;
10759
2.30k
                start = cur;
10760
2.30k
                continue;
10761
2.30k
            }
10762
7.64k
        }
10763
602M
        else if (state == ' ') {
10764
10.1k
            if (*cur == '>') {
10765
373
                ctxt->checkIndex = 0;
10766
373
                ctxt->endCheckState = 0;
10767
373
                return(1);
10768
373
            }
10769
9.74k
            if (!IS_BLANK_CH(*cur)) {
10770
3.55k
                state = 0;
10771
3.55k
                start = cur;
10772
3.55k
                continue;
10773
3.55k
            }
10774
9.74k
        }
10775
602M
        else if (state != 0) {
10776
586M
            if (*cur == state) {
10777
81.2k
                state = 0;
10778
81.2k
                start = cur + 1;
10779
81.2k
            }
10780
586M
        }
10781
15.8M
        else if (*cur == '<') {
10782
134k
            if ((cur[1] == '!') &&
10783
80.3k
                (cur[2] == '-') &&
10784
16.4k
                (cur[3] == '-')) {
10785
15.7k
                state = '-';
10786
15.7k
                cur += 4;
10787
                /* Don't treat <!--> as comment */
10788
15.7k
                start = cur;
10789
15.7k
                continue;
10790
15.7k
            }
10791
134k
        }
10792
15.6M
        else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) {
10793
104k
            state = *cur;
10794
104k
        }
10795
10796
604M
        cur++;
10797
604M
    }
10798
10799
    /*
10800
     * Rescan the three last characters to detect "<!--" and "-->"
10801
     * split across chunks.
10802
     */
10803
1.01M
    if ((state == 0) || (state == '-')) {
10804
35.1k
        if (cur - start < 3)
10805
3.19k
            cur = start;
10806
31.9k
        else
10807
31.9k
            cur -= 3;
10808
35.1k
    }
10809
1.01M
    index = cur - ctxt->input->cur;
10810
1.01M
    if (index > LONG_MAX) {
10811
0
        ctxt->checkIndex = 0;
10812
0
        ctxt->endCheckState = 0;
10813
0
        return(1);
10814
0
    }
10815
1.01M
    ctxt->checkIndex = index;
10816
1.01M
    ctxt->endCheckState = state;
10817
1.01M
    return(0);
10818
1.01M
}
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
3.54M
xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) {
10829
3.54M
    int ret = 0;
10830
3.54M
    size_t avail;
10831
3.54M
    xmlChar cur, next;
10832
10833
3.54M
    if (ctxt->input == NULL)
10834
0
        return(0);
10835
10836
3.54M
    if ((ctxt->input != NULL) &&
10837
3.54M
        (ctxt->input->cur - ctxt->input->base > 4096)) {
10838
6.16k
        xmlParserShrink(ctxt);
10839
6.16k
    }
10840
10841
12.3M
    while (ctxt->disableSAX == 0) {
10842
12.3M
        avail = ctxt->input->end - ctxt->input->cur;
10843
12.3M
        if (avail < 1)
10844
59.7k
      goto done;
10845
12.2M
        switch (ctxt->instate) {
10846
369k
            case XML_PARSER_EOF:
10847
          /*
10848
     * Document parsing is done !
10849
     */
10850
369k
          goto done;
10851
100k
            case XML_PARSER_START:
10852
                /*
10853
                 * Very first chars read from the document flow.
10854
                 */
10855
100k
                if ((!terminate) && (avail < 4))
10856
9.96k
                    goto done;
10857
10858
                /*
10859
                 * We need more bytes to detect EBCDIC code pages.
10860
                 * See xmlDetectEBCDIC.
10861
                 */
10862
90.5k
                if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) &&
10863
3.33k
                    (!terminate) && (avail < 200))
10864
2.30k
                    goto done;
10865
10866
88.2k
                xmlDetectEncoding(ctxt);
10867
88.2k
                ctxt->instate = XML_PARSER_XML_DECL;
10868
88.2k
    break;
10869
10870
218k
            case XML_PARSER_XML_DECL:
10871
218k
    if ((!terminate) && (avail < 2))
10872
247
        goto done;
10873
218k
    cur = ctxt->input->cur[0];
10874
218k
    next = ctxt->input->cur[1];
10875
218k
          if ((cur == '<') && (next == '?')) {
10876
        /* PI or XML decl */
10877
148k
        if ((!terminate) &&
10878
142k
                        (!xmlParseLookupString(ctxt, 2, "?>", 2)))
10879
130k
      goto done;
10880
18.3k
        if ((ctxt->input->cur[2] == 'x') &&
10881
14.9k
      (ctxt->input->cur[3] == 'm') &&
10882
13.6k
      (ctxt->input->cur[4] == 'l') &&
10883
9.84k
      (IS_BLANK_CH(ctxt->input->cur[5]))) {
10884
9.49k
      ret += 5;
10885
9.49k
      xmlParseXMLDecl(ctxt);
10886
9.49k
        } else {
10887
8.86k
      ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10888
8.86k
                        if (ctxt->version == NULL) {
10889
10
                            xmlErrMemory(ctxt);
10890
10
                            break;
10891
10
                        }
10892
8.86k
        }
10893
69.6k
    } else {
10894
69.6k
        ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10895
69.6k
        if (ctxt->version == NULL) {
10896
91
            xmlErrMemory(ctxt);
10897
91
      break;
10898
91
        }
10899
69.6k
    }
10900
87.8k
                if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10901
87.8k
                    ctxt->sax->setDocumentLocator(ctxt->userData,
10902
87.8k
                            (xmlSAXLocator *) &xmlDefaultSAXLocator);
10903
87.8k
                }
10904
87.8k
                if ((ctxt->sax) && (ctxt->sax->startDocument) &&
10905
87.8k
                    (!ctxt->disableSAX))
10906
86.1k
                    ctxt->sax->startDocument(ctxt->userData);
10907
87.8k
                ctxt->instate = XML_PARSER_MISC;
10908
87.8k
    break;
10909
1.80M
            case XML_PARSER_START_TAG: {
10910
1.80M
          const xmlChar *name;
10911
1.80M
    const xmlChar *prefix = NULL;
10912
1.80M
    const xmlChar *URI = NULL;
10913
1.80M
                int line = ctxt->input->line;
10914
1.80M
    int nbNs = 0;
10915
10916
1.80M
    if ((!terminate) && (avail < 2))
10917
251
        goto done;
10918
1.80M
    cur = ctxt->input->cur[0];
10919
1.80M
          if (cur != '<') {
10920
7.03k
        xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10921
7.03k
                                   "Start tag expected, '<' not found");
10922
7.03k
                    ctxt->instate = XML_PARSER_EOF;
10923
7.03k
                    xmlFinishDocument(ctxt);
10924
7.03k
        goto done;
10925
7.03k
    }
10926
1.79M
    if ((!terminate) && (!xmlParseLookupGt(ctxt)))
10927
1.18M
                    goto done;
10928
613k
    if (ctxt->spaceNr == 0)
10929
0
        spacePush(ctxt, -1);
10930
613k
    else if (*ctxt->space == -2)
10931
97.5k
        spacePush(ctxt, -1);
10932
515k
    else
10933
515k
        spacePush(ctxt, *ctxt->space);
10934
613k
#ifdef LIBXML_SAX1_ENABLED
10935
613k
    if (ctxt->sax2)
10936
452k
#endif /* LIBXML_SAX1_ENABLED */
10937
452k
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
10938
160k
#ifdef LIBXML_SAX1_ENABLED
10939
160k
    else
10940
160k
        name = xmlParseStartTag(ctxt);
10941
613k
#endif /* LIBXML_SAX1_ENABLED */
10942
613k
    if (name == NULL) {
10943
7.94k
        spacePop(ctxt);
10944
7.94k
                    ctxt->instate = XML_PARSER_EOF;
10945
7.94k
                    xmlFinishDocument(ctxt);
10946
7.94k
        goto done;
10947
7.94k
    }
10948
605k
#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
605k
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
10955
144k
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
10956
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
10957
605k
#endif /* LIBXML_VALID_ENABLED */
10958
10959
    /*
10960
     * Check for an Empty Element.
10961
     */
10962
605k
    if ((RAW == '/') && (NXT(1) == '>')) {
10963
129k
        SKIP(2);
10964
10965
129k
        if (ctxt->sax2) {
10966
108k
      if ((ctxt->sax != NULL) &&
10967
108k
          (ctxt->sax->endElementNs != NULL) &&
10968
108k
          (!ctxt->disableSAX))
10969
108k
          ctxt->sax->endElementNs(ctxt->userData, name,
10970
108k
                                  prefix, URI);
10971
108k
      if (nbNs > 0)
10972
28.4k
          xmlParserNsPop(ctxt, nbNs);
10973
108k
#ifdef LIBXML_SAX1_ENABLED
10974
108k
        } else {
10975
21.2k
      if ((ctxt->sax != NULL) &&
10976
21.2k
          (ctxt->sax->endElement != NULL) &&
10977
21.2k
          (!ctxt->disableSAX))
10978
21.1k
          ctxt->sax->endElement(ctxt->userData, name);
10979
21.2k
#endif /* LIBXML_SAX1_ENABLED */
10980
21.2k
        }
10981
129k
        spacePop(ctxt);
10982
475k
    } else if (RAW == '>') {
10983
359k
        NEXT;
10984
359k
                    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
10985
359k
    } else {
10986
115k
        xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED,
10987
115k
           "Couldn't find end of Start Tag %s\n",
10988
115k
           name);
10989
115k
        nodePop(ctxt);
10990
115k
        spacePop(ctxt);
10991
115k
                    if (nbNs > 0)
10992
19.7k
                        xmlParserNsPop(ctxt, nbNs);
10993
115k
    }
10994
10995
605k
                if (ctxt->nameNr == 0)
10996
18.1k
                    ctxt->instate = XML_PARSER_EPILOG;
10997
587k
                else
10998
587k
                    ctxt->instate = XML_PARSER_CONTENT;
10999
605k
                break;
11000
613k
      }
11001
7.90M
            case XML_PARSER_CONTENT: {
11002
7.90M
    cur = ctxt->input->cur[0];
11003
11004
7.90M
    if (cur == '<') {
11005
954k
                    if ((!terminate) && (avail < 2))
11006
4.59k
                        goto done;
11007
949k
        next = ctxt->input->cur[1];
11008
11009
949k
                    if (next == '/') {
11010
67.5k
                        ctxt->instate = XML_PARSER_END_TAG;
11011
67.5k
                        break;
11012
882k
                    } else if (next == '?') {
11013
62.0k
                        if ((!terminate) &&
11014
53.3k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11015
29.2k
                            goto done;
11016
32.8k
                        xmlParsePI(ctxt);
11017
32.8k
                        ctxt->instate = XML_PARSER_CONTENT;
11018
32.8k
                        break;
11019
820k
                    } else if (next == '!') {
11020
265k
                        if ((!terminate) && (avail < 3))
11021
1.20k
                            goto done;
11022
264k
                        next = ctxt->input->cur[2];
11023
11024
264k
                        if (next == '-') {
11025
133k
                            if ((!terminate) && (avail < 4))
11026
999
                                goto done;
11027
132k
                            if (ctxt->input->cur[3] == '-') {
11028
132k
                                if ((!terminate) &&
11029
128k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11030
22.2k
                                    goto done;
11031
109k
                                xmlParseComment(ctxt);
11032
109k
                                ctxt->instate = XML_PARSER_CONTENT;
11033
109k
                                break;
11034
132k
                            }
11035
132k
                        } else if (next == '[') {
11036
129k
                            if ((!terminate) && (avail < 9))
11037
890
                                goto done;
11038
128k
                            if ((ctxt->input->cur[2] == '[') &&
11039
128k
                                (ctxt->input->cur[3] == 'C') &&
11040
128k
                                (ctxt->input->cur[4] == 'D') &&
11041
128k
                                (ctxt->input->cur[5] == 'A') &&
11042
128k
                                (ctxt->input->cur[6] == 'T') &&
11043
128k
                                (ctxt->input->cur[7] == 'A') &&
11044
128k
                                (ctxt->input->cur[8] == '[')) {
11045
128k
                                if ((!terminate) &&
11046
120k
                                    (!xmlParseLookupString(ctxt, 9, "]]>", 3)))
11047
110k
                                    goto done;
11048
18.3k
                                ctxt->instate = XML_PARSER_CDATA_SECTION;
11049
18.3k
                                xmlParseCDSect(ctxt);
11050
18.3k
                                ctxt->instate = XML_PARSER_CONTENT;
11051
18.3k
                                break;
11052
128k
                            }
11053
128k
                        }
11054
264k
                    }
11055
6.95M
    } else if (cur == '&') {
11056
107k
        if ((!terminate) && (!xmlParseLookupChar(ctxt, ';')))
11057
42.3k
      goto done;
11058
64.8k
        xmlParseReference(ctxt);
11059
64.8k
                    break;
11060
6.84M
    } 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
6.84M
        if (avail < XML_PARSER_BIG_BUFFER_SIZE) {
11074
277k
      if ((!terminate) && (!xmlParseLookupCharData(ctxt)))
11075
50.0k
          goto done;
11076
277k
                    }
11077
6.79M
                    ctxt->checkIndex = 0;
11078
6.79M
        xmlParseCharDataInternal(ctxt, !terminate);
11079
6.79M
                    break;
11080
6.84M
    }
11081
11082
556k
                ctxt->instate = XML_PARSER_START_TAG;
11083
556k
    break;
11084
7.90M
      }
11085
122k
            case XML_PARSER_END_TAG:
11086
122k
    if ((!terminate) && (!xmlParseLookupChar(ctxt, '>')))
11087
56.3k
        goto done;
11088
66.4k
    if (ctxt->sax2) {
11089
51.6k
              xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
11090
51.6k
        nameNsPop(ctxt);
11091
51.6k
    }
11092
14.8k
#ifdef LIBXML_SAX1_ENABLED
11093
14.8k
      else
11094
14.8k
        xmlParseEndTag1(ctxt, 0);
11095
66.4k
#endif /* LIBXML_SAX1_ENABLED */
11096
66.4k
    if (ctxt->nameNr == 0) {
11097
3.29k
        ctxt->instate = XML_PARSER_EPILOG;
11098
63.1k
    } else {
11099
63.1k
        ctxt->instate = XML_PARSER_CONTENT;
11100
63.1k
    }
11101
66.4k
    break;
11102
559k
            case XML_PARSER_MISC:
11103
667k
            case XML_PARSER_PROLOG:
11104
677k
            case XML_PARSER_EPILOG:
11105
677k
    SKIP_BLANKS;
11106
677k
                avail = ctxt->input->end - ctxt->input->cur;
11107
677k
    if (avail < 1)
11108
4.30k
        goto done;
11109
673k
    if (ctxt->input->cur[0] == '<') {
11110
663k
                    if ((!terminate) && (avail < 2))
11111
1.54k
                        goto done;
11112
661k
                    next = ctxt->input->cur[1];
11113
661k
                    if (next == '?') {
11114
87.2k
                        if ((!terminate) &&
11115
79.4k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11116
34.3k
                            goto done;
11117
52.8k
                        xmlParsePI(ctxt);
11118
52.8k
                        break;
11119
574k
                    } else if (next == '!') {
11120
515k
                        if ((!terminate) && (avail < 3))
11121
894
                            goto done;
11122
11123
514k
                        if (ctxt->input->cur[2] == '-') {
11124
146k
                            if ((!terminate) && (avail < 4))
11125
729
                                goto done;
11126
146k
                            if (ctxt->input->cur[3] == '-') {
11127
146k
                                if ((!terminate) &&
11128
144k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11129
35.2k
                                    goto done;
11130
110k
                                xmlParseComment(ctxt);
11131
110k
                                break;
11132
146k
                            }
11133
367k
                        } else if (ctxt->instate == XML_PARSER_MISC) {
11134
367k
                            if ((!terminate) && (avail < 9))
11135
6.56k
                                goto done;
11136
360k
                            if ((ctxt->input->cur[2] == 'D') &&
11137
360k
                                (ctxt->input->cur[3] == 'O') &&
11138
360k
                                (ctxt->input->cur[4] == 'C') &&
11139
360k
                                (ctxt->input->cur[5] == 'T') &&
11140
360k
                                (ctxt->input->cur[6] == 'Y') &&
11141
360k
                                (ctxt->input->cur[7] == 'P') &&
11142
360k
                                (ctxt->input->cur[8] == 'E')) {
11143
360k
                                if ((!terminate) && (!xmlParseLookupGt(ctxt)))
11144
314k
                                    goto done;
11145
46.0k
                                ctxt->inSubset = 1;
11146
46.0k
                                xmlParseDocTypeDecl(ctxt);
11147
46.0k
                                if (RAW == '[') {
11148
36.3k
                                    ctxt->instate = XML_PARSER_DTD;
11149
36.3k
                                } else {
11150
9.75k
                                    if (RAW == '>')
11151
7.74k
                                        NEXT;
11152
                                    /*
11153
                                     * Create and update the external subset.
11154
                                     */
11155
9.75k
                                    ctxt->inSubset = 2;
11156
9.75k
                                    if ((ctxt->sax != NULL) &&
11157
9.75k
                                        (!ctxt->disableSAX) &&
11158
9.33k
                                        (ctxt->sax->externalSubset != NULL))
11159
9.33k
                                        ctxt->sax->externalSubset(
11160
9.33k
                                                ctxt->userData,
11161
9.33k
                                                ctxt->intSubName,
11162
9.33k
                                                ctxt->extSubSystem,
11163
9.33k
                                                ctxt->extSubURI);
11164
9.75k
                                    ctxt->inSubset = 0;
11165
9.75k
                                    xmlCleanSpecialAttr(ctxt);
11166
9.75k
                                    ctxt->instate = XML_PARSER_PROLOG;
11167
9.75k
                                }
11168
46.0k
                                break;
11169
360k
                            }
11170
360k
                        }
11171
514k
                    }
11172
661k
                }
11173
11174
69.6k
                if (ctxt->instate == XML_PARSER_EPILOG) {
11175
3.76k
                    if (ctxt->errNo == XML_ERR_OK)
11176
61
                        xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL);
11177
3.76k
        ctxt->instate = XML_PARSER_EOF;
11178
3.76k
                    xmlFinishDocument(ctxt);
11179
65.9k
                } else {
11180
65.9k
        ctxt->instate = XML_PARSER_START_TAG;
11181
65.9k
    }
11182
69.6k
    break;
11183
1.05M
            case XML_PARSER_DTD: {
11184
1.05M
                if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt)))
11185
1.01M
                    goto done;
11186
34.6k
    xmlParseInternalSubset(ctxt);
11187
34.6k
    ctxt->inSubset = 2;
11188
34.6k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
11189
28.5k
        (ctxt->sax->externalSubset != NULL))
11190
28.5k
        ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
11191
28.5k
          ctxt->extSubSystem, ctxt->extSubURI);
11192
34.6k
    ctxt->inSubset = 0;
11193
34.6k
    xmlCleanSpecialAttr(ctxt);
11194
34.6k
    ctxt->instate = XML_PARSER_PROLOG;
11195
34.6k
                break;
11196
1.05M
      }
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
12.2M
  }
11203
12.2M
    }
11204
3.54M
done:
11205
3.54M
    return(ret);
11206
3.54M
}
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
4.19M
              int terminate) {
11237
4.19M
    size_t curBase;
11238
4.19M
    size_t maxLength;
11239
4.19M
    size_t pos;
11240
4.19M
    int end_in_lf = 0;
11241
4.19M
    int res;
11242
11243
4.19M
    if ((ctxt == NULL) || (size < 0))
11244
0
        return(XML_ERR_ARGUMENT);
11245
4.19M
    if ((chunk == NULL) && (size > 0))
11246
0
        return(XML_ERR_ARGUMENT);
11247
4.19M
    if ((ctxt->input == NULL) || (ctxt->input->buf == NULL))
11248
0
        return(XML_ERR_ARGUMENT);
11249
4.19M
    if (ctxt->disableSAX != 0)
11250
656k
        return(ctxt->errNo);
11251
11252
3.54M
    ctxt->input->flags |= XML_INPUT_PROGRESSIVE;
11253
3.54M
    if (ctxt->instate == XML_PARSER_START)
11254
101k
        xmlCtxtInitializeLate(ctxt);
11255
3.54M
    if ((size > 0) && (chunk != NULL) && (!terminate) &&
11256
3.46M
        (chunk[size - 1] == '\r')) {
11257
4.75k
  end_in_lf = 1;
11258
4.75k
  size--;
11259
4.75k
    }
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
3.54M
    pos = ctxt->input->cur - ctxt->input->base;
11266
3.54M
    res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk);
11267
3.54M
    xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11268
3.54M
    if (res < 0) {
11269
774
        xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11270
774
        return(ctxt->errNo);
11271
774
    }
11272
11273
3.54M
    xmlParseTryOrFinish(ctxt, terminate);
11274
11275
3.54M
    curBase = ctxt->input->cur - ctxt->input->base;
11276
3.54M
    maxLength = (ctxt->options & XML_PARSE_HUGE) ?
11277
1.15M
                XML_MAX_HUGE_LENGTH :
11278
3.54M
                XML_MAX_LOOKUP_LIMIT;
11279
3.54M
    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
3.54M
    if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0))
11285
35.0k
        return(ctxt->errNo);
11286
11287
3.50M
    if (end_in_lf == 1) {
11288
4.71k
  pos = ctxt->input->cur - ctxt->input->base;
11289
4.71k
  res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r");
11290
4.71k
  xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11291
4.71k
        if (res < 0) {
11292
12
            xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11293
12
            return(ctxt->errNo);
11294
12
        }
11295
4.71k
    }
11296
3.50M
    if (terminate) {
11297
  /*
11298
   * Check for termination
11299
   */
11300
48.0k
        if ((ctxt->instate != XML_PARSER_EOF) &&
11301
28.7k
            (ctxt->instate != XML_PARSER_EPILOG)) {
11302
19.5k
            if (ctxt->nameNr > 0) {
11303
12.0k
                const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
11304
12.0k
                int line = ctxt->pushTab[ctxt->nameNr - 1].line;
11305
12.0k
                xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
11306
12.0k
                        "Premature end of data in tag %s line %d\n",
11307
12.0k
                        name, line, NULL);
11308
12.0k
            } else if (ctxt->instate == XML_PARSER_START) {
11309
351
                xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
11310
7.19k
            } else {
11311
7.19k
                xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
11312
7.19k
                               "Start tag expected, '<' not found\n");
11313
7.19k
            }
11314
28.4k
        } else {
11315
28.4k
            xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
11316
28.4k
        }
11317
48.0k
  if (ctxt->instate != XML_PARSER_EOF) {
11318
28.1k
            ctxt->instate = XML_PARSER_EOF;
11319
28.1k
            xmlFinishDocument(ctxt);
11320
28.1k
  }
11321
48.0k
    }
11322
3.50M
    if (ctxt->wellFormed == 0)
11323
1.75M
  return((xmlParserErrors) ctxt->errNo);
11324
1.75M
    else
11325
1.75M
        return(0);
11326
3.50M
}
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
89.7k
                        const char *chunk, int size, const char *filename) {
11358
89.7k
    xmlParserCtxtPtr ctxt;
11359
89.7k
    xmlParserInputPtr input;
11360
11361
89.7k
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11362
89.7k
    if (ctxt == NULL)
11363
83
  return(NULL);
11364
11365
89.6k
    ctxt->options &= ~XML_PARSE_NODICT;
11366
89.6k
    ctxt->dictNames = 1;
11367
11368
89.6k
    input = xmlNewPushInput(filename, chunk, size);
11369
89.6k
    if (input == NULL) {
11370
86
  xmlFreeParserCtxt(ctxt);
11371
86
  return(NULL);
11372
86
    }
11373
89.5k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11374
16
        xmlFreeInputStream(input);
11375
16
        xmlFreeParserCtxt(ctxt);
11376
16
        return(NULL);
11377
16
    }
11378
11379
89.5k
    return(ctxt);
11380
89.5k
}
11381
#endif /* LIBXML_PUSH_ENABLED */
11382
11383
/**
11384
 * Blocks further parser processing
11385
 *
11386
 * @param ctxt  an XML parser context
11387
 */
11388
void
11389
37.7k
xmlStopParser(xmlParserCtxt *ctxt) {
11390
37.7k
    if (ctxt == NULL)
11391
0
        return;
11392
11393
    /* This stops the parser */
11394
37.7k
    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
37.7k
    if (ctxt->errNo == XML_ERR_OK) {
11403
3.38k
        ctxt->errNo = XML_ERR_USER_STOP;
11404
3.38k
        ctxt->lastError.code = XML_ERR_USER_STOP;
11405
3.38k
        ctxt->wellFormed = 0;
11406
3.38k
    }
11407
37.7k
}
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
2.14k
                const xmlChar *publicId, const xmlChar *systemId) {
11475
2.14k
    xmlDtdPtr ret = NULL;
11476
11477
2.14k
    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
2.14k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11484
5
        xmlFreeInputStream(input);
11485
5
        return(NULL);
11486
5
    }
11487
11488
2.14k
    if (publicId == NULL)
11489
1.78k
        publicId = BAD_CAST "none";
11490
2.14k
    if (systemId == NULL)
11491
0
        systemId = BAD_CAST "none";
11492
11493
2.14k
    ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0");
11494
2.14k
    if (ctxt->myDoc == NULL) {
11495
5
        xmlErrMemory(ctxt);
11496
5
        goto error;
11497
5
    }
11498
2.13k
    ctxt->myDoc->properties = XML_DOC_INTERNAL;
11499
2.13k
    ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none",
11500
2.13k
                                       publicId, systemId);
11501
2.13k
    if (ctxt->myDoc->extSubset == NULL) {
11502
19
        xmlErrMemory(ctxt);
11503
19
        xmlFreeDoc(ctxt->myDoc);
11504
19
        goto error;
11505
19
    }
11506
11507
2.11k
    xmlParseExternalSubset(ctxt, publicId, systemId);
11508
11509
2.11k
    if (ctxt->wellFormed) {
11510
90
        ret = ctxt->myDoc->extSubset;
11511
90
        ctxt->myDoc->extSubset = NULL;
11512
90
        if (ret != NULL) {
11513
90
            xmlNodePtr tmp;
11514
11515
90
            ret->doc = NULL;
11516
90
            tmp = ret->children;
11517
1.74k
            while (tmp != NULL) {
11518
1.65k
                tmp->doc = NULL;
11519
1.65k
                tmp = tmp->next;
11520
1.65k
            }
11521
90
        }
11522
2.02k
    } else {
11523
2.02k
        ret = NULL;
11524
2.02k
    }
11525
2.11k
    xmlFreeDoc(ctxt->myDoc);
11526
2.11k
    ctxt->myDoc = NULL;
11527
11528
2.14k
error:
11529
2.14k
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
11530
11531
2.14k
    return(ret);
11532
2.11k
}
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
11.8k
               const xmlChar *systemId) {
11598
11.8k
    xmlDtdPtr ret = NULL;
11599
11.8k
    xmlParserCtxtPtr ctxt;
11600
11.8k
    xmlParserInputPtr input = NULL;
11601
11.8k
    xmlChar* systemIdCanonic;
11602
11603
11.8k
    if ((publicId == NULL) && (systemId == NULL)) return(NULL);
11604
11605
11.8k
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
11606
11.8k
    if (ctxt == NULL) {
11607
44
  return(NULL);
11608
44
    }
11609
11.8k
    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
11610
11611
    /*
11612
     * Canonicalise the system ID
11613
     */
11614
11.8k
    systemIdCanonic = xmlCanonicPath(systemId);
11615
11.8k
    if ((systemId != NULL) && (systemIdCanonic == NULL)) {
11616
6
  xmlFreeParserCtxt(ctxt);
11617
6
  return(NULL);
11618
6
    }
11619
11620
    /*
11621
     * Ask the Entity resolver to load the damn thing
11622
     */
11623
11624
11.7k
    if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL))
11625
11.7k
  input = ctxt->sax->resolveEntity(ctxt->userData, publicId,
11626
11.7k
                                   systemIdCanonic);
11627
11.7k
    if (input == NULL) {
11628
10.3k
  xmlFreeParserCtxt(ctxt);
11629
10.3k
  if (systemIdCanonic != NULL)
11630
10.1k
      xmlFree(systemIdCanonic);
11631
10.3k
  return(NULL);
11632
10.3k
    }
11633
11634
1.44k
    if (input->filename == NULL)
11635
0
  input->filename = (char *) systemIdCanonic;
11636
1.44k
    else
11637
1.44k
  xmlFree(systemIdCanonic);
11638
11639
1.44k
    ret = xmlCtxtParseDtd(ctxt, input, publicId, systemId);
11640
11641
1.44k
    xmlFreeParserCtxt(ctxt);
11642
1.44k
    return(ret);
11643
11.7k
}
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
11.8k
xmlParseDTD(const xmlChar *publicId, const xmlChar *systemId) {
11656
11.8k
    return(xmlSAXParseDTD(NULL, publicId, systemId));
11657
11.8k
}
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
33.0k
                            int hasTextDecl, int buildTree) {
11669
33.0k
    xmlNodePtr root = NULL;
11670
33.0k
    xmlNodePtr list = NULL;
11671
33.0k
    xmlChar *rootName = BAD_CAST "#root";
11672
33.0k
    int result;
11673
11674
33.0k
    if (buildTree) {
11675
33.0k
        root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL);
11676
33.0k
        if (root == NULL) {
11677
59
            xmlErrMemory(ctxt);
11678
59
            goto error;
11679
59
        }
11680
33.0k
    }
11681
11682
33.0k
    if (xmlCtxtPushInput(ctxt, input) < 0)
11683
46
        goto error;
11684
11685
32.9k
    nameNsPush(ctxt, rootName, NULL, NULL, 0, 0);
11686
32.9k
    spacePush(ctxt, -1);
11687
11688
32.9k
    if (buildTree)
11689
32.9k
        nodePush(ctxt, root);
11690
11691
32.9k
    if (hasTextDecl) {
11692
10.3k
        xmlDetectEncoding(ctxt);
11693
11694
        /*
11695
         * Parse a possible text declaration first
11696
         */
11697
10.3k
        if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
11698
1.31k
            (IS_BLANK_CH(NXT(5)))) {
11699
1.27k
            xmlParseTextDecl(ctxt);
11700
            /*
11701
             * An XML-1.0 document can't reference an entity not XML-1.0
11702
             */
11703
1.27k
            if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
11704
1.02k
                (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
11705
77
                xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
11706
77
                               "Version mismatch between document and "
11707
77
                               "entity\n");
11708
77
            }
11709
1.27k
        }
11710
10.3k
    }
11711
11712
32.9k
    xmlParseContentInternal(ctxt);
11713
11714
32.9k
    if (ctxt->input->cur < ctxt->input->end)
11715
3.40k
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
11716
11717
32.9k
    if ((ctxt->wellFormed) ||
11718
26.9k
        ((ctxt->recovery) && (!xmlCtxtIsCatastrophicError(ctxt)))) {
11719
26.9k
        if (root != NULL) {
11720
26.9k
            xmlNodePtr cur;
11721
11722
            /*
11723
             * Unlink newly created node list.
11724
             */
11725
26.9k
            list = root->children;
11726
26.9k
            root->children = NULL;
11727
26.9k
            root->last = NULL;
11728
162k
            for (cur = list; cur != NULL; cur = cur->next)
11729
135k
                cur->parent = NULL;
11730
26.9k
        }
11731
26.9k
    }
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
82.8k
    do {
11738
82.8k
        ctxt->input->cur = ctxt->input->end;
11739
82.8k
        xmlParserShrink(ctxt);
11740
82.8k
        result = xmlParserGrow(ctxt);
11741
82.8k
    } while (result > 0);
11742
11743
32.9k
    if (buildTree)
11744
32.9k
        nodePop(ctxt);
11745
11746
32.9k
    namePop(ctxt);
11747
32.9k
    spacePop(ctxt);
11748
11749
32.9k
    xmlCtxtPopInput(ctxt);
11750
11751
33.0k
error:
11752
33.0k
    xmlFreeNode(root);
11753
11754
33.0k
    return(list);
11755
32.9k
}
11756
11757
static void
11758
39.5k
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
11759
39.5k
    xmlParserInputPtr input;
11760
39.5k
    xmlNodePtr list;
11761
39.5k
    unsigned long consumed;
11762
39.5k
    int isExternal;
11763
39.5k
    int buildTree;
11764
39.5k
    int oldMinNsIndex;
11765
39.5k
    int oldNodelen, oldNodemem;
11766
11767
39.5k
    isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY);
11768
39.5k
    buildTree = (ctxt->node != NULL);
11769
11770
    /*
11771
     * Recursion check
11772
     */
11773
39.5k
    if (ent->flags & XML_ENT_EXPANDING) {
11774
1.35k
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
11775
1.35k
        goto error;
11776
1.35k
    }
11777
11778
    /*
11779
     * Load entity
11780
     */
11781
38.1k
    input = xmlNewEntityInputStream(ctxt, ent);
11782
38.1k
    if (input == NULL)
11783
5.08k
        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
33.0k
    oldMinNsIndex = ctxt->nsdb->minNsIndex;
11791
33.0k
    if (buildTree)
11792
33.0k
        ctxt->nsdb->minNsIndex = ctxt->nsNr;
11793
11794
33.0k
    oldNodelen = ctxt->nodelen;
11795
33.0k
    oldNodemem = ctxt->nodemem;
11796
33.0k
    ctxt->nodelen = 0;
11797
33.0k
    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
33.0k
    ent->flags |= XML_ENT_EXPANDING;
11817
11818
33.0k
    list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree);
11819
11820
33.0k
    ent->flags &= ~XML_ENT_EXPANDING;
11821
11822
33.0k
    ctxt->nsdb->minNsIndex = oldMinNsIndex;
11823
33.0k
    ctxt->nodelen = oldNodelen;
11824
33.0k
    ctxt->nodemem = oldNodemem;
11825
11826
    /*
11827
     * Entity size accounting
11828
     */
11829
33.0k
    consumed = input->consumed;
11830
33.0k
    xmlSaturatedAddSizeT(&consumed, input->end - input->base);
11831
11832
33.0k
    if ((ent->flags & XML_ENT_CHECKED) == 0)
11833
19.6k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
11834
11835
33.0k
    if ((ent->flags & XML_ENT_PARSED) == 0) {
11836
19.6k
        if (isExternal)
11837
9.63k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
11838
11839
19.6k
        ent->children = list;
11840
11841
155k
        while (list != NULL) {
11842
135k
            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
135k
            if (list->doc != ent->doc)
11850
0
                xmlSetTreeDoc(list, ent->doc);
11851
11852
135k
            if (list->next == NULL)
11853
13.3k
                ent->last = list;
11854
135k
            list = list->next;
11855
135k
        }
11856
19.6k
    } else {
11857
13.3k
        xmlFreeNodeList(list);
11858
13.3k
    }
11859
11860
33.0k
    xmlFreeInputStream(input);
11861
11862
39.5k
error:
11863
39.5k
    ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED;
11864
39.5k
}
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
161k
{
12999
161k
    xmlParserInputPtr input;
13000
13001
161k
    if (ctxt == NULL)
13002
0
        return;
13003
13004
161k
    while ((input = xmlCtxtPopInput(ctxt)) != NULL) { /* Non consuming */
13005
0
        xmlFreeInputStream(input);
13006
0
    }
13007
161k
    ctxt->inputNr = 0;
13008
161k
    ctxt->input = NULL;
13009
13010
161k
    ctxt->spaceNr = 0;
13011
161k
    if (ctxt->spaceTab != NULL) {
13012
149k
  ctxt->spaceTab[0] = -1;
13013
149k
  ctxt->space = &ctxt->spaceTab[0];
13014
149k
    } else {
13015
12.0k
        ctxt->space = NULL;
13016
12.0k
    }
13017
13018
13019
161k
    ctxt->nodeNr = 0;
13020
161k
    ctxt->node = NULL;
13021
13022
161k
    ctxt->nameNr = 0;
13023
161k
    ctxt->name = NULL;
13024
13025
161k
    ctxt->nsNr = 0;
13026
161k
    xmlParserNsReset(ctxt->nsdb);
13027
13028
161k
    if (ctxt->version != NULL) {
13029
15.1k
        xmlFree(ctxt->version);
13030
15.1k
        ctxt->version = NULL;
13031
15.1k
    }
13032
161k
    if (ctxt->encoding != NULL) {
13033
846
        xmlFree(ctxt->encoding);
13034
846
        ctxt->encoding = NULL;
13035
846
    }
13036
161k
    if (ctxt->extSubURI != NULL) {
13037
2.32k
        xmlFree(ctxt->extSubURI);
13038
2.32k
        ctxt->extSubURI = NULL;
13039
2.32k
    }
13040
161k
    if (ctxt->extSubSystem != NULL) {
13041
250
        xmlFree(ctxt->extSubSystem);
13042
250
        ctxt->extSubSystem = NULL;
13043
250
    }
13044
161k
    if (ctxt->directory != NULL) {
13045
15.6k
        xmlFree(ctxt->directory);
13046
15.6k
        ctxt->directory = NULL;
13047
15.6k
    }
13048
13049
161k
    if (ctxt->myDoc != NULL)
13050
0
        xmlFreeDoc(ctxt->myDoc);
13051
161k
    ctxt->myDoc = NULL;
13052
13053
161k
    ctxt->standalone = -1;
13054
161k
    ctxt->hasExternalSubset = 0;
13055
161k
    ctxt->hasPErefs = 0;
13056
161k
    ctxt->html = ctxt->html ? 1 : 0;
13057
161k
    ctxt->instate = XML_PARSER_START;
13058
13059
161k
    ctxt->wellFormed = 1;
13060
161k
    ctxt->nsWellFormed = 1;
13061
161k
    ctxt->disableSAX = 0;
13062
161k
    ctxt->valid = 1;
13063
161k
    ctxt->record_info = 0;
13064
161k
    ctxt->checkIndex = 0;
13065
161k
    ctxt->endCheckState = 0;
13066
161k
    ctxt->inSubset = 0;
13067
161k
    ctxt->errNo = XML_ERR_OK;
13068
161k
    ctxt->depth = 0;
13069
161k
    ctxt->catalogs = NULL;
13070
161k
    ctxt->sizeentities = 0;
13071
161k
    ctxt->sizeentcopy = 0;
13072
161k
    xmlInitNodeInfoSeq(&ctxt->node_seq);
13073
13074
161k
    if (ctxt->attsDefault != NULL) {
13075
1.40k
        xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator);
13076
1.40k
        ctxt->attsDefault = NULL;
13077
1.40k
    }
13078
161k
    if (ctxt->attsSpecial != NULL) {
13079
1.87k
        xmlHashFree(ctxt->attsSpecial, NULL);
13080
1.87k
        ctxt->attsSpecial = NULL;
13081
1.87k
    }
13082
13083
161k
#ifdef LIBXML_CATALOG_ENABLED
13084
161k
    if (ctxt->catalogs != NULL)
13085
0
  xmlCatalogFreeLocal(ctxt->catalogs);
13086
161k
#endif
13087
161k
    ctxt->nbErrors = 0;
13088
161k
    ctxt->nbWarnings = 0;
13089
161k
    if (ctxt->lastError.code != XML_ERR_OK)
13090
15.6k
        xmlResetError(&ctxt->lastError);
13091
161k
}
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
438k
{
13132
438k
    int allMask;
13133
13134
438k
    if (ctxt == NULL)
13135
200
        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
437k
    allMask = XML_PARSE_RECOVER |
13145
437k
              XML_PARSE_NOENT |
13146
437k
              XML_PARSE_DTDLOAD |
13147
437k
              XML_PARSE_DTDATTR |
13148
437k
              XML_PARSE_DTDVALID |
13149
437k
              XML_PARSE_NOERROR |
13150
437k
              XML_PARSE_NOWARNING |
13151
437k
              XML_PARSE_PEDANTIC |
13152
437k
              XML_PARSE_NOBLANKS |
13153
437k
#ifdef LIBXML_SAX1_ENABLED
13154
437k
              XML_PARSE_SAX1 |
13155
437k
#endif
13156
437k
              XML_PARSE_NONET |
13157
437k
              XML_PARSE_NODICT |
13158
437k
              XML_PARSE_NSCLEAN |
13159
437k
              XML_PARSE_NOCDATA |
13160
437k
              XML_PARSE_COMPACT |
13161
437k
              XML_PARSE_OLD10 |
13162
437k
              XML_PARSE_HUGE |
13163
437k
              XML_PARSE_OLDSAX |
13164
437k
              XML_PARSE_IGNORE_ENC |
13165
437k
              XML_PARSE_BIG_LINES |
13166
437k
              XML_PARSE_NO_XXE |
13167
437k
              XML_PARSE_UNZIP |
13168
437k
              XML_PARSE_NO_SYS_CATALOG |
13169
437k
              XML_PARSE_CATALOG_PI;
13170
13171
437k
    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
437k
    ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0;
13188
437k
    ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0;
13189
437k
    ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0;
13190
437k
    ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0;
13191
437k
    ctxt->loadsubset |= (options & XML_PARSE_SKIP_IDS) ? XML_SKIP_IDS : 0;
13192
437k
    ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0;
13193
437k
    ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0;
13194
437k
    ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1;
13195
437k
    ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1;
13196
13197
437k
    return(options & ~allMask);
13198
438k
}
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
11.8k
{
13216
11.8k
#ifdef LIBXML_HTML_ENABLED
13217
11.8k
    if ((ctxt != NULL) && (ctxt->html))
13218
0
        return(htmlCtxtSetOptions(ctxt, options));
13219
11.8k
#endif
13220
13221
11.8k
    return(xmlCtxtSetOptionsInternal(ctxt, options, 0));
13222
11.8k
}
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
426k
{
13267
426k
    int keepMask;
13268
13269
426k
#ifdef LIBXML_HTML_ENABLED
13270
426k
    if ((ctxt != NULL) && (ctxt->html))
13271
0
        return(htmlCtxtUseOptions(ctxt, options));
13272
426k
#endif
13273
13274
    /*
13275
     * For historic reasons, some options can only be enabled.
13276
     */
13277
426k
    keepMask = XML_PARSE_NOERROR |
13278
426k
               XML_PARSE_NOWARNING |
13279
426k
               XML_PARSE_NONET |
13280
426k
               XML_PARSE_NSCLEAN |
13281
426k
               XML_PARSE_NOCDATA |
13282
426k
               XML_PARSE_COMPACT |
13283
426k
               XML_PARSE_OLD10 |
13284
426k
               XML_PARSE_HUGE |
13285
426k
               XML_PARSE_OLDSAX |
13286
426k
               XML_PARSE_IGNORE_ENC |
13287
426k
               XML_PARSE_BIG_LINES;
13288
13289
426k
    return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask));
13290
426k
}
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
490
{
13307
490
    if (ctxt == NULL)
13308
0
        return;
13309
490
    ctxt->maxAmpl = maxAmpl;
13310
490
}
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
311k
{
13325
311k
    xmlDocPtr ret = NULL;
13326
13327
311k
    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
311k
    while (ctxt->inputNr > 0)
13335
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13336
13337
311k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13338
33
        xmlFreeInputStream(input);
13339
33
        return(NULL);
13340
33
    }
13341
13342
311k
    xmlParseDocument(ctxt);
13343
13344
311k
    ret = xmlCtxtGetDocument(ctxt);
13345
13346
    /* assert(ctxt->inputNr == 1); */
13347
626k
    while (ctxt->inputNr > 0)
13348
314k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13349
13350
311k
    return(ret);
13351
311k
}
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
163k
{
13369
163k
    xmlParserCtxtPtr ctxt;
13370
163k
    xmlParserInputPtr input;
13371
163k
    xmlDocPtr doc = NULL;
13372
13373
163k
    ctxt = xmlNewParserCtxt();
13374
163k
    if (ctxt == NULL)
13375
110
        return(NULL);
13376
13377
163k
    xmlCtxtUseOptions(ctxt, options);
13378
13379
163k
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) cur, encoding,
13380
163k
                                      XML_INPUT_BUF_STATIC);
13381
13382
163k
    if (input != NULL)
13383
161k
        doc = xmlCtxtParseDocument(ctxt, input);
13384
13385
163k
    xmlFreeParserCtxt(ctxt);
13386
163k
    return(doc);
13387
163k
}
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
25.7k
{
13452
25.7k
    xmlParserCtxtPtr ctxt;
13453
25.7k
    xmlParserInputPtr input;
13454
25.7k
    xmlDocPtr doc = NULL;
13455
13456
25.7k
    if (size < 0)
13457
0
  return(NULL);
13458
13459
25.7k
    ctxt = xmlNewParserCtxt();
13460
25.7k
    if (ctxt == NULL)
13461
0
        return(NULL);
13462
13463
25.7k
    xmlCtxtUseOptions(ctxt, options);
13464
13465
25.7k
    input = xmlCtxtNewInputFromMemory(ctxt, url, buffer, size, encoding,
13466
25.7k
                                      XML_INPUT_BUF_STATIC);
13467
13468
25.7k
    if (input != NULL)
13469
25.7k
        doc = xmlCtxtParseDocument(ctxt, input);
13470
13471
25.7k
    xmlFreeParserCtxt(ctxt);
13472
25.7k
    return(doc);
13473
25.7k
}
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
50.3k
{
13595
50.3k
    xmlParserInputPtr input;
13596
13597
50.3k
    if (ctxt == NULL)
13598
71
        return(NULL);
13599
13600
50.2k
    xmlCtxtReset(ctxt);
13601
50.2k
    xmlCtxtUseOptions(ctxt, options);
13602
13603
50.2k
    input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13604
50.2k
    if (input == NULL)
13605
8.83k
        return(NULL);
13606
13607
41.4k
    return(xmlCtxtParseDocument(ctxt, input));
13608
50.2k
}
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
83.1k
{
13629
83.1k
    xmlParserInputPtr input;
13630
13631
83.1k
    if ((ctxt == NULL) || (size < 0))
13632
0
        return(NULL);
13633
13634
83.1k
    xmlCtxtReset(ctxt);
13635
83.1k
    xmlCtxtUseOptions(ctxt, options);
13636
13637
83.1k
    input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding,
13638
83.1k
                                      XML_INPUT_BUF_STATIC);
13639
83.1k
    if (input == NULL)
13640
74
        return(NULL);
13641
13642
83.0k
    return(xmlCtxtParseDocument(ctxt, input));
13643
83.1k
}
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
17
{
13702
17
    xmlParserInputPtr input;
13703
13704
17
    if (ctxt == NULL)
13705
0
        return(NULL);
13706
13707
17
    xmlCtxtReset(ctxt);
13708
17
    xmlCtxtUseOptions(ctxt, options);
13709
13710
17
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
13711
17
                                  encoding, 0);
13712
17
    if (input == NULL)
13713
17
        return(NULL);
13714
13715
0
    return(xmlCtxtParseDocument(ctxt, input));
13716
17
}
13717